Skip to main content
Premium Trial:

Request an Annual Quote

AGBT: WashU Applies Genomic Testing to Predict Cancer Relapse, Drug Response, Vaccine Epitopes


MARCO ISLAND, Fla. (GenomeWeb) – At the Advances in Genome Biology and Technology meeting here this week, Elaine Mardis, co-director of the Genome Institute at Washington University in St. Louis, provided an update on how her institution has been implementing genomic testing for different cancer types in both clinical research and diagnosis.

In breast cancer, WashU researchers have focused on patients with estrogen receptor-positive tumors. Some of these relapse after many years, and the goal is to find biomarkers in primary biopsies that can predict the long-term outcome. To that end, the researchers sequenced a panel of 83 genes with relevance to ER-positive breast cancer in samples from 650 patients who had undergone tamoxifen therapy for five years and for whom 10 years of follow-up data on their relapse status was available. The average age of the formalin-fixed samples was 25 years.

No matched normal DNA was available, but the researchers were able to use filtering approaches to identify multiple mutations in many of the genes, which they correlated with the outcome data to identify predictors of good or bad prognoses. Two genes in particular correlated with poor outcome, Mardis reported. Going forward, she and her colleagues plan to sequence the panel in additional cohorts, and to identify combinations of mutations that can predict long-term outcome.

For acute myeloid leukemia, genomic testing is already moving into the clinic at Washington University Medical Center. Some AML patients don't respond to treatment at all, and others relapse within the first six months. While a number of predictors already exist for the treatment response of AML patient – such as age and cytogenetics – they are “very imperfect”, Mardis said, and the researchers want to test if genomic analyses allow for better predictions. The plan is to sequence the exome of every new AML patient coming to WashU, using DNA from both a bone marrow biopsy and from normal skin. The exome will be augmented with additional probes for 259 genes that are known from previous studies to be recurrently mutated in AML. In addition, transcriptome sequencing will be able to identify gene fusions.

In patients with other tumor types who have run out of treatment options after undergoing the standard of care, WashU already performs comprehensive genomic testing that integrates whole-genome sequencing, exome sequencing, and RNA-seq in order to identify potentially druggable targets. As an example, Mardis mentioned the case of a pediatric patient with a low-grade glioma. The analysis of her tumor genome revealed 33 somatic point mutations in coding regions and four somatic indel mutations, including a small insertion in the BRAF gene. Based on this result, she was initially enrolled in a trial for a MEK inhibitor, which acts in the same pathway as BRAF. After her cancer progressed after eight months, she was enrolled in a trial for a BRAF inhibitor, and her treatment is ongoing.

To help with the interpretation of somatic mutations in patient tumors, the WashU researchers are drawing on a number of computational tools. One is the Database of Curated Mutations (DoCM), which currently contains 488 variants from 63 genes that have been implicated in many cancer types. Another one, the Clinical Interpretations of Variants in Cancer (CIViC) database, is an open-access, open-source web resource that is currently still under development. Mardis said WashU plans to launch CIViC at the Annual Society of Clinical Oncology meeting this spring, and plans to allow experts to contribute to the resource, which aims to enrich the clinical knowledge about cancer variants. A third database the researchers are using is the Drug Gene Interaction Database (DGIdb), which provides information on cancer drugs and clinical trials associated with mutations in certain genes.

A final area of cancer where genomic analyses are proving helpful is for the discovery of immuno-epitopes for personalized cancer vaccines. In melanoma patients, for example, the researchers have sequenced skin biopsies of tumor and normal samples to identify somatic mutations and have analyzed the data for potential immuno-epitopes, peptides that are most likely to be immunogenic in a vaccine. They plan is to apply a similar approach to patients with triple-negative breast cancer and lung adenocarcinoma, Mardis said.