J Inherit Metab Dis (2010) 33 (Suppl 2):S255–S261 DOI 10.1007/s10545-010-9117-3
NEWBORN SCREENING
Cystic fibrosis newborn screening: using experience to optimize the screening algorithm Jaime E. Hale & Richard B. Parad & Henry L. Dorkin & Robert Gerstle & Allen Lapey & Brian P. O’Sullivan & Terry Spencer & William Yee & Anne Marie Comeau
Received: 19 January 2010 / Revised: 16 April 2010 / Accepted: 20 April 2010 / Published online: 3 June 2010 # SSIEM and Springer 2010
Abstract Newborn screening (NBS) for cystic fibrosis (CF) offers the opportunity for early diagnosis and improved outcomes in patients with CF and has been universally available in the state of Massachusetts since 1999 using an immunoreactive trypsinogen (IRT)-DNA algorithm. Ideally, CF NBS is incorporated as part of an integrated NBS system that allows for comprehensive Communicated by: Bridget Wilcken Competing interest: None declared. J. E. Hale : A. M. Comeau (*) New England Newborn Screening Program, UMass Medical School, 305 South Street, Jamaica Plain, MA 02130, USA e-mail:
[email protected] R. B. Parad Brigham and Women’s Hospital, Boston, MA, USA H. L. Dorkin : T. Spencer Children’s Hospital, Boston, MA, USA R. Gerstle Baystate Medical Center, Springfield, MA, USA A. Lapey Massachusetts General Hospital, Boston, MA, USA B. P. O’Sullivan : A. M. Comeau Department of Pediatrics, UMass Medical School, Worcester, MA, USA W. Yee Tufts Medical Center, Boston, MA, USA
and coordinated education, laboratory screening, clinical follow-up, and evaluation so that evidence-based data can be used to maximize quality improvements and optimize the screening algorithm. The New England Newborn Screening Program (NENSP) retrospectively analyzed Massachusetts’s CF newborn screening data that yielded decisions to eliminate a screen-positive category, maintain the IRT cutoff value that prompts the second tier DNA testing, and communicate CF relative risk to primary care providers (PCPs) based on categorization of positive CF NBS results.
Introduction Newborn screening (NBS) for cystic fibrosis (CF) offers the opportunity for early diagnosis and improved outcomes in patients with CF. Following the endorsement from the United States Centers for Disease Control and Prevention (CDC) that newborn screening for CF (CF NBS) is justified (Grosse et al. 2004) and the recommendation from the United States CF Foundation (US CFF) endorsing CF NBS (CFF 2009), CF was included among the 29 core disorders proposed in 2005 for the United States national uniform panel by the American College of Medical Genetics (ACMG)/ Health Resources and Services Administration (HRSA) (Watson et al. 2006). Presently, all 50 U.S. states are screening infants for CF (Therrell 2009). CF NBS has been universally available in the state of Massachusetts (MA) since 1999 through the New England Newborn Screening Program (NENSP) and a statewide pilot research program offering optional expanded NBS services (Atkinson et al. 2001). In February 2009, after 795,206 MA infants had undergone CF NBS through the MA pilot program, CF NBS became mandatory (Code of Massachusetts Regulations 2008).
S256
J Inherit Metab Dis (2010) 33 (Suppl 2):S255–S261
Ideally, CF NBS is incorporated as part of an integrated NBS system that allows for comprehensive and coordinated education, laboratory screening, clinical follow-up, and evaluation to maximize quality improvements (Comeau et al. 2007). The specific protocols of the MA CF NBS program algorithm (pre-analytic, analytic, and postanalytic) have evolved as a result of such on-going quality assurance. The MA CF NBS program employs a two-tiered laboratory-screening algorithm with an initial immunoreactive trypsinogen (IRT) assay and subsequent multiple CFTR mutation testing (IRT-DNA) (Comeau et al. 2004) (Fig. 1). Infants with a positive CF NBS result are referred for diagnostic testing for sweat chloride concentration by quantitative pilocarpine iontophoresis (sweat testing) and appropriate genetic counseling and/or therapeutic services at a CF center accredited by the US CFF. Rigorous followup of positive CF NBS results provides data for evidencebased algorithm assessment and optimization including optimization of positive predictive value (PPV) without an excessive increase in the false negative rate, and clarification of result interpretation in communications to primary care providers (PCPs) and families. We report a retrospective evaluation of 9 years of CF NBS data yielding Initial Dried Blood Spot Specimen1 IRT assay2 Elevated IRT3
SCREEN NEGATIVE
DNA assay4
STOP
0 mutation detected
1 or 2 mutations detected
failsafe protocol 5 CF SCREEN POSITIVE IRT < ultrahigh CF-SCREEN NEGATIVE
IRT>ultrahigh CF-SCREEN POSITIVE
STOP
evidence for (1) the decision to maintain the IRT value prompting the second tier DNA testing at the 95th percentile, (2) the categorization of risk of CF relative to specific CF NBS results, (3) the utility of expanding multimutation panels, and (4) the decision to eliminate the “failsafe” screen-positive category from the MA CF NBS algorithm.
