INTRODUCTION

Screening of newborn infants for genetic disease has been in existence for 37 years with the objective of identifying disorders while they are presymptomatic and amenable to preventive treatment. It began in 1962, when Robert MacCready, Director of the Diagnostic Laboratories in the Massachusetts Department of Public Health, together with Robert Guthrie, the founder of newborn screening, organized the collection of a simple filter paper dried blood specimen (Guthrie specimen) from every newborn infant in Massachusetts and tested these specimens for phenylketonuria (PKU), using the bacterial assay for phenylalanine that Guthrie had developed (52, 102). By the late 1960s, routine testing of neonates for PKU had spread to almost every state and to several countries in Europe. Many of the programs had also begun testing for other genetic diseases such as galactosemia, maple syrup disease, and homocystinuria (89). In the mid-1970s a radioimmunoassay for thyroxine (T₄) was adapted to the Guthrie specimen for the identification of congenital hypothyroidism (34). Currently, a number of additional assays to detect genetic disease can be applied to the Guthrie specimen, including an immunoassay for 17-hydroxyprogesterone to identify congenital adrenal hyperplasia (132), an enzyme assay to detect biotinidase deficiency (56), and hemoglobin electrophoresis to detect sickle cell anemia (42). Immunoassays to detect infectious diseases such as congenital toxoplasmosis (50) and HIV (58) have also been applied to the newborn specimen.

While these additions to PKU screening have improved the outcome of the disorders, only that for congenital hypothyroidism has had a major impact on the performance of newborn screening and the prevention of disease. However, a new era in newborn screening is approaching (92). Tandem mass spectrometry, a technology recently adapted to the Guthrie specimen, is substantially enhancing the screening process and expanding coverage to many treatable genetic disorders heretofore not identifiable by newborn screening. Moreover, the ability to examine DNA in the Guthrie specimen is offering the possibility of molecular screening for an even greater expansion.

This review summarizes the current status of genetic screening of the newborn, explains the new possibilities, and examines several issues that excite some and trouble others.

ORGANIZATION OF SCREENING

Guthrie developed both the methodology of newborn screening and its organization (54). He also promoted the passage of state laws that mandated screening for PKU, which often also required that the screening be conducted by or under the direction of state health departments (1). This led to control of screening within each state that improved laboratory performance. However, this control is now restricting progress. Specifically, there is wide variation in the number of disorders covered, efficiency of the testing, and quality of follow-up among the states. Regionalization of screening can address these problems, but this requires a degree of interstate cooperation that has been possible only in the Pacific Northwest and New England. Current laboratory and programmatic methodology could provide all of the screening in the United States within a few regional laboratories. Screening outside the United States has generally followed this US pattern of “Balkanization.” One notable exception is Japan, which has established a national program that includes a standard for the entire population (124).

Blood Specimen

The newborn specimen is obtained as originally described by Guthrie & Susi (55). An area of the foot lateral to the heel is lanced, and capillary blood drops on a filter paper card, soaking several circles. The blood is dried in air and then delivered by courier or mail to the testing laboratory. On occasion, the specimen is obtained by a venipuncture of the hand (85) or from a catheter used for access to a vein or artery during intensive care of an infant. The metabolite values are not substantially different from one source of blood to another (100).

Timing of Specimen Collection

A major concern in newborn screening has been the age of the infant at the time of specimen collection. Over 25 years ago Holtzman et al (63) reported lower increases in phenylalanine with earlier blood collection in phenylketonuric infants. They believed that infants with PKU could be missed because of normal or near-normal blood phenylalanine levels in specimens obtained before 48 hours of age. McCabe et al (110) came to a similar conclusion from a regression analysis of blood phenylalanine levels largely obtained after 48 hours of age. This concern seemed unwarranted, however, after it was shown that infants with PKU had increased blood phenylalanine levels within the first 24 hours of age (30, 116). Nevertheless, the concern has resurfaced and has extended to the general reliability of newborn screening with the frequent collection of the Guthrie specimen at 24 hours of age (91). Lowering the cutoff level for phenylalanine seems to maintain the reliability of PKU detection in even very early specimens (71). This may also apply to other genetic disorders (119, 136).

