Population-Based Newborn Screening for Genetic Disorders When Multiple Mutation DNA Testing Is Incorporated: A Cystic Fibrosis Newborn Screening Model Demonstrating Increased Sensitivity but More Carrier Detections
* New England Newborn Screening Program of University of Massachusetts Medical School, Boston, Massachusetts
The Children's Hospital, Boston, Massachusetts
New England Medical Center, Boston, Massachusetts
|| Baystate Medical Center Children's Hospital, Springfield, Massachusetts
¶ Massachusetts General Hospital, Boston, Massachusetts
# University of Massachusetts Memorial Health Care, Worcester, Massachusetts
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objectives. Newborn screening for cystic fibrosis (CF) provides a model to investigate the implications of applying multiple-mutation DNA testing in screening for any disorder in a pediatric population-based setting, where detection of affected infants is desired and identification of unaffected carriers is not. Widely applied 2-tiered CF newborn screening strategies first test for elevated immunoreactive trypsinogen (IRT) with subsequent analysis for a single CFTR mutation (
F508), systematically missing CF-affected infants with any of the >1000 less common or population-specific mutations. Comparison of CF newborn screening algorithms that incorporate single- and multiple-mutation testing may offer insights into strategies that maximize the public health value of screening for CF and other genetic disorders. The objective of this study was to evaluate technical feasibility and practical implications of 2-tiered CF newborn screening that uses testing for multiple mutations (multiple-CFTR-mutation testing).
Methods. We implemented statewide CF newborn screening using a 2-tiered algorithm: all specimens were assayed for IRT; those with elevated IRT then had multiple-CFTR-mutation testing. Infants who screened positive by detection of 1 or 2 mutations or extremely elevated IRT (>99.8%; failsafe protocol) were then referred for definitive diagnosis by sweat testing. We compared the number of sweat-test referrals using single- with multiple-CFTR-mutation testing. Initial physician assessments and diagnostic outcomes of these screened-positive infants and any affected infants missed by the screen were analyzed. We evaluated compliance with our screening and follow-up protocols. All Massachusetts delivery units, the Newborn Screening Program, pediatric health care providers who evaluate and refer screened-positive infants, and the 5 Massachusetts CF Centers and their affiliated genetic services participated. A 4-year cohort of 323 506 infants who were born in Massachusetts between February 1, 1999, and February 1, 2003, and screened for CF at 
2 days of age was studied.
Results. A total of 110 of 112 CF-affected infants screened (negative predictive value: 99.99%) were detected with IRT/multiple-CFTR-mutation screening; 2 false-negative screens did not show elevated IRT. A total of 107 (97%) of the 110 had 1 or 2 mutations detected by the multiple- CFTR-mutation screen, and 3 had positive screens on the basis of the failsafe protocol. In contrast, had we used single-mutation testing, only 96 (87%) of the 110 would have had 1 or 2 mutations detectable by single-mutation screen, 8 would have had positive screens on the basis of the failsafe protocol, and an additional 6 infants would have had false-negative screens. Among 110 CF-affected screened-positive infants, a likely "genetic diagnosis" was made by the multiple-CFTR-mutation screen in 82 (75%) versus 55 (50%) with F508 alone. Increased sensitivity from multiple-CFTR-mutation testing yielded 274 (26%) more referrals for sweat testing and carrier identifications than testing with F508 alone.
Conclusions. Use of multiple-CFTR-mutation testing improved sensitivity and postscreening prediction of CF at the cost of increased referrals and carrier identification.
Key Words: newborn screening • genetic screening • population screening • cystic fibrosis screening • cystic fibrosis • multiple-mutation testing • DNA screening
Abbreviations: CF, cystic fibrosis • MA-CF, Massachusetts CF newborn screening program • IRT, immunoreactive trypsinogen • CFTR, cystic fibrosis transmembrane conductance regulator • CFF, Cystic Fibrosis Foundation • NENSP, New England Newborn Screening Program • [IRT], IRT concentration • Cl–, chloride • QNS, quantity not sufficient • NPV, negative predictive value PPV, positive predictive value • CI, confidence interval
After reported beneficial outcomes for screened-positive, cystic fibrosis (CF)-affected infants from a randomized clinical trial of CF newborn screening in which nutritional benefits are clearly demonstrated but long-term pulmonary outcomes remain under evaluation,1 the state of Massachusetts implemented a statewide and elective CF newborn screening service in February 1999.2 The dual purposes of the ongoing Massachusetts-CF newborn screening (MA-CF) program are 1) to provide a service with potential for long-term clinical benefit while awaiting the outcomes of a longer term prospective clinical trial3 and retrospective studies4–7 and 2) to provide the service universally as part of a statewide evaluation of technical feasibility for population-based DNA assay applications, using CF newborn screening as a model. In keeping with the recommendations of a 1997 Centers for Disease Control and Prevention-sponsored national workshop on newborn screening for CF,8 ** the supplementary service is optional, requiring consent.
