Abstract 4569: Digitizing single-cell expression patterns in urine for prostate cancer detection

Abstract

Abstract Gene expression profiles of microdissected tissue are frequently used to stratify cancer subtypes and to predict patients' clinical outcome. This type of analysis usually operates under the assumption that uniform gene expression is present in a cancer cell population. However, high fluctuation of gene expression in single cells stemming from intra-clonal heterogeneity within a cancer cell population can lead to misinterpretation of the data. An important challenge then lies in developing approaches to decipher this expression chaos for cancer diagnostic and/or prognostic strategies. We have developed a novel digitizing approach to analyze single-cell gene expression in exfoliated prostate cells isolated from urine sediment. In analyzing 1220 single cells, we observed a panel of genes with dichotomous expression patterns (DEP) and further digitized their expression levels to either 1 for high or 0 for low expression. Of 35 androgen-responsive genes examined, 6 genes exhibited DEP characteristics. When aligning the dichotomous numbers of these genes in the order of CXCL6-TGFBR2-GSK3B-CDKN1C-GATA3-EIF4EBP1, we identified 64 (26) types of code; e.g., 1-0-1-0-1-0. With the conversion from original expression levels to a series of dichotomous numbers, these new codes not only reduced the complexity associated with single cell expression, they also enabled us to decipher expression heterogeneity in the samples by considering multiple gene expression codes simultaneously. Importantly, these digitized codes identified distinct clonal prostate cell populations in the urine samples. At the single-cell level, high volumes of cells with the same code (common codes) in a collection of cells suggested the presence of a large clonal cell population. By examining the types of code and the size of clonal populations, we found that increasing clonal size of cells carrying common codes was associated with high-grade cancers. Certain common codes were also found in cells from patients diagnosed with benign prostate hyperplasia and prostatic intraepithelial neoplasia. On the other hand, digitized code types appeared to be highly diverse in normal cell populations. Our data suggest that digitizing single-cell expression can stratify benign, low grade and high grade prostate cancer. This analysis pipeline has a wider application for disease diagnosis/prognosis using cancer cells isolated by non-invasive means from biological fluids or washes from biopsy needles and surgical blades. Citation Format: Chun-Lin Lin, Chun-Liang Chen, Chiou-Miin Wang, Joseph Liu, Susan Huang, Tim Huang. Digitizing single-cell expression patterns in urine for prostate cancer detection. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4569. doi:10.1158/1538-7445.AM2014-4569