March 2010
Feature Story

William Check, PhD

In Atlanta-Hartsfield Airport, after passing through security, travelers hear a cheery female voice welcoming them to the “transportation mall.” “There are three ways to reach your destination,” the voice announces happily. “All are free of charge.”

While molecular genetic testing for cystic fibrosis is not free, it does provide three ways to reach an important public health goal—to get CF-affected infants into treatment as early as possible. Genetic testing is used to screen couples of reproductive age for carrier status, to screen newborns, and to test the pregnancies of carrier parents.

Within CF newborn screening programs, which are in place in all states in the U.S., in most Canadian prov­inces and European countries, and in Australia and New Zealand, there are also three general approaches. “In the U.S., every newborn screening algorithm for CF starts with immunoreactive trypsinogen [IRT] and ends with a sweat test,” says Philip M. Farrell, MD, PhD, professor of pediatrics and population health sciences at the University of Wisconsin School of Medicine and Public Health. Some programs use only IRT values, typically doing two measurements. Others add a mutation panel when IRT is above a preset cutoff value. “There are im­portant variations in where to set the IRT cutoff and how many mu­tations to include in the panel,” says Dr. Farrell, who has been associated with Wisconsin’s CF newborn screening effort since its in­ception in 1985. Typically, an infant with an elevated IRT level and one or two mutations on the panel is sent for a sweat test, still used by many programs as the gold standard for CF diagnosis. In the third and newest algorithm, a specimen with elevated IRT and one mutation on the screening panel is sent for a gene scan to look for a second mutation, which positively discriminates carriers from likely affected individuals. “California is the only program in the world with this additional genetic step,” Dr. Farrell says.

Another distinctive feature of the California program is its mutation panel, which was designed for the state’s large ethnic population. For instance, more than half of the approximately 560,000 infants born in California each year are Hispanic. “We will need five to 10 years of followup to do an adequate assessment of our mutation panel,” says Martin Kharrazi, PhD, chief of the program research and demonstration section in the Genetic Disease Screening Program of the California Department of Public Health. “But at this point we feel we have done a fantastic job in selecting these mutations.”

Programs that have not yet moved beyond IRT-only screening concern Dr. Farrell. “As I recall, there are 14 states still using IRT/IRT algorithms,” he says. “This is a clearly inferior and, in my opinion, worrisome strategy. The problem is really the likelihood of false-negative results, which cause missed or delayed diagnoses.” False-negative results also occur with IRT/DNA algorithms, he acknowledges. “But a good IRT/DNA algorithm can achieve 96 percent to 98 percent sensitivity. In my judgment,” he says, “molecular genetic testing is the most remarkable technology to come along. The quality of that technology has far surpassed our ability to manage communication and information transfer. Communication is now our No. 1 challenge.” All states that now have IRT/IRT algorithms are reconsidering this approach. “Eventually genetic methods will be ubiquitous,” Dr. Farrell predicts. “It is just a matter of time. With IRT/DNA algorithms, we can approach 100 percent sensitivity. The key is the lowest possible IRT cutoff—95th to 96th percentile is optimal—and a good mutation panel.”

Cystic fibrosis newborn screening programs have existed for more than a quarter century, since shortly after the 1979 discovery that IRT is elevated in infants born with CF. Measurement of IRT formed the basis of Colorado’s program, which debuted in 1982, and Wisconsin’s, which started three years later. Wisconsin became the first state in 1991 to add a molecular genetic component to its program, just two years after the CFTR (CF transmembrane conductance regulator) gene was cloned and mutations in this gene were identified as the cause of CF. A November 2003 workshop sponsored by the Centers for Disease Control and Prevention and the Cystic Fibrosis Foundation found that “The peer-reviewed evidence ... supports the clinical utility of newborn screening for CF” and recommended that states consider incorporating CF into newborn screening programs (Grosse SD, et al. MMWR Recomm Rep. 2004;53[RR-13]:1–36). Guidelines for implementing CF newborn screening programs have been published (Comeau AM, et al. Pediatrics. 2007;119:e495–e518).