The purpose of CF newborn screening is identification of CF-affected infants; thus, when protocols that use DNA assays are implemented, care must be taken to minimize the detection of infants who are unaffected carriers. Strategies used by CF newborn screening programs have included measuring for elevated levels of immunoreactive trypsinogen (IRT; which is an indirect measure of pancreatic injury that is present at birth in most newborns who have CF) on serial dried blood spot specimens9 or measuring for elevated IRT followed by assaying for the single most common CF transmembrane conductance regulator (CFTR) mutation, F508, on the same dried blood spot (2-tier algorithm).10 Unlike population screening for carrier identification, such newborn screening protocols were developed so that assays for CFTR mutations are limited to a subset of infants' specimens (up to 10% of total population). However, within this subset, observation of any 1 mutation from the second-tier DNA component of the screen yields an interpretation that the infant's risk of CF is high enough to refer the infant for diagnostic evaluation. Thus, for newborn screening programs that were operational in 1998/1999, it was assumed that the screen would identify most CF-affected infants even when the DNA component included just the most common CFTR mutation, F508 (observed in 71.25% of Wisconsin CF chromosomes11). To increase sensitivity, a small number of European regional programs developed optimized mutation panels for their specific populations.6,12 Our preimplementation analysis of genotype data available from the Cystic Fibrosis Foundation (CFF) on patients seen in Massachusetts CF centers13 suggested that the presence of F508 in only 60% of MA-CF chromosomes would yield a maximum 85%14 sensitivity and that testing for a limited set of multiple-CFTR mutations was required to achieve 95% sensitivity with the DNA component of the screen.15 The ideal number and choice of mutations to be included in a newborn screening panel has yet to be determined16 and ultimately may be different from that recommended for prenatal carrier screening, which does not include the functional assay for IRT.
The MA-CF newborn screening strategy uses a modification of the 2-tier laboratory algorithm, incorporating assays for multiple-CFTR mutations and a "failsafe" protocol whereby infants who have the highest IRT values and do not have CFTR mutations detected are referred for sweat testing to maximize CF detection among racial and ethnic populations whose mutations are not represented on common mutation panels (Fig 1). Furthermore, our screen recommends a follow-up protocol in which all screened-positive newborns are referred to a CFF-approved center where sweat testing and definitive diagnosis occur and where genetic counseling is offered to all families of CFTR-mutation-carrying infants.
|
We sought to compare the predictive values and sweat-test referral patterns of screening with single mutation (
F508 alone) with those of screening with multiple-CFTR mutations. We also sought to evaluate the utility of the provision for sweat testing of infants with the highest IRT values in the absence of a detected CFTR mutation and to be able to evaluate the utility of multiple-CFTR-mutation testing when this "failsafe" was in operation. Our comprehensive follow-up model and involvement of all Massachusetts CF center directors facilitated patient tracking through sweat testing and genetic counseling, validation of screening results, tracking diagnostic outcomes, and ensuring appropriate services for screened-positive families. This report presents data and analysis from the first 4 years of implementation. |
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
After the public process in which expansion of the number of disorders for which Massachusetts screens its newborns was deliberated,17 the Massachusetts Department of Public Health promulgated regulations2 in which statewide optional CF newborn screening was added, effective February 1, 1999. For this endeavor, the New England Newborn Screening Program (NENSP) invited directors of the 5 Massachusetts CFF-certified centers (responsible for the care of affected newborns identified by the screen) to its CF Workgroup; directors from all centers have been active participants in review of CF-newborn-screening and follow-up protocols, implementation, and outcomes assessment.
Study Subjects
The report presents analyses of all infants enrolled in MA-CF newborn screening from implementation February 1, 1999, through date of birth January 31, 2003. The approximate proportions of race and ethnicity represented in this birth cohort were as follows: white non-Hispanic, 74.3%; Hispanic, 11.4%; black non-Hispanic, 7.2%; Asian, 5.6%; other, 1.3%; and unknown, 0.2%.18 Enrollment was by an informed consent process approved by the Human Subjects Review Boards of the University of Massachusetts Medical School and the Massachusetts Department of Public Health. Verbal consent was obtained from parents for their respective infants (80 000 per year) with documentation as described below. Health care providers trained in the specific consent procedure administered the consent protocol after in-service training using instruction manuals and instruments provided by NENSP. Specifically, prenatal and neonatal care providers distributed an educational brochure19 (available in 9 languages) that describes routine newborn screening, the optional program for cystic fibrosis screening, and the process for enrolling in the optional screen. (Optimally, distribution of the brochure is prenatal, and at a minimum, distribution follows the birth.) Before collection of the specimen, the care provider asked a parent for his or her decision about the optional CF screening. The only documentation required was documentation of "declines participation," which was recorded in a "CF study" box printed on the newborn-screening specimen card or requisition. The parent was then given a copy of the completed requisition. At the NENSP, specimens with "declines participation" were excluded from MA-CF newborn screening.
Laboratory Screening Algorithm
Overview
The laboratory used a 2-tiered IRT/DNA protocol. Detection of CFTR mutations or the additional failsafe protocol (extremely high IRT) was used to determine CF screen-positive results. During the first 4 years, 2 protocol revisions were implemented: 1) initially, the prompt for DNA analysis was an IRT concentration ([IRT]) >90th percentile of the values obtained on the day the specimen was analyzed (daily percentile); after review of our data from the first 9 months, this IRT threshold was adjusted to >95th daily percentile for increased specificity; 2) initially, the multiple-CFTR-mutation panel included 16 mutations; after 21 months, this was increased to 27 mutations as a result of availability of commercial reagents.
