Should VUS ever be used in clinical decisions?

The ThinkerIt is universally accepted that too many ‘variants of uncertain significance’ (VUS) are being managed inappropriately as pathogenic. This is leading to widespread, unacceptable, avoidable harms and intolerable pressures on health systems.

The ACMG guidelines state that a VUS should not be used in clinical decision-making. If this guideline was fully adopted clinical harms would certainly be reduced. But there has been reluctance from all sides to make it quite so simple. Why?


Most VUS are benign

I have said it many times before, but I repeat it again here. Most VUS are benign. We all have many rare variants. The vast majority are not causing any appreciable problems.

So, overall, the recommendation to not use VUS in clinical decision-making is the best and safest approach. In the absence of a more sophisticated process it is definitely better than the ad-hoc VUS over-management system we have induced.

But we should be aiming to deliver a more refined approach.


Some VUS will become pathogenic

A small minority of VUS are pathogenic. Most are currently indistinguishable from the ocean of VUS that are benign. It will take new knowledge for us to know they are pathogenic. We have no option other than to treat these variants in the same way as VUS overall and we should not use them in clinical management.

We needn’t feel any anxiety or defensiveness about this. It is not possible to know the unknowable. We should ensure robust systems are in place to implement variant classification updates and reissue of reports, when they occur. But the guideline to not use VUS in clinical decision-making is appropriate for these variants. When they become pathogenic in the future, they will be managed accordingly.


We are already suspicious about some VUS

Most over-management of VUS occurs because people think VUS are variants we are suspicious are pathogenic.

The variants the current system doesn’t handle well are the tiny group of VUS that we are already suspicious are pathogenic. They don’t quite meet the formal evidence requirements to be called likely pathogenic, but we would willingly stake a decent wager that they will.

Much of the over-management of VUS occurs because people think all VUS are variants we are suspicious are pathogenic.

But these suspicious VUS form a tiny minority of the variants we include within the VUS category. Typically they are not explicitly noted as ‘suspicious’ as we do not have an approved, consistent process for doing this. If one adheres to the guidelines as they stand these variants should also not be considered in clinical decisions. But is this always appropriate? Are there situations when it might be reasonable to act on a VUS we believe could be pathogenic?


A case study

We have recently been involved in case where we did think it was reasonable to use a VUS in clinical decision-making. I’ve changed the names and a few details to preserve confidentiality, but the essence is as follows. Lucy developed ovarian cancer at 45. She responded well to surgery and platinum chemotherapy. She relapsed four years later but again responded well to platinum chemotherapy. This is a typical pattern for BRCA-related ovarian cancer. Lucy also has a family history of breast and ovarian cancer. Unfortunately, she recently relapsed again and during review a BRCA test was ordered; if she has a pathogenic BRCA variant she can be treated with a PARP inhibitor. The BRCA test reported a VUS and we were asked to review.

It turns out this is no ordinary VUS. There is a high suspicion that it is pathogenic. Not quite sufficient to call it likely pathogenic according to some labs. These labs call it a VUS – that is the only other option available.


Use in cancer management but not prevention

We decided on the following management. For Lucy, we used the variant in clinical decision-making and managed it as a pathogenic variant. We believe formal evidence to support this will become available, but Lucy needs a decision now so that her treatment can be decided. Her personal history, family history and genetic result are all consistent with her ovarian cancer being due to BRCA, and she is likely to respond well to a PARP inhibitor.

For Sarah, Lucy’s 26 year old daughter, we made a different decision. We decided to not use the variant in clinical decision-making at this point. We have not offered predictive testing. Sarah is entirely healthy. Her risk of cancer today is very low even if she has a pathogenic BRCA mutation, because she is only 26. Over the next few years we should be able to get evidence to prove the variant is pathogenic (assuming people share their data). We will then be on much firmer ground in estimating Sarah’s risk of cancer and in helping her to make complex decisions about her options, should she have inherited the variant.

So for cancer management, using a suspicious VUS in clinical decision-making may be considered appropriate. For cancer prevention in a healthy individual the bar is, and should be higher. It is much less likely that using a VUS in clinical decision-making will be appropriate.


Variants Under Surveillance?

We need flexibility to use some VUS in clinical management and a system that explicitly handles this.

There are many other genetic conditions where we need to retain flexibility to use some VUS in clinical management. Until we adapt the system to explicitly handle this issue it will remain very difficult to prevent the mismanagement of variants.

A simple fix might be to change VUS to be the Variants Under Suspicion or Variants Under Surveillance. We have ample evidence that people assume this is what a VUS is. If we put the variants that are appropriate to be considered in this way into that category we will have less of a mismatch between what we mean and what others think we mean.

To make this work we will have to focus attention on deciding the parameters that would define a variant as being appropriate to be ‘under surveillance’. There will be general principles that can be applied to any gene or variant. But there will also be gene-specific, variant-specific and disease-specific considerations that will need to be integrated. The variant prioritisation and knowledge integration strategies I have discussed before would lend themselves well to making these decisions.


Focusing on the right issues

Whatever we choose to do, it is clear we need to focus on new ways of dealing with these tricky variants. They are the most challenging to handle within our current systems, both for those trying to classify variants and those trying to clinically manage variants. Our inability to develop a process for them is at the root of many of our variant interpretation problems.

And importantly, if we could be more explicit and more attentive to this tricky group of variants it would greatly accelerate harnessing the evidence we need to formally classify them.




Image by Drflet, via Wikimedia Commons, CC-BY-SA