Prospective transcriptomic pathway analysis of human lymphatic vascular insufficiency: identification and validation of a circulating biomarker panel.

ABSTRACT

BACKGROUND: In our previous transcriptional profiling of a murine model, we have identified a remarkably small number of specific pathways with altered expression in lymphedema. In this investigation, we utilized microarray-based transcriptomics of human skin for an unbiased a priori prospective candidate identification, with subsequent validation of these candidates through direct serum assay. The resulting multi-analyte biomarker panel sensitively should sensitively discriminate human lymphedema subjects from normal individuals. METHODS AND FINDINGS: We enrolled 63 lymphedema subjects and 27 normals in our attempt to discover protein analytes that can distinguish diseased individuals from controls. To minimize technical and biologically irrelevant variation, we first identified potential candidates by performing transcriptional microarray analysis on paired diseased and normal skin specimens sampled from the same individuals. We focused our attention on genes with corresponding protein products that are secreted and took these candidates forward to a protein multiplex assay applied to diseased and normal subjects. We developed a logistic regression-based model on an eventual group of six proteins and validated our system on a separate cohort of study subjects. The area under the receiver operating characteristic curve was calculated to be 0.87 (95% CI 0.75 to 0.97). CONCLUSIONS: We have developed an accurate bioassay utilizing proteins representing four central pathogenetic modalities of the disease lymphangiogenesis, inflammation, fibrosis, and lipid metabolism, suggesting that these proteins are directly related to the pathogenesis of the tissue pathology in lymphatic vascular insufficiency. Further studies are warranted to determine whether this newly-identified biomarker panel will possess utility as an instrument for in vitro diagnosis of early and latent disease; the ultimate applicability to risk stratification, quantitation of disease burden, and response to therapy can easily be envisioned.