IRT
Wallac DELFIA kits (Turku, Finland) were used to measure [IRT] according to the manufacturer's instructions. [IRT]s were ranked in descending order, and specimens in the top 10th or 5th percentile underwent DNA testing.
DNA
Amplification and colorimetric detection on linear array strips with Analyte Specific Reagents for a 16-mutation assay (gift from Roche Molecular Systems, Alameda, CA) and a 27-mutation assay (Linear Array CF-31; Roche Molecular Biochemicals, Indianapolis, IN) were used.20 For both panels, the DNA assay assessed only CFTR mutations; detection of polymorphisms was incorporated as a reflex test for confirmation of putative F508 homozygotes (assay for F508C, I506V, and I507V) or for genotype elucidation on detection of 2 mutations including R117H (assay for IVS8polyT 5/7/9T). The 16-mutation panel included F508, R117H, G551D, G542X, W1282X, N1303K, R334W, 621 + 1G>T, R553X, I507, 1717-1G>A, R347P, R560T, 3849 + 10kbC>T, A455E, and S549N. The 27-mutation panel included all of the 16 except S549N and the following additional mutations: 3120 + 1G>A, 3659delC, A559T, R1162X, S1255X, 405 + 3A>C, 711 + 1G>T, 2789 + 5G>A, G480C, 2307insA, G85E, and 1078delT.
Screening Result Definition
For specimens with [IRT] below the DNA prompt, final results were CF-screen negative. For specimens in which DNA testing was prompted, final results were either CF-screen negative (IRT-prompted DNA analysis and no mutations detected and IRT <99.8 percentile [see below]) or 1 of 3 categories of CF-screen positive. The 3 CF-screen-positive categories were 1) IRT-prompted DNA analysis and 2 mutations detected; 2) IRT-prompted DNA analysis and 1 mutation detected; and 3) IRT-prompted DNA analysis and none of the panel mutations detected but "[IRT] 
150 ng/mL or [IRT] determined to be in top 0.2% monthly percentile" and at the time of reporting, no subsequent specimen obtained before 30 days of life showed an [IRT] ranked 
95th daily percentile.
CF-Screen-Positive Reporting Algorithm
All CF-screen-positive results were telephone-reported to the infant's primary care provider. The report included a result-specific estimate of empirically determined relative risk for CF and guidance on the recommended follow-up. The telephone report was accompanied by a fax of 1) report of results; 2) a letter to the provider, explaining the screening algorithm; 3) CF center/genetics contact, scheduling, and location information; and 4) sweat-test procedure information for providers to deliver to parents. When the primary care provider identified a specific CF center or when a specific CF center identified itself as the referral center for a particular infant, detailed laboratory and medical provider contact information was faxed from NENSP to that center. In the absence of center identification, NENSP staff contacted the primary care provider to track protocol adherence and diagnosis/treatment.
Follow-up Protocol for CF-Screen-Positive Results
Overview
The follow-up protocol required 1) sweat testing for all infants whose CF newborn screening result was CF-screen positive, 2) recommendation for genetic counseling referral to all families of infants whose newborn screening results revealed the presence of 1 or more CFTR mutations, and 3) referral to a CF specialist for all infants whose sweat-test results were consistent with CF or whose CF newborn screening result included the detection of 2 mutations (regardless of sweat-test result). The NENSP reporter encouraged follow-up to take place at a CF center because these centers have standardized quantitative protocols that meet National Committee for Clinical Laboratory Standards guidelines and participate in ongoing quality control, the Massachusetts CF centers with their institutional affiliates can provide all services including genetic counseling, and at least 1 of the Massachusetts CF centers is within a 2-hour drive of all Massachusetts residents.
Sweat Testing
Quantitative pilocarpine iontophoresis (sweat test) was performed by standard methods.21 Sweat chloride ([Cl–]) results were considered negative when the [Cl–] was <30 mEq/L, borderline when the [Cl–] was 30 to 59 mEq/L22, and positive when the [Cl–] was 60 mEq/L (all on at least 75 mg or 15 µL of sweat from at least 1 arm). When <75 mg or <15 µL sweat was obtained from both arms, the quantity was considered not sufficient (QNS) for reporting accurate results and the protocol required repeating the test, as below.
Case Definition and Post-Sweat-Test Protocol
CF diagnosis was established by a positive sweat-test result or by a CF specialist according to the criteria set by the CFF.23 For sweat-test-negative results, infants whose CF-screen-positive results showed 0 or 1 mutation with negative sweat-test results were considered to be unaffected. Families of infants with 1 mutation were offered genetic counseling standardized to address the context of newborn screening. Infants whose CF-screen-positive results showed 2 mutations with negative sweat-test results were evaluated by a CF specialist for diagnosis in consideration of a particular genotype that might be associated with a negative sweat test or the possibility of a laboratory or specimen error. For sweat-test-borderline or QNS results, families of infants with borderline and QNS results were counseled to have the sweat test repeated every 4 to 8 weeks until resolution. For the infants whose sweat-test-borderline result might be explained by the presence of a particular CFTR mutation known to be associated with borderline values, a consultation with a CF specialist was recommended. When borderline values persisted in infants whose CF-positive screen showed 0 or 1 mutation, more extensive mutation analysis was recommended. Sweat-test-positive results, infants with positive sweat-test results were considered to be affected and were referred immediately to a CF specialist. Additional CFTR mutation testing was recommended when the CF-positive screen had detected only 1 or 0 CFTR mutations.