Much of the peer-reviewed evidence presented at the CDC/CF Foundation workshop came from a nine-year randomized controlled trial conducted within Wisconsin’s program under a grant from the National Institutes of Health. “We found that the most important benefit by far and the strongest evidence in favor of newborn screening relates to preventing malnutrition,” says Dr. Farrell, who was one of the investigators in the trial. Vitamin E de­ficiency has significant adverse effects, such as hemolytic anemia. They found evidence, too, of cog­nitive dysfunction related to prolonged early vitamin E deficiency and protein-calorie malnutrition.

In a talk at the 2009 meeting of the Association for Molecular Pathology, Iris Schrijver, MD, associate professor of pathology and director of the Molecular Pathology Laboratory at Stanford University Medical Center, described the California CF newborn screening program and presented outcomes data provided by the California Department of Public Health from its first two years. Dr. Schrijver’s laboratory performs the mutation panel for the California program. Addressing the justification for CF newborn screening programs, she cited the 2003 CDC/CF Foundation workshop, which found evidence that early management reduces hospitalizations and improves growth and survival (Farrell PM, et al. J Pediatr. 2005;147[3 suppl]:S30–36; Koscik RL, et al. J Pediatr. 2005;147[3 suppl]:S51-56; Grosse SD, et al. J Pediatr. 2006;149:362–366). “In essence,” Dr. Schrijver told CAP TODAY, “long-term nutritional benefits based on weight, height, and cognitive function alone are enough to justify CF newborn screening. Added to this is the benefit of educating parents in terms of subsequent reproductive risk.”

Infants with the top 1.5 percent of IRT values are reflexed to the 40-mutation panel. Dr. Schrijver says the panel is not a screen, but more like a definitive test; when two disease-causing mutations are found, a diagnosis of CF can be made. Sweat testing is considered the gold standard but is primarily done for confirmation. All 95 infants with two mutations on the panel during the two-year study period received a diagnosis of CF on the sweat test.

“It is important to realize that the CFTR gene has hundreds of genetic changes that are thought to be associated with disease,” says Dr. Schrijver, chair of the CAP/ACMG Biochemical and Molecular Genetics Resource Committee. Some panels look for only the most common mutation, the deltaF508 deletion. “With limited panels, it is much more likely that mutations are being missed,” Dr. Schrijver says. “With larger, well-defined panels, you have a better chance of fully genotyping the child. Depending on the panel, however, it is possible to find mild muta­tions as well, so you could identify chil­dren with milder disease. In general, this is a good thing. It allows a program to make genotype-phenotype correlations and to follow children from an early age to ensure that they maintain good nutritional status.”

In the California program, when only one mutation is found on the panel, the sample is sent for a gene scan using temporal temperature gel electrophoresis, or TTGE, and followed up by focused DNA sequencing. “The great majority of specimens with one mutation [on the panel] will be carriers, so they too receive a definitive diagnosis through the California program,” Dr. Schrijver says. Of 933 infants with one mutation on the panel, 64 percent did not have a second mutation on the gene scan and were classified as carriers. Parents of carriers are offered genetic counseling via telephone. Is it good to find CF carriers at birth, when it won’t have an impact until reproductive age? “One can argue both ways,” Dr. Schrijver says. Importantly, only a subset of all carriers (those with a high IRT value) will be detected in the newborn screening program. Having the information early on may make the family and the individual more comfortable with it. On the other hand, it takes away that person’s right not to know his or her status. “In principle, each person should decide whether to have carrier screening,” Dr. Schrijver says.

Through the first two years of the program, 145 infants with CF were identified. Since 11 cases of CF were later diagnosed among babies who were negative on newborn screening, program sensitivity can be expressed as 93 percent (145/156). Another way of calculating sensitivity is the number of cases found through newborn screening relative to cases expected based on CF prevalence. Taking into account the demographic mix in California, this latter number is 172, yielding a sensitivity of 84 percent.

