DNA

A New Cystic Fibrosis Newborn Screening Algorithm: IRT/IRT1[/DNA Marci K. Sontag, PhD, Dan Wright, James Beebe, PhD, Frank J. Accurso, MD, and Scott D. Sagel, MD Objective To evaluate an immunoreactive trypsinogen (IRT) IRT/IRT1[/DNA algorithm, aimed at improving sensitivity while decreasing cystic fibrosis (CF) carrier identification.

Study design New technologies allow the measurement of the second IRT level solely in infants with an elevated first IRT level. Specimens with an elevated second IRT level undergo mutation analysis. We tested the projected efficacy with retrospective data from Colorado. Results All known infants with CF would have been identified with our proposed IRT cutoff points, and 3 would have been missed with our mutation panel. Two of 3 missed cases would have been identified by using a failsafe method (IRT >99.9th percentile), yielding a sensitivity rate of 99.7% (95% CI, 98.4-99.9). Estimated reduction in carrier detection was 80% compared with IRT/DNA. Conclusion IRT/IRT1[/DNA appears to improve cystic fibrosis newborn screen sensitivity while decreasing carrier identification, providing an alternative to IRT/IRT in states that obtain 2 blood spots. (J Pediatr 2009;155:61822). See editorial, p 605

A

ll cystic fibrosis (CF) newborn screening is based on immunoreactive trypsinogen (IRT) testing of blood spots from the Guthrie card.1 Two algorithms are followed by most screening programs for newborns with CF. IRT/IRT was the first to be widely adopted; when IRT determinations from both the initial and a second blood spot (collected approximately 2 weeks apart) are higher than the predetermined cutoff points, the infant is called back for a diagnostic sweat test.2,3 In the United States, 11 of the 50 states routinely collect a second newborn screen on every infant.4 IRT/DNA was introduced after the identification of the cystic fibrosis transmembrane conductance regulator (CFTR) gene.5 When the IRT level on the initial bloodspot is elevated, that same blood spot is tested for $1 mutations from a pre-defined panel.6,7 Variations on these 2 proposals have been discussed, but most programs continue to screen infants with IRT/IRT or IRT/DNA.8,9 Although both of these algorithms have been successful in identifying newborns with CF, each has its drawbacks.10 The initial IRT cutoff is often lower in IRT/DNA than in IRT/IRT, resulting in a potential better sensitivity rate.11 More than 25 years of experience with IRT/IRT in Colorado has demonstrated acceptable sensitivity, albeit lower than that of IRT/DNA.12 IRT/DNA comes with the burden of identifying infants who do not have CF but have 1 CFTR mutation. The parents of those infants need to be offered genetic counseling. We developed a new screening algorithm that might improve sensitivity compared with IRT/IRT, while decreasing the carrier identification rate compared with IRT/DNA. We tested this algorithm by using data obtained from the Colorado newborn screening program from 1982 to 2007, representing >1 million infants screened. We examined IRT/IRT1[/DNA for different IRT cutoff points and 2 commonly used mutation panels. IRT/IRT1[/DNA appears to improve sensitivity while decreasing carrier identification and may be a better alternative to IRT/IRT in states that collect 2 blood spots.

Methods IRT/IRT1[/DNA Method We have developed a new algorithm to screen newborns for CF (Figure 1). All infants born in Colorado receive a first newborn screen before they are discharged from the hospital (at approximately 48 hours of life) and another screen at 2 weeks. We measure the second IRT solely in specimens from infants with an elevated first IRT level (>60ng/mL, approximately 97th percentile). When the IRT level is not elevated on the first newborn screening (NBS) specimen, the second specimen is not tested, and the results are considFrom the Department of Epidemiology, Colorado School ered complete. When the second IRT level is also elevated (>60 ng/mL), the secof Public Health, University of Colorado Denver, Aurora,

ACMG CF CFTR IRT NBS

American College of Medical Genetics Cystic fibrosis Cystic fibrosis transmembrane conductance regulator Immunoreactive trypsinogen Newborn screening

