Newborn screening (NBS) is a public health programme for early diagnosis of treatable, mostly genetic, diseases, such as inherited metabolic diseases (IEM). Early treatment in a latent disease stage can prevent disease manifestations, which for diseases on many NBS panels include irreversible brain and organ damage and death [1-3]. Internationally, NBS was started > 50 years ago and in Denmark for phenylketonuria (PKU) in 1975. NBS for PKU became the paradigm for a well-functioning, cost-effective screening. It enabled expansion of the disorders screened for: in Denmark, hypothyroidism was added in 1977 and expanded newborn screening (eNBS) was added in 2002. Today, the panel includes 17 diseases (Table 1); see also Figure 1 for a Danish NBS timeline. Some countries screen for > 60 diseases, type and number varying, depending on demography and differences in how to select diseases for NBS [1, 4]. However, most refer to the Wilson and Jungner criteria (Table 2) [5]. This manuscript focuses on eNBS for IEM in Denmark using mostly tandem mass spectrometry and presents current experiences after screening of nearly a million newborns.

Methods

We included all children born from 1 February 2002 to 12 February 2019 [6]. The period included a pilot period until 2 February 2009, with eNBS done following written informed consent, followed by eNBS done in a routine NBS programme with informed dissent, i.e. active withdrawal from screening after information was provided (Figure 1). The percentage screened changed during the period from 65% at the start to 85% at the end of the pilot period and during the routine period 99.85%. A total of 967,780 newborns were screened with eNBS in the full period with 82,930 (8.6%) remaining unscreened.

We included a historic cohort of children born from 1 January 1992 to 31 December 2001, diagnosed clinically with a disease from the eNBS panel.

The diseases included in the screening panel changed during the reported period as shown in Figure 1 and Table 3. Decision on the current panel was based on results from the pilot period and a review including scoring of the screening potential for selected diseases [7]. All diseases with a scoring > the 75th centile (and some from the 50-75th centile) were reviewed further and a final list of panel diseases was made, which was effective from February 2009 (Figure 1 and Table 1) [7].

Samples consisted of capillary blood collected at local hospitals by heel prick, spotted on filter paper, dried and sent to the screening laboratory (Statens Serum Institut (SSI)). Samples were taken postpartum on day 4-9 in the pilot period (the time of sampling was used routinely until 2009) and at 48-72 hours during routine eNBS [7]; a result was available 2-6 days after sampling. Filter paper blood samples were stored in the Danish Newborn Screening Biobank, SSI. The Center of Inherited Metabolic Diseases, Copenhagen (CIMD) reported newborns with screen-positive results to the local paediatric department, which admitted the child and obtained samples for confirmatory testing. The analyses used at screening and for confirmatory testing are listed in Table 3. Using this infrastructure, the minimal time from birth to start of treatment was 11 days. However, newborns with convincing results and an acutely presenting disorder, like tyrosinemia, were referred directly to the CIMD once results from the SSI became available, making treatment possible at day five.

True-positive results came from newborns in whom or in whose mothers the suspected disease was diagnosed by confirmatory testing. False-positive results were from newborns in whom the suspected disease was not confirmed in the child or the mother. False-negative results were from children with negative results by eNBS who were born in the screening period and had been diagnosed clinically or otherwise, e.g. by family studies, with a disease in the eNBS panel.

Management of children with true-positive results or from the historic cohort took place at the CIMD, serving all of Denmark, the Faroe Islands, and Greenland, making it likely that most children born in this area and diagnosed with diseases in the screening panel are known to the authors. Evaluation of clinical status at last follow-up aimed at disclosing clinically significant sequelae to the disease (Table 4). The clinical evaluation was done longitudinally by the same metabolic physician throughout the period. The evaluation was performed in collaboration with local paediatricians – for further details, see [7].

Screening of the first half million newborns has been published previously, including analytical details [7]. The study was approved by the Ethics Committee (KF 01-152/98).

Trial registration: not relevant.

Results

A total of 603 samples from 967,780 newborns were screen positive. Of the 603 samples, 249 were true positives and 354 were false positives (Table 3). Patients with true-positive results included 222 newborns and 27 mothers, corresponding to 1:3,887 screened newborns. Biallelic variants were confirmed in relevant genes in all patients with a true positive result. Most variants were convincingly pathogenic; for MCADD, 20% had an uncertain/mild genotype (like the c.199C > T mutation) [2, 7]; for IVA, the “mild” c.932C > T mutation was found in three of four children; in biotinidase deficiency, 36 of 45 children had the “mild” c.1330 G > C mutation and a partial enzyme deficiency.

