The art (and science) of variant interpretation

By Jill Polk, Clinical Science Liaison, Invitae

When it comes to genetic testing in healthcare, it’s important that a laboratory excels not only at identifying genetic variants, but also at evaluating and clinically classifying the variants that it finds. In other words, genetic testing is not only about finding the differences in a person’s DNA—it’s also about figuring out if those differences are likely to affect the person’s health.

At Invitae, we take the challenge associated with finding the more difficult variants quite seriously. Yet variant detection is just one piece of the puzzle. Variant classification is the other area where the hard work comes in. It’s often not easy to determine whether a variant is likely benign/benign (normal human variation not related to disease), is likely pathogenic/pathogenic (variation related to disease), or is a variant of uncertain significance or “VUS” (there is not enough information available to support a definitive classification).

So how does Invitae approach variant interpretation? 

Our experienced team of clinical genomics scientists created a system that we named semiquantitative, hierarchical evidence-based rules for locus interpretation, known as Sherloc for short. In the spirit of transparency and scientific excellence, we took it a step further and published our carefully constructed classification framework in a peer-reviewed article in Genetics in Medicine

Additional details about the ins and outs of Sherloc can be found on our website, but for now, these are the key points to keep in mind: 

  • Sherloc is a systematic process designed to be accurate and reproducible
  • It’s based on five types of evidence
  • It uses objective, rule-based scoring
  • It relies on point thresholds to classify a variant using the standards and guidelines established by the American College of Medical Genetics (ACMG)

We follow this standardized method of variant classification consistently which, in turn, gives patients highly accurate results each time we analyze a sample. 

In addition, Invitae developed our own unique functional modeling platform (FMP), which enables us to give definitive results more often by reducing VUS rates when it is possible and responsible to do so. (Learn more about FMP here.)

However, this leads to another important consideration: protecting against “over-calling.” Over-calling is a situation in which a laboratory may classify a variant as disease-causing when the available evidence may not sufficiently justify that classification. In such instances, that laboratory may have to later re-classify the variant back to VUS or even benign. As you can imagine, this can be extremely detrimental for patients and providers. In an extreme case, a provider may have performed preventive surgery on a patient based on the “positive” genetic test result. 

Reclassifications are sometimes unavoidable as new evidence comes to light; however, at Invitae, we aim to be as careful as possible about our variant calls. Invitae’s Sherloc framework prevents us from putting too much emphasis on any one particular line of evidence. 

To check the accuracy of our variant calls, the Invitae team regularly monitors something we call prospective performance. This means we track our variant calls over time to see if they agree with the general consensus among other reputable labs that aggregate evidence for those same variants from their testing. We see this as critical for monitoring the quality of our variant classifications as we incorporate more sophisticated data modeling approaches; between Sherloc and FMP, Invitae’s prospective performance exceeds 99%.

Overall, variant interpretation and transparent data sharing are key factors that help distinguish “great” labs from “good” labs. At Invitae, we continuously strive to meet, and often establish, the highest standards in clinical variant interpretation in genetic testing.

Our goal is for every patient and healthcare provider to have the utmost confidence in the medical decisions they make based on the genetic test results that we provide.