Clinical Interpretation of Cancer Variants: How Complicated Is It? How Can We Tackle It?
Part 1: The Challenges
Next-generation sequencing (NGS) has paved the way for fast and high-throughput molecular analysis of clinical samples. In addition, advancements in technologies have enabled more accurate and precise identification of genetic variations in cancer, as well as in targeting them with appropriate drugs. However, the practices for NGS analysis, variant classification, annotation and reporting vary significantly between different laboratories. In view of these differences, a joint consensus recommendation was issued by the Association for Molecular Pathology, American Society of Clinical Oncology, College of American Pathologists, and American College of Medical Genetics, setting standards and guidelines for the interpretation and reporting of sequence variants in cancer.1,2 Even though a concordance rate of 34% was seen in variant classifications between laboratories after the 2015 guidelines for germline variant interpretation1 were issued. This high discordance was mainly due to how individual laboratories interpreted the data.3 Why is this so? In order to find a solution to the problem, we need to understand the challenges in clinical interpretation of cancer variants. This two part series will address the challenges and possible solutions to tackle them.
There are several categories of cancer variants. More are constantly emerging. How can you keep up?
Several large-scale projects such as the International Cancer Genome Consortium (ICGC), The Cancer Genome Atlas (TCGA) and the Catalogue of Somatic Mutations in Cancer (COSMIC) have curated information on somatic mutations. These databases include millions of variants identified across millions of samples spanning different cancer types.4 In addition to the known variants there are many more novel variants that will be encountered as more samples are sequenced. Of these, some are significant and actionable and some are not. And then, there are variants with clinical significance and others with potential clinical significance, but not investigated enough; variants that are significant in one cancer group, but which lack evidence supporting clinical significance in others; and then, clinically significant, but rare variants. Some have different implications based on whether they are present alone or in combination with another mutation. For example, EGFR L858R has different treatment implications based on if it is present by itself or if T790M is also present.5 Add to the mix, the unclassified and uncurated variants. But how meaningful are they? Do they have pathogenicity in cancer? What is their clinical relevance? How actionable are they?
There are numerous databases and resources available. Each one provides different information. How do you determine which has the most relevant information needed for a particular variant?
Several databases and resources, (such as ClinVar, CIViC, COSMIC, DoCM, OMIM, to name a few), collate information on variants. Each one may provide different details on the specific variant you are looking for. So, the preferred way is to review all the sources and collate the information.
How do you account for regional differences in variant classification? How easy is it to understand and keep up with global, regional and privacy regulations associated with variants and determine what actions can be taken with a specific variant?
There are regional differences in how certain mutations are classified in terms of whether treatment options are approved based on a local drug agency or local medical guidelines. Some mutations are classified with high clinical significance in some regions and not others. For example, the MET exon 14 skipping mutation would be classified as an AMP Tier I, Level A in the USA and Tier II, Level C in Europe, Canada and the UK. These variations in classification depend on the local drug label or local medical guidelines. A lab will have to keep up with all these regulations and be conversant with them.
How much time and how many resources are needed to curate all the information and determine clinical relevance and actionability?
In order to have up-to-date and correct information in a high volume testing lab, a team of curation scientists who understand the significance and relevance of cancer variants may be needed. In addition to perusing the literature constantly, they also have to manually obtain relevant information from various databases. Since clinical labs may employ a large selection of panels and interrogating many genes, this would necessitate a significant number of scientists dedicated curate. Finally, all this information has to be summarized into a report with possible patient management options based on the mutations identified that can be taken by a clinician. Owing to the differences in information available from various sources, and the manual nature of this process, accuracy and reproducibility can suffer.
How do you tackle all these challenges?
Solutions that can mitigate some of these challenges are available. In part two of this series, we will review an automated solution that can make clinical interpretation of cancer variants efficient, reproducible and meaningful.