Improving test properties for neonatal cystic fibrosis screening in the Netherlands before the nationwide start by May 1st 2011

High throughput techniques made neonatal screening for many diseases technically possible after the year 2000. New technical possibilities need attunement by all parties involved, such as scientists in laboratories and clinics, physicians and other professionals in (public) health care, patients (organizations) and regulatory, advisory and governmental agencies (Achterbergh et al. 2007).

In The Netherlands the Health Council installed a commission to study the pros and cons of potential neonatal screening possibilities in 2005, from the point of view of a diversity of backgrounds, including paediatrics, genetics, ethics and law (Health Council of the Netherlands. Neonatal Screening. The Hague: Health Council of the Netherlands 2005). The report led to the extension of the program to 17 disorders, not including cystic fibrosis. The fact that cystic fibrosis treatment confers a substantial health gain was considered beyond dispute. The question was, however, how much screening immediately after birth contributes to this health gain? The committee considered cystic fibrosis a “borderline case”, as the evidence of health gain was less compelling than in the case of disorders such as phenylketonuria and hypothyroidism. The advantages of neonatal screening for cystic fibrosis, namely a better nutritional status, prevention of an often protracted and aggravating diagnostic process and a decrease in the number of incidents of sickness and hospital admissions were considered clear (Health Council of the Netherlands. Neonatal Screening. The Hague: Health Council of the Netherlands 2005). The imperfections of the screening were however the main reason not to advise to include cystic fibrosis in the neonatal screening program, but rather to undertake research into better screening methods. In the 2005 Health Council report (Health Council of the Netherlands. Neonatal Screening. The Hague: Health Council of the Netherlands 2005), an IRT/DNA screening program was estimated to lead to the diagnosis of 50-60 patients per year, 600 infants referred for sweat test, and 400 heterozygotes diagnosed (carriers of CF). The Health Council recommended to include cystic fibrosis in neonatal heel prick screening as soon as a test method would become available with a high specificity (as low specificity leads to a great number of clinical investigation of unaffected neonates). A study to improve the test properties of cystic fibrosis neonatal screening was indeed initiated: CHOPIN (Cystic fibrosis heelprick screening in a newborn population in the Netherlands).

A pilot study to improve test properties was performed in 2008 and 2009 in four provinces of the Netherlands: Utrecht, Gelderland, Noord-Brabant and Limburg. The study was led by two of us (JD and AV). Two screening strategies were compared: measuring the serum concentration of immunoreactive trypsinogen (IRT) followed by pancreatitis associated protein (PAP) (Sarles et al. 2005) vs IRT followed by a DNA test panel of 35 mutations in the cystic fibrosis transmembrane regulator (CFTR) gene¹ and, if one CFTR mutation was found, extended gene analysis (EGA; DNA sequencing of all coding exons and intron/exon boundaries of the CFTR gene). In 2011 the results of the study have been accepted for publication (Vernooij-van Langen et al. 2011), but a short summary of the results was already presented in a 2010 Health Council report (Health Council of the Netherlands. Neonatal screening for cystic fibrosis. The Hague: Health Council of the Netherlands 2010). Apart from the two strategies followed in the pilot, a third scenario has been calculated, combining the two strategies. IRT/PAP leads to a relatively large number of referrals (119 as compared to 20 in the IRT/CFTR mutation panel/EGA strategy). The specificity of IRT/CFTR mutation panel/EGA is higher but this strategy will also identify carriers and less severe forms of CF, as can be seen in Table 1. PAP as a second step added to the IRT/CFTR mutation panel/EGA strategy will reduce the number of carriers and less severe forms. Thus IRT/PAP/CFTR mutation panel/EGA combines the best screening test properties.

The Health Council again asked a committee to advise on CF newborn screening after the CHOPIN study had been finished (Health Council of the Netherlands. Neonatal screening for cystic fibrosis. The Hague: Health Council of the Netherlands 2010). Apart from the strategies above, the committee discussed the IRT limit (≥50 μg/l = the 2.43% highest concentrations vs. ≥60 μg/l = the highest 1.03%), and considered that an increase of the limit for IRT to 60 μg/l produces a sharp drop in the number of non-classical CF patients identified by the screening. Also, in case of IRT ≥ 100 μg/l and PAP ≥ 1.6 μg/l, sequencing of the CFTR gene was included even if none of the 35 mutations was present, since then the risk of a rare mutation is relatively high. A four step protocol was proposed for the national program (Fig. 1):

The PAP cut off levels are identical to the ones earlier reported by Sarles et al. (2005). However, recently the manufacturer of the PAP kits has recognized a systematic error in his calibration procedure. This four step protocol uses the corrected values (Dagorn 2011).