Positive predictive value (PPV) calculated for the entire screening period was 41%; 15% at start of eNBS and 62% in 2018. The false positive rate (FPR) was 0.036% overall (1:2,781 screened); 0.05% at the initiation of eNBS; and 0.024% (1:4,166 screened) in 2018. Variations in PPV and FPR were seen for different diseases; e.g., PPV was 100% for LCHADD and 11 % for MMA/PA (Table 3). A total of 16 children had false-negative results, 11 of whom had a fatty acid oxidation disease (FAOD), mostly CTD in the beginning of eNBS screening for which the screening algorithm was adjusted [7].

The overall prevalence of the diseases included in the programme was 1:3,900. In the decade preceding eNBS, the prevalence for the same diseases was 1:8,300. The increase in prevalence was uneven between diagnoses: MCADD had a prevalence of 1:9,875 compared with 1:35,513 in the decade before eNBS [2, 7]; VLCADD, 3-MCCD, GA1, BIOTD and ASLD had less pronounced increases, while the prevalence of LCHADD, MMA/PA and HT1 was unchanged.

Eighteen of the 222 affected newborns (7.3%) had symptoms at the time of screening, of whom two died (MCADD and MADD), and one had a cardiac arrest with full recovery (MCADD) (Table 4); all newborns with MSUD and citrullinemia and 50% with MMA/PA were symptomatic. On follow-up, 16 children (6.5%) had symptoms related to the disease. Thus, at the last follow-up, 93% of children had no clinically significant sequelae (except for biochemical abnormalities and infrequent rhabdomyolysis and asymptomatic retinopathy in LCHADD). Most children with FAOD, BIOTD, ASLD and HT1 are healthy. A significant percentage of patients with MMA/PA (n = 3/7, 40%) and GA1 (n = 3/11, 30%) developed complications (Table 4).

Among the 82,930 non-screened newborns, 27 had eNBS panel diseases (Table 3). Of these, two had MCADD, both presenting in metabolic crises; two had LCHADD, one of whom died, and one had severe cardiomyopathy; one GA1 child had a dystonic crisis and one PA child a metabolic decompensation, both with subsequent permanent brain damage.

A maternal IEM was found in 27 cases, dominated by 19 with CTD and seven with 3-MCC. All mothers were asymptomatic.

A total of 29 children with an elevated marker for 3-MCCD (C5OH) were not reported after 3-MCCD was removed from the panel in 2009 (3-MCCD is a non-disease in > 90% of children). However, C5OH is also a marker for holocarboxylase synthase deficiency (HLCSD), common in the Faroe Islands, and a good screening target. Our strategy is to sequence the HLCSD gene in all newborns with a raised C5OH and only report those with two pathogenic mutations in that gene. One of the 29 non-reported children has presented clinically in a 3-MCCD metabolic crisis.

Discussion

The overall frequency of the metabolic diseases included in eNBS was 1:3,900, which represents a 50% increase compared with the frequency during the decade before eNBS. This is similar in other eNBS programmes [8-11]. We believe that the frequencies observed are accurate as all children in Denmark with an IEM are diagnosed and treated in the same centre. For MCADD, a more than three-fold higher incidence after onset of screening was found [2, 7]. VLCADD, 3-MCCD, CTD, BIOTD and ASLD are also found more frequently by screening than by clinical presentation. A proportion of patients may remain asymptomatic throughout life, as illustrated by the 27 asymptomatic mothers with CTD, 3-MCCD or MCADD. On the other hand, many patients with the same diseases will develop symptoms, and the low frequency in the period before eNBS may, in part, be explained by early death or incomplete diagnostic evaluation. Some children died before screening results became available, and these might have remained undiagnosed in the pre-screening era. For MCADD, IVA and BIOTD, mutations associated with mild disease could be documented as one reason for the increased frequency. The incidences of LCHADD, MMA/PA and HT1 were similar to those in the pre-screening era, probably due to a high clinical penetrance.