Data Collection and Analyses
CF centers reported dates and results of all sweat testing, dates of genetic counseling, and subsequent genotype results to the NENSP. CF centers reported positive sweat-test results of any patient who was born after February 1, 1999 and seemed to have been missed by the MA-CF. The program's technical feasibility was assessed by concordance of screening results with diagnostic outcomes. Practical implications were measured by rates of referral to CF centers and protocol compliance. All data were centralized at the NENSP. Statistics were calculated using Stata (Stata Corp, College Station, TX).
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Detection and Classification of CF-Affected Infants
During the first 4 years of CF newborn screening in Massachusetts, 113 infants who had received a diagnosis of having CF were born (as of August 5, 2003). The MA-CF detected 110 of the 112 affected infants whose parents requested CF newborn screening (Fig 2, Table 1), with a negative predictive value (NPV) of 99.99%. The positive predictive value (PPV) of the later screen ([IRT] >95%) was 9.4% (95% confidence interval [CI]: 7.6–11.4). As expected, a majority of the 113 affected infants were non-Hispanic white, but the screening algorithm did identify CF in 9 Hispanic and 3 non-Hispanic nonwhite infants within a birth cohort that included
37 000 Hispanics and 47 000 non-Hispanics nonwhites.24 One of the 2 affected infants who were not detected by screening was a nonwhite infant who presented with meconium ileus, a presentation that has been associated with lower IRT values and false-negative screens.25
|
|
The initial clinical presentation of the 110 CF-affected infants reported to NENSP by the primary care provider was varied. At the time of the screening report, more than two thirds of the CF-screened-positive infants (76 of 110) had only nonspecific signs or no signs at all to prompt clinical suspicion of CF (Table 2). Concordance between "affected classification" and sweat results was high. Of the 98 CF-screened-positive affected infants for whom sweat results have been reported, 84 (86%) had values that were clearly positive and 13 of the remaining 14 had sweat-test results that are consistent with expectations for a particular CF genotype (Table 3).
|
|
Analysis by the particular category of CF-screen-positive result demonstrated that the highest risk of CF is observed when 2 mutations are detected in the screen (PPV: 100%) and that infants who show 0 mutations with extremely elevated IRT carry substantial risk of being affected (Table 1). Furthermore, within the CF-screen-positive category in which 1 mutation is detected, the highest risk of CF seemed to be in infants whose specimens also had a particularly high [IRT] (12 of 35 infants whose [IRT] was >150 ng/mL, yielding a risk of 0.34 [95% CI: 0.19–0.52]). Infants with CF-screen-positive results showing 0 mutations were frequently infants with other perinatal stresses as noted previously.11
Evaluation of IRT Values for DNA Prompt
A PPV of 8.2% is derived when data from the 2 IRT algorithms are combined. As expected, optimization at the first tier of the screening protocol (restricting the IRT prompt for DNA analysis from the top 10% of [IRT] to the top 5%) improved the PPV of the CF newborn screen (Fig 2) from 5.2% (95% CI: 3.2%–8.0%) to 9.4% (95% CI: 7.6%–11.4%).
False-Negative Results
One of the 2 infants who were not detected by the >95th daily percentile screen had an IRT value at the 93.9th percentile and meconium ileus (positive sweat test; G542X/unknown); the other infant missed by the screen had an IRT value at the 84th percentile and presented at 2 months with failure to thrive and upper respiratory tract infection (positive sweat test; F508/R117H).
Use of Multiple Mutations in the Screening Algorithm
In our cohort of 112 CF-screened and CF-affected infants, 15 (13%) did not carry F508 on either chromosome (Table 4). Implementation of a screening protocol that tested for multiple-CFTR mutations resulted in the early identification of 11 of these 15 affected infants (5 of whom carried 2 non-F508-CFTR mutations from the multiple-CFTR-mutation panel). The failsafe protocol (CF-screen-positive when IRT was >99.8%) identified another 3 affected infants. The 15th infant's IRT did not prompt a DNA assay; subsequent diagnostic testing revealed no F508 mutation but detected another (G542X) that is included in our multiple-CFTR-mutation testing panel. The multiple-mutation screening algorithm resulted in 274 (26%) more referrals for sweat testing than a single-mutation screening algorithm would have (Table 5).
|
|
By using the multiple-CFTR-mutation testing panel, 27 infants who are compound heterozygotes (infants who carry 2 different mutations) were identified as such directly by the screen. These infants were appropriately reported to the pediatrician as carrying 2 mutations and being at the highest risk for CF (Table 5). Of the residual infants who were shown by the screen to have
F508 with the multiple-CFTR-mutation panel, only 1 of 33 would be shown to be affected, compared with 1 of 16 when screening was done by F508 alone.
