Identification of markers that discriminate between cervical cancer subtypes in formalin-fixed paraffin-embedded (FFPE) tissues


Year:

Session type:

John Hall1, Hui Sun Leong2, Crispin Miller2, Catharine West1
1Translational Radiobiology Group, The Paterson Institute for Cancer Research, Manchester, United Kingdom,2Applied Computational Biology and Bioinformatics Group, The Paterson Institute for Cancer Research, Manchester, United Kingdom

Background

A primary concern with using FFPE for transcription profiling analysis is the presence of chemically-modified or fragmented transcripts, which may adversely affect the reliability and interpretation of microarray results. Squamous cell carcinoma (SCC) and adenocarcinoma (AC) are distinct histological subtypes of invasive cervical cancer. The molecular mechanism underlying the different clinical features between these tumours is still unclear. The key objective of this work is therefore to address whether FFPE materials can be used to determine differentially expressed genes pertinent to the classification of the different histological subtypes innate to carcinoma of the cervix.

Method

RNA was extracted from 28 FFPE human cervix samples (19 SCC and 9 AC) and hybridised to Affymetrix Human Exon 1.0 ST arrays. Global gene expression changes between SCC and AC were identified using Limma with an FDR <0.01.

Results

Exon array analysis revealed 1224 differentially expressed genes, of which 1068 showed significantly higher expression levels in SCC relative to AC. This includes SCC markers, which have been reported in the biomedical literature, such as TP63, KRT5, SFN, TRIM29 and S100A9 that are known to be involved in epidermal development and differentiation. On the contrary, only 156 genes were found to be specifically up-regulated in AC including TFF3, PROM1 and MUC3A. Cross-validation of these results using an independent dataset derived from fresh-frozen human non-small cell lung cancer (NSCLC) with known histology data showed that gene expression profiles identified from our cervix FFPE data are robust enough to stratify the histological subtypes of AC and SCC of NSCLC.

Conclusion

Our results demonstrate that clinically-relevant gene expression profiles can be obtained from FFPE tissues using our optimised microarray profiling protocols. Selected SCC/AC markers identified in this study will be followed-up to further establish their potential in discriminating the different histological subtypes of cervix carcinoma.