However, there is a suggestion within the California data that projecting the incidence of CF in newborns based on historical data may no longer be accurate. Among Hispanics, 69 of 72 (95 percent) expected cases were found, compared with only 77 of 93 (83 percent) expected cases among non-Hispanic whites. Dr. Kharrazi raises the possibility that this discrepancy could be due to non-uniform uptake of carrier screening and prenatal testing. “Many non-Hispanic whites are taking advantage of prenatal testing,” Dr. Kharrazi says, “and CF cases are being found. Often parents elect not to take those pregnancies to term.” He believes this is the chief reason for the discrepancy. “We have not studied this possibility statistically,” he cautions. “It deserves scientific attention.”

A few years ago data were reported from the Massachusetts CF newborn screening program that could reflect a similar phenomenon. “From 2003 to 2006 we saw a lower number of CF babies being identified on newborn screening,” says Anne Marie Comeau, PhD, deputy director of the New England Newborn Screening Program and associate professor of pediatrics at the University of Massachusetts Medical School. Specifically, there was a significant reduction in deltaF508 homozygotes, the genotype most typically associated with severe disease (Hale JE, et al. N Engl J Med. 2008;358:973–974). “We are quite sure that our sensitivity has not changed,” Dr. Comeau says, “so it seems to reflect a true change in the number of babies being born with that genotype.” In the absence of further data, Dr. Comeau declines to speculate on the cause(s) of this decrease. Her group reported the finding because it anticipated the kind of discrepancy found in the California program. “We were lucky to have begun CF newborn screening in 1999 and to have projections of what our sensitivity should be, and we were right on,” she says. Calculated sensitivity was 96 percent. “If another state would implement CF newborn screening more recently, their observed number of cases might be inconsistent with projections and they might think they had inadequate sensitivity. I have had other programs thank me, saying they had found exactly that.”

Dr. Schrijver reported in her AMP talk that 93 percent of cases identified with two mutations on the panel began treatment at less than 60 days of age, the goal defined by the Wisconsin data. However, only 45 percent of cases identified with the third step, the gene scan, began treatment by this benchmark. “The state is looking to shorten that time,” Dr. Schrijver says. Dr. Kharrazi concurs, calling it a “major emphasis.” One possibility is to combine the mutation panel and the gene scan in one laboratory to avoid the time it takes to ship specimens.

In the California CF newborn screening program, the 40-mutation panel was selected to have high sensitivity in that state’s ethnically diverse population. In addition, Dr. Kharrazi says, “We focused on mutations that cause severe disease.” The selection process was a several-year effort involving CF samples from many sources. For example, they assembled a statewide registry aided by the CF Foundation and CF Care Centers in the state and consulted death records to gather specimens banked from newborns from 1982 to 2000. If there was no information on mutations, they pulled those samples and did a stepped approach to determine genetic changes. “We started with a 31-mutation panel made by ABI,” Dr. Kharrazi says. If both mutations could not be found, then next they tested samples on an 87-mutation Genzyme panel, followed by sequencing. “Many labs helped us with sequencing,” Dr. Kharrazi says, including Quest, Ambry, and Johns Hopkins. “It was a giant effort.” In addition, with the assistance of CF Care Centers, they recruited black and Hispanic CF patients born before 1982 or outside the state, or both, and analyzed their CFTR genes. “During this process we found over a dozen novel mutations [in Hispanics], one of which turned out to be very frequent,” Dr. Kharrazi says.

To construct a mutation panel that would detect 90 percent of cases with the California algorithm required finding one mutation in each affected person. “So we needed to detect 78 percent of mutations in each of our race-ethnic groups,” Dr. Kharrazi says. To reach that goal, in addition to the deltaF508 deletion, five mutations were needed for non-Hispanic whites and a further eight for African-Americans. For Hispanics, 23 mutations were added. Three mutations were eventually included for other reasons; hence the 40-mutation panel now being used. “No other state went through a comprehensive genotyping effort in CF cases to select their panel,” Dr. Kharrazi says. “It has lived up to our expectations and then some.” All but one of the babies with two mutations on the panel had a sweat test above 60 mmol/L of chloride.

