Sherrie Rice
October 2025—Bringing equity to cystic fibrosis carrier and newborn screening was the aim of expert groups that have released their recommendations for both.
Carrier screening for 23 CFTR variants, which had been the recommended practice since 2004, was working well, “but only if a person was of white European or Ashkenazi Jewish ancestry,” said Karen Raraigh, MGC, CGC, assistant professor of genetic medicine at Johns Hopkins University. “It wasn’t working all that well because it was not an equitable test.” For people of Asian American and African American ancestry, she said, the detection rate was lower.
Raraigh and others set out in 2020 to bring equity to CF carrier screening and, in a CAP TODAY webinar (https://bit.ly/CAP_061125-WOD) made possible by a special educational grant from Devyser, she explained how (Deignan JL, et al. Genet Med. 2023;25[8]:100867).
“Because we now have defined the framework for how to do this,” she said, “the list can be refined as CFTR2 and gnomAD and other data sets increase in size and complexity.”
Raraigh set forth, too, the new recommendations for screening newborns, also determined to be inequitable in the U.S., where the likelihood of CF detection depended on the state of birth and CF-affected infants missed by newborn screening were disproportionately Black, Asian, or Hispanic (McGarry ME, et al. Int J Neonatal Screen. 2025;11[2]:24).
CFTR2 is an international initiative for the clinical and functional translation of CFTR.
Its goals are to analyze CFTR variants reported in cystic fibrosis patients worldwide, to determine and assign disease liability to the variants, and to provide expert-reviewed variant interpretations to researchers, physicians and other providers, and patients and their families.
The process to do so begins with populating the CFTR2 data set with genotype and clinical information from CF patient registries, then listing the variants reported in that collection of patients in order of frequency, “to prioritize interpretation from most frequent to least frequent,” Raraigh said. Next, each variant undergoes three evaluations: clinical (sweat chloride), functional (CFTR function, splicing), and population or penetrance (frequency, nonpenetrance). Interpreted variants then are placed into one of three categories: CF-causing, varying clinical consequences, and non-CF-causing.
“As of the latest update in September 2024, we’ve interpreted over 1,000 variants as CF-causing, 55 as variants that lead to varying clinical consequences, and 27 as non-CF-causing variants,” Raraigh said.
CF is a recessive condition, and thus two variants are assumed to be necessary for disease. A CF-causing variant is expected to result in CF when found in trans with another CF-causing variant, and disease severity can vary. A non-CF-causing variant is not expected to result in CF when found in trans with a CF-causing variant. And a variant of varying clinical consequences results in CF in some but not in others when found in trans with a CF-causing variant. “The ‘not others’ in this category may be asymptomatic or have other milder diagnoses or designations like CFTR-related disorder or CRMS/CFSPID,” Raraigh said, referring to CFTR-related metabolic syndrome and a CF screen-positive but inconclusive diagnosis in newborns.
The first iteration of CFTR2 was released online in 2012 with 40,000 individuals with CF, and the second was released in 2015 with 89,000 individuals. The most recent release came at the end of 2024 with 123,000 individuals, representing 55 countries, regions, and territories. “We have thus far been able to interpret 1,167 CFTR variants,” she said, all available for search at https://cftr2.org.
The same concepts for variant evaluation have been used since the start of CFTR2.
The clinical evaluation consists of looking for evidence that the variant occurs in someone in the data set (or in several people) who appear to have CF. “Ideally, we look for people whose genotype includes our variant under evaluation—for example, Q237E. And the other allele in the genotype should be severe CF-causing or a null variant. That way, essentially no function is coming from that null allele,” Raraigh explained, “so we can just assess the effects of the variant we’re interested in looking at.”
Within this group, they evaluate the mean sweat chloride and the prevalence of pancreatic insufficiency. “We’re looking for clinical evidence that people with this specific variant or genotype have CF,” she said.
The functional evaluation is typically time-consuming and labor-intensive. “It sounds simple—you have a genetic variant, figure out what it does,” she said. But they have to consider all the ways the CFTR protein could potentially be disrupted by a genetic variant. “Some of these are potential effects on the RNA, while some are effects on the protein itself,” Raraigh said, noting there are several types of functional tests that are done, based on the variant type and what the impact may be.
Lastly, they evaluate each CFTR variant for evidence of nonpenetrance—people with the variant of interest who do not have CF. They also determine the frequency of the variant within CFTR2 and in general population data sets. “Variants that are more common in the CF population tend to be pathogenic, and variants that are more common in the general population tend to lean more toward benign,” she explained.
