Rational Cancer Diagnostics Using Rare Mutation Detection Technology With Massively Parallel Sequencing

Worldwide, cancers remain a leading cause of death. The judicious use of cancer diagnostics -- broadly defined as tests for cancer -- has great potential to reduce disease morbidity and mortality. Impeding this potential is the difficulty of creating effective new tests, as the techniques successful for one type of cancer frequently cannot be generalized to another. Although the ability to detect cancer-specific DNA mutations at the low levels commonly encountered in clinical specimens would yield a promising, broadly applicable diagnostic strategy, existing technologies have been unacceptably limited in throughput or accuracy. Here we describe the development and application of a scalable, generalizable DNA sequence-based technology for the reliable detection of mutations. By drastically reducing artifacts introduced through sample preparation and massively parallel sequencing, rare mutations arising from cancer cells – when present – can be confidently discriminated from a large excess of non-mutant DNA. The technology can be directed to virtually any genomic region, affording rational test design. When applied to routinely collected Pap specimens, our approach detected cancer-specific mutations in 41% (9 of 22) and 100% (24 of 24) of women harboring various stages of ovarian and endometrial cancers, respectively. Our approach was highly specific, as no false positives were detected in a cohort of Pap specimens collected from women without gynecologic cancer. We also demonstrate how the urine of patients with urothelial carcinoma can be utilized to predict disease recurrences. Eighty-eight percent (7 of 8) of patients with a detectable mutation had recurrences while none were detected in the six patients without recurrent disease (P