CO (M.S.); Laboratory Services Division, Colorado Department of Public Health and Environment, Denver, CO (D.W., J.B.); and Department of Pediatrics, University of Colorado Denver School of Medicine and The Children’s Hospital, Aurora, CO (F.A., S.S.) Supported by the Cystic Fibrosis Foundation SONTAG07AO, NIDDK RO1 DK61886, NHLBI U01 HL081335. The authors declare no real or perceived conflicts of interest. 0022-3476/$ - see front matter. Copyright Ó 2009 Mosby Inc. All rights reserved. 10.1016/j.jpeds.2009.03.057

618

Vol. 155, No. 5  November 2009 performed through infant’s name, mother’s name, and date of birth. IRT determinations are made on the 2-week specimen on samples with an unmatched first screen. Linking the 2 samples, and the new database technology available through SpecimenGate, allows the second specimen to be tested only when the first blood spot had an elevated IRT level. A second screen is submitted and linked to a first screen in 88.1% of cases (95% CI, 87.7-88.5). Linking is successful in 96% of the samples (95% CI, 95.7-96.1).

NBS Blood Spot Obtained and Sent to Lab

Normal < 97th percentile

Elevated > 97th percentile

Normal CF Screen

Repeat IRT on 2nd Blood Spot

Elevated > 97th percentile

Repeat - Normal < 97th percentile

CFTR Mutation Analysis

Normal CF Screen

2 mutations

1 mutation

0 mutations

Diagnosis CF

At risk for CF

Both IRTs >150ng/ml

Confirmatory

Diagnostic

30 – 59 mmol/L Intermediate Consult with CF Center

Low Risk for CF Normal CF Screen

Sweat Test

≤ 60mmol/L CF Refer to CF Center

At least 1 IRT<150ng/ml

≤ 29 mmol/L CF Very Unlikely

Genetic Counseling

Figure 1. Flowchart depicting IRT/IRT1[/DNA screening algorithm and failsafe algorithm to identify infants with 2 extremely elevated IRT levels and no CF mutations. Recommended sweat chloride reference values in infancy are in concordance with current diagnostic guidelines.18

ond specimen is tested for CFTR mutations with the Asuragen Signature CF 2.0 ASR assay with 43 mutations, including mutations specific to the Hispanic community. Infants with 1 or 2 CFTR mutations undergo a sweat test to confirm or rule out the diagnosis of CF. Infants with 2 extremely elevated IRTs (>150 ng/mL, approximately 99.9th percentile) and 0 mutations are considered to be at risk for CF and are referred to undergo a sweat test, with a variation of a failsafe model first adopted in Massachusetts.13 Linking of First and Second Specimens A new laboratory information system (SpecimenGate, Perkin-Elmer Wallec, Turku, Finland) with linking of first and second screening specimens for each baby is in use in the Colorado Department of Public Health and Environment newborn screening laboratory. A unique identification number links the cards for the first and second collection. The mother of the infant is given the card for the second collection and is instructed to bring it back at the 2-week well baby check. Supplemental cards are available for use when the first card is not returned, and then the linking is

Determination of IRT Cutoffs Data between Mar 1, 2006, and Jul 1, 2008, from the Colorado program were used to calculate approximate IRT percentiles from all babies with a newborn screen collected between 1 and 3 days of age. The percentiles were comparable with cutoff points used by other programs.13,14 The projected number of children who would be recalled for sweat tests and identified as heterozygote carriers of a CFTR mutation, by using each of 3 algorithms (IRT/IRT, IRT/DNA, and IRT/IRT1[/DNA) were compared.12-14 The IRT cutoff for IRT/IRT[/DNA was determined with retrospective data from 2005 to 2006, choosing a level that was less than that of identified cases, resulting in a cutoff level of 60 ng/mL, approximately 97th percentile. This is comparable with other programs, and this level would diagnosed CF in all infants identified through the newborn screening to date, including missed cases.

