Prenatal screens: Beware misleading results
© 2016 Roy Benaroch, MD
A few years ago, I met a lovely couple at a prenatal visit. Although we usually do these visits in groups, this time we had arranged to get together privately, alone, when the office was closed, when there wouldn’t be other families and children around. I wanted to meet the family, but I didn’t want to be cruel. This wasn’t a typical visit, and I didn’t think they were going to be comfortable sitting in a room full of pregnant women, or a room with children running around. This couple expected their baby to die.
Prenatal screening, the parents told me, had shown that their baby had Trisomy 13. This is a genetic condition involving an extra copy of one chromosome. Most babies who have this do not survive long enough to leave the hospital; the few that do are severely ill, and have multiple health complications. The family had been offered an abortion, but had decided to proceed with the pregnancy. They were well-educated, and certainly realistic – though they had hope, they knew that there was very little chance that their baby would do well. Still, that was the chance they wanted to take. They had heard that I had some experience taking care of children with this condition, and wanted me on board if by some miracle their baby was able to survive, even briefly.
A few months later, Sally was born. And she was normal. She didn’t have Trisomy 13, or really anything else wrong with her, and she was able to leave the hospital in a few days. She’s about 18 months old now, and she’s a great little kid, with red hair, a picky appetite, and a little bit of a temper. Imagine what the parents went through, expecting her to die; and imagine how many parents, told of the grim news, would have elected to abort what would have been a healthy baby.
So what happened? Was the test just wrong? No – the test was done correctly, and the result of the test was, technically, correct. But it was misunderstood, by both the doctors and the parents.
The fundamental misunderstanding was that screening tests are not diagnostic tests. This comes up again and again in medicine, and every week I have to explain to parents why their screening tests for allergies do not necessarily mean that their child is allergic, or that the vision screening test doesn’t mean their child needs glasses. An abnormal mammogram does not mean a woman has breast cancer, and an abnormal prostate antigen blood test doesn’t mean that grandpa has prostate cancer. These are only screening tests, meant to estimate risk—they do not, ever, diagnose disease.
Let me go back to the test done on mom during her pregnancy. It’s called a “cell free DNA” (cfDNA) test, done on blood drawn from mom. During pregnancy, a few cells from the placenta burst, releasing their DNA into mom’s circulation. It’s a tiny amount of DNA, and it’s quickly broken down and cleared, but with some very clever genetic tools that miniscule fraction of fetal DNA can be isolated in mom’s blood, and measured. And it’s that DNA that’s tested to provide the results of the test. The cfDNA can detect conditions where the as-yet-unborn baby has an incorrect number of chromosomes, including Down Syndrome (that’s an extra chromosome 21), Trisomy 13 (an extra 13), other trisomies, or conditions involving the sex chromosomes. These conditions have tremendous implications for a baby’s health, so advance screening is desirable. cfDNA screening can be done easily and painlessly, on a small sample of mom’s blood, with no risk to the baby. Pretty slick.
But that’s the key word, screening. Though it’s testing genetic material from the baby, cfDNA is still a screening test. To really understand why, we’re going to have to do some math (feel free to skip this and the next paragraph if you want to just take my word for it.) The accuracy of a test is expressed by two terms: sensitivity (the number of positive tests divided by the number of people who truly have disease), and the specificity (the proportion of negative tests among a group of people without disease). The sensitivity and specificity of cfDNA testing is excellent – over 99%, which is the “accuracy” figure often quoted in marketing literature about these tests.
But in real life, what we really want to know isn’t the specificity or sensitivity of a screening test. What we need to know is its positive predictive value – that is, in a woman with a positive cfDNA test, what is the chance that her baby will truly have one of these health conditions? To figure that out, you need to apply Bayes Theorem, which requires not only the figures for the accuracy of the test, but the “pre-test probability”. In a screening population with an overall low risk of disease, even a very accurate test is going to have plenty of false positives. If you don’t believe me, follow that link and do some math examples.
Bottom line: for cfDNA testing in a 38 year old woman, the positive predictive value of a “high risk” screen for Trisomy 13 is 37%. (The pre-test probability depends on mom’s age – older moms have an increased risk of babies with chromosomal disorders. You can calculate both the positive and negative predictive values for cfDNA based on the age of mom here.) In other words, even with an abnormal cfDNA screen, the chance of this mom having a baby with Trisomy 13 was 37% — with a 63% chance of the baby being fine.
What should have happened after the abnormal screening test is what should always happen after an abnormal screening test – or, better yet, before the test is even done. Patients need to understand that a positive screen means the condition is “at risk.” Better yet, if there’s solid data, the actual risk percentage should be shared (37%, in this case.) Then the family could decide what to do next. After an abnormal cfDNA test, what should usually be recommended is a diagnostic test, to get a genuine sample of fetal DNA (typically though amniocentesis or chorionic villus sampling.) These diagnostic tests are very, very accurate – and in Sally’s case, if these were done, they would have shown that she did not have Trisomy 13.
But the amnio wasn’t done, in part because the doctors told mom, incorrectly, that the cfDNA was like an amnio, and that the result was conclusive. The doctors fully expected this baby to have Trisomy 13 and even made plans to not do a c-section if the baby ended up in distress. After all, she was going to die, anyway.
The results of screening tests should never be described as “positive” or “negative.” The best way to express the result is “high risk” or “low risk”. The language, here, is really important – and using the right language helps both doctors and patients understand what test results mean. One of the reasons I’m skeptical of patient-ordered tests is that patients may not understand what the results mean (though, admittedly, in this case her doctors didn’t seem to understand the results, either.)
Prenatal screening is a good idea, and cfDNA testing is a good tool. If you’re having these kinds of tests done, make sure you understand what the results mean, and make sure that you have a confirmatory, diagnostic test before you make any decisions that can’t be changed later.
Sally’s story is completely true (other than her name). Thanks to her parents for giving me permission to share it.