Molecular characterization of ovarian cancers in PLCO: Evaluating subgroups and prognostic signatures
PLCO is part of the Ovarian Cancer Cohort Consortium (OC3) which includes over 25 cohorts with 1.6 million women and 8,000 ovarian cancers. In the first analysis in the OC3, we demonstrated substantial heterogeneity of most established ovarian cancer risk factors by histologic subtype, supporting the strong biological heterogeneity of ovarian carcinomas. Importantly, most risk factors showed only weak or no associations with HG serous cancers.
Relying purely on morphologic assessment may have limitations in identifying biologically diverse subtypes. In OC3, we are now evaluating other approaches to identify subgroups of ovarian cancer, such as tumor aggressiveness and tumor dominance. In a subgroup of cohorts with tissue specimens (including PLCO, the Nurses’ Health Study, and several others), we seek to evaluate risk factor associations by molecular subgroups to better understand the etiologic heterogeneity of ovarian cancer.
Various approaches have been proposed for molecular classification of ovarian carcinomas, including panels of immunohistochemical markers and mRNA signatures. We recently conducted genome-wide methylation profiling of ovarian carcinomas and found a methylation signature that identifies ovarian carcinoma subgroups.
High grade serous ovarian cancers are the most common and fatal subtype. TCGA has identified molecular subgroups of HG serous cancers with prognostic differences. An mRNA signature has been replicated in different populations that can differentiate prognostically relevant HG serous subgroups. Similarly, we recently found a methylation signature that identifies subgroups of HG serous cancers with prognostic differences.
We are working with two large consortia, the OC3 and the Ovarian Cancer Association Consortium (OCAC), to evaluate prognostic differences in HG serous subtypes and to understand the biologic and epidemiologic underpinnings of these differences. This will allow us to replicate new findings in well-powered studies and contribute PLCO to ongoing studies of molecular classification and prognosis.
As a PLCO-specific goal, we propose to examine molecular differences between tumors that were related to CA-125/TVU abnormalities (screen-detected) versus those that were not but diagnosed less than 9 months from a screening (interval cancers). Such differences could explain why CA-125/TVU works only for a subset of cancers and may lead to identification of biomarkers for interval cancers.
For the proposed work, paraffin tissue cores are suitable sources for RNA and DNA that allow measuring mRNA (Nanostring) and methylation signatures, as well as somatic mutations. TMA sections are suitable to stain for relevant IHC markers.
For all ovarian cancer cases in PLCO with available tissue cores and TMAs, we propose to extract nucleic acids (combined RNA and DNA extraction) from a single tissue core and cut 12 sections for immunohistochemistry (11 markers plus H&E). Based on preliminary information from IMS, 217 cases with tissue should be available from PLCO. For all cases, we propose to run a Nanostring panel, conduct methylation profiling, and test for a panel of somatic mutations. In addition, we will conduct IHC staining on TMA sections. We have successfully extracted nucleic acids from comparable tissue cores and run methylation arrays, Nanostring, and mutation panels. This approach is highly efficient to evaluate several established and new molecular classification approaches for ovarian cancers. PLCO is already strongly invested in both consortial efforts and an important contributor of risk factor and genetic data.
1. Evaluate molecular subgroups of ovarian cancers
a. Replicate a methylation signature for ovarian cancer subtypes
In the Polish Case Control Study and in the SEER discard study, we developed a signature for ovarian cancer subtypes. We will apply the signature and evaluate the agreement with morphologic subtype based on the study result, based on an expert panel, and in relation to other molecular markers. If the signature replicates in PLCO, we will further replicate in the two consortia.
b. Replicate a Nanostring mRNA signature for ovarian cancer subtypes
The Ovarian Cancer Tissue Group in OCAC has developed a Nanostring mRNA panel for ovarian cancer subtyping. Similar to 1a, we will evaluate the Nanostring tumor classification in PLCO and contribute PLCO data to the consortial evaluation of the signature.
c. Evaluate IHC markers for classification and prognosis: ARID1A, CDKN2A, DKK1, ER, HNF1B, MDM2, PGR, TP53, TFF3, VIM, and WT1.
These markers have been widely evaluated for classification of ovarian cancers. Several of these markers also have prognostic relevance for specific ovarian cancer subtypes. We will contribute the PLCO results to consortial efforts of classification and prognosis.
d. Evaluate a somatic mutation panel in ovarian cancer subtypes based on histologic and molecular classification
2. In the OC3, evaluate risk factor and biomarker associations by molecular subtypes of ovarian cancers (using classifications evaluated in #1)
3. Evaluate prognostic markers for high grade serous carcinomas
a. Nanostring mRNA survival signature
b. Methylation survival signature
4. Explore molecular differences between screen-detected vs. interval cancers in the PLCO intervention arm using mRNA, metylation, and somatic mutations