A methodology for utilization of predictive genomic signatures in FFPE samples

A Method to Correlate mRNA Expression Datasets Obtained from Fresh Frozen and Formalin-Fixed, Paraffin-Embedded Tissue Samples: A Matter of Thresholds

Background

Gene expression profiling of tumors is a successful tool for the discovery of new cancer biomarkers and potential targets for the development of new therapeutic strategies. Reliable profiling is preferably performed on fresh frozen (FF) tissues in which the quality of nucleic acids is better preserved than in formalin-fixed paraffin-embedded (FFPE) material. However, since snap-freezing of biopsy materials is often not part of daily routine in pathology laboratories, one may have to rely on archival FFPE material. Procedures to retrieve the RNAs from FFPE materials have been developed and therefore, datasets obtained from FFPE and FF materials need to be made compatible to ensure reliable comparisons are possible.

Aim

To develop an efficient method to compare gene expression profiles obtained from FFPE and FF samples using the same platform.

Methods

Twenty-six FFPE-FF sample pairs of the same tumors representing various cancer types, and two FFPE-FF sample pairs of breast cancer cell lines, were included. Total RNA was extracted and gene expression profiling was carried out using Illumina’s Whole-Genome cDNA-mediated Annealing, Selection, extension and Ligation (WG-DASL) V3 arrays, enabling the simultaneous detection of 24,526 mRNA transcripts. A sample exclusion criterion was created based on the expression of 11 stably expressed reference genes. Pearson correlation at the probe level was calculated for paired FFPE-FF, and three cut-off values were chosen. Spearman correlation coefficients between the matched FFPE and FF samples were calculated for three probe lists with varying levels of significance and compared to the correlation based on all measured probes. Unsupervised hierarchical cluster analysis was performed to verify performance of the included probe lists to compare matched FPPE-FF samples.

Results

Twenty-seven FFPE-FF pairs passed the sample exclusion criterion. From the profiles of 27 FFPE and FF matched samples, the best correlating probes were identified for various levels of significance (Pearson P<0.01, n = 1,432; P<0.05, n = 2,530; and P<0.10, n = 3,351 probes). Unsupervised hierarchical clustering of the 27 pairs using the resulting probes yielded 25, 21, and 19 correctly clustered pairs, respectively, compared to 1 pair when all probes were used.

Conclusion

The proposed method enables comparison of gene expression profiles of FFPE and/or FF origin measured on the same platform.

Cross-platform prediction of gene expression signatures

Gene expression signatures can predict the activation of oncogenic pathways and other phenotypes of interest via quantitative models that combine the expression levels of multiple genes. However, as the number of platforms to measure genome-wide gene expression proliferates, there is an increasing need to develop models that can be ported across diverse platforms. Because of the range of technologies that measure gene expression, the resulting signal values can vary greatly. To understand how this variation can affect the prediction of gene expression signatures, we have investigated the ability of gene expression signatures to predict pathway activation across Affymetrix and Illumina microarrays. We hybridized the same RNA samples to both platforms and compared the resultant gene expression readings, as well as the signature predictions. Using a new approach to map probes across platforms, we found that the genes in the signatures from the two platforms were highly similar, and that the predictions they generated were also strongly correlated. This demonstrates that our method can map probes from Affymetrix and Illumina microarrays, and that this mapping can be used to predict gene expression signatures across platforms.

Formalin fixation in the '-omics' era: a primer for the surgeon-scientist

Formalin is the most commonly used tissue fixative worldwide. While it offers excellent morphological preservation for routine histology, it has detrimental effects on nucleic acids. Most studies of nucleic acids have therefore used fresh frozen tissue, the collection and storage of which is resource intensive. The ability to use modern genomic, transcriptomic and epigenomic methods with nucleic acids derived from formalin-fixed, paraffin-embedded (FFPE) tissues would allow enormous archives of routinely stored tissues (usually with well-annotated clinical data) to be used for translational research. This paper outlines the effects of formalin on nucleic acids, describes ways of minimizing nucleic acid degradation and optimizing extraction, and reviews recent studies that have used contemporary techniques to analyse FFPE-derived nucleic acids (with a focus on malignant tissue sources). Simple tips are also offered to ensure the utility of your institution's samples for future studies, and broadly applicable guidelines are listed for those contemplating their own study using FFPE-derived material.