When he teaches medical genetics to first-year medical students, Wayne W. Grody, MD, PhD, likes to mention the 1997 film “Gattaca.” In its near-future scenario, every newborn receives a complete genome scan, and the scan of the protagonist, played by Ethan Hawke, shows him to be at risk for coronary artery disease in later life. He finds himself barred from pursuing his ambition to be an astronaut.
The film touches on almost all the ethical debates we deal with now in molecular testing—employment discrimination, misuse of genetic data, privacy, and the testing of children for disease—says Dr. Grody, who is a professor of pathology and laboratory medicine, pediatrics, and human genetics at the UCLA School of Medicine. These issues have blazed to life with the first—and to date the only—disease for which population-based molecular genetic screening is routinely done in the U.S.: cystic fibrosis. But the clinical decisions that must be made in cystic fibrosis diagnosis and treatment have turned the surrounding maze of ethical issues into something more like a morass.
In a presentation on clinical and molecular aspects of cystic fibrosis at the CAP ’08 meeting in September, Dr. Grody outlined several ethical issues created by genetic research advances, especially the increasing number of CF mutations for which tests are available. The rise in that number has made him an active advocate against what he calls “mutation screening panel inflation.” And it is one reason why cystic fibrosis screening remains as controversial today as it was 15 years ago, he said, when the discussion of whether it should be done began.
Medical genetics is a “bench to bedside” aspect of molecular pathology that results from research in the field being applied to the clinical setting, Dr. Grody said, and its applications are in two broad areas: gene level diagnostics—the main focus of his own research—and gene level therapeutics, or gene replacement in therapy.
“Although there has been a lot of hype and excitement about gene replacement, in general it has been pretty much a disappointment,” he said. “It’s just been a very difficult nut to crack, mainly because, at this point, we don’t have optimal vectors to deliver the gene to the target organ and make sure it gets in the right cells.” So far, in genetic and metabolic disorders and cancers, there have been some very negative outcomes including induced malignancies and even patient deaths from gene therapy.
Cystic fibrosis has been a proposed target for gene therapy for a long time, and may be one of its first successes. “But it’s important to remember, certainly once we are at the point of doing gene therapy, that diagnostics and therapy will go hand in hand. The molecular pathology laboratory will have to demonstrate the precise site of the molecular lesion before we can even conceive of correcting the gene.”
In CF and other genetic testing, “the vast majority of mutations are what we call point mutations—a single nucleotide change or micro-deletion. Whether Southern blot or PCR or sequencing is used,” he said “they are much more stringent-type assays than the immunological-based assays we may be more familiar with, because we can really detect a single nucleotide alteration out of 3.5 billion nucleotides of the human genome.”
Since CF is an inherited genetic disease, it has, more than any other area of molecular pathology, pushed pathology into national-level regulatory and legal issues. “The issues of doing the test go well beyond simply confirming a clinical diagnosis. In fact, for the vast majority of individuals who get the test, there is no diagnosis because they are healthy, they are not symptomatic, and we may be looking at a single copy of a genetic mutation.”
“Even if we find the mutation, it will not cause disease on its own because it’s recessive. But it will have important implications for the person’s risk of CF children if he or she has a baby with another carrier.”
In such a case, medical genetics specialists offer prenatal diagnosis, and that’s where many ethical issues enter the picture. “The natural outcome, if the fetus is found to be affected and to carry both parental mutations, is to terminate the pregnancy.” That decision is left to the individual couple, of course. “But without the option of termination, probably the whole screening program would just stop, because there would be no point exerting all that money and effort if avoiding the affected birth were no longer an option,” he said.
In fact, if there would be no change in management, “we wouldn’t even do the amniocentesis,” he noted, “because why expose the fetus to the needle and the risk of miscarriage?” These difficulties are compounded for predictive genetic testing, which is usually for later-onset dominant genetic disorders like Huntington’s disease or BRCA1 and BRCA2. “There, you’re telling a healthy young adult that they have their parent’s dominant mutation and in 10 or 20 years they have a high risk of getting the same disease the parent had.”
