Updated: May 18
September 24, 2014. We recently developed a candidate assay for early detection of Ovarian Cancer. Early detection for cancer has the potential to dramatically alter outcomes in Oncology. Like many topics in oncology, controversies remain over the magnitude of benefit of early detection (e.g. last week’s Salvo in the Wall Street Journal), however, when combined with molecular profiling (discussed below) a robust non-invasive cancer detection assay holds great promise for saving lives.
Ovarian cancer is one tumor type for which early detection is likely to be most beneficial. Figure 1 shows Ovarian cancer statistics from www.seer.cancer.gov which collects epidemiological data on cancer detection and survival. Due to its internal location and relative lack of symptoms, only 15% of ovarian cancer is detected when localized to the primary site, whereas more than 60% of ovarian cancer has already metastasized at diagnosis. The difference in 5 year survival is striking: 92% survival if localized; 27% if metastatic. The result is that ovarian cancer is one of the more lethal tumors with only 45% of patients surviving 5 years. Compare that to prostate cancer at 99%, breast cancer at 89% or Colon cancer at 65%. The most brutal is Pancreatic Adenocarcinoma at 7% (numbers according to SEER).
One concern regarding interpretation of the SEER data, and the source of most early detection controversy is the possibility of a selection bias in the detection of localized tumors. Perhaps a larger proportion of early tumors would have never progressed; therefore, treatment and/or removal of all early lesions may be unnecessary and may actually increase risk to the patient.
The recent availability of large molecular datasets in oncology provides the opportunity for robust molecular characterization of tumors, and has already led to RNA based prognostic signatures for several tumor types (e.g. Genomic Health’s Oncotype DX breast, colon and prostate tests GHI prognostics test page). The presence of prognostic signatures in localized tumors demonstrates the utility of RNA expression profiles to distinguish between benign and malevolent early lesions.
Twenty to Thirty RNAs can define high grade serous ovarian cancer
We recently applied our analytical framework and tools to ovarian cancer and identified a concise set of genes that are highly enriched in most high grade serous ovarian cancer, relative to either other tumor types or a panel of normal tissues (Figure 2A, B).
The most studied ovarian tumor marker CA125, originally identified by Bob Bast and coworkers over thirty years ago (Reactivity of a monoclonal antibody with human ovarian carcinoma), exhibits relatively poorer RNA expression enrichment in either tumor panels (Figure 3). Note that each individual gene in the ovarian signature exhibits greater ovarian cancer specific expression than the RNA for CA125. Identification of CA125 was a remarkable feat in 1981, the new data should enable the creation of greatly improved marker sets as shown here.
To demonstrate the power of the analytical tools we’ve created, we also generated signatures for Pancreatic Adenocarcinoma (figure 4A, B) and 6 other tumor types: colorectal, endometrial, kidney (spanning clear cell, chromophobe and papillary), non small cell lung (spanning adenocarcinoma and squamous), melanoma and prostate (figure 4C).
This ovarian cancer-specific signature may represent a novel tool for the early blood based detection of malignant ovarian tumors. The signature, however, represents only the first rate-limiting step towards the development of a blood based test for early ovarian cancer.
Looking for partners to help test the ovarian (and potential other) signature(s)
We are interested in identifying committed collaborators to help test the signature:
Measuring signature genes in appropriate subsets of circulating nucleic acid compartments, e.g. exosomes, cell free nucleic acids.
Determining specificity and limit of detection in archived samples of both patients and unaffected individuals.
Testing the signature in characterized archived frozen serum samples using a “prospective-retrospective” design used successfully for previous expression based diagnostic tests.