Determination of Projected Sensitivity with Examination of Earlier Colorado Data Retrospective data from children with CF born in Colorado identified with newborn screening (and known missed cases in which CF was diagnosed conventionally) were studied with the new algorithm. Projected missed cases for IRT/ IRT and IRT/IRT1[/DNA algorithms with IRT cutoff levels ranging from 55 (approximately 96 percentile) to 105 (approximately 99.7 percentile) and 3 different mutation panels (F508, 23 mutations recommended by American College of Medical Genetics [ACMG]15, and 43 Asuragen mutation panel) were calculated. The projected number of sweat tests per year for IRT/IRT was approximated with the historic averaged Colorado recall rate of 0.2% for infants with positive results on first and second screenings.12 The projected number of sweat tests for IRT/IRT1[/DNA was approximated by using 3% for the recall rate of the first screen and estimating 20% of children with an elevated first screen level would have an elevated second screen level by using historic data, with an estimated carrier rate of 1 of 20.8,10,11,16 A higher carrier rate could be hypothesized in infants with persistent hypertrypsinogenemia, so a carrier rate of 1 of 15 was used to project the number of carriers that would be identified with IRT/IRT[/DNA. Approval from the Colorado Multiple Institutional Review Board was obtained for these analyses. 619

THE JOURNAL OF PEDIATRICS



www.jpeds.com

Results The primary goal for a screening algorithm for CF is to identify cases, therefore minimizing the number of false negative results. To meet this goal, we investigated a lower initial IRT cutoff from a fixed value of 105 ng/mL. Examining the IRTs of the 312 infants with CF (non-meconium ileus) diagnosed with CF in Colorado with the IRT/IRT-based newborn screening approach between 1982 and 2007 and 20 additional infants who were initially missed by the newborn screening, we determined that a cutoff level of 60 ng/mL (approximately 97th percentile) would have identified all known cases. During the same time, CF was diagnosed in 68 infants (17% of all infants with CF) after presentation with meconium ileus. These infants are not included in this report. More than 1 134 000 infants were screened during this period. Projected missed cases and the number of corresponding sweat tests on the basis of each proposed screening algorithm are presented in the Table and Figure 2. All infants who have been identified with CF in the state of Colorado had IRTs >60 ng/mL, resulting in no known missed cases on the basis of the IRT cutoff level of 60 ng/mL with IRT/IRT1[/DNA. Limiting the DNA panel to only the DF508 mutation would result in a missed case rate of at least 8%, using a cutoff of 60ng/ml, and up to 13% with a cutoff of 105 ng/mL. Expanding the panel to 43 mutations and lowering the IRT cutoff level improves the sensitivity. Three infants of 288 genotyped, non-meconium ileus infants (1.0%; 95% CI, 0.3-2.7) who were identified with the current IRT/IRT protocol would not have been identified with the mutation panel proposed in the new IRT/ IRT1[/DNA algorithm. Two of these missed cases are Hispanic, and the third has Native American ancestry. The genotypes of these infants are 1898 + 5 G->T/Unidentified, 3500-2A->G/Unidentified, and I506 T/Unidentified. Two of the missed cases would have been identified with the ‘‘failsafe’’ model,13 by which all infants who have 2 IRT levels >150 ng/mL for both of the 2 screens would undergo a sweat test. The cutoff level of 150 ng/mL is approximately the 99.9th percentile on the first screen, and on the second screen, 150 ng/mL is approximately the 99.4th percentile. Only 0.2% of children without CF who had IRT levels >150 ng/mL on the first screen had

Vol. 155, No. 5 an IRT level >150 ng/mL on the second screen, corresponding to <1 child per 10 000 screened every year that will be tested with the failsafe protocol. This IRT/IRT1[/ DNA approach with a failsafe model would result in a projected missed case rate of <0.3% (95% CI, 0.07-1.6), significantly better than the historic missed case rate of IRT/IRT of 5.2% (95% CI, 3.1-8.0). The costs of IRT/IRT[/DNA per 10 000 births was less than either IRT/IRT or IRT/ DNA (Table).

Discussion Our projected results indicate that IRT/IRT1[/DNA provides a better alternative to IRT/IRT in states with 2 screening tests and has advantages compared with both IRT/IRT and IRT/DNA. The sensitivity of IRT/IRT1[/ DNA with an initial cutoff level of 60 ng/mL (approximately 97 percentile) is projected to be >99.5%, which corresponds to 1 potential missed case of every 700 000 live births. This is an improvement compared with the missed case rate for IRT/ IRT in Colorado of at least 5%,12 and other missed cases may have yet to be diagnosed. The projected missed-case rate of <1% includes the benefits of the failsafe model. The population in Colorado is enriched with Hispanic subgroups that make up >30% of current births in the state.17 The mutation panel we are using was chosen to have better sensitivity for detecting CFTR mutations seen in the Hispanic population; however, the coverage of this panel in the Hispanic population (72%) is lower than the standard American College of Medical Genetics (ACMG) panel in the Caucasian poulation (88.1%; personal communication, Martin Kharrazi, PhD, 6/20/08). All projected missed cases in Colorado with IRT/IRT1[/DNA are in minority groups, resulting in a biased screening test. Other newborn screening programs are currently considering moving away from a failsafe model after unacceptable referral-to-case ratios (personal communication, Anne Comeau, PhD, Boston, Masschusetts, 6/20/08). We have shown that assessing IRT twice before performing CFTR mutation analysis in Colorado decreases the referral rate to <1 infant expected per every 10 000 births. This additional referral rate and sweat testing