“We’re not doing that in CF because it’s not a dominant disease,” Dr. Grody said. “But there are some predictive aspects with CF as well.”
Since the Human Genome Project was launched in 1990, three percent to five percent of its budget has been devoted to the Ethical, Legal and Social Implications Program, or ELSI. For two and a half years in the 1990s, Dr. Grody served as the CAP’s representative to an ELSI project called the NIH-DOE (Department of Energy) Task Force on Genetic Testing, which subsequently issued a book of recommendations.
ELSI has helped bring about laws to prevent some of the discrimination depicted in “Gattaca,” he explained. For example, “Genetic employment discrimination is now illegal because of the Genetic Information Non-Discrimination Act.” The NIH task force also discouraged testing of children for adult-onset diseases when no intervention or treatment is available. “That’s why we don’t do Huntington’s predictive tests on 10-year-olds, because there is no treatment, and they’re not going to get the disease for 30 to 40 years anyway. Similarly with BRCA1 and BRCA2, we wouldn’t do the test for anyone under 18.”
CF is not a dominant adult-onset disease, but similar testing restrictions apply. “I have a very strong rule—I will do CF mutation testing only on a symptomatic child suspected of having CF. We won’t do it on a totally healthy child. We often get requests from couples with one CF child, and they want to know if the younger sibling got the same mutation.” The UCLA laboratory used to do that but now refuses. “The chances are that child may just be a carrier, and there’s really no need to burden that child with that knowledge until they’re old enough to have their own children.”
It’s actually an accident of history that CF became the first nationwide gene test, and it’s unfortunate because it is not a simple disorder, Dr. Grody said. With sickle cell disease, for example, “every patient with the disease has the same nucleotide substitution. Unfortunately, most of the genetic diseases we deal with now—and CF is really the prototype of these—are diseases where we do know the gene, but we don’t know, going into the test, what mutations we are looking for.”
That, and the clinical variability of CF, have been two of the biggest sources of ethical challenges. “CF is a multiorgan, multisystem disorder; many, but not all, patients have pancreatic insufficiency, the babies can have failure to thrive, while other patients are walking around as adults and don’t even know they have CF. They might just have chronic sinusitis.”
For a long time, Dr. Grody noted, the only laboratory test available to diagnose CF was the sweat test, which shows elevated sodium and chloride concentrations. “In some ways, sweat chloride is still the best test because you don’t have to worry about which one of the mutations your patient has. However, there are some atypical cases of CF where the sweat chloride is not elevated.”
Because the CF carrier frequency is so high (about one in 30 in the Caucasian population of North America, and somewhat lower levels in minority groups), CF was the most sought-after gene after BRCA1 and BRCA2. So when it was discovered in 1989, the year before the genome project was launched, “it was a very big deal in genetics, and people immediately started talking about the potential for population screening.”
Although a recessive gene, CF was a prime candidate for genetic screening because the carriers are completely normal. “There’s no way to detect carriers walking around in everyday life by any other means except mutations. Most individuals who are carriers have no family history of the disorder. The first time they find out they’re a carrier is when they have an affected child, and usually their partner didn’t know they were a carrier either.”
The goal of the population screening program is to offer the test to all couples of reproductive age, identify all couples at risk—who have a one in four risk of having an affected child—and offer them prenatal diagnosis by amniocentesis, at their option. But screening didn’t get going immediately after the defective gene was described by Francis Collins in a 1989 issue of Science.
The patients the Collins team studied all had the same mutation, called ΔF508. “And if this had turned out to be the only mutation of the gene, I think we would have been doing population carrier screening since 1989.” But the gene, with 250,000 nucleotides, is very large and turned out to be very hard to work with, Dr. Grody said. Within months of the Science article, people started publishing tables of many other mutations they were finding by sequencing the DNA of other CF patients. The second most common mutation, G542X, is found in two percent of Caucasians. Then the others go down to one percent and much less than one percent.
