Browsing by Author "Bathe, Oliver F"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
- ItemOpen AccessA quantitative multimodal metabolomic assay for colorectal cancer(2018-01-04) Farshidfar, Farshad; Kopciuk, Karen A; Hilsden, Robert; McGregor, S. E; Mazurak, Vera C; Buie, W. D; MacLean, Anthony; Vogel, Hans J; Bathe, Oliver FAbstract Background Early diagnosis of colorectal cancer (CRC) simplifies treatment and improves treatment outcomes. We previously described a diagnostic metabolomic biomarker derived from semi-quantitative gas chromatography-mass spectrometry. Our objective was to determine whether a quantitative assay of additional metabolomic features, including parts of the lipidome could enhance diagnostic power; and whether there was an advantage to deriving a combined diagnostic signature with a broader metabolomic representation. Methods The well-characterized Biocrates P150 kit was used to quantify 163 metabolites in patients with CRC (N = 62), adenoma (N = 31), and age- and gender-matched disease-free controls (N = 81). Metabolites included in the analysis included phosphatidylcholines, sphingomyelins, acylcarnitines, and amino acids. Using a training set of 32 CRC and 21 disease-free controls, a multivariate metabolomic orthogonal partial least squares (OPLS) classifier was developed. An independent set of 28 CRC and 20 matched healthy controls was used for validation. Features characterizing 31 colorectal adenomas from their healthy matched controls were also explored, and a multivariate OPLS classifier for colorectal adenoma could be proposed. Results The metabolomic profile that distinguished CRC from controls consisted of 48 metabolites (R2Y = 0.83, Q2Y = 0.75, CV-ANOVA p-value < 0.00001). In this quantitative assay, the coefficient of variance for each metabolite was <10%, and this dramatically enhanced the separation of these groups. Independent validation resulted in AUROC of 0.98 (95% CI, 0.93–1.00) and sensitivity and specificity of 93% and 95%. Similarly, we were able to distinguish adenoma from controls (R2Y = 0.30, Q2Y = 0.20, CV-ANOVA p-value = 0.01; internal AUROC = 0.82 (95% CI, 0.72–0.93)). When combined with the previously generated GC-MS signatures for CRC and adenoma, the candidate biomarker performance improved slightly. Conclusion The diagnostic power for metabolomic tests for colorectal neoplasia can be improved by utilizing a multimodal approach and combining metabolites from diverse chemical classes. In addition, quantification of metabolites enhances separation of disease-specific metabolomic profiles. Our future efforts will be focused on developing a quantitative assay for the metabolites comprising the optimal diagnostic biomarker.
- ItemOpen AccessImmunohistochemical phenotyping of T cells, granulocytes, and phagocytes in the muscle of cancer patients: association with radiologically defined muscle mass and gene expression(2019-09-14) Anoveros-Barrera, Ana; Bhullar, Amritpal S; Stretch, Cynthia; Dunichand-Hoedl, Abha R; Martins, Karen J B; Rieger, Aja; Bigam, David; McMullen, Todd; Bathe, Oliver F; Putman, Charles T; Field, Catherine J; Baracos, Vickie E; Mazurak, Vera CAbstract Background Inflammation is a recognized contributor to muscle wasting. Research in injury and myopathy suggests that interactions between the skeletal muscle and immune cells confer a pro-inflammatory environment that influences muscle loss through several mechanisms; however, this has not been explored in the cancer setting. This study investigated the local immune environment of the muscle by identifying the phenotype of immune cell populations in the muscle and their relationship to muscle mass in cancer patients. Methods Intraoperative muscle biopsies were collected from cancer patients (n = 30, 91% gastrointestinal malignancies). Muscle mass was assessed histologically (muscle fiber cross-sectional area, CSA; μm2) and radiologically (lumbar skeletal muscle index, SMI; cm2/m2 by computed tomography, CT). T cells (CD4 and CD8) and granulocytes/phagocytes (CD11b, CD14, and CD15) were assessed by immunohistochemistry. Microarray analysis was conducted in the muscle of a second cancer patient cohort. Results T cells (CD3+), granulocytes/phagocytes (CD11b+), and CD3−CD4+ cells were identified. Muscle fiber CSA (μm2) was positively correlated (Spearman’s r = > 0.45; p = < 0.05) with the total number of T cells, CD4, and CD8 T cells and granulocytes/phagocytes. In addition, patients with the smallest SMI exhibited fewer CD8 T cells within their muscle. Consistent with this, further exploration with gene correlation analyses suggests that the presence of CD8 T cells is negatively associated (Pearson’s r = ≥ 0.5; p = <0.0001) with key genes within muscle catabolic pathways for signaling (ACVR2B), ubiquitin proteasome (FOXO4, TRIM63, FBXO32, MUL1, UBC, UBB, UBE2L3), and apoptosis/autophagy (CASP8, BECN1, ATG13, SIVA1). Conclusion The skeletal muscle immune environment of cancer patients is comprised of immune cell populations from the adaptive and innate immunity. Correlations of T cells, granulocyte/phagocytes, and CD3−CD4+ cells with muscle mass measurements indicate a positive relationship between immune cell numbers and muscle mass status in cancer patients. Further exploration with gene correlation analyses suggests that the presence of CD8 T cells is negatively correlated with components of muscle catabolism.
- ItemOpen AccessPerformance of variable selection methods using stability-based selection(2017-04-04) Lu, Danny; Weljie, Aalim; de Leon, Alexander R; McConnell, Yarrow; Bathe, Oliver F; Kopciuk, KarenAbstract Background Variable selection is frequently carried out during the analysis of many types of high-dimensional data, including those in metabolomics. This study compared the predictive performance of four variable selection methods using stability-based selection, a new secondary selection method that is implemented in the R package BioMark. Two of these methods were evaluated using the more well-known false discovery rate (FDR) as well. Results Simulation studies varied factors relevant to biological data studies, with results based on the median values of 200 partial area under the receiver operating characteristic curve. There was no single top performing method across all factor settings, but the student t test based on stability selection or with FDR adjustment and the variable importance in projection (VIP) scores from partial least squares regression models obtained using a stability-based approach tended to perform well in most settings. Similar results were found with a real spiked-in metabolomics dataset. Group sample size, group effect size, number of significant variables and correlation structure were the most important factors whereas the percentage of significant variables was the least important. Conclusions Researchers can improve prediction scores for their study data by choosing VIP scores based on stability variable selection over the other approaches when the number of variables is small to modest and by increasing the number of samples even moderately. When the number of variables is high and there is block correlation amongst the significant variables (i.e., true biomarkers), the FDR-adjusted student t test performed best. The R package BioMark is an easy-to-use open-source program for variable selection that had excellent performance characteristics for the purposes of this study.