This protocol was expected to identify 25 CF patients on an annual basis, additional to four infants already diagnosed because of meconium ileus (Health Council of the Netherlands. Neonatal screening for cystic fibrosis. The Hague: Health Council of the Netherlands 2010). Only 12 carriers of CF would be identified. Out of the 25 CF patients 23 were expected to have only mutations from the panel and two to be identified by extended mutation analysis. This protocol was proposed by the Health Council for national implementation, and accepted by the ministry of Health.

Due to the screening program CF patients will have a better nutritional status. An often protracted and aggravating diagnostic process is prevented, a decrease in hospital admissions anticipated and other benefits seem likely. The specificity is >99.99%.

While the specificity of CF neonatal screening was the main test property to be improved, according to the Health Council advice in 2005, also the sensitivity needed improvement. In view of the large number of rare mutations not included in the CFTR mutation panel a failsafe procedure was built in the protocol. Earlier a failsafe procedure was performed in Massachusetts using a multiple CF mutation screen, which added somewhat to the sensitivity, however at the cost of increased referrals and carrier identification (Comeau et al. 2004). In the current Netherlands procedure, the CFTR gene is sequenced in the last step. The procedure is especially important for infants with migrant ancestors, since the sensitivity of the CFTR mutation panel was only 44% for Turkish migrants and 69% for North African migrants in Europe (Health Council of the Netherlands. Neonatal screening for cystic fibrosis. The Hague: Health Council of the Netherlands 2010; Lakeman et al. 2008). The CF carrier frequency in Dutch inhabitants of Turkish and North African ancestry can only be roughly approximated, but is estimated to be around 1 in 50. The annual number of neonates with CF in these population groups is estimated to be 1-2. The failsafe procedure consisted of sequencing the CFTR gene in case of an IRT concentration of ≥100 μg/l and PAP ≥ 1,6 μg/l, also if no mutation was identified using the CFTR mutation panel. It is hard to estimate the sensitivity, but this is not expected to be 100%, due to IRT and PAP. The consequence of pursuing an extremely high specificity is in general a less optimal sensitivity. Based on the CHOPIN study the sensitivity might be ±95%.

While the goal of neonatal screening is to identify infants with two CFTR mutations, that lead to the classical form of cystic fibrosis, some infants will be diagnosed as CF carriers, having only one CFTR mutation. If the newborn child is a carrier, then it follows that one, or both, parents (and possibly other children) are carriers (Health Council of the Netherlands. Neonatal screening for cystic fibrosis. The Hague: Health Council of the Netherlands 2010). As this may be relevant information to the parents in connection with future family planning, they are informed of this secondary finding. Parents in this situation can be referred for genetic counselling. At the moment the heel prick is performed, parents can opt-out of receiving carrier status information. To allow parents to make an informed decision on (not) receiving carrier status information, the parents should be alerted to these possible outcomes prior to screening. Thus the information and counselling given to parents during pregnancy becomes more complex. After birth, information of this kind can, in practice, give rise to misunderstandings with regard to the health of the carriers.

In theory, genetic screening can be performed in different phases of life: before conception, during pregnancy, in the newborn or later in life. Before pregnancy (in preconceptional screening) more reproductive options exist, such as choosing not to have children (or adopting a child), preimplantation genetic diagnosis (embryo selection), prenatal diagnosis and termination of affected foetuses, choosing a different partner or using donor gametes (e.g. artificial insemination by donor sperm). However, in most countries preconceptional screening for cystic fibrosis is not available in health care.

The Punnett squares in Fig. 2 show the information that can be derived from the fact that a child of healthy parents is a CF carrier. If the child is a carrier, at least one parent is carrier as well. If none of the parents have the CF phenotype, the carrier parent must also have a normal CFTR gene. If the carrier frequency in a population is 1:30, the allele frequency is 1:60, and for each next pregnancy the risk of a CF affected infant is 1/4x1/60 = 1/240, much higher than the birth prevalence would be (1/3600). A CF carrier test in the parents might provide certainty. However, as the 2005 Health Council report (Health Council of the Netherlands. Neonatal Screening. The Hague: Health Council of the Netherlands 2005) mentions, “… it is not always possible to determine for certain whether only one parent is a carrier (as is the case with, for example, cystic fibrosis, where not all mutations are known).” This holds true when CF testing is based on a CFTR mutation panel. Sequencing of the CFTR genes of the parents will reduce the residual risk to a figure far lower than the population risk.

After the decision to start a national CF screening program, there are several questions that remain to be answered. Further evaluation of the sensitivity and the specificity achieved in the current four steps is needed. In an a different IRT/PAP study published last year the IRT/PAP strategy according to Sarles et al. (2005) showed some limitations, since not all CF patients with severe forms were found: PAP values of 3 out of 13 newborns with CF (23%) were equal to or below the cut-off level (Sommerburg et al. 2010). Further evaluation might lead to changes of the cut-off levels.