Section 1 -

Gene Expression Profiling from Paraffin-embedded Samples Dennis Sgroi Director of Breast Pathology, Massachusetts General Hospital Associate Professor of Pathology, Harvard Medical School

Histopathological criteria, along with clinical staging, have been the primary approaches to defining prognosis in patients newly diagnosed with breast cancer. However, these markers provide not only imperfect predictors of the indication for adjuvant therapy, but also poor predictors of the risk of early relapse in the setting of adjuvant therapy. Contemporary advances in microarray-based gene expression profile analysis are likely to revolutionize the classification of breast cancer, providing a comprehensive analysis of all expressed genes, leading to sophisticated markers of tumor biology and clinical behavior. Recently, the use of such technologies has resulted in the discovery of a gene expression signature that outperformed all currently used clinicopathological parameters in predicting disease outcome in women with breast cancer. These results suggest for the first time that the analysis of breast cancer gene expression may become a clinical reality that may eventually guide the clinical decision-making process. The application of high-throughput DNA microarray technologies to the analysis of clinical breast cancer specimens holds great promise in producing substantial advances in our understanding of the biological and clinical behavior of human breast cancer. Several exciting studies in a small number of human cancer types have demonstrated the powerful utility of gene expression technologies in identifying clinically relevant prognostic categories [1, 2, 3, 4, 5]. The most clinically useful data will come from analysis of patient cohorts treated in a uniform fashion, allowing specific correlation of microarray results with outcomes following particular chemotherapeutic interventions. In addition, ideal studies will require correlation of expression profiles with long-term disease-free and overall survival. The current application of gene expression technologies to clinical samples is limited to RNA derived from fresh tissue or cells, or from archival fresh frozen tissues. Given this technical limitation and the fact that large archival frozen tissue repositories with long term clinical outcome data are scarce worldwide, the rapid and widespread use of gene expression profiling technologies to address many clinically and biologically relevant issues in human breast cancer has been significantly hampered. One potential solution to this problem is the use of formalin fixed, paraffin-embedded (FFPE) resected breast tissue samples. FFPE resected breast tissues have been collected worldwide throughout decades of routine histopathological examination and are a potentially invaluable resource for investigative gene expression profile studies. Given the wide availability of FFPE breast tissue blocks along with linked long-term clinical follow-up data, clinical and biological paradigms can be rapidly studied retrospectively. Therefore, the use of FFPE breast cancer tissues corresponding to well-defined clinical cohorts with long term follow-up data provides a major leap in our ability to expedite the comprehensive evaluation of gene expression technologies in the clinical setting of breast cancer treatment. Although previous studies have demonstrated that nuclei acids can be retrieved from FFPE tissue samples [6, 7], demonstration of reliable quantitation of gene expression from such specimens has been limited to the use of real-time quantitative Taqman reverse-transcriptase polymerase chain reaction (RTQ-PCR) technologies [8]. Through the use of the highly specific and accurate method of RTQ-PCR, Specht et al. [8] demonstrated no significant difference in gene expression levels when using frozen tissue and FFPE samples derived from the identical tissue source. Importantly, this group also demonstrated that delayed formalin fixation of liver, an RNase rich tissue, did not alter transcript levels of 7 genes with half-lives ranging from 2-10 hours. Furthermore, they reported the successful extraction and analysis of RNA from FFPE samples of greater than 20 years of age. In light of the observations by Specht et al, and the obvious clinical and scientific advantages afforded by the use of large FFPE breast tissue repositories, we sought to determine the feasibility of generating reliable gene expression profiles using RNA derived from FFPE clinical breast tissue samples. First, we sought to determine if global gene expression patterns generated with RNA derived from FFPE tissues is consistently reproducible, and second we sought to assess RNA quality of FFPE breast cancer tissue blocks that have been generated and stored, over a 10 year period, using standard anatomical pathology techniques. Using specimen-matched frozen and FFPE clinical breast tissues obtained from the Massachusetts General Hospital (MGH) department of pathology, we laser capture microdissected ~2,500 malignant invasive breast carcinoma cells from the frozen and FFPE specimens as described, respectively [5]. The RNA was extracted using a modified version of the previously described protocol [5], then subjected to two rounds of T7-based linear RNA amplification using the RiboAmp™ kit (Arcturus) as described [5]. Given potential variations in fixation time of clinical breast tissue specimens, we sought to determine the effect of formalin fixation time on our ability to extract and amplify RNA from FFPE tissues. As seen in Fig 1, RNA can be consistently and reproducibly, extracted and amplified from