Of the 11 CF-screen-positive CF-affected infants who do not carry F508 but who do carry other CFTR mutations included in the multiple-CFTR-mutation testing panels, only 5 would have been identified by the failsafe protocol as at-risk for CF; 6 would have false-negative screens, but 1 of those 6 presented clinically with meconium ileus.
All 3 CF-affected infants who were detected only because of the failsafe protocol (IRT >99.8%) were Hispanic. One of these carries a mutation (G85E) that was not included on the 16-mutation panel (used at the time of testing) but is on the 27-mutation panel. This infant also carries a mutation (R117C) that is not present on either panel or on the population-screening panel recommended by the American College of Medical Genetics.26 The other 2 CF-affected infants presumably have CFTR mutations not yet identified. Thirteen of the 27 CFTR mutations in our panels have yet to be detected in any of our CF-affected infants; if our multiple-mutation-CFTR panel had been composed of only 9 mutations, then the screen would have a 98% sensitivity (110 of 112 affected infants have at least 1 of the 9 mutations). Follow-up genetic testing of infants or their parents by other laboratories confirmed all mutations detected by the screen. Among infants whom we reported as CF-screen-positive, there were 11 whose CF chromosomes had mutations identified by subsequent DNA diagnostic testing (8 independent mutations were observed). Finally, 20 (9%) CF chromosomes from CF-affected infants in our cohort have mutations that have yet to be identified despite additional mutation analysis.
MA-CF Newborn Screening Follow-up
Of the 1338 infants with CF-screen-positive results, 31 had died before sweat testing (3 of whom received a diagnosis of CF and 1 of whom could have had clinical signs that might be associated with CF but died without a diagnosis), 11 families of infants refused sweat testing, 1 infant had trisomy 18, and 26 (2.2%) had incomplete follow-up (including 12 who were lost, after all tracking efforts, including certified letters to parents, failed). Of the remaining infants, 95% were documented to have completed their sweat-test appointments as of this report (Table 6). Late or noncompliance with sweat testing was often observed with infants who had a stay in a neonatal intensive care unit; these infants would need to be stabilized and/or gain sufficient weight before the sweat test. For a small number of affected infants, no sweat appointments had yet been completed because the diagnosis was substantiated by a CF specialist according to CFF guidelines without the sweat test. Of the 858 families who were offered genetic counseling, 70% accepted (Table 6). Observed median time interval from birth to diagnosis or rule out, as it related to observed number of mutations, showed infants with 2 mutations to have the shortest and those with 0 mutations to have the longest interval (2 weeks vs 6 weeks; Table 6).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Implementation of population-based CF newborn screening enhances early detection of CF-affected infants, regardless of prenatal care or perceptions of risk on the basis of race or ethnicity. The 2-tiered screening protocol that we describe incorporates a limited set of common CFTR mutations and seems to detect the vast majority of CF-affected infants in our heterogeneous population. Incorporation of a "failsafe" for detecting affected infants whose mutations are uncommon enhanced sensitivity, increasing it from 96% to 98% (107–110 of 112 affected infants).
Although still early in our evaluation and with the possibility that a few more affected infants in this cohort may present clinically at a later time, we are reasonably confident that the current denominator of 112 is close to the total likely to be found because 1) our screening program is composed of an integrated system with centralized tracking of affected infants within the population; 2) the observed frequency of CF in Massachusetts infants (1/2888) is significantly higher (P = .002) than observed in other population-based programs (1/4982)27; and 3) our detection rate of 28 affected newborns per year is consistent with data in the national CFF Registry showing that in each of years 1990, 1991, and 1992, a total of 27, 30, and 34 CF patients were born in Massachusetts and presented clinically over an 8- to 11-year course (ie, the number of infants diagnosed from birth years 1990–1992 [91/268 364 births] was not significantly different from number detected through the MA-CF in 1999–2003 [112/323 506; P = .94 by Fisher exact test]). In addition, within our cohort of 112 CF-affected infants, the observed frequency of the disorder among Hispanic infants (1/4115) was nearly double the expected (1/8500).24,28 This higher frequency might be attributable to statistical variation, enhancement by screening of the spectrum of disease, or a true regional difference.
Comparison of a screening algorithm that includes single- with multiple-CFTR-mutation testing shows similar predictive values (9.8 vs 8.2% PPV and 99.9% NPV for both). However, use of a limited set of common CFTR mutations clearly increases sensitivity. As predicted,14 a panel composed of 12 common CFTR mutations yielded >95% sensitivity in our population. Addition of population-specific CFTR mutations might increase sensitivity further in particular populations, but it seems that addition much beyond the 11 or 12 most common CFTR mutations in mixed populations yields diminishing returns: 13 of the 27 CFTR mutations on our panels have yet to be detected in any of our CF-affected infants, and 9% of CF chromosomes identified by the screen have mutations yet to be identified despite additional mutation analysis.