Of the 11 CF cases missed in the program’s first two years, eight had IRT levels below the cutoff. “In seven, IRT levels were so low that no screening program in the world would have found them,” Dr. Kharrazi says. Two were missed because neither mutation was on the panel. One was missed in the third step; it was a duplication that cannot be identified with the TTGE scan. “We don’t see any reason to change anything on our panel,” Dr. Kharrazi concludes.

Another important finding Dr. Kharrazi describes concerns the 340 infants identified in the first two years of the program with one (severe) mutation on the panel and a second mutation on the TTGE scan. Some variants found on the scan are well-established CF-causing mutations while others are associated with CFTR-related disorders that are less severe, such as sinusitis, pancreatitis, and male infertility. “We have found that even though some infants came back with intermediate [30 to 59 mmol/L] or even normal sweat test results, if we followed them for six to 12 months or longer, we started to see changes indicative of CF,” Dr. Kharrazi says. Slow growth or recurrent infections were typical symptoms. “In other programs these babies would be considered carriers because they have one mutation on the panel and a normal or borderline sweat test result,” he says. “Because the California program recommends that CF Care Centers follow up these babies regardless of their sweat test value,” he adds, “they are in the right place to be cared for quickly and correctly if their health starts to deteriorate.”

They are beginning to understand, Dr. Kharrazi says, that there are subtle forms of CF that a sweat test is unable to identify. So far 105 of the 340 infants with a second variant on TTGE are in this category, which the California program is calling CFTR-related metabolic syndrome (CRMS), a term proposed in a recent publication to describe “infants with hypertrypsinogenemia on NBS [newborn screening] who have sweat chloride values <60 mmol/L and up to 2 CFTR mutations, at least 1 of which is not clearly categorized as a ‘CF-causing mutation”’ (Borowitz D, et al. J Pediatr. 2009;155:S106–116). Dr. Kharrazi notes that his conclusions are preliminary, saying, “This is not yet an accepted notion.” The information they are gathering about ambiguous mutation combinations, he believes, will be useful in settling this question.

In the Massachusetts CF newborn screening program, the IRT prompt for DNA testing is the 95th percentile. “Setting a correct prompt is quite important,” Dr. Comeau says. “We are publishing a paper in which we examined retrospective data to see what would happen if we set the prompt at the 99th percentile. Of 200 CF-affected infants found by NBS [newborn screening], close to 14 percent would have been missed; they had IRT values at or below the 99th percentile. This is a huge percentage to miss, especially considering that most of these in the 14 percent have classic CF.” Many European programs set their IRT cutoff at the 98th or 99th percentile. “It is really difficult to compare one program’s IRT prompt with another,” Dr. Comeau cautions. ”We believe there are most likely varying targets of the specific IRT that people are testing.”

An elevated IRT prompts DNA analysis with a 39-mutation panel. Early in the program, when they used a 16-mutation panel, Dr. Comeau and her colleagues asked what the difference would be if they looked only for the deltaF508 mutation, as many programs do. They found that one-third of CF cases diagnosed by the panel—so-called genetic diagnoses—would have been missed (Comeau AM, et al. Pediatrics. 2004;113:1573–1581).

“We report all results that show either one or two mutations as a CF-positive screen,” Dr. Comeau says. In general, their reports are made by telephone to the primary care provider followed by fax with additional information. “We usually insist on talking to the person who will talk to the parents, not just whoever picks up the phone,” she says. “It is very important to avoid missed pieces of information.”

In the phone report the physician is told that one mutation means the baby is most likely a carrier, with a CF risk of one in 34. If the IRT is highly elevated, that information is provided as well. “In that case we tell the primary care provider that the risk of this baby having CF is one in three,” Dr. Comeau says. Regardless of the IRT value, all newborns with a positive screen are recommended for sweat testing. “We also tell the primary care physician when scheduling the sweat test to tell the CF center that this is because of one mutation on the genetic screen,” Dr. Comeau says. “Then the family is scheduled for genetic counseling.”