“This led to a lot of debate within the genetics community. Should we offer a test that’s only 75 percent sensitive? In most areas of the clinical laboratory, we wouldn’t consider that acceptable. But others said: Is it fair to deny this test to people who could benefit from it, when perhaps we could avoid some CF births?”
With a series of seven pilot screening studies, ELSI initiated a project to find out how this imperfect test would play out in the real world, Dr. Grody said. Two of the studies looked at siblings and relatives of CF patients, and they came up with results that were in some ways counterintuitive. “They found their subjects, mainly the parents of affected children, were less interested in carrier testing and prenatal diagnosis than the naïve population.”
“With this horrible disease, why wouldn’t they want screening? What was finally concluded was that they are living with a child with CF whom they love, and on the next pregnancy they psychologically just can’t face a termination, because to them it’s like saying to the living child, We wish you had never been born, which I can certainly understand.”
The five other studies focused on the “naïve” population who knew nothing about CF at all. “We agreed we would test for the six best-known mutations at that time; this would now be considered well below the standard of care,” he pointed out. The four-year study led by Dr. Grody looked at thousands of multi-ethnic patients in Los Angeles area prenatal clinics, using DNA probes developed by a biotechnology company that is now part of Roche (Grody WW, et al. Am J Hum Genet. 1997;60:935–947).
Because of the limited scope of the test, the researchers felt that educating people was important. “We gave them questionnaires on their genetic knowledge, because the vast majority were going to test negative and it’s important for them to know we had only tested them for six mutations—so they could still have any of the other 1,500 that we didn’t test.”
Despite the prevalence of carrier status, in this multiethnic population the carrier rate was one in 60 for the women, and after their partners were tested, “you’re down to a very small number of couples” where both test positive. In the first group, only one high-risk couple was found. They happened to be of Northern European ancestry, and were both ΔF508. “They did opt for prenatal testing. By bad luck they had a homozygous fetus, and they chose to terminate.”
The study’s small number of at-risk couples prevented it from proving the cost-effectiveness of screening one way or the other. However, Dr. Grody said, “I can tell you with most CF patients living into their 30s now, with modern antibiotics and lung transplants for some of them, the average cost of care for a lifetime is about $2 million. We could argue—since my grant was far smaller than that—that we spent less than that to prevent this one CF birth. But there are so many variables that go into it, including the fact that this couple was interested in preventing the birth in the first place.”
Many couples do not wish to prevent the birth, he pointed out. “Remember, not everyone with the disease is so severely affected. If you could tell who’s going to only have chronic sinusitis, you probably wouldn’t abort that baby. Unfortunately, the genotype-phenotype correlation in this gene is terrible. So there’s no way to tell if you’ll have a severe or a mild case.” In fact, some have objected to the screening program all along for that reason.
It’s also possible that the disease may be treatable someday, Dr. Grody said. True, the gene therapies that have been tried so far have failed. But with the average life expectancy of patients now being in the 30s, “with the pace of advances in biology, probably 30 years from now we will have some gene therapy. That’s a good bet, so how do you justify terminating that fetus if the baby eventually could be treated for the disease? I don’t know the answer to that.”
Continued study of the disease has turned up other considerations. “It was found that one of the more common mutations, R117H, depending on what polymorphisms are coupled with it in the gene, may not cause classical CF with lung disease. It may cause no lung disease at all, and no pancreatic problem—just a single isolated congenital malformation of the vas deferens which causes only infertility. In many cases such patients have fathered children, because they do make sperm but it can’t get out because the tube is blocked. I doubt very many couples would terminate based on that.”
Researchers now believe there are modifier genes, some already tentatively identified, that could impinge on the action of the CF gene. “Even for ΔF508, which is the classical one that causes pancreatic insufficiency, you can’t absolutely say for sure there is going to be a classic presentation.”