By using the multiple-CFTR-mutation panel, a screening result with a genetic "diagnosis" of CF was made in 75% of screened-positive CF-affected infants, compared with 50% had we used F508 alone (P = .002), thus facilitating more rapid referral and intervention. More than two thirds of the affected infants had no recorded signs prompting CF clinical concern before the screening result. The difference in rapid referral for 75% versus 50% may be significant, especially as new early interventions for CF become available. In addition, the risk of being affected for a screened-positive infant who showed only F508 by the multiple-CFTR-mutation screen was cut in half compared with what could be predicted from a screen using F508 alone.
Another important feature of the multiple-CFTR-mutation testing was our ability to predict the potential for inconclusive sweat-test results on the basis of the screening genotype. Consultation with a CF specialist was indicated at the initial visit for all CF-screen-positives with 2 mutations, and ready explanation for some borderline sweat-test results was available at the initial visit to a CF center.
Of interest is that the only CF-affected infants known to have been missed by the MA-CF were missed because their [IRT] did not prompt DNA analysis, not because the DNA analysis with a limited set of multiple-CFTR mutations excluded them from follow-up. A primary DNA screen with our multiple-CFTR mutations would have identified both of the missed infants. Such an observation calls into question whether a primary DNA screen would be more effective or whether the IRT prompt for DNA analysis should be lowered. However, a primary DNA screen in the absence of a functional assay such as IRT would have missed the 3 Hispanic infants who were identified by the failsafe protocol. In addition, a primary DNA screen without IRT would provide insufficient data to distinguish the 25 CF-affected CF-screen-positive infants with 1 mutation from the up to 11 981 carriers in the cohort who would also be identified. A primary DNA screen using high-throughput sequencing would allow such differentiation, but detection of uncharacterized sequence variants and undefined genotype-phenotype relationships would pose significant challenges. Parallel testing of IRT with DNA or lowering the IRT prompt to the lowest observed in a CF-affected infant ([IRT] 84%) would identify all CF-affected infants but would still require some follow-up for more of the 11 981 carriers in the cohort—a scenario that could overwhelm provider systems, resulting in increased health care costs and increased incidents of parental anxiety.
The purpose of the newborn screen is to select infants who are most at risk and refer them for diagnostic follow-up. For each disorder added to the list, the sensitivity and specificity of the screen must be balanced with what is known about the need for early intervention. For disorders with a rapid irreversible clinical progression, the goal of the screen is 100% sensitivity. At present, there are insufficient data to suggest that CF is such a disorder. Like all disorders on newborn screening panels, there must be an understanding that the screen is not a diagnostic test and will not detect 100% of affected infants. A significant clinical concern should result in a referral for a sweat test regardless of the screening result.
Massachusetts is not the first state to offer CF newborn screening to its population but has done so as a supplemental test after the 1997 Centers for Disease Control and Prevention-sponsored national workshop recommendations8 and is the first state to use an algorithm inclusive of multiple-CFTR mutations. Other states that are in the process of or planning to implement CF newborn screening may have genetically different and more or less diverse populations. Given the experience in a heterogeneous population such as Massachusetts', others might also benefit from a protocol with a limited panel of CFTR mutations that incorporates a failsafe to detect the affected infants who do not carry common mutations. States with more homogeneous populations may choose to implement protocols with only the most common mutation and make educational materials available to providers that reflect appropriate expectations from the screening algorithm.
We have demonstrated technical feasibility with a comprehensive model that includes education, laboratory screening, diagnostic workup, and genetic counseling. We have shown that multiple-CFTR-mutation testing increased diagnostic sensitivity, was better for predicting postscreening risk of CF, but resulted in 26% more carrier identifications and referrals. We have further demonstrated the utility of including a "failsafe" provision for referral for sweat testing to maximize CF detection among ethnic populations whose mutations are not represented on common mutation panels. Cost-effectiveness analyses in light of available screening options as clinical outcome data become available are warranted.
|
ACKNOWLEDGMENTS |
|---|
This work was supported in part by Program Funds of the New England Newborn Screening Program of the University of Massachusetts Medical School and in part by the Health Resource and Services Administration Grant 5 H46 MC 00198-02.
We acknowledge the work of Jacalyn Gerstel-Thompson, Karin Riddle, Galina Mushinsky, Dennis Coffey, Marina Burkatovskava, Janice Parrin, Rosalie Hermos, Jane Griffin, Kathleen McIntosh, Rosemarie Collard, and Judith Russo at the NENSP for dedication to the timely turnaround of laboratory and reporting data; Monica Ulles, Nancy Shotola, Anita St. John, Joan Murphy, Carol Mason, and Adam Courchaine for tracking particularly large numbers of sweat-test data elements; Pamela Hawley, Patricia Wheeler, Laurie Demmer, Gabriel Cohn, and Susan Pauker for screening-linked genetic counseling of the many families; Thomas Martin, current CF center director at NEMC, nursing and technical staff at each of the CF center's sweat laboratories and clinics, and staff at each of the institution's Genetics Divisions and the Genetic Division of Harvard Pilgrim Health care; all of the pediatric health care providers for their role in communicating with families; all of the Massachusetts families who consented to participate in this evaluation project; and Cecilia Larson, George F. Grady, and Alfred DeMaria for critical review of the manuscript.
|
FOOTNOTES |
|---|
Received for publication Aug 25, 2003; Accepted Oct 23, 2003.
