by Kirk, Edwin;
This means that this truly is a screen. It’s often called ‘non-invasive prenatal testing’ (NIPT), but I strongly prefer ‘non-invasive prenatal screening’ for this reason. There is good reason to believe that some obstetricians have misunderstood NIPS reports and acted on that result alone, without doing a confirmatory test.16 If so, there would certainly have been some terminations of pregnancy that were done based on a mistake, with the baby not having had any chromosomal abnormality at all — a disturbing thought.
[16 Ideally the confirmatory test would be an amniocentesis, because chorionic villus sampling also looks at the placenta rather than directly at the baby.]
Don’t get me wrong — NIPS is very good at what it does. It misses far fewer affected pregnancies than the other screening options, and a positive result is much less likely to be wrong. In the time since NIPS first became available, we’ve seen a sharp fall in the number of prenatal tests being done, thanks to this better performance. Recently, however, the trend has been reversing, because of mission creep.
Most NIPS is provided by private companies, and there are limited ways that they can compete with each other. Price is one option, obviously, but making the case that their test is better in some way is another. One way to do that is to test for more conditions than the competition. NIPS started out looking for extra copies of chromosomes 13, 18, and 21 as well as extra or missing copies of the sex chromosomes. As we’ve seen (in chapter 4), testing for sex chromosomes has the potential to produce results that are not straightforward. Still, overall, this approach produced a relatively small number of false-positive results — more for chromosomes 13 and 18 than for 21, but, even so, a NIPS result that says there is a high chance that the fetus has an extra copy of chromosome 13 has a reasonable chance of being correct.
Adding in extra targets has been … problematic. Some companies have added in screening for relatively common, but still rare, abnormalities, such as the deletion on chromosome 22 that causes velocardiofacial syndrome (or Sedláčková syndrome if you prefer). Others have taken to reporting differences affecting other chromosomes if they happen to show up in the data, such as an extra copy of chromosome 10.
One of my jobs in the lab is to issue reports about prenatal diagnostic tests. So far, I have only seen one ‘extra target’ NIPS result turn out to be a true positive that we confirmed at amniocentesis, a large chunk missing from one chromosome. All of the follow-up tests for smaller deletions that I’ve reported on so far have been negative, although presumably at some point that will change.
The reason for all these false positives is mainly a simple quirk of statistics. It’s one that can affect all sorts of laboratory tests, and it can be summed up like this: the rarer the condition, the more likely a positive test result for that condition will be wrong. Unintuitively, the exact same test can give you different results, depending on whom you are testing.
NIPS test reports often include a statement similar to: ‘this test is 99.9 per cent sensitive and 99.9 per cent specific for the detection of Down syndrome’. That sounds very impressive, and it is: 99.9 per cent sensitive means that if 1,000 women who were carrying a baby with Down syndrome had the test, 999 would have a positive result, and only one would be missed. That’s really good, way better than the best of the previous options, which would detect 900 and miss 100, rather than just one. The problem lies in the other number, the specificity: 99.9 per cent specific means that if 1,000 women carrying a baby without Down syndrome had the test, one would have a false positive.
One out of a thousand doesn’t seem so bad, does it? To see why this isn’t as good as it sounds, here are some simple numbers that are made-up but make the point. Let’s say you have two groups of women. Group A have a 1 in 100 chance of having a baby with Down syndrome, because of their age. Group B are younger and only have a 1 in 1,000 chance. A thousand women from both groups have NIPS done, using a test like the one described above.
In the 1,000 women from Group A, with a 1 in 100 chance, there are 10 affected pregnancies. The test is so sensitive that all of them are detected. There is also one false positive (the 1 in 1,000 chance comes up). That means there are 11 positive results, of which 10 are correct. For these women, a positive result has a 10 out of 11 chance (91 per cent) of being correct.17
[17 This is called the positive predictive value (PPV) of the test.]
Now we test 1,000 women from Group B. There is one affected pregnancy, with a (correct) positive result. There is also one false positive. For these women, a positive result has only a 1 in 2 chance (50 per cent) of being correct. A positive result from exactly the same test has a different meaning in the two groups.
As you can see, the rarer a condition is in the population that’s being tested, the lower the chance that a positive result will be correct. If the specificity is a bit lower (because you are looking for a smaller target than a whole chromosome, for instance) things will get worse. Because the tests were originally designed to look for the most common problems, each extra condition you add is going to be rarer than the last, and the test will perform worse as a result. In the last couple of years, we’ve started to see the number of invasive tests in pregnancy rise again. Is more better? Is it a good idea to extend these tests out to look for rarer conditions? If you get a true positive for a rare condition, giving you options you wouldn’t otherwise have had, you might think it is. If you have a miscarriage due to an invasive test that was done because of a result that then turned out to be a false positive, you may think otherwise.
*
At about the time Rachael and Jonny Casella were learning the terrible news about their daughter Mackenzie’s diagnosis, the parents of two other children who were patients of Sydney Children’s Hospital were hearing similar news, and they were hearing it from me. Twice in the space of ten days, I had much the same conversation, gave the same stark message. We know now why your child has been having these symptoms. This condition is not curable, it will steadily get worse, and within a few years it will be fatal.