SCREENED GENETIC DISORDERS

Phenylketonuria

Phenylketonuria (PKU) constitutes the paradigm of the disorders to which newborn screening is applied. It is caused by a deficiency of the enzyme phenylalanine hydroxylase (PAH) leading to the accumulation of phenylalanine in the blood and phenylalanine metabolites in the urine (Figure 1). If untreated, patients with PKU develop mental retardation as well as other neurological abnormalities. The average incidence of this disorder is approximately 1:10, 000 live births (155).

The degree of PAH deficiency determines the degree of hyperphenylalaninemia (HPA). An elevation of the blood phenylalanine level to 1200 μmol/l or higher (normal level is <120 μmol/l) is referred to as classic PKU and is associated with virtually undetectable PAH activity. A phenylalanine concentration of 600–1200 μmol/l is classified as mild PKU, while a level of 180–600 μmol/l indicates non-PKU mild hyperphenylalaninemia (MHP). Both mild PKU and MHP are associated with some residual PAH activity. Restriction of dietary phenylalanine is required to ensure normal or near-normal cognitive development in classic and mild PKU, whereas individuals with MHP may not require dietary treatment (189).

The gene for PAH has been cloned and mapped to chromosome 12q24.1. More than 400 mutations at the PAH locus have been linked to PAH (http://www.mcgill.ca/pahdb). Genotype and biochemical phenotype are closely correlated; null mutations are associated with classical PKU, and other mutations are associated with mild PKU or MHP (52, 73). Nevertheless, variability in biochemical phenotype has been observed within a number of the mutations (73).

Newborn screening began with the Guthrie test for PKU (55). It is still the most commonly used method for the identification of PKU in newborn screening, although, in some programs, it has been replaced by a fluorometric assay that is quantitative and more sensitive (31). Tandem mass spectrometry is now beginning to replace these methods (see Recent Advances/Tandem Mass Spectrometry, below). It has greater sensitivity and a very low false-positive rate for detection of PKU within the first 24 hours of life (15).

The opportunity to begin dietary treatment for PKU within the first weeks of life as a result of newborn screening detection has led to a revolutionary change in the cognitive development of these children. Newborn screening has virtually eliminated mental retardation from PKU (103). In optimally treated patients, the mean IQ is similar to that in the general population (167). Nevertheless, even early-treated and well-controlled patients with PKU may have subtle signs of neurological impairment (137, 149). The best developmental scores seem to be achieved when the diet begins within the first 3 weeks of life, blood phenylalanine levels are controlled in the range of 120–360 μmol/l (167), and the diet is maintained at least through the first 10 years of life (7, 62), if not for lifetime (152).

Approximately 1%–5% of the infants identified with HPA by newborn screening have secondary HPA from a deficiency of the tetrahydrobiopterin (BH₄) cofactor of PAH rather than an intrinsic defect in PAH (Figure 1). Any of several defects in the pterin pathway for BH₄ synthesis can result in BH₄ deficiency. Differentiating these children from those with PKU is critical since BH₄ is also the cofactor for the other two aromatic amino acid hydroxylases, tyrosine hydroxylase and tryptophan hydroxylase, which are required for biosynthesis of the neurotransmitters dopamine, norepinephrine, and serotonin. If not properly treated, BH₄ deficiency leads to mental retardation and severe neurological impairment. The treatment includes BH₄ and neurotransmitter–enhancing medications rather than only dietary therapy. Determining pterin metabolites in urine is the most frequently used method for differentiating a pterin defect from PKU in infants detected by newborn screening (31).

A specific increase of the phenylalanine level in the newborn period may be transient. It can be associated with prematurity, although, more often, no cause is identified. Secondary HPA due to liver disease, most notably in tyrosinemia type I and galactosemia, may also be identified by newborn screening (31).

Congenital Hypothyroidism

Congenital hypothyroidism (CH) is the most frequent disorder identified by newborn screening, with an overall incidence of approximately 1:4000 (28). Over 90% of the infants with CH have primary sporadic hypothyroidism due to thyroid agenesis or ectopia. The remaining cases include inborn errors of thyroid hormone biosynthesis, thyroid hormone resistance (28, 49), and defects in thyroid stimulating hormone (TSH) (4) or the TSH receptor (5) and in thyroid peroxidase (6). The major clinical features of untreated CH are retardation of growth and delayed neurocognitive development eventuating in mental deficiency (49).

