For Variant Interpretation, a New Generation of Cancer Knowledgebases Can Help

The earliest days of interpreting cancer-associated genetic variants weren’t terribly overwhelming. With limited genomic data available and minimal genetic testing being conducted, researchers or clinical specialists could quickly review the literature to determine whether a variant was likely benign or pathogenic. This manual process, often powered by PubMed and Google searches, was imperfect but feasible.

Today, that process is hardly recognizable. Whether it’s for research or clinical lab testing purposes, variant interpretation for oncology has become a massive undertaking. The affordability of DNA sequencing technologies has led not only to more cancer samples being sequenced, but also to exome- or whole genome-scale analysis even in the clinic. What was once a trickle of cancer genomic data is now a tsunami.

Make no mistake: this represents massive progress, both for our understanding of the biology of cancer and our ability to tailor treatment to patients for significantly improved outcomes. All of these data have made it possible to realize the potential of precision medicine in the oncology field. Patients can be matched to the treatment most likely to work for their cancer, or connected to the most relevant clinical trial. On the discovery and development side, variant analysis is streamlining the process of clinical trials and approval for new drug candidates since they can be targeted to specific biomarkers rather than tested in a random population.

There is enormous pressure to improve and scale the variant interpretation process, which in too many labs is still conducted through a tedious manual process. Literature review alone can take hours for a single variant, with new publications coming out every day that must be added to the mix. Checking genomic and proteomic function databases adds another layer of complexity. On top of that, it is often essential for variant interpretation specialists to keep up with drug approvals, clinical trial enrollment status, the universe of targeted therapies and much more.

For a truly scalable approach, researchers and molecular pathologists alike need a single source where all of this information is pulled together into a handy dashboard to streamline the interpretation process. Many tools have tried unsuccessfully to meet these needs in the past, but now, a new generation of cancer knowledgebases may finally be sophisticated enough to serve as one-stop shops for variant interpretation.

Hurdles to clear

Reaching this point has not been easy. Myriad challenges over the years have made it very difficult for scientists to choose and deploy an effective tool for variant interpretation. As more and more products claimed to solve the interpretation problem, the crowded landscape of offerings became its own challenge. Researchers may find a promising solution, only to realize the tool was abandoned by its developers and hasn’t been updated with relevant information in years. In other cases, bespoke tools can be helpful for a very specific need, but they have to be cobbled together with other tools to form a hodgepodge workflow to support the entire interpretation process. The end result tends to be unstable, at risk of failure from any of the components in that workflow. Even when the process works, it forces users to consult many different data sources along the way, slowing the interpretation considerably.

An ongoing stress point for all of these tools comes from the rapidly growing scientific and clinical literature. Products designed to work well with the volume of cancer variant data that existed, say, in 2010 may not be able to process the volume of data available today. With new results from large-scale genomic studies and countless case reports published on a regular basis, any interpretation solution has to be built to manage exponentially growing data repositories.

For variant interpretation specialists in clinical labs, another point of contention has come from the decision about whether to outsource the cancer genome analysis process or to perform it in-house. Outsourcing has the obvious benefit of making another entity responsible for setting up a robust interpretation workflow, but it can come with higher costs and long turnaround times. Doing things in-house makes it easier to control costs and reporting time, but requires the clinical lab team to handle extensive analysis and interpretation needs on their own.

Priority features

The latest interpretation offerings come in the form of cancer knowledgebases that promise to serve as one-stop shops for the entire workflow. These tools are significantly better than their predecessors, but some perform better than others. Here’s a quick rundown of the most important factors to consider in any evaluation of these products.

Curation and maintenance protocols are easy to overlook but are among the most critical for how a cancer knowledgebase will function. First, look for tools that rely on expert curation for scientific evidence and other relevant data; ideally, a team of MD- or PhD-level physicians or scientists should be reviewing information before it is incorporated into the knowledgebase. Second, check the update schedule. With information evolving so quickly in the cancer space, the best knowledgebases are updated daily to get the latest scientific data, professional guidelines, regulatory status for drugs and clinical trial availability.

Next up: how extensive is the content held in the knowledgebase? For research purposes, missing a single relevant publication in the interpretation process would be frustrating, but for clinical variant reporting, it could be a nightmare that leads to inaccurate recommendations for a patient’s care. Any worthy cancer knowledgebase should be as comprehensive as possible, including complete data about targeted therapies, drug regulatory status, clinical trials with detailed enrollment information, and evidence of therapeutic efficacy. In addition, the knowledgebase should allow users to search or browse all of the genes and variants known to be associated with cancer, ideally with complex molecular profiles to fully represent their functional impact.

Last but not least, don’t forget to check for ease of use. The sophisticated inner workings of a cancer knowledgebase should be wrapped in a seamless and intuitive interface with great search features for a positive user experience.

A cancer knowledgebase that meets all of these needs can be a powerful asset in the lab or the clinic, streamlining the variant interpretation process and ensuring high-quality results while keeping up with the onslaught of new data.

About the authors

Sara Patterson, PhD, serves as associate director of clinical genome informatics products at The Jackson Laboratory. 

Cara Statz, PhD, serves as senior clinical analyst at The Jackson Laboratory.