Reprint requests to (A.M.C.) New England Newborn Screening Program, University of Massachusetts Medical School, 305 South St, Jamaica Plain, MA 02130. E-mail: anne.comeau{at}umassmed.edu
Dr Dorkin is currently at Massachusetts General Hospital, Boston, Massachusetts.
** Specific recommendations relevant to this initiative include: "[T]he workshop participants concluded that sufficient evidence exists to recommend pilot state-based demonstration programs. Public health research programs should use a multidisciplinary team. ... Pilot CF screening programs for newborn should be approached and promoted as research endeavors, for which participation is not mandatory and informed consent is emphasized, and as community intervention programs that refer patients to accredited CF care centers. Pilot CF screening programs for newborn should include the following design attributes: a) genetic counseling for families with newborns who have CF or those who are carriers, b) referral to accredited CF care centers, and c) clinician education about the need for informed consent for testing newborns for CF." 
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
- Farrell PM, Kosorok MR, Laxova A, et al. Nutritional benefits of neonatal screening for cystic fibrosis.
N Engl J Med.1997; 337
:963
–969
[Abstract/Free Full Text] - Code of Massachusetts Regulations: Testing of Newborn Children for Treatable Diseases, 105 CMR 270.001-270.009; 1998
- Farrell PM, Kosorok MR, Rock MJ, et al. Early diagnosis of cystic fibrosis through neonatal screening prevents severe malnutrition and improves long-term growth.
Pediatrics.2001; 107
:1
–13
[Abstract/Free Full Text] - Waters DL, Wilcken B, Irwing L, et al. Clinical outcomes of newborn screening for cystic fibrosis.
Arch Dis Child Fetal Neonatal Ed.1999; 80
:F1
–F7
[Abstract/Free Full Text] - Merelle ME, Schouten JP, Gerritsen J, Dankert-Roelse JE. Influence of neonatal screening and centralized treatment on long-term clinical outcome and survival of CF patients.
Eur Respir J.2001; 18
:306
–315
[Abstract/Free Full Text] - Assael BM, Castellani C, Ocampo MB, Iansa P, Callegaro A, Valsecchi MG. Epidemiology and survival analysis of cystic fibrosis in an area of intense neonatal screening over 30 years.
Am J Epidemiol.2002; 156
:397
–401
[Abstract/Free Full Text] - Wang SS, O'Leary LA, FitzSimmons SC. The impact of early cystic fibrosis diagnosis on pulmonary function in children. J Pediatr.2002; 141 :804 –810[CrossRef][Web of Science][Medline]
- Cono J, Qualls N, Khoury M, Hannon W. Newborn screening for cystic fibrosis: a paradigm for public health genetics policy development. Proceedings of a 1997 Workshop. MMWR Recomm Rep.1997; 46(RR-16) :1 –24
- Hammond KB, Abman SH, Sokol RJ, Accurso FJ. Efficacy of statewide neonatal screening for cystic fibrosis by assay of trypsinogen concentrations. N Engl J Med.1991; 325 :769 –774[Abstract]
- Gregg RG, Wilfond BS, Farrell PM, Laxova A, Hassemer D, Mischler EH. Application of DNA analysis in a population-screening program for neonatal diagnosis of cystic fibrosis (CF): comparison of screening protocols. Am J Hum Genet.1993; 52 :616 –626[Web of Science][Medline]
- Gregg RG, Simantel A, Farrell PM, et al. Newborn screening for cystic fibrosis in Wisconsin: comparison of biochemical and molecular methods.
Pediatrics.1997; 99
:819
–824
[Abstract/Free Full Text] - Scotet V, de Braekeleer M, Roussey M, et al. Neonatal screening for cystic fibrosis in Brittany, France: assessment of 10 years' experience and impact on prenatal diagnosis. Lancet.2000; 356 :789 –794[CrossRef][Web of Science][Medline]
- Cystic Fibrosis Foundation. Patient Registry, 1990–1994 Annual Data. Bethesda, MD: Cystic Fibrosis Foundation; 1996
- Hardy GH. Mendelian proportions in a mixed population.
