The Proportion of Contaminating Host Organ RNA within Clinical Biopsies of Metastatic Breast Cancer Is Minimal
Hongxia Sun, Christos Hatzis, Chunxiao Fu, Rony Avritscher, Kamran Ahrar, Alda Tam, Steven Curley, Hui Yao, Jean-Nicholas Vauthey, Vicente Valero, Daniel Booser, Dipen Maru, Stanley Hamilton, Raja Luthra, Rosanna Lau, Rebekah E Gould, Lajos Pusztai, Michael Wallace, W Fraser Symmans. MD Anderson Cancer Center, Houston, TX; Nuvera Biosciences, Woburn, WA
Background: Host organ contamination is a limitation for DNA-based testing of small biopsies from metastatic sites that is currently addressed by histologic assessment of tumor cellularity and/or microdissection. Relatively higher transcriptional activity in tumor cells might favor RNA-based testing. Therefore, we used breast cancer within liver as a model to predict the extent of host organ RNA contamination in gene expression microarrays of radiologically-guided clinical biopsies from liver metastases of breast cancer.
Design: Microarray-based (Affymetrix U133A, Santa Cruz, CA) multi-gene expression indices were developed to measure genes that either discriminate normal liver from breast cancer or have expression levels that calibrate to the proportion of liver RNA spiked into breast cancer RNA. These discriminating and calibrating multi-gene liver indices, and the reverse transcription-polymerase chain reaction (RT-PCR) measurements of a single discriminating liver gene (CYP2E1) were tested in a blinded validation study cohort of 30 known mixtures of breast cancer and liver RNA. Finally, the proportion of liver RNA contamination was estimated in clinical liver biopsies from 23 patients with metastatic breast cancer.
Results: The correlation coefficient for concordance (CCC) between predicted and actual percent liver RNA in the blinded validation study was higher for the RT-PCR assay for CYP2E1 (0.894) than for the discriminating 16-gene index (0.679) or the calibrating 10-gene index (0.663) measured from microarrays. All three assays slightly overestimated the percent liver RNA in a mixture. Of 23 clinical biopsies, the better microarray-based predictor (calibrating index) predicted <5% liver RNA in 13/23, <10% in 16/23, and <20% in 21/23. Of 17 biopsies with sufficient remaining RNA, the RT-PCR assay predicted <1% liver RNA in 10/17, <5% liver RNA in 16/17, and 12.6% liver RNA in one case.
Conclusions: It is likely that the majority of radiologically-guided clinical biopsies of metastatic breast cancer in the liver contain >90% breast cancer RNA. This was bested estimated using an RT-PCR assay that could become a quality control procedure. The relatively low proportion of contaminating host organ RNA might offer an advantage to RNA-based, rather than DNA-based, genomic tests of metastatic samples.
Category: Special Category - Pan-genomic/Pan-proteomic approaches to Cancer
Tuesday, March 5, 2013 9:30 AM
Poster Session III # 237, Tuesday Morning