The higher frequencies of “mild” genotypes in screened as opposed to unscreened patients and the unclear natural histories for some diseases complicate assessment of eNBS benefits [12]. The benefit is most evident for MCADD with a 1% mortality and 97% living healthy lives at the last follow-up (Table 4). This is far better than historic studies with mortality rates of 25% and neurologic sequelae in 25% [13]. Evidence has been presented for GA1 that eNBS markedly improves prognosis [3]; 73% of the children with GA1 in our cohort were clinically normal. Moreover, of the three children with symptoms, two had very mild dystonia only and no history of precipitating metabolic crises. All children with BIOTD, LCHADD or ASLD had no sequelae. In the entire group of children with true-positive results, three (1.2%) died (two before screening results became available), and 93% were without sequelae at the last follow-up. In the cohort of unscreened children, one with LCHADD died; two with MCADD and one with LCHADD had severe metabolic crises; and one with GA1 and one with PA had metabolic crises with subsequent brain damage. Compared with the course of screened children, the course of disease of these six children supports the belief that eNBS is associated with a lower morbidity than diagnosis after clinical presentation. In a study from Boston, 2% of 189 children found by eNBS and 42% of 142 children diagnosed clinically had severe outcomes [14]. Though the precise extent of improved outcome offered by eNBS cannot be reliably evaluated at present, data certainly seem to indicate a substantial clinical benefit.

eNBS diagnosed 27 mothers with 3-MCCD, CTD or MCADD, which is well known from the literature [8, 15]. All mothers with 3-MCCD or CTD were asymptomatic; 3-MCCD has now been removed from the panel, and the clinical significance of CTD in asymptomatic mothers has been questioned [16]. However, symptoms, including cardiac ones, in mothers with CTD found by screening have been observed [15]. These data and sudden death in previously asymptomatic adult patients with CTD [17] argue for continued screening. The high frequency of IEM in mothers of newborns with positive results by eNBS underscores that follow-up investigations should be done of mothers of newborns that were screen positive for these disorders.

During the Danish routine eNBS, FPR was low and PPV high, except in a short period following the introduction of earlier sampling from 2009. This may be explained by an increase in the finding of heterozygotes for FAOD. CTD had the highest FPR. CTD is common in the Faroes Islands [18], and taken together with a high false-negative rate, we had to accept a high FPR. Now a Faroese second screening at 3-4 weeks postpartum is done, enabling a lower FPR. The maternal use of carnitine-depleting pivalic acid-containing antibiotics also contributed to the high FPR.

Confirmation of a flagged result as false or true positive was done within five days after information of the family. Although short, this period is stressful for the family. International data for MCADD indicate that use of health services in families with false-positive results exceeds predicted use during the first year of life [19]. We have no indication of this occurring in Denmark, but it should be studied systematically. Information given in an adequate and sensitive manner is key to reducing anxiety. Apart from a folder about NBS, in Denmark this information is given by the local paediatric department via information sheets from the CIMD, and this seems to function well.

The fact that 99.85% of Danish newborns are screened documents parental trust in NBS and the adequacy of the informed dissent procedure. Few are not screened: very few parents actively decline; some children have died early; some families may not have understood procedures because of language barriers. The frequency of diseases within the eNBS panel was 28 times higher among non-ethnic Danes [20], underscoring the need for thorough information to all ethnic groups.

This up-date on eNBS in Denmark corroborates our first study [7]. Along with international data, it confirms that eNBS is a successful preventive public health programme. Early treatment in a latent disease phase is effective, and NBS should be extended to other diseases not currently in the panel. Selection of new diseases is not easy. We removed some diseases from our initial panel because of their high FPR, new evidence of a benign natural history, or presentation before screening. Limited knowledge about natural history, spectrum of severity and effect of treatment is inherent in rare diseases, and evidence about benefit may be obtained only during prospective screening, making gradual adjustments to the NBS programme necessary. Focus should be on treatable diseases; a process of gradual evaluation of screening performance; and relevant psychological and social support to the families, including an acceptance of support also regarding preventive measures in an asymptomatic child.

Correspondence: Allan Lund. E-mail: allan.lund@regionh.dk

Accepted: 7 November 2019

Conflicts of interest: none. Disclosure forms provided by the authors are available with the full text of this article at Ugeskriftet.dk/dmj

Acknowledgments: We are indebted to the staffs of the obstetric and paediatric departments for participating in the recruitment and diagnostic work-up. Finally, we express our gratitude to Committee for Clinical Genetics and Screening, The Danish Paediatric Society and The Danish Health Authority for excellent cooperation.