No matter how carefully or kindly, no matter how well you deliver news like this, you know that it is a hammer blow. For the parents, this will forever be remembered as one of the worst days of their lives.
At some point — often in the very hour that they learn the diagnosis — parents who receive news like this will ask the same question that Rachael and Jonny asked. Why did this happen to our child, to us? Wasn’t there something that could have been done? These days, they have often had screening for chromosomal disorders during the pregnancy, and it’s common that they mistakenly think that this was a test for all genetic conditions. So there’s another question — we had all the tests, why wasn’t this picked up?
For almost everyone who has a child affected by an autosomal recessive condition, there is no family history, no prior warning of the risk. For X-linked conditions, there may be a family history, but often there isn’t. This means that the only way to find out if you’re a carrier before having an affected child would be to have a carrier screening test. For most of my career, there simply haven’t been such tests available for most conditions, for most people. When I gave parents this type of bad news, I could at least look them in the eye and tell them there was no way they could have known in advance that this could happen to their child.
Now, though … things are changing.
Several years ago, when it first became evident that massively parallel sequencing was going to become available for reasonable prices, I started to think about carrier screening in earnest. In that, I was well behind the times.
Usually, being a carrier for a recessive condition is neither here nor there. It does you no harm, but it also does you no good. There are exceptions, however, and the best known of these has to do with malaria and a group of conditions that affect red blood cells. The various forms of the malaria parasite (species from the genus Plasmodium) have a life cycle that shuttle
s back and forth between humans and mosquitoes. In humans, the parasite spends part of its life living inside red blood cells (which are consumed by mosquitoes, which subsequently infect other humans, and so on). There is a group of blood conditions, the thalassaemias, in which the oxygen-carrying protein, haemoglobin, is abnormal, and in turn the red blood cells that contain haemoglobin are abnormal. They are fragile and don’t last very long in the bloodstream after they form; at worst, affected children would be lucky to make it to their third birthdays unless they have regular blood transfusions.
For carriers, though, the news is rather better. Their mildly fragile red cells do a good-enough job of carrying oxygen through the body, and usually this causes no health problems at all. But from the point of view of a Plasmodium,18 those mildly abnormal cells are an uncomfortable place to be. This gives partial protection against the effects of malaria, and especially against its most severe forms. Malaria remains a serious killer: according to the ongoing Global Burden of Disease study, 619,827 people19 died of malaria in 2017, out of 56 million deaths in total. Reduce your risk of death from malaria, and you increase your chances of living long enough to have children, and pass on your genes to the next generation. That’s called a selective pressure — if people with a particular type of genetic variation are more likely to successfully reproduce, that variation will become increasingly common in the population.
[18 To the extent that a Plasmodium has a point of view.]
[19 Yes, this is a suspiciously precise number. In case you’re interested, the top three killers were cardiovascular diseases with 17.8 million, cancers with 9.6 million, and respiratory diseases with 3.9 million. Malaria killed more people than many other causes, including murder (405,346), drowning (295,210), terrorism (26,445), and disasters (9,603). Malaria killed 23 people, mostly small children, for every one who died due to terrorism in that year. This is why the front page of your newspaper is forever running banner headlines about the ongoing malaria disaster. It isn’t? … nor mine, for some reason.]
The mosquitoes that can carry malaria like it hot, or at least warm: before people started trying to do something about it, malaria was found as far south as 32° of latitude, and as far north as 64° (!) … but it was always most concentrated in a belt around the equator, and has largely remained so. Where malaria is or was, you can expect to find thalassaemia and related conditions, and carrier frequencies can be very high indeed.20 This includes most of the countries around the Mediterranean — which brings us to Italy, to Greece, and then to Cyprus.
[20 You might think that a condition that is fatal when you have two copies of a faulty gene would be prevented by that from becoming common in the population. Consider though, that if 1 in 10 people is a carrier for a condition, in only 1 in 100 couples are both partners carriers (1/10 times 1/10), and therefore only 1 in 400 children are affected (1 in 100 times the 1 in 4 chance for each baby born to a carrier couple of being affected). If there is a benefit to carriers, 1 in 10 people benefits, but only 1 in 400 suffers negative consequences.]
In 1955, the Italians Ida Bianco and Enzo Silvestroni suggested the possibility of preventive counselling: if you identify couples who are carriers, you could counsel them against having children. George Stamatoyannopoulos, fresh out of medical school, took on the challenge. In 1966, he went to Orchomenos, a village in Greece with a population of 5,000, and started screening for sickle cell disease (a variant of thalassaemia that is common in Africa but also crops up in some other places, including parts of Greece). There were a lot of carriers: nearly a quarter of the population. About 1 in 100 babies born in the village was affected. Stamatoyannopoulos advised unmarried carriers to steer clear of each other and choose non-carriers to marry, but, when he returned to the village, he found that his advice had been ignored. In that sense, the effort had not been a success — but the first attempt at carrier screening for reproductive purposes had been made, and a great deal had been learned. In 1971, when the World Health Organization called on Stamatoyannopoulos to visit Cyprus and advise about the problem of thalassaemia on that island, he was well prepared.