Two approaches are used in newborn screening for CH. One is primary screening for a low level of thyroxine (T₄) with secondary screening for the increased level of TSH that results from reduced thyroxine feedback inhibition of TSH secretion. The other is primary screening for an increased level of TSH, often with secondary screening for a low T₄ level. Both methods readily identify CH (77).

Treatment of CH consists of replacement with pharmacologic doses of thyroxine. Early initiation of this treatment has had a great impact on the developmental outcome of these children. Before newborn screening, treatment often did not begin before 3 months of age, too late to prevent intellectual impairment (78). Since the establishment of newborn screening for CH in the 1970s (34), intellectual outcome has been normal or near-normal (49). Nevertheless, outcome is related to the severity of the hypothyroidism. In cases of severe hypothyroidism due to thyroid agenesis with very low T₄ levels (<40 nmol/l), deficits in IQ as well as mild impairment in educational and motor attainment may occur despite early treatment (165).

Quite frequently in newborn screening for CH, increased levels of TSH and/or low T₄ levels are noted without thyroid dysgenesis or dyshormonogenesis. These screening findings normalize within a few days or weeks without substitutive therapy (28). Transient changes such as these have been observed in up to 85% of premature infants and correlate with gestational age (135). They have been associated with poor neurodevelopmental outcome (146), but this does not seem to be improved by early replacement of thyroid hormone (178). Transient hypothyroidism can also result from iodine deficiency in the mother (27), which is important in areas with endemic goiter since loss in intellectual capacities has been observed in these children (8).

Galactosemia

Three enzyme defects in galactose metabolism can produce genetic disorders (Figure 2). These are deficiencies of galactokinase (GALK), galactose-1-phosphate uridyltransferase (GALT), and uridine diphosphate galactose-4-epimerase (epimerase).

Classic galactosemia, the most severe of the galactose metabolism disorders, is caused by a deficiency of GALT that leads to the accumulation of galactose and galactose-1-phosphate. In most cases, there is essentially no detectable GALT activity in red blood cells or in vivo oxidation of galactose (160). The disease typically presents in the neonatal period with failure to thrive, vomiting, and liver disease. Death from bacterial sepsis, usually due to Escherichia coli, occurs in a high percentage of untreated neonates (94). Long-term complications in the untreated state include liver cirrhosis, cataracts, and mental retardation. The average incidence is 1:62,000 (93).

The gene for GALT has been cloned (145) and well characterized (87). More than 130 different mutations have been identified in association with galactosemia. The most commonly observed mutation in classic galactosemia, present in approximately 70% of patients, is Q188R (36). The N314D mutation is prevalent in the general population and produces the benign Duarte variant (35). The S135L mutation is found predominantly in African-Americans and South Africans and produces a relatively mild form of classic galactosemia (104). In a recently performed study, six mutations (Q188R, K285N, S135L, N314D, L195P, and Y209C) accounted for 87.5% of the mutant alleles seen in galactosemic children (36).

Galactosemia is included in most newborn screening programs. It is detected through either a metabolite assay that measures galactose and galactose-1-phosphate or a spot enzyme assay that measures GALT activity (90). The metabolite assay identifies all defects in galactose metabolism, whereas the enzyme assay identifies only classic galactosemia (77). Since the Duarte/galactosemia genetic compound variant of GALT deficiency is associated with low GALT activity (approximately 25% of normal) and transient increases in galactose and galactose-1-phosphate, interpretation of a positive screening result requires differentiation between classic galactosemia and this common benign variant (95). Screening for the N314D mutation as a second tier to newborn screening for galactosemia facilitates this differentiation (177).

Treatment of galactosemia is the elimination of galactose from the diet. Early dietary treatment prevents or reverses the neonatal complications (94). Nevertheless, long-term effects develop despite early identification and treatment (59). Studies in the United States and Europe have disclosed a decline in IQ with age, delay in speech development, and deficits in speech expression (154, 180, 181). Particularly troubling in females has been the occurrence of ovarian failure (72). This has led some to question whether newborn screening for galactosemia is beneficial (161), although screening does prevent neonatal death from sepsis (94).

