Molecular detection of TMPRSS2-ERG gene fusion in pT2–pT3 prostate cancer specimens and biopsies

Abstract

<h3>Aims</h3> Several recurrent gene re-arrangements participate in the aetiology of prostate cancer, all involving members of the <i>ETS</i> family of transcriptional factors. We addressed the detection sensitivity of the most common gene fusion, <i>TMPRSS2-ERG</i> on chromosome 21q22, in paraffin-embedded tissues derived from prostatectomies and from core needle biopsies. <h3>Methods</h3> Tissues included blocks (171 total) of sliced prostatic glands from <i>n</i>=22 patients treated surgically at The Ottawa Hospital, as well as 25 blocks containing core biopsies from seven patients. An average of eight blocks per case were analysed. RNA was isolated after Histosolve extraction of 10 µm sections followed by proteinase K digestion and column purification. The presence of <i>TMPRSS2-ERG</i> fusion products was detected by reverse transcription polymerase chain reaction using several forward primers located in <i>TMPRSS2</i> exon 1, and reverse primers located in <i>ERG</i> exon 3 to 6. The predicted products (110–541 bp), were separated by polyacrylamide gel electrophoresis, stained and imaged. <h3>Results</h3> All selected primer pairs had the potential to identify nearly all of >15 <i>TMPRSS2-ERG</i> fusion subtypes characterised so far. The correct products [T1/E4, when exon 1 (<i>TMPRSS2</i>) is fused to exon 4 (<i>ERG</i>)], were identified in the VCAP but not the LnCAP cell line, even with exon 6 reverse primers. However, with these same reagents, the largest products (>300 bp) resulting from breakpoints in introns 2 or 3 of <i>TMPRSS2</i> and introns 5 and 6 of ERG were not discernable in paraffin-embedded tissues. The majority of prostatic specimens also expressed selected exons of the <i>ERG</i>, <i>ETV1</i> and <i>TMPRSS2</i> genes. The T1/E4 fusion was detected as a 110–258 bp product with exon 4 reverse primers, in one or more regional sections of the prostatic gland in 15 of 22 (68%) cancer patients and in the core needle biopsies of 7 of 7 patients. The presence of the T1/E4 fusion in different sections was associated with the presence of malignant cells in histologically stained slides. There was no significant correlation with tumour stage (comprising 55% pT2, 41% pT3) nor grade (comprising 55%, 36% and 9% Gleason scores of 6, 7 and 8 respectively). <h3>Conclusion</h3> RNA fragmentation in fixed tissues limits the molecular diagnosis of prostate cancer to only a subset of genomic breakpoints responsible for oncogenic <i>TMPRSS2-ERG</i> gene fusions.