Methods Figure 1 shows the laboratory screening algorithm with a recording of algorithm changes implemented throughout the MA CF NBS pilot program. Each infant’s CF NBS data from the NENSP were provided to the infant’s healthcare provider as previously described (Comeau et al. 2004, 2007) and in compliance with the United States Health Insurance Portability and Accountability Act and state privacy regulations. CF centers provided specific outcome data to the NENSP (including diagnostic laboratory values), where the individuals’ outcome data were linked to the individuals’ screening values. For the duration of the pilot program, NENSP staff retrospectively analyzed centralized identified data, and aggregate reports were presented to the CF Workgroup (NENSP and CF center directors) for advice and decision-making. For the current report, data generated since February 1999 were analyzed for outcomes as though all specimens had been processed using the IRT/DNA algorithm in which only the top 5% of IRT percentiles prompted DNA testing. Enrolled subjects included 712,232 of 718,071 infants whose newborn screening samples were collected through the MA NBS program from February 1, 1999, through January 31, 2008, and whose mothers did not opt out of the CF NBS option. The reported racial characteristics of the birth cohort between 1999 and 2008 show a majority of non-Hispanic whites (75% in 1999 to 68% in 2007) and other racial trends as follows: non-Hispanic blacks (7 to 8%); Asians (5 to 7%) and Hispanics (11 to 14%) (Massachusetts Department of Public Health 2009).
Refer to CFF Center
Fig. 1 Evolution of the IRT/DNA algorithm in Massachusetts cystic fibrosis (CF) newborn screening (NBS) program. 1 Massachusetts guidelines recommend obtaining the NBS specimen at between 24 and 72 h of age. 2 Wallac DELFIA kits (Turku, Finland). 3 From February 1999 to November 1999, cutoff set at ≥90th percentile; from November 1999 cutoff set at ≥95th percentile. 4 From February 1999 to November 2000, linear array CF-16 (Roche Diagnostics); from November 2000 to September 2003, linear array CF-31 (Roche Diagnostics); from September 2003 to November 2004, linear array CF-gold (Roche Diagnostics); from November 2004 to present, Tag-It 39+4 hid (TM BioSciences and Luminex). 5 Terminated Feb. 1, 2009. February 1999 to November 2003 ultrahigh cutoff set at 99.8th percentile across values from previous 30 days; from November 2003 through January 31, 2009, ultrahigh cutoff set at 99.9th percentile across values from previous 30 days
Results Cumulative screening yield From February 1, 1999, through January 31, 2008, 712,232 infants were screened for CF by the MA CF NBS Program (Fig. 1). Of these, 709,471 (99.6%) infants received inrange (negative) CF NBS results and 2,761(0.4%) received out-of-range (positive) CF NBS results. Among all infants with positive CF NBS results, 141 had two CFTR mutations detected by the CF NBS, 1,976 had one mutation detected, and 644 had “failsafe” results.