This disorder results in increased galactose-1-phosphate. Galactose may be slightly increased as a secondary effect (Figure 2). There are two forms of the disorder: a generalized enzyme defect with clinical symptoms indistinguishable from classic galactosemia, and an asymptomatic form in which the enzyme defect is limited to red blood cells (160). Most infants with epimerase deficiency identified by newborn screening have had the benign form (129). Treatment of children with generalized epimerase deficiency using a galactose-free diet has not prevented mental retardation (183).

This disorder produces a specific accumulation of galactose (Figure 2). It is very rare, with an estimated incidence of 1:1,000,000 live births (67). The only known complication is cataracts due to lenticular accumulation of galactitol, an alcoholic metabolite of galactose. Notably, these patients do not have the liver and brain involvement of classic galactosemia. Dietary treatment is identical to that of classic galactosemia and has prevented the cataracts (67, 76). For 17 years, we have followed a child identified by newborn screening and treated from infancy. He has remained intellectually normal and free of cataracts (HL Levy, unpublished data).

Homocystinuria

Several known defects in methionine degradation produce an increase in homocyst(e)ine, but only one, cystathionine-β-synthase (CBS) deficiency (usually referred to as homocystinuria), also results in an increase in methionine (Figure 3). Thus, newborn screening for increased methionine can be used to identify this genetic disorder.

CBS is a vitamin B₆ (pyridoxine)–dependent enzyme (Figure 3). Accordingly, there are two forms of homocystinuria: one that is vitamin B₆-responsive and clinically the milder form, and a more severe B₆-nonresponsive form. These forms seem to be equally divided in the affected population (122).

Individuals with homocystinuria are clinically normal at birth but, if untreated, usually become mentally retarded and have dislocation of the lens, osteoporosis with bone deformities, and thromboembolism (120). The incidence of B₆-nonresponsive homocystinuria varies widely; it is 1:65,000 in Ireland (205) and at least 1:157,000 in New England (136). The frequency of B₆-responsive homocystinuria is unknown. The worldwide frequency of all forms of homocystinuria has been estimated at only 1:344,000, but this is almost certainly an underestimate since homocystinuria is often undiagnosed for many years (120).

The gene for CBS has been cloned (82) and mapped to chromosome 21q22.3 (123). The most frequently identified mutations are I278T that almost always associates with B₆-responsiveness (163) and G307S that is associated with the B₆-nonresponsive form of homocystinuria. This latter mutation is the leading cause of homocystinuria in Ireland and in patients of Celtic origin elsewhere (41).

Newborn screening for homocystinuria is currently conducted in 15 US states, Japan and many European countries (175). Most programs employ the Guthrie bacterial assay for methionine (53). Unfortunately, the B₆-responsive form is not detected by newborn screening, and it is likely that even B₆-nonresponsive infants are missed. Reducing the cutoff level for methionine from 134 μmol/l to 67 μmol/l has substantially increased the frequency of identified infants in New England (136). This is best accomplished by employing a newer technology such as tandem mass spectrometry, which has greater sensitivity for methionine than the bacterial assay (10).

Early treatment is highly effective in preventing long-term complications (184, 205). Treatment of the B₆-nonresponsive form includes dietary methionine restriction and L-cystine supplementation. Betaine, a methyl donor that stimulates the methylation of homocysteine to methionine (Figure 3), thereby reducing the level of homocysteine, is often used either in conjunction with diet or instead of diet (197) . The B₆-responsive form can be treated with pharmacological doses of vitamin B₆ combined with folic acid (184).

Newborn screening for homocystinuria also identifies isolated hypermethioninemia, a disorder usually caused by methionine adenosyltransferase (MAT) deficiency (Figure 3). Individuals with this disorder have generally been normal, although in a few instances reduced cognitive function has been reported (121).

Neonatal liver disease, especially metabolic liver disease such as occurs in tyrosinemia type I or galactosemia, can cause elevated methionine levels, but this is usually accompanied by elevations of other amino acids, notably tyrosine. High protein diets can produce transient hypermethioninemia, again usually accompanied by other amino acid elevations (97).

Biotinidase Deficiency

Biotin functions as a coenzyme for four carboxylases: pyruvate carboxylase, propionyl-CoA carboxylase, β-methylcrotonyl-CoA-carboxylase, and acetyl-CoA carboxylase (201). It is conserved by biotinidase cleavage of the biocytin (biotinyllysine) released from the degraded carboxylases, yielding free biotin (Figure 4). A deficiency of biotinidase produces intracellular biotin deficiency and an increase in biocytin. The deficiency can be partial (10%–30% of mean normal serum activity) or profound (<10% of normal serum activity). The estimated combined incidence of profound and partial biotinidase deficiency is 1:61,000 (202).