Table. Projected results from 3 different cystic fibrosis newborn screening mechanisms per 10 000 births

IRT/IRT† IRT/DNAz IRT/IRT1[/DNAz

Number of IRT determinations (1st screen/ 2nd screen)

Number of mutation analyses performed

Number of sweat tests performed

Heterozygotes identified

Costs per 10 000 births*

10 000/10 000 10 000/NA 10 000/300

0 300 45

9 18x 6{

NA 15 3

$36 350 $38 446 $21 662

NA, Not applicable. *Costs estimated on the basis of 2007 to 2008 actual costs in US dollars: $1.75 per IRT, $60.82 per mutation analysis, $150 per sweat test. †Cutoff levels: 105 ng/mL (99.7th percentile) and 70 ng/mL. zCutoff level at 97th percentile (estimated from 2005-2006 Colorado data). xCarrier rate approximately 1 of 20 infants plus 3 expected infants with CF per year. {Carrier rate approximately 1 of 15 infants plus 3 expected infants with CF per year.

620

Sontag et al

ORIGINAL ARTICLES

November 2009

14%

100 Missed Cases IRT/IRT↑/F508

10%

80

IRT/IRT↑/ACMG 23 8%

IRT/IRT↑/Asuragen 43 60

IRT/IRT 6%

40 4% 20 2%

Number of Sweat Tests per 10000 live births

IRT/IRT

12%

Missed Case Rate (%)

120

Number of Sweat Tests IRT/IRT/DNA

0

0% 55

60

65

70

75

80

85

90

95

100

105

IRT(ng/ml) CutOff

Figure 2. Projected missed case rate for Colorado with 4 screening algorithms (IRT/IRT, IRT/IRT[/F508 only, IRT/IRT[/ ACMG 23 panel, IRT/IRT[/Asuragen 43), with IRT cutoff levels varying from 60 to 105 ng/mL, and associated number of annual sweat tests for an IRT/IRT algorithm or overall IRT/ IRT1[/DNA algorithm. (The number of sweat tests was similar for all mutation panel algorithms and therefore were combined.)

that will be performed is acceptable to minimize the missed cases in minorities. IRT/IRT1[/DNA identifies a smaller number of heterozygotes compared with IRT/DNA.13,14 We project that we will identify <3 heterozygotes per year of 10 000 infants screened, compared with 13 with IRT/DNA. Although some in the community with CF view the identification of carriers as beneficial, it is not an intended outcome of newborn screening. The considerable number of carriers identified with IRT/DNA puts a significant burden on the genetic counseling community, because carriers are identified typically at a rate of 1/20-1/25 positive IRTs,8,10,11,16 which is higher than the population carrier rate of 1/30-1/35. Population screening currently offered to couples prenatally or preconception is a preferred mechanism for identifying carriers.15 To perform IRT/IRT1[/DNA, we have shifted costs from IRT assays to mutation analysis. Historically, Colorado has tested the IRT levels of every newborn screening specimen (both the first and the second). With new technology, we perform IRT determinations on all first-screen samples, but only 3% of the second samples. The savings from the IRT analysis are redirected into CFTR mutation analysis and have resulted in a cost savings >$10 000 for every 10 000 babies screened. The IRT/IRT1[/DNA model also decreases the sweat tests and counseling required by recalling approximately 5 infants per 10 000 live births.