J Inherit Metab Dis (2010) 33 (Suppl 2):S255–S261
Compliance with diagnostic recommendations Of the 2,761 infants with positive CF NBS results, 91% were evaluated at a CF center, 2% died prior to evaluation, and 7% did not complete a diagnostic evaluation [1% refusal, 1% had tests with insufficient sweat (quantity not sufficient, QNS) that were not re-done, and 5% pending or lost to follow up]. A CF diagnosis was confirmed with a sweat chloride≥60 MEq/L or by a CF clinician. Cumulative screening outcomes Of the 2,761 infants with positive results, 200 have been diagnosed with CF, yielding a case-to-referral ratio of 1 in 14. Of the 709,471 infants with negative screening results, 5 were later diagnosed with CF, yielding a negative predictive value of 99.99%; an additional 4 were followed at a CF center but do not carry a CF diagnosis. Data used to evaluate IRT value that prompts for DNA analysis Table 1 shows that all five false-negative CF NBS results were attributable to the IRT value of the infant being below the 95th percentile prompt for DNA testing; none were attributable to the DNA testing panel. The one infant who had a false-negative CF NBS with an IRT value close to the 95th percentile also had meconium ileus, a neonatal clinical presentation requiring a diagnostic evaluation. All other infants with false-negative CF NBS results had IRT values
S257
well below the 95th percentile and at a value too low to use in NBS while maintaining reasonable positive predictive values. Table 1 also shows the additional infants followed at CF center who do not yet carry a diagnosis of CF. Worldwide, CF NBS programs utilize a variety of IRT values to prompt DNA analysis. We investigated the feasibility of increasing the IRT value prompting DNA analysis to increase the specificity of the screen. Table 2 demonstrates that increasing the IRT percentile value used to prompt DNA analyses decreases screening sensitivity. Using a prompt at the 99th percentile, an additional 27 CF affected infants would have had false-negative CF NBS results (9 original + 27 = 36 total ) and missed the opportunity for early diagnosis; 20 of these 27 infants had sweat chloride values or genotypes that would meet the US CFF 2008 diagnostic criteria for CF. Only 4 of these 20 infants meeting CFF diagnostic criteria presented with meconium ileus. The MA CF Working Group determined that we should keep the IRT prompt at 95%. Data used to enhance communication of relative risk for the CF NBS results showing elevated IRT and one CFTR mutation Of the 2,761 infants identified with a positive CF NBS, the vast majority (1,976) had CF NBS results showing an elevated IRT concentration and one mutation. These results can be among the most challenging to communicate to both PCPs and families because most of these infants will not have CF and ultimately will be shown to be carriers. Extensive follow-up
Table 1 Children who are followed at a cystic fibrosis (CF) center who were not identified by CF newborn screening (NBS) Presentation
Status at last update
NBS IRT%, age at dx
Genotype
Sweat [Cl−] (MEq/L)a
84.2%, 3 months
DF508/R117H
67
Five CF infants with false-negative CF NBS results FTT, upper respiratory infections, chronic cough
Pancreatic sufficient, sinus disease, positive cultures for Staph. aureus and H. flu
Meconium ileus
93.9%, birth
G542X / unknown
57.7, 67.4
FTT, recurrent pneumonia, asthma
62.3%, 4 years
D828G / 3271+18 C or T
62
Asthma
78.6%, 3 years
D1270N / R74W
86.5
Chronic cough and sinusitis
74.1%, 4 years
R75Q / unknown (second mutation not identified by sequencing)
82, 68
39.7%, 5 years
DF508 / unknown
39
94.6%, 3 years
DF508/R117H
56
21.3%, 71.2% (sib)
DF508 / R117H
20, not done
Four additional infants followed at CF center who do not (yet) carry a CF diagnosis Chronic cough Chronic cough; sweat-tested and genotyped after parents found to be carriers during pregnancy with younger sibling Two siblings who are well; genotyped for family history
Pancreatic sufficient, asthma, moderate Staph. aureus and H. flu Does not carry CF diagnosis, pancreatic sufficient, exercise-induced asthma, normal PFTs, cultures Staph. aureus Positive cultures for Staph. aureus and H.flu
IRT Immunoreactive trypsinogen, FTT failure to thrive, PFT pulmonary function test a
Value(s) reported from independent visits
S258
J Inherit Metab Dis (2010) 33 (Suppl 2):S255–S261
Table 2 Implications of using particular IRT percentiles as prompts for DNA analyses IRT percentile prompting DNA analysis
Screen positivesa
True positivesb
False negativesc Total CF affected
CF affected by CFF 2008 criteria
Identified as CF affected but do not meet CFF 2008 criteria
25 (16) 16 (10) 11 (6) 7 (4)
11 9 8 6
5 (3)
4
≥99 ≥98 ≥97 ≥96
1,396 1,890 2,286 2,596
173 184 190 196
36 25 19 13
≥95
2,761
200
9 (4)
(17) (12) (9) (6)
IRT Immunoreactive trypsinogen, CF cystic fibrosis, CFF Cystic Fibrosis Foundation a
Of 712,232 infants screened, number of MA infants with positive CF NBS relative to IRT percentile
b
Number of CF cases who would be identified by the CF NBS relative to IRT percentile
c
Number of MA CF-affected infants who would have had false-negative CF newborn screening results relative to the IRT percentile prompting DNA analyses (% false negative)
of infants with positive CF NBS results has allowed the MA CF NBS program to incorporate communication of relative risk of CF following a positive NBS result that is based upon combined consideration (multi-analyte profiling) of both the IRT concentration and the screening-genotype results. Of these 1,976 infants, 1,872 (95%) had a CF diagnostic evaluation outcome reported to the MA CF NBS program by January 4, 2010. Although the overall risk that a particular infant with one mutation detected through CF NBS will have CF is 0.03 (1 in 34) (95% CI: 0.02–0.04), the actual risk for a particular multi-analyte profile ranges from 0.34 to 0.01, based on categorization into one of five strata of IRT concentrations (>150, 126–150, 101–125, 75–100, or