"NIPT works by taking advantage of the fact that residual material from cells in the placenta, and thus reflecting the foetus, can be found in the mother's blood. We then test the DNA of this material, known as cell-free DNA, to assess whether the foetus is at risk of a chromosomal condition,” says Erik Sistermans, professor of Human Genetics at Amsterdam UMC.
In addition to the DNA sequence, cell-free DNA also contains a second layer of information known as 'fragmentomics'. This information is encapsulated in multiple factors, such as the concentration, size, distribution, and patterns at the extremes of the DNA fragments. "What is striking is that these 'fragmentation' properties vary slightly between different people. In the past, this variation was always explained by the fact that the composition of cells from which the cell-free DNA originates also varies,” says Jasper Linthorst, researcher at Amsterdam UMC. “In this study, we show that many genes are involved and that a large part of this variation can be explained by the genetic background of the mother,” he adds.
A crucial part of NIPT testing revolves around estimating how much of the captured DNA is derived from the placenta. This task is generally performed by an AI algorithm which uses the aforementioned fragmentation properties to predict the proportion of foetal DNA in a plasma sample. This is critical to NIPT as it informs the healthcare professional about the test’s reliability to detect chromosomal abnormalities in the foetus.
"A variant in a gene called DNASE1L3, a molecular pair of scissors that cuts up pieces of DNA into smaller chunks, seems to have the greatest effect, and it is relatively common in the Netherlands – we find it in about 7% of the population. We have found that in people with this variant, the NIPT fails more often because there seems to be less DNA present in the plasma. Even if there is enough DNA available for a test, in these pregnant women the AI algorithms greatly overestimate the amount of DNA that comes from the placenta which can compromise the reliability of their NIPT results,” says Linthorst.
Cell-free DNA is increasingly used as a biomarker in other fields. These findings may therefore also have consequences for the accuracy of those tests, including those currently being developed to test for cancer. "These tests make use of comparable AI prediction models, we've shown here for the first time that these models are affected by genetic background and that may have big consequences for the reliability of other cell-free DNA tests as well,” concludes Sistermans.