As in the California program, the Massachusetts program focuses on finding a broad spectrum of CF disease. “We have been very careful to follow infants with an intermediate sweat test for the first six months of life,” Dr. Comeau says. “If their sweat test becomes negative, that’s good. If it turns positive, we have found a CF baby.”

In any baby with a positive or intermediate sweat test, the clinician orders sequencing to find the second mutation, provided the family agrees. “Typically in an infant with a positive sweat test and only one mutation on the panel, the family opts for sequencing,” Dr. Comeau says. “That step informs us about mutations we might be missing.” If one mutation kept coming up, they would know they had missed a hot spot and would consider adding it to the panel. “That has not happened,” Dr. Comeau says.

The program has an additional benefit: identifying carrier parents and other affected children. Parents of newborns who have one mutation and a sweat test consistent with CF receive genetic counseling focused on carrier risk for the parent. During one period early in the program, 82 percent of parents offered CF carrier testing chose to have it. In five couples (of 78), both parents were carriers (Wheeler PG, et al. Genet Med. 2001;3:411–415). “One of these couples [whose newborn was only a carrier] had an older child who was unexpectedly found to have CF,” Dr. Comeau and her colleagues reported.

In her talk, Dr. Schrijver reported a similar collateral benefit. “Three non-screened siblings of three screen-positive CF patients have been diagnosed with CF and have subsequently begun treatment as a direct result of the [California NBS] program,” she said.

While people working in CF newborn screening programs strongly believe in their value, not all medical geneticists share their enthusiasm. “For many years and even now, I have been a little skeptical of the need for newborn screening for CF,” says Wayne W. Grody, MD, PhD, professor of pathology and laboratory medicine, pediatrics, and human genetics at the UCLA School of Medicine. “In my training I came from more of a metabolic disease perspective. We offered newborn screening for disorders where the infant would suffer irreversible damage if you didn’t pick them up right away, such as phenylketonuria. I have never been convinced that applies totally to CF, where symptoms develop later.” Referring to Dr. Schrijver’s presentation, Dr. Grody says, “Iris presented the latest data showing that newborn screening for CF can prevent failure to thrive. That’s useful because you could spend months to years trying to find a cause for it. I was a little more convinced after her talk. But I’m still not totally convinced we need to know from day one for all newborns.”

The carrier issue concerns Dr. Grody also. “We generally don’t do carrier screening for recessive diseases in healthy children,” he points out. “I worry how those are reported out. Do parents understand the difference between carriers and affected children?”

Similar concerns are voiced by W. Edward Highsmith Jr., PhD, co-director of the molecular gene­tics laboratory at the Mayo Clinic and associate professor of laboratory medicine and pathology and of medical genetics at Mayo College of Medicine. “One of the things I struggle with about having both CF population-based carrier screening and newborn screening is that it seems like a lot of time, money, and effort are devoted to this one disease,” Dr. Highsmith says. “As genetic diseases go, CF is quite common. Still, it seems like a lot. There are cogent arguments on each side.” If he were going to select one, he would select carrier screening “because it happens before the fact. If you know you have a high-risk pregnancy, you can accomplish the same goal and get the baby into treatment early.” On the other hand, Dr. Highsmith says, “You can make a really good argument that carrier screening is not universal. Insurance coverage and uptake among people offered this test vary a lot.” For example, uptake is fairly low in the South and Midwest of the U.S. and higher on both coasts.

For Dr. Highsmith, too, severity is an issue. “The benefit from newborn screening for CF is nowhere near as dramatic as with PKU or congenital hyperthyroidism,” he says. He also cites MCAD (medium chain acyl-CoA dehydrogenase) deficiency, which, he says, “kills children.”