The situation in Cyprus was not unlike that of Orchomenos, although it was at the scale of an entire country rather than a single village. The carrier frequency for thalassaemia and the frequency of affected babies were just a little lower than for sickle cell disease in Orchomenos, but the impact on peoples’ lives and on the health system was considerable. Nearly half of the output of the blood bank in the capital, Nicosia, was being used for keeping people with thalassaemia alive, and, at the end of that decade, 6 per cent of the entire budget of the Ministry of Health was being spent on a single drug, desferrioxamine, which is needed to treat people for the dangerous overload of iron that comes with frequent blood transfusions.
It took much of the 1970s to figure out how best to do carrier screening and get it working effectively. An early effort to persuade carrier couples not to marry each other was just as unsuccessful in Cyprus as it had been in Greece. By 1977, it had become possible to do prenatal diagnosis, and attention was focused on screening people of reproductive age in order to give them information on which they could consider acting. This won strong support from the Church of Cyprus, because of the realisation that there were fewer terminations of pregnancy as a result of prenatal testing. Up to that point, many people who knew they had a 1 in 4 chance of having an affected baby were choosing termination of pregnancy rather than take that chance. For those couples, prenatal diagnosis meant that three out of four pregnancies could be shown to be unaffected, and thus would continue.
In 1979, the expected number of affected babies born in Cyprus (based on historical figures) was 77, and the actual number was only 18. Given a choice, couples were choosing to take steps not to have affected children.
The story of carrier screening in the decades since then has been one of mixed successes and mostly slow progress, until a recent boom. Carrier screening for the thalassaemias and related conditions, targeted at people from populations where these are common, is cheap and generally quite effective in many countries around the world. There are targeted screens for some other conditions, too.
In this regard, Israel is the undisputed world leader. There are a number of recessive conditions that are more common in people of Jewish ancestry, especially the Ashkenazi Jews (those who trace their ancestry to Central Europe). Most infamous of these is Tay-Sachs disease, another lysosomal storage disease that affects the brain. In the classical form of Tay-Sachs, most affected children do not reach their fourth birthday. There are community-led screening programs for Jewish people in various parts of the world, but in Israel the Ministry of Health offers free screening to everyone who is planning a pregnancy or is early in a pregnancy. It’s all targeted, to a very fine level of detail. There’s a list of genes recommended for most of the population, then different sets depending on ancestry — for Ashkenazi Jews, for Jews of North African origin (except Morocco), for Jews of Moroccan origin, and so on; and also for those from other populations: there’s a specific list just for Bedouins in the Negev region, for instance.
In much of the rest of the world, in order to access carrier screening, you need two things that not everyone has: information and money. You need to know that the tests exist, and you need to be able to afford to pay for them (or have health insurance that will pay for them).
Currently available tests range from covering as few as three conditions to as many as hundreds. ‘Three conditions’ in this case usually means SMA, which we’ve encountered already, cystic fibrosis (CF), and fragile X syndrome. CF is a complex condition in which the body’s secretions are thicker than they should be. That might not sound so bad, but it really is — untreated, this causes children to have progressive, serious lung infections and severe nutritional deficiencies due to failure of the pancreas; in the past, most affected children did not make it to adulthood. Modern treatment makes a lot of difference
to the condition, with greatly improved life expectancy, but is burdensome on child and family. Fragile X syndrome is a common cause of intellectual disability. If someone is offering a screen for hundreds of conditions, they will almost always include these three.
This was the news that shocked Rachael and Jonny Casella. Had they known about SMA carrier testing, they could easily have found out that they were carriers before having a pregnancy — opening up choices, including PGT and prenatal diagnosis.
They became powerful advocates for screening, starting by writing to every member of the Australian Federal Parliament, as well as to New South Wales state politicians. They met with the state health minister, with the federal deputy health minister, and finally with the federal minister for health, Greg Hunt.
By a remarkable and fortunate coincidence, at much the same time that the Casellas began their advocacy, a group of researchers were talking to the government about the same topic. Nigel Laing, an internationally renowned expert on the genetics of muscle diseases who had advocated for carrier screening for many years, had convened a meeting of interested Australian doctors and scientists in late 2016. I attended the meeting and spoke about some research I had led, studying screening in couples who were related to one another. Because we share genes with our relatives, such couples have a much higher chance of having children affected by recessive conditions than those who are unrelated. We showed that screening using a very large panel of genes, about a quarter of the exome, worked well in such couples and was acceptable to them.
Thanks to Nigel’s leadership, the nucleus of an Australian team of carrier screening researchers was already in place in 2017, when I had to give terrible news to parents twice in ten days. That experience prompted me to reach out to the group, and together we decided to write to the federal health department to suggest that this was an area which needed attention.