Since the launch of CFTR2 in 2012, they’ve reclassified only 12 variants. Six variants of varying clinical consequences were reclassified as non-CF-causing, two as CF-causing, and two as of unknown significance (the latter is a category that’s no longer used for now). The two that were classified as of unknown significance were further reclassified: one to CF-causing and one to non-CF-causing. “Notably, we also had a handful of variants that started out as variants of unknown significance and were reclassified. That should be a major goal,” she said.
CFTR2 is having an impact on policy decisions, Raraigh said, in that sequencing is used more often now than DNA variant panels.
The risk with sequencing is that the ability to identify the variants can outpace the ability to interpret them, she noted. “Luckily, CFTR is one of the most sequenced genes and we know a fair bit about the variation within it.”
Thus, for those who make decisions about screening protocols, CFTR2 can provide worldwide CFTR variant representation, in addition to information about whether a CFTR variant causes CF and its frequency in the CF population. “So CFTR2 can provide information that informs people designing screening protocols about which variants should be reported as part of screening,” Raraigh said.
The American College of Medical Genetics and Genomics in 2004 recommended a panel of 23 variants as the minimum to be included in a carrier screen, and it wasn’t until 2023 that the carrier panel was updated to 100 variants. This came shortly after CFTR2 had released its expanded list of 485 variant interpretations.
Raraigh was a member of the ACMG committee convened in 2020 to update the list to make it more equitable and sensitive for all groups. The task was to identify all variants that might be included and then how many of them will achieve an acceptable level of carrier detection in all groups.
They considered multiple questions, among them what to do if there are conflicting data or interpretations within a data set, how the definition of disease is formed within the data set, where to draw the line in the severity spectrum, what to do about rare variants that may be in the disease data set but absent from the population data set, and which data set to use. “After putting together all the nuances and pros and cons of each data set, we landed on CFTR2 to source the list of pathogenic variants to consider for carrier screening,” and on gnomAD to determine population frequency, Raraigh said.
They used the ACMG list of 23 and the list from CFTR2 at that time to create an initial list of 416 CF-causing variants to consider for carrier screening. A review of the list brought it down to 197. Complex alleles, structural variants, and absence from gnomAD would mean automatic exclusion, for example. With the revised list of 197, they did a frequency evaluation of each variant in gnomAD and a coverage evaluation. For the latter, “we looked at the major U.S. ancestral groups and figured out how many variants would have to be included from each population to achieve 95 percent sensitivity in that population,” she said. They merged the 95 percent variant lists from each ancestry to achieve a non-overlapping set of variants. “We landed on a nice round number of 100 variants.”
The Cystic Fibrosis Foundation released its first newborn screening guideline in April. It wasn’t until 2010 that all states in the U.S. screened newborns for CF. Two main protocols are used: IRT-DNA and IRT-DNA-SEQ.
All states begin by measuring immunoreactive trypsinogen, or IRT, from dried blood spots. It’s a pancreatic enzyme precursor that is typically elevated in infants with CF, but it’s also elevated in infants without CF (owing to traumatic birth or prematurity, for example). Specimens found to have elevated IRT are sent to tier two testing, which uses a DNA component. (In some states, the IRT is performed twice.)
“Most states are still using just the IRT-DNA protocol and perform only DNA variant panel testing,” Raraigh said, noting the panels vary in size. In states using the IRT-DNA-SEQ protocol, specimens with only one CFTR variant found on the panel will progress to CFTR gene sequencing so the second variant can be identified, given that infants with CF are expected to have two identifiable variants. Results from the panel or the panel plus sequencing will determine whether an infant is given a positive CF newborn screening result and sent for sweat testing.
In the IRT-DNA protocol, infants who have one or two CFTR variants on the panel are sent for sweat testing. “Babies who have two panel variants usually have CF and we typically see a positive sweat test,” she said. If the sweat test is positive in an infant with only one variant, usually the CF clinic will order additional DNA testing to try to identify the second variant, Raraigh said. If the sweat test is normal in an infant with only one variant, the infant is considered a CF carrier and usually discharged from follow-up. “Typically, many, many carriers are sweat tested to find just a handful of affected infants.”
In the sequencing protocol, infants with one variant are typically called out as carriers but not sweat tested; only infants with two variants are sent for sweat testing and diagnostic follow-up.
