Everyone has genetic variation….and lots of it. It’s part of what makes us each unique. Genetic variation is defined by differences in our own genome and a reference genome. (The fact that there is only one reference genome selected to compare to all of our collective, rich, human diversity is a topic for another post.) Another source of genetic variation is acquired mutations, or variants. The concept of genetic mutants has a scary ring to it: cue images of flies with legs in place of antennae. But, at the heart of it, variants aren’t all bad. In fact, the vast majority of them are “benign,” or harmless. The science of teasing out whether a variant is pathogenic or benign is called variant interpretation and it’s a big part of what we do at Clinical Genetic Testing companies. The process involves gathering and considering a bunch of evidence, which should lead to a verdict. But when the pieces of evidence don’t add up, or when they conflict with each other, you have a “variant uncertain significance” or a VUS (the pronunciation of which is still hotly debated [a little nod to my Genetics friends across clinical labs]).

The Mission to Resolve VUS

The reasons a variant might be of uncertain significance can be complex, but the most frequent (and least interesting) cause is simply due to a lack of information for rare variants. Historic (and unfortunately still-current) inequities in access to genetic testing have led to further disparities in the genetic variants accessible in repositories like ClinVar and projects like GnomAD, which are heavily biased towards a single population. These databases are rich resources of variant interpretation evidence and also frequently serve as fodder for further laboratory-based research which, in turn, provides yet another source of variant interpretation evidence. Combine that with the previously mentioned problem related to a single reference genome, and a catalyst for increased VUS rates in underserved populations is established. No one likes a variant of uncertain significance because it creates…well…uncertainty. Is the patient at increased risk or not? Most VUSs eventually end up being reclassified as benign (but not ALL of them!). So what to do about this VUS problem?

Enter functional studies! Many functional studies hold tremendous power for variant interpretation. Not only can they be assigned a strong line of evidence when well-conceived, meticulously executed and properly validated, but virtually any variant can be made and tested at any time, even if it has never been seen in a patient (or deposited to ClinVar)! The downside is that until recently, functional studies tended to be expensive and time-consuming. But we are now on the precipice of being flooded with functional data from Multiplexed Assays of Variant Effects (MAVE). A MAVE is a type of functional assay that is ultra-high-throughput and can produce data for virtually every variant that could possibly arise (at least the small single-nucleotide or small deletion ones, which are the biggest source of VUS, anyway). Because this data can be at the variant interpretation expert’s fingertips the moment a novel variant is detected, MAVEs have the power to erase some of the VUS disparities observed in underserved populations. 

Trickier Than It Looks

Problem solved right? Not quite. Remember, the conditions for us to be able to harness data from functional studies is that they need to be well-conceived, meticulously executed and properly validated. Study conception and execution are elements that can be navigated by the thoughtful investigator. As to the proper validation, however, the investigator (or variant interpretation scientist) is at the mercy of existing data—which, as outlined above, can be very sparse. The current recommendation for validation of a functional study is to compare the function of variants included in the assay with known pathogenic and known benign variants as controls. Ideally, the validation variants are of the same variant type (usually missense variants: small changes to the protein most often brought about by a single letter change in the DNA: sometimes with big effects), and ideally there are a whole lot of them. 

But herein lies the rub: many genes do not have established pathogenic missense variation. For most cancer predisposition genes, we’re certain that if you eliminate the protein (by premature truncation that leads to complete degradation of the RNA), it’s bad. In many of these genes, some small missense changes are bad, too. In others, missense changes are largely tolerated. When variation in genes can cause clinical presentations that are common, like breast or ovarian cancer in a family, it is exceedingly difficult to determine if missense changes in hereditary cancer testing cohorts are harmful or tolerated. Most of these occurrences of cancer are not explained by a genetic predisposition or exposure—they are what’s known as sporadic disease. There are several genes that are linked with cancer predisposition where missense pathogenicity remains to be determined, most famously (at least in my world) PALB2, RAD51C and RAD51D. These genes suffer a particularly high VUS rate because missense changes are neither convicted nor acquitted of being disease-causing. 

Advancing Research to Solve Clinical Dilemmas

Circling back to the question at hand: if there are no known pathogenic missense variants, how can you say that your functional readout (of missense variants) suggests clinical relevance? A recent publication on RAD51C function proposes a novel approach to circumvent this conundrum. The research recently published from Dr. Fergus Couch’s group in Cancer Research outlines a method that will likely be instrumental for any loss of function gene, especially those with MAVE studies in genes where missense pathogenicity is yet to be established. This publication uses a two-pronged approach to overcoming this problem. First, functional validation is undertaken with loss-of-function (those early terminating) variants. Second, the excuse whereby a loss-of-function variant would overshoot the pathogenic mark for a missense pathogenic variant (with possibly milder effect) is mitigated by an additional layer of analysis afforded by a case-control analyses. If you’re in the game and know a little about case-control analyses, you know that you need a lot of cases and controls to get a statistically significant result (if there is one). But remember that we’re talking about rare missense variants, and you’d be lucky to have even one carrier of a single variant in your cases or controls, let alone the 10-plus required for this approach. To get around this problem, the Couch group pools the variants based on their functional readout. Using this method, they were able to show increased odds ratios (this can be interpreted as increased risk) for breast and ovarian cancer for the misbehaving missense variants (functionally deleterious) and that these odds were similar to nasty loss-of-function (truncating) variants. The opposite was true for the functionally neutral variants: they did not have increased odds for breast and ovarian cancer.

To conclude, this RAD51C paper serves as an important milestone for variant interpretation which can be extrapolated to other such genes with elusive missense pathogenicity: 1) it provides an approach for establishing (or refuting) missense variation as a mechanism of pathogenicity for loss-of-function genes; 2) it provides variant-level functional data that can be combined with other lines of evidence for interpretation of individual missense variants; and 3) it provides a set of clinically relevant missense variants that can serve as the benchmark for future validation of novel functional studies.

___

Source: Hu C, Nagaraj AB, Shimelis H, et al. Functional and Clinical Characterization of Variants of Uncertain Significance Identifies a Hotspot for Inactivating Missense Variants in RAD51C. Cancer Res. 2023 Aug 1;83(15):2557-2571. doi: 10.1158/0008-5472.CAN-22-2319