Mitochondrial DNA mutations in pancreatic cancer.

Abstract

Background: Somatic mutations of mitochondrial DNA (mtDNA) are increasingly being recognized in many human cancers, but automated sequencing of 16.5 kb of DNA poses an onerous task. We have recently described an oligonucleotide microarray (MitoChip) for rapid and accurate sequencing of the entire mitochondrial genome (Zhou et al., J Mol Diagnostics, 8: 9_14, 2006), greatly facilitating the analysis of mtDNA mutations in cancer. In this report, we perform a comprehensive cataloging of somatic mutations in the mitochondrial genome of human pancreatic cancers using our novel array-based approach.

Materials and methods: MitoChip analysis was performed on DNA isolated from 15 histologically confirmed resection specimens of pancreatic ductal adenocarcinomas. In all cases, matched nonneoplastic pancreatic tissue was obtained as germline control for mtDNA sequence. DNA was extracted from snap-frozen cryostat-embedded specimens and hybridized to the sequencing microarray after appropriate polymerase chain reaction amplification and labeling steps. The vast majority of somatic mutational analyses of mtDNA in human cancers utilize lymphocyte DNA as germline control, without excluding the potential for organ-specific polymorphisms. Therefore, we also examined a series of 15 paired samples of DNA obtained from nonneoplastic pancreata and corresponding EBV-immortalized lymphoblastoid cell lines to determine whether lymphocyte DNA provides an accurate surrogate for the mtDNA sequence of pancreatic tissue.

Results: We sequenced 497,070 base pairs of mtDNA in the 15 matched samples of pancreatic cancer and nonneoplastic pancreatic tissue, and 467,269 base pairs (94.0%) were assigned by the automated genotyping software. All 15 pancreatic cancers demonstrated at least one somatic mtDNA mutation compared to the control germline DNA with a range of 1-14 alterations. Of the 71 somatic mutations observed in our series, 18 were nonsynonymous coding region alterations (i.e., resulting in an amino acid change), 22 were synonymous coding region alterations, and 31 involved noncoding mtDNA segments (including ribosomal and transfer RNAs). Overall, somatic mutations in the coding region most commonly involved the ND4, COI, and CYTB genes; of note, an A-G transition at nucleotide position 841 in the 12sRNA was observed in three independent samples. In the paired analysis of nonneoplastic pancreata and lymphoblastoid cell line DNA, 14 nucleotide discrepancies were observed out of 226,876 nucleotide sequences (a concordance rate of 99.99%), with 9 samples demonstrating a perfect match across all bases assigned.

Conclusions: Our findings confirm that somatic mtDNA mutations are common in pancreatic cancers, and therefore, have the potential to be a clinically useful biomarker for early detection. Further, our studies confirm that lymphocyte DNA is an excellent, albeit not perfect, surrogate for nonneoplastic pancreatic tissues in terms of being utilized as a germline control. Finally, our report confirms the utility of a high-throughput array-based platform for mtDNA mutational analyses of human cancers.