Science.1908; 28
:49
–50
[Free Full Text] - Parad R, Eaton R, Comeau AM. Lower predicted false negative CF detections are sought using knowledge of local allele frequencies and an expanded CFTR mutation panel. 4th Meeting of the International Society for Neonatal Screening; Stockholm, Sweden; June 13–16, 1999
- Bobadilla JL, Farrell MH, Farrell PM. Applying CFTR molecular genetics to facilitate the diagnosis of cystic fibrosis through screening. Adv Pediatr.2002; 49 :131 –190[Medline]
- Atkinson K, Zuckerman B, Sharfstein J, Levin D, Blatt RJR, Koh H. A public health response to emerging technology: expansion of the Massachusetts newborn screening program. Public Health Rep.2001; 116 :122 –131[CrossRef][Web of Science][Medline]
- Massachusetts Birth Data 1999, 2000, 2001 (averaged). Available at: www.state.ma.us/DPH/. Accessed October 17, 2003
- Answers to common questions about newborn screening. Available at: www.umassmed.edu/nbs/. Accessed October 17, 2003
- Package insert. ASR Linear Array CF-31. Available at: biochem.roche.com/pack-insert/3017443a.pdf. Accessed October 17, 2003
- LeGrys V. Sweat testing: Sample collection and quantitative analysis; Approved Guideline 2nd ed, NCCLS Document C34-A2. Wayne, PA: National Committee for Clinical Laboratory Standards; 2000
- Padoan R, Bassotti A, Seia M, Corbetta C. Negative sweat test in hypertrypsinaemic infants with cystic fibrosis carrying rare CFTR mutations. Eur J Pediatr.2002; 161 :212 –215[CrossRef][Web of Science][Medline]
- Rosenstein BJ, Cutting GR. The diagnosis of cystic fibrosis: a consensus statement. Cystic Fibrosis Foundation Consensus Panel. J Pediatr.1998; 132 :563 –565[CrossRef][Web of Science][Medline]
- Massachusetts Birth Data 1999, 2000, 2001 (extrapolation from 3 year to 4 year birth cohort). Available at: www.state.ma.us/DPH/. Accessed October 17, 2003
- Wilcken B, Wiley V, Sherry G, Ursala BHA. Neonatal screening for cystic fibrosis: a comparison of two strategies for case detection in 1.2 million babies. J Pediatr.1995; 127 :965 –970[CrossRef][Web of Science][Medline]
- American College of Medical Genetics Technical Standards and Guidelines for CFTR Mutation Testing Approved by the Board of Directors of the American College of Medical Genetics January 26, 2002. In press: Genetics in Medicine 2002;3 (5). Available at: www.acmg.net/Pages/ACMG_Activities/stds-2002/cf.htm Section CF 3.3.1. Accessed October 17, 2003
- National Newborn Screening and Genetic Resource Center Data. Available at: genes-r-us.uthscsa.edu/resources/newborn/99/ch10complete.pdf Table 10.7. Accessed October 17, 2003
- Cystic Fibrosis Foundation. Patient Registry, 2001 Annual Data. Bethesda, MD: Cystic Fibrosis Foundation; 2002
PEDIATRICS (ISSN 1098-4275). ©2004 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Kloosterboer, G. Hoffman, M. Rock, W. Gershan, A. Laxova, Z. Li, and P. M. Farrell Clarification of Laboratory and Clinical Variables That Influence Cystic Fibrosis Newborn Screening With Initial Analysis of Immunoreactive Trypsinogen Pediatrics, February 1, 2009; 123(2): e338 - e346. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. E. Hale, R. B. Parad, and A. M. Comeau Newborn Screening Showing Decreasing Incidence of Cystic Fibrosis N. Engl. J. Med., February 28, 2008; 358(9): 973 - 974. [Full Text] [PDF]
|
|
|
|
|
|
R. Giusti, A. Badgwell, and A. D. Iglesias Newborn Screening for Cystic Fibrosis in New York State: In Reply Pediatrics, July 1, 2007; 120(1): 241 - 243. [Full Text] [PDF]
|
|
|
|
|
|
R. Giusti, A. Badgwell, A. D. Iglesias, and and the New York State Cystic Fibrosis Newborn Scr New York State Cystic Fibrosis Consortium: The First 2.5 Years of Experience With Cystic Fibrosis Newborn Screening in an Ethnically Diverse Population Pediatrics, February 1, 2007; 119(2): e460 - e467. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M. Comeau, F. J. Accurso, T. B. White, P. W. Campbell III, G. Hoffman, R. B. Parad, B. S. Wilfond, M. Rosenfeld, M. K. Sontag, J. Massie, et al. Guidelines for Implementation of Cystic Fibrosis Newborn Screening Programs: Cystic Fibrosis Foundation Workshop Report Pediatrics, February 1, 2007; 119(2): e495 - e518. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. S. Green, S. M. Dolan, and T. H. Murray Newborn Screening: Complexities in Universal Genetic Testing Am J Public Health, November 1, 2006; 96(11): 1955 - 1959. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V. Scotet, M.-P. Audrezet, M. Roussey, G. Rault, A. Dirou-Prigent, H. Journel, V. Moisan-Petit, V. Storni, and C. Ferec Immunoreactive Trypsin/DNA Newborn Screening for Cystic Fibrosis: Should the R117H Variant Be Included in CFTR Mutation Panels? Pediatrics, November 1, 2006; 118(5): e1523 - e1529. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. E. van den Akker-van Marle, H. M. Dankert, P. H. Verkerk, and J. E. Dankert-Roelse Cost-effectiveness of 4 Neonatal Screening Strategies for Cystic Fibrosis Pediatrics, September 1, 2006; 118(3): 896 - 905. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. P. O'Sullivan, R. G. Zwerdling, H. L. Dorkin, A. M. Comeau, and R. Parad Early Pulmonary Manifestation of Cystic Fibrosis in Children With the {Delta}F508/R117H-7T Genotype Pediatrics, September 1, 2006; 118(3): 1260 - 1265. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. M. Farrell Cystic Fibrosis Newborn Screening: Shifting the Key Question From "Should We Screen?" to "How Should We Screen?" Pediatrics, June 1, 2004; 113(6): 1811 - 1812. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