IRT/IRT recently has been adopted by other screening programs. We strongly recommend that programs consider this modification of IRT/IRT and add CFTR mutation analysis to their programs. IRT/IRT1[/DNA can improve sensitivity of the newborns with CF screen, while minimizing identification of carriers. n Submitted for publication Aug 29, 2008; last revision received Jan 30, 2009; accepted March 26, 2009. Reprint requests: Marci K. Sontag, PhD, The Children’s Hospital, 13123 E 16th Ave, B395, Aurora, Colorado, 80045. E-mail: [email protected].

References 1. Crossley JR, Elliott RB, Smith PA. Dried-blood spot screening for cystic fibrosis in the newborn. Lancet 1979;1:472-4. 2. Wilcken B, Brown AR, Urwin R, Brown DA. Cystic fibrosis screening by dried blood spot trypsin assay: results in 75 000 newborn infants. J Pediatr 1983;102:383-7. 3. 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;325:769-74. 4. National Newborn Screening and Genetics Resource Center. Available at: http://genes-r-us.uthscsa.edu/. Accessed Aug 1, 2007. 5. Riordan JR, Rommens JM, Kerem B, Alon N, Rozmahel R, Grzelczak Z, et al. Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA [published erratum appears in Science 1989 Sep 29;245(4925):1437]. Science 1989;245:1066-73. 6. Farrell PM, Aronson RA, Hoffman G, Laessig RH. Newborn screening for cystic fibrosis in Wisconsin: first application of population-based molecular genetics testing. Wis Med J 1994;93:415-21. 7. Ranieri E, Ryall RG, Morris CP, Nelson PV, Carey WF, Pollard AC, et al. Neonatal screening strategy for cystic fibrosis using immunoreactive trypsinogen and direct gene analysis. BMJ 1991;302:1237-40. 8. Corbetta C, Seia M, Bassotti A, Ambrosioni A, Giunta A, Padoan R. Screening for cystic fibrosis in newborn infants: results of a pilot programme based on a two tier protocol (IRT/DNA/IRT) in the Italian population. J Med Screen 2002;9:60-3. 9. Pollitt RJ, Dalton A, Evans S, Hughes HN, Curtis D. Neonatal screening for cystic fibrosis in the Trent region (UK): two-stage immunoreactive trypsin screening compared with a three-stage protocol with DNA analysis as an intermediate step. J Med Screen 1997;4:23-8. 10. Wilcken B, Wiley V, Sherry G, Bayliss U. Neonatal screening for cystic fibrosis: a comparison of two strategies for case detection in 1.2 million babies. J Pediatr 1995;127:965-70. 11. 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;52:616-26. 12. Sontag MK, Hammond KB, Zielenski J, Wagener JS, Accurso FJ. Twotiered immunoreactive trypsinogen-based newborn screening for cystic fibrosis in Colorado: screening efficacy and diagnostic outcomes. J Pediatr 2005;147:S83-8. 13. Comeau AM, Parad RB, Dorkin HL, Dovey M, Gerstle R, Haver K, et al. 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. Pediatrics 2004;113:1573-81. 14. Rock MJ, Hoffman G, Laessig RH, Kopish GJ, Litsheim TJ, Farrell PM. Newborn screening for cystic fibrosis in Wisconsin: nine-year experience with routine trypsinogen/DNA testing. J Pediatr 2005;147:S73-7. 15. Watson MS, Cutting GR, Desnick RJ, Driscoll DA, Klinger K, Mennuti M, et al. Cystic fibrosis population carrier screening: 2004

A New Cystic Fibrosis Newborn Screening Algorithm: IRT/IRT1[/DNA

621

THE JOURNAL OF PEDIATRICS



www.jpeds.com

revision of American College of Medical Genetics mutation panel. Genet Med 2004;6:387-91. 16. Scotet V, De Braekeleer M, Audrezet MP, Lode L, Verlingue C, Quere I, et al. Prevalence of CFTR mutations in hypertrypsinaemia detected through neonatal screening for cystic fibrosis. Clin Genet 2001; 59:42-7.

622

Vol. 155, No. 5 17. Colorado Births and Deaths 2006. Health Statistics Section, Colorado Department of Public Health and Environment; Denver, Colorado, 2007. 18. Farrell PM, Rosenstein BJ, White TB, Accurso FJ, Castellani C, Cutting GR, et al. Guidelines for diagnosis of cystic fibrosis in newborns through older adults: Cystic Fibrosis Foundation consensus report. J Pediatr 2008;153:S4-14.

Sontag et al