specimen-matched tissue that has been fixed in 10% neutral buffer formalin for up to 8 days. We and others have demonstrated that gene expression profile data generated with amplified RNA (aRNA) derived from duplicate microdissected frozen tissue samples is highly reproducible and lacks significant representational bias as judged by microarray and real-time quantitative PCR validation studies. Given this success with frozen tissue, we sought to extend these technologies to FFPE tissues and determine the reproducibility of array data generated with aRNA derived from such tissues. RNA amplicons derived from FFPE tissues consistently average 250bps, whereas RNA amplicons derived from frozen tissues consistently average 500bps (Fig 1); the length of the RNA derived from our FFPE samples is consistent with that observed by others [6]. Figure 1. Formalin Fixation Time Course. Formaldehyde RNA gel of aRNA from frozen tissue (lane 6) and from FFPE tissue fixed for 1 day (lanes 1&2), 4 days (lane 3) and 8 days (lane 4&5). Far left is RNA ladder. Importantly, gene expression data generated using aRNA derived from duplicate microdissected FFPE tissue samples is as reproducible as that produced using aRNA derived from duplicate frozen tissue samples (Fig 2). The correlation coefficient (r) of microarray data generated from multiple pairs of duplicate microdissected FFPE samples is consistently above 0.93. Figure 2. Scatter plots of microarray data for duplicate hybridations of amplified RNA from three different FFPE breast samples to an Affymetrix X3P GeneChip. The correlation (r values) for duplicates in FFPE sample A, r = 0.930; in FFPE sample B, r = 0.971; and for FFPE sample C, r = .968. We have also demonstrated that the gene expression profiles from matched frozen and FFPE breast cancer samples correlate with each other. More specifically, in 6 frozen and FFPE matched pairs the profiles from the FFPE and frozen samples from the same breast tumor had greater similarity to each other than to profiles generated from tumor samples from different patients (Figure 3). Inspection of tumor matched FFPE and frozen tumor-derived profiles demonstrate similar gene expression patterns. Some differences do exist, however the high degree of similarity supports the notion that RNA from FFPE samples can be used for gene expression profiling. Figure 3. Heat map of the overall expression profile of 11255 genes among 6 matched FFPE and frozen samples (from 6 patients). Matched tissue samples (frz and ffpe represent frozen and formalin fixed paraffin embedded tissues, respectively; numbers represent different tumor samples) are on the Y-axis and the 11255 genes are along the X-axis. The FFPE and frozen samples from the same breast cancer clustered together (having the highest similarity to each other). Representative examples of clusters of highly similar gene expression within an FFPE-Frozen tissue pair are depicted as white rectangles. These results are promising and it is hoped that this approach will expedite translational and basic science research in human breast cancer. However, caution should be exercised in the interpretation of microarray-based gene expression signatures identified from FFPE tissues of varying age. Such signatures should be rigorously validated in independent cohort using other gene expression technologies such as real-time quantitative PCR. References Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A, Boldrick JC, Sabet H, Tran T, Yu X, Powell JI, Yang L, Marti GE, Moore T, Hudson, Jr J, Lu L, Lewis DB, Tibshirani R, Sherlock G, Chan WC, Greiner TC, Weisenburger DD, Armitage JO, Warnke R, Levy R, Wilson W, Grever MR, Byrd JC, Botstein D, Brown PO, Staudt LM. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 403:503-11, 2000. van't Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AAM, Mao M, Peterse H, van der Kooy K, Marton MJ, Witteveen AT, Schreiber GJ, Kerkhoven RM, Roberts C, Linsley PS, Bernards R, Friend SH. Gene expression profiling predicts clinical outcome of breast cancer. Nature 415:530-6, 2002. Van de Vijver MJ, He YD, van't Veer LJ, Dai H, Hart AAM, Voskuil DW, Schreiber GJ, Peterse JL, Roberts C, Marton MJ, Parrish M, Atsma D, Witteveen A, Glas A, Delahaye L, van der Velde T, Bartelink H, Rodenhuis S, Rutgers ET, Friend SH, Bernards R. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 347:1999-2009, 2002. Sørlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H, Hastie T, Eisen MB, van de Rijn M, Jeffrey SS, Thorsen T, Quist H, Matese JC, Brown PO, Botstein D, Lønning PE, Børresen-Dale A-L. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci 98:10869-74, 2001. Ma XJ, Wang Z, Ryan PD, Isakoff SJ, Barmettler A, Fuller A, Muir B, Mohapatra G, Salunga R, Tuggle JT, Tran Y, Tran D, Tassin A, Amon P, Wang W, Wang W, Enright E, Stecker K, Estepa-Sabal E, Smith B, Younger J, Balis U, Michaelson J, Bhan A, Habin K, Baer TM, Brugge J, Haber DA, Erlander MG and Sgroi DC. A two-gene expression ratio predicts clinical outcome in breast cancer patients treated with tamoxifen. Cancer Cell, 2004; 5: 607-616. Lehmann U, Bock O, Glockner S, Kreipe H. Quantitative molecular analysis of laser-microdissected paraffin-embedded human tissues. Pathobiology 68:202-8, 2000. Ren Z-P, Sallstrom J, Sundstrom C, Nister M, Olsson Y. Recovering DNA and optimising PCR conditions from microdissected formalin-fixed and paraffin-embedded materials. Pathobiology 68:215-7, 2000. Specht K, Richter T, Müller U, Walch A, Werner M, Höffler H. Quantitative gene expression analysis in microdissected archival formalin-fixed and paraffin-embedded tumor tissue. Am J Pathol 158:419-29, 2000.