Dr. Highsmith agrees that one “extremely good aspect” of CF newborn screening programs is that when an affected baby is identified and brought to a CF center, “you certainly get clinical improvement.” However, he asks, “What happens when you find carriers? Although IRT screening is not meant to enrich for carriers, experience has shown that there is an excess of carriers in IRT-positive cases. And how do you distinguish babies who are authentically positive but in whom you didn’t find a second mutation? It seems to me the answer is that you keep an eye on that child, to see if it later develops CF. But what is the cost of that strategy from both a medical and a parental stress viewpoint?” The California program addresses that “pretty well,” says Dr. Highsmith, who adds that no other state does it.

In the end, he admits his objections are probably of little practical importance. “The fact is that offering carrier screening to pregnant women is the standard of practice, and all 50 states are doing newborn screening,” he says. “I don’t think that’s going to change.”

Dr. Farrell of the University of Wisconsin ­doesn’t accept the charge that the impact of CF newborn screening is minimal. “When malnutrition develops, it can be rapidly progressive,” he says. “It can be very challenging to rescue these infants. You need to get this disease diagnosed by two months.” Moreover, he adds, “The data suggest rather strongly that about five percent of children with CF will die undiagnosed in the absence of newborn screening.”

Some critics have pointed out that outcomes measures for the screened and control groups in the Wisconsin randomized controlled trial converged as they got older. “That’s true,” Dr. Farrell says, “but so far the control group has never caught up with the screened group.” They are now looking at the adolescent growth spurt in the two groups. “We have preliminary evidence that if these children don’t grow normally in the first couple of years, they may not experience a normal adolescent growth spurt. It is almost as if you get programmed for your adolescent growth spurt in the early years of life.”

As for psychological harm to babies identified as having CF, Dr. Farrell says they have studied that side of the program carefully. “We have found no evidence that infants are harmed through early diagnosis. Some parents are angry to be put through the stress and made to worry that their baby will have CF. And fingerpointing can occur between parents—‘It’s not in my family.’ Other parents are happy to have the information and to be able to make more-informed reproductive decisions.” The reactions seen in any screening program will vary, he notes. “You do screening mammography and PSA testing to find people who have cancer early for better prognosis,” he says. “It’s exactly the same for CF.”

It’s true that parents can have trouble discriminating between having CF and being a carrier, Dr. Farrell says. “Some parents feel that if you have one mutation rather than two you might have ‘a little bit of CF,’” he says. “So we have to devote extra effort to educating them. But according to our studies most counseling and education seems to be going well.”

Dr. Comeau, too, believes that genetic counseling determines whether carrier detection is good or bad. “I absolutely think it depends on how it is handled by genetic counselors,” she says.

Cost is an issue the California program has considered. “Our program tends to base decisions on societal cost,” Dr. Kharrazi says. “You have to look at overall costs, such as the cost to track people when a sweat test is required, the cost to provide genetic counseling for all babies brought in for sweat tests, and the cost to families to get to the CF centers, especially if a second sweat test is required, which happens in five percent to 10 percent of cases.” A gene scan is required in less than 0.1 percent of babies screened (about seven percent of those with an elevated IRT), and two-thirds of the time the scan on the blood spot, which is already available, makes it unnecessary for a family to bring its baby to a CF Care Center. “When you add all the costs, we think that extra step is economically well worth it,” Dr. Kharrazi says. “And from a societal point of view it is likely to be cheaper.” California has negotiated a below-retail price for the gene scan, though the cost is still high.

While Dr. Comeau agrees that the medical benefits of CF newborn screening programs for identified infants are important and adequately justify the efforts, it’s her opinion that the activity has provided an additional advantage to the overall health care system. “From my personal and professional education point of view,” she explains, “the way we do CF newborn screening has been a nice introduction for primary care practitioners to become practiced in delivering genetic information. In the beginning when we started speaking with providers, they had varying states of preparedness for delivering genetic information to families.” The situation is much improved, in her experience. “And some of that might be attributable to CF newborn screening programs,” she says. “It has been a good way to start because cystic fibrosis is a disease that most primary care practitioners are familiar with.” They have grown more familiar with the genetic component over time. “That’s important as we move into areas where they will have much more genetic information coming through them.”