“At nearly every stage of the CF newborn screening algorithm, there is wide variability in how states choose to perform their testing,” Raraigh said. When infants have a negative screening result but then are diagnosed with CF, “about half of those cases are missed at the IRT level because the state has a cutoff that’s higher than the baby’s IRT was.” And about the other half are missed because their CFTR variants are not detected at the DNA stage, she said, noting they’re usually not on the panel.
Raraigh calls the variability among states in the panels they’re using a patchwork quilt, with many using a panel of 23 CFTR variants, others using panels of several hundred to more than 1,000, and everything in between.
A focus of the investigation into newborn screening practices has been on whether states’ choices disproportionately affect infants from different ancestral populations.
Between 2007 and 2020, the white non-Hispanic population made up 84.6 percent of the CF population but accounted for 70 percent of the group missed by newborn screening. Hispanic individuals made up 8.6 percent of the CF population but accounted for 15 percent of the infants missed by screening. Similarly, Black/African American individuals made up 3.6 percent of the CF population but accounted for 6.5 percent of those missed by screening (McGarry ME, et al. Pediatr Pulmonol. 2023;58[2]:465–474).
“Similar story with delay in diagnosis,” Raraigh said, adding, “Almost a quarter of those with a delayed diagnosis are Hispanic,” for example.
To remedy the problem, the CF Foundation convened a committee to address timeliness, sensitivity, and equity in newborn screening. “As a group,” she said, “we decided that all newborn screening programs should aim for 95 percent sensitivity or true positive detection across the ancestral groups tested by the state.”
All seven consensus, evidence-based recommendations had 100 percent agreement from the more than a dozen members of the committee.
No. 1 is to use a floating immunoreactive trypsinogen cutoff over a fixed cutoff. “The evidence we reviewed suggested that floating cutoffs are preferred because they account for some of the natural variability that might be present due to seasonality,” Raraigh said. IRT values can be affected by temperature or kit-to-kit variations. “But whatever cutoff is chosen is still important. A floating cutoff that’s too high will still miss a lot of babies,” she said.
No. 2 is to use a very high immunoreactive trypsinogen referral strategy in CF newborn screening programs whose variant panel does not include all CF-causing variants in CFTR2 or that do not have a variant panel that achieves at least 95 percent sensitivity in all ancestral groups within the state.
This referral strategy, she said, is a way to catch infants with rare variants. Even if the panel identifies no variants, specimens with a very high IRT level above a very high threshold are considered positive on screening or are sent to the sequencing step to ensure rare CFTR variants were not missed. “Infants from minoritized populations tend to have more rare variants, so this strategy is sometimes called a fail-safe to make sure they don’t slip through the cracks,” Raraigh
said.
The likelihood of an infant with a very high IRT and no CFTR variants having CF is low—IRT can be elevated for many reasons—but some states are choosing to keep this test as part of the algorithm to increase sensitivity. “And this recommendation tells states they should keep this referral strategy if their DNA panels are small enough that they’re not reaching 95 percent sensitivity in their population,” she said.
The remaining recommendations are as follows:
- CF newborn screening algorithms should not limit CFTR variant detection to the F508del variant or variants included in the ACMG-23 panel.
- CF newborn screening programs should screen for all CF-causing CFTR variants as identified by CFTR2.
- CFTR variant screening should be conducted twice weekly or more frequently as resources allow. “The evidence showed that doing the testing more frequently was associated with an earlier first CF visit and an earlier age of diagnosis,” Raraigh explained.
- A CFTR sequencing tier should follow IRT and CFTR variant panel testing to improve the specificity and positive predictive value of newborn screening. “Once sequencing is implemented,” she said, “only babies with two CFTR variants or more should be referred for a sweat test.”
- Both the primary care provider and CF specialist should be notified of abnormal screening results. Notifying both raises the likelihood that CF risk is assessed correctly and that accurate information about CF is shared.
Sequencing is at the core of improved carrier and newborn screening, Raraigh said.
“When we recommend that each state now look at all the CF-causing variants in CFTR2, that’s over 1,000. When we talk about a refined carrier screening list of 100 variants, it’s pretty difficult to design a panel that’s not sequence based to include all those variants. And so increasingly labs are relying on DNA sequencing technology to keep up with these changing lists.”
Now, when necessary, she said, states are helping other states. “There’s an initiative to share DNA laboratories,” with one laboratory sending its specimens to be sequenced in a laboratory in another state. “There’s a push to make this happen,” she said, because the states too have seen how important early diagnosis and treatment can be.
“A lot of changes have been made already, and we’re going to see that continue to snowball.”
Sherrie Rice is editor of CAP TODAY.