Using NHS testing data to understand genetic variation in cancer-prone families

With the exception of identical twins, all human beings have a unique DNA sequence. If we take two unrelated humans and sequence their 3 billion DNA base pairs, we can see that their DNA sequences are around 99.8% identical. At a basic biological level, our individuality as human beings comes down to variations in just 0.2% of our DNA.

In the main, DNA sequence variation is a good thing. It underlies evolution, is responsible for the glorious diversity we see amongst plants and animals in the natural world, and makes us all unique as human beings. But some sequence variants are more problematic: if a DNA sequence variant prevents a gene from working properly, then a genetic disease can result. The harmful genetic variant can be passed from parents to children, via eggs and sperm.

One class of genetic conditions are known as the ‘cancer predisposition syndromes’. People with a harmful variant in particular genes have an increased risk of developing certain types of cancer – and tumours in these susceptible people often develop at a younger age than in most people with that type of cancer. A well-known example is people with harmful (pathogenic) variants in the BRCA1 or BRCA2 genes, who have elevated risks of developing breast and ovarian cancer.

It's important to identify people with a genetic predisposition to cancer, so that they can be offered genetic counselling and testing, and suitable screening to detect cancers at an early stage. For some conditions, medical or surgical interventions can reduce these people’s cancer risk. People and families making these very personal decisions about their medical care need to be given the very best information we have – and this involves genetics professionals sharing knowledge and intelligence on different genetic variants.

A gene is essentially a recipe to make a specific protein. If we think about introducing a typo into a recipe, we can imagine the impact that will have. Perhaps our recipe instructs us to ‘add one egg and the ham’. If instead it says ‘add one egg and the jam’, we might end up with something a bit different, but which still works perfectly well to feed a hungry family – i.e. a normal (benign) variation. The problem comes if we see ‘add one egg and the hat’. This recipe is clearly not going to produce anything edible, and, in DNA terms, is what geneticists would call a ‘pathogenic’ (disease-causing) variant.

One of the big challenges for geneticists is how to tell the difference between a damaging gene variant causing a cancer predisposition syndrome, and a harmless change in the gene. Genetic variants that we do not yet understand are known as ‘variants of uncertain significance’ (VUS). These VUSs are very tricky for geneticists and families alike. It’s impossible to give accurate genetic counselling to a family when we don’t understand whether the DNA variant is harmful. There have been rare cases in the past where people have undergone radical risk-reducing surgery because of carrying a genetic variant that, as knowledge moved on, turned out to be benign.

Each month, geneticists from the NHS labs and clinics hold a virtual meeting of the Cancer Variant Interpretation Group (CanVIG-UK), to discuss VUSs. The process of ‘variant interpretation’ involves bringing together several highly technical strands of evidence to decide by expert community consensus whether a variant is pathogenic or benign. For this to be successful, it’s crucial that geneticists share knowledge and data with one another. However, sharing data is challenging both technically (because different labs record their data in different formats) and legally (because of data protection and confidentiality laws).

At the National Disease Registration Service (NDRS), we realised that we could help here. As England’s national cancer registry, we are legally allowed to collect and process medically sensitive personal data at population level, we have the necessary IT security and encryption in place, and we have the expertise to deal with data in many different formats. We were therefore well-placed to collect lab data on genetic tests, with the added bonus that we could also link this data to a person’s cancer registration record.

Over the course of a few years, we in NDRS collected data from the sixteen NHS genetic testing labs and restructured the data into a common format. In some cases, the labs’ data went as far back as 1997, when the genes were newly discovered! We counted the number of times each different genetic variant had been seen in independent genetic tests, and calculated a ‘variant frequency’ by dividing this count by the total number of independent genetic tests that had been performed. We then returned this anonymised information to the NHS Cancer Variant Interpretation Group.

By bringing the collated NHS labs’ variant data to the table, it was possible for the first time to compare the frequency of each variant in our independently tested NHS cancer families with that in online population control genomic databases. If the variant is more common in our high-risk cancer population, then that is significant evidence for it being pathogenic. This insight can be combined with other threads of evidence to decide whether each variant is harmful or benign.

In the early years of setting up the data project, we were contacted by a genetic counsellor from Cambridge, who was coming up for retirement. She had three families on her books, all of whom had the same variant of uncertain significance in the BRCA2 gene – and she was keen to find a definitive answer for these families before she retired. We discussed the variant at the CanVIG-UK meeting – within the NDRS data, we were able to identify 19 individuals with this variant, all but one of whom had a registered cancer (mainly breast or ovarian). The variant did not appear at all in the online genomics population database. By combining this intelligence with key scientific and clinical details from some of the labs, we were able to reclassify this variant as pathogenic, and our genetic counsellor colleague in Cambridge was able to offer the most appropriate management to the three families under her care. Other clinical genetics services around the country re-contacted each of their families with this variant to tell them of the new knowledge and to ensure they were offered screening. We were subsequently contacted by one of the labs, who told us that nineteen family relatives had come forward for genetic testing for this one variant.

The national cancer genetics data project has taken a huge amount of commitment, creativity and sheer hard work by many different experts, both in the NHS labs and in NDRS. The work has involved scientists, analysts, bioinformaticians and IT developers: it is so rewarding to see the fruit of several scientific disciplines collaborating to solve a complex problem. We’re incredibly proud to be the only country in the world to have built this kind of data repository on a national scale – and we’re incredibly privileged to have the opportunity to undertake such interesting and meaningful work. As a team, we know that our work has led to improved genetic counselling and increased certainty for families, and has contributed to the global genomics knowledge base on variation in cancer predisposition genes.