Since CF carrier population screening of pregnant couples began in 2001, there is little question that it has reduced the incidence of CF. A 2003 summary of results at Kaiser Permanente showed the HMO screened 50,000 patients, detected 1,776 carriers, discovered 50 at-risk couples, and had 63 pregnancies. The incidence of CF in that population was reduced by 50 percent, Dr. Grody said (Witt D, et al. ASHG 2003. Abstract 255).
However, postmarket surveillance has indicated problems. Since 2001, the American College of Obstetricians and Gynecologists has said that CF screening is the standard of care, but the most recent survey has shown only about 60 percent of obstetricians are offering it routinely to patients. “We generally have a good partnership with ACOG, but I think some obstetricians really don’t like this test because it’s too complicated to explain in a short visit,” Dr. Grody said, adding that he tends to agree. However, there are several lawsuits now pending, filed against obstetricians by couples with no history of CF who found out afterward they were supposed to be given a carrier test. “I don’t think it’s defensible because it’s been the standard of care since 2001,” Dr. Grody said.
Even more significant is the question of which mutations to test for. “It was one thing that the NIH panel didn’t tell us,” he said. “I already knew this was a mess because at the time I was chairing CAP’s Biochemical and Molecular Genetics Resource Committee, which was sending out proficiency testing for CF and other diseases.” In a survey of recipients, the committee asked people how many CF mutations they were screening for. The results: one laboratory was testing only for ΔF508, one laboratory was doing 70 mutations, and the median was 12 to 13. “This is what happens in the absence of national guidelines,” he pointed out.
An ACMG committee that Dr. Grody chaired developed a series of recommendations for implementing population CF carrier screening, which were published in 2001 and available on the ACMG Web site. “We came up with a panel of 25 mutations that would be the minimum used for screening.” The committee selected this number based on the patient registry of the Cystic Fibrosis Foundation. “We set our cutoff for any mutation that accounted for at least 0.1 percent of total mutations. That’s how we came to 25 mutations.”
As an example of the kinds of dilemmas that arise, he pointed to R117H. “We debated a lot about keeping that in there, because it’s the one that may not cause CF, but it is one of the more common mutations. So the recommendation says, if you find R117H, you have to do another second-tier test before you do the full genetic counseling.”
“Another mutation, I148T, turned out to be 100 times more common than we thought, and on further study it turned out not to be CF-causing at all. It’s a completely benign polymorphism. The patients who had this in their DNA actually had another mutation somewhere in the gene.” This variant has since been dropped from the screening panel.
Dr. Grody has spent much of the past year advocating against “mutation screening panel inflation” (Grody WW, et al. Genet Med. 2007;9:739–744). When one commercial lab announced it would offer a screening test with 90 mutations so clinicians would get a better pickup rate, that caused other commercial companies to raise their number of mutations. “Otherwise the obstetricians were not sending them tests. So the obstetricians think a 90-mutation panel is four times more sensitive than a panel of 25, but I think it’s actually about one percent more sensitive, because those added mutations are extremely rare.” In some cases, the commercial panels have retained mutations of dubious clinical significance, he added.
It has created an unseemly competition among laboratories, as well as the natural extreme: actually sequencing the whole gene. Two or three laboratories now offer full sequencing, Dr. Grody said, and it’s something he has ordered on babies with atypical presentations. “We strongly feel this is a great technique, but you’d never do it for carrier screening; it’s way too expensive and picks up too many variants of uncertain significance.”
Not only do accurate tests of certain mutations cause counterproductive outcomes, but also “clinical laboratorians know there’s kind of a contradiction in predictive value as you go looking for more variants,” he said. “In other words, when you’re trying to detect an extremely rare event, when you do find it, there’s actually a greater chance it’s a technical false-positive than a real positive.”
The impact of CF population carrier screening, and the volume of testing being performed, “have transformed testing for this disease from a sort of esoteric backwater of molecular diagnostics to possibly the premier molecular test,” Dr. Grody said. But he contends that the focus on quantity that has produced the cystic fibrosis mutation “arms race” is a negative trend, and he urged reliance on the standards that have been hammered out and monitored over several years by professional consensus groups in the field.
Anne Paxton is a writer in Seattle.