Introduction

Quantification methods for gene expression have become important tools both in the understanding of the molecular events underlying human breast cancer and in the identification of diagnostic and therapeutic targets. Microarray expression profiling has provided an exciting new technology for attempting to identify gene-based classifiers that correlate with breast cancer diagnosis, disease prognosis, or prediction of response to treatment (1,2,3,4). A useful technique to confirm or use such classifiers is quantitative reverse transcription–polymerase chain reaction (RT-qPCR) assays of the selected genes (5,6,7).

The development of molecular tests for clinical use has been limited by the lack of available clinical samples for validation of candidate biomarkers. Fresh frozen (FF) samples are difficult to collect for large scale studies, complicated to process, and expensive to store. Formalin-fixed, paraffin-embedded samples (FFPE) are stable at room temperature and easily storable, and—most important—they constitute a widely available archive of clinical samples linked to precious clinical and follow-up information. Set against these advantages is the fact that RNA isolated from FFPE is considered a poor material for gene expression analysis, owing to its extensive degradation. While microarray-based studies are highly sensitive to RNA degradation, RT-qPCR appears to be more robust and tolerates partial degradation of RNA (8). Although RNA degradation leads to a loss of amplifiable templates, optimized normalization strategies could effectively compensate for this bias (9,10,11). Normalization is essential to control for experimental errors, such as the inherent variability of RNA, variability of extraction protocols that may co-purify inhibitors, and different reverse transcription and PCR efficiencies (12).

In this study, we analyzed the correlation in gene expression measurements by RT-qPCR between breast cancer FF and FFPE tissues, and evaluated the performance of different normalization methods in compensating for the effect of RNA degradation. We also investigated the factors that could influence obtaining reliable results from FFPE samples.

Materials and methods

Tissue specimens

Matching pairs of FF and FFPE biopsies of breast tumors from 30 patients (60 samples in total) were retrieved from the Department of Pathology of Hospital Universitario La Paz (Madrid, Spain). Tissue samples were procured between 1991 and 1998. The histopathological features of each sample were reviewed by an experienced breast pathologist to confirm diagnosis and similar tumor content (≥70%). Approval from the Ethical Committee of Hospital Universitario La Paz (HULP code PI-405) was obtained for the conduct of the study.

Isolation of RNA and cDNA synthesis

For extraction of RNA from FFPE tissue, 15 5-µm sections were cut from each archival block. Paraffin was removed by xylene extraction followed by ethanol washes. RNA was isolated from tissue slices using the MasterPure RNA Purification Kit (Epicenter Biotechnologies, Madison, WI, USA). RNA was isolated from 10 10-µm sections of FF tissue with TRIzol Reagent (Invitrogen, Carsbald, CA, USA) and cleaned up with Qiagen RNeasy spin columns. Total RNA isolated was quantified and qualitatively assessed using spectrophotometer OD₂₆₀ measurements and capillary electrophoresis (Agilent 2100 Bioanalyzer, Agilent, Santa Clara, CA, USA). We normalized to total RNA input; therefore, first-strand cDNA was synthesized from 1 µg total RNA according to the High Capacity cDNA Reverse Transcription Kit protocol (Applied Biosystems, Foster City, CA, USA).

Quantitative RT-PCR

RT-qPCR amplifications were performed with TaqMan Gene Expression Assays products in an ABI PRISM 7900 HT Sequence Detection System (Applied Biosystems). Reactions were carried out using TaqMan Low Density Arrays (TLDAs; Applied Biosystems) containing 50 µL TaqMan Universal PCR Master Mix (Applied Biosystems) and 50 µL cDNA template corresponding to 100 ng total RNA per channel of the microfluidic card.

We used two configurations of TLDAs containing reference and breast cancer prognosis–related genes. The first one was configured to measure the expression of 95 genes in duplicate for two samples per card. The second one was configured to analyze expression of 63 genes in triplicate, also for two samples per card. For a complete list of the Gene Expression Assays included in each configuration, see Supplementary Table S1. We followed the Minimum Information for Publication of Quantitative Real-time PCR Experiments (MIQE) guidelines (13). Raw Cq data, experimental annotation and sample annotation are available in the RDML data format (Supplementary Table S2).