Profound biotinidase deficiency leads to developmental delay, hypotonia, seizures, ataxia, alopecia, skin rash, and neurosensory hearing loss (203). Organic aciduria due to deficient activities of the carboxylases (increased lactate, propionate, and β-methylcrotonate) as well as ketoacidosis can develop late in the course of the disease (67). Partial biotinidase deficiency may be asymptomatic, although in situations of stress neurological symptoms may occur (113).

The cDNA for human biotinidase has been cloned and sequenced (20), and at least 20 mutations have been identified in biotinidase deficiency (138). The most frequent of these mutations in profound deficiency is Q456H (130); in partial biotinidase deficiency it is D444H (172).

Newborn screening for biotinidase deficiency is performed by a semiquantitative colorimetric assay for biotinidase activity applied to an eluate of the Guthrie specimen (56). The method can identify both the partial and profound forms of biotinidase deficiency (86). Currently, screening for biotinidase deficiency is performed in 22 US states, many countries in Europe, and Japan (175).

Treatment consists of pharmacologic doses of biotin. If it begins within the newborn period, the clinical sequelae are prevented (86, 185). Late initation of treatment can reverse some of the complications of the disorder such as skin rash and hair loss, but not the neurological deficits (187). It is unclear whether treatment is required in partial biotinidase deficiency (113).

Maple Syrup Disease¹

Maple syrup disease (MSD) results from a defect in the branched-chain α-ketoacid dehydrogenase complex (BCKAD) leading to the accumulation of the branched-chain amino acids (BCAA; leucine, isoleucine, and valine) and their respective α-ketoacids (Figure 5). The enzyme complex consists of four protein components; a defect in any one is capable of producing MSD. Four genes for the protein subunits of the complex (E1α, E1β, E2, E3) have been cloned, and a number of mutations have been identified at these four loci. However, the genotypes do not closely correlate with the clinical phenotypes (19).

There are at least five clinical phenotypes in MSD. Classic MSD, first described by Menkes and colleagues in 1954 (115), is the most severe form, with a residual BCKAD activity of only 0–2% of normal (19). Affected infants usually present within the first week of life with feeding intolerance, vomiting, lethargy, and severe ketoacidosis, often rapidly progressive to coma and death. The typical odor resembling maple syrup is recognized earliest in cerumen and later in urine, but may not have become evident at the time the infant first develops clinical symptoms (77). The average incidence of classic MSD is 1:185,000 but is as high as 1:176 in the Mennonites (19).

Newborn screening for MSD is performed by determination of an increased level of leucine in the Guthrie specimen. A Guthrie bacterial assay for leucine is the most frequently used screening test, although tandem mass spectrometry is beginning to replace this and other metabolic screening tests (see Recent Advances/Tandem Mass Spectrometry, below). MSD screening is currently included in the programs of 21 US states, most European countries, and Japan (175).

Therapy is based on a diet that is low in the branched-chain amino acids. Care is required to avoid deficiencies of isoleucine and valine during therapy by providing supplements of these two BCAAs, particularly in infancy. Intensive care with peritoneal dialysis or hemodialysis may be necessary during periods of metabolic crisis associated with profound ketoacidosis and coma. If treatment does not begin in early infancy, neurological impairment or death invariably occurs (125).

Since the establishment of screening, the outcome in MSD has greatly improved. With early diagnosis and treatment before the tenth day of life, the outcome can be normal (119, 125, 127). Nevertheless, despite optimal therapy, fatal cerebral edema during severe ketoacidotic episodes or severe neurological impairment can occur (125, 150).

The milder MSD variants may be missed in newborn screening as a result of a normal leucine level in the newborn period (77). These patients generally have higher residual BCKAD activities, ranging from 3% to as high as 40% of normal, and usually present after the neonatal period, often within the second year of life (19). The intermediate variant produces developmental delay and ataxia without ketoacidosis and much lower elevations of the BCAAs than the classic disorder. In the intermittent variant, patients develop normally and have normal levels of the BCAAs, but during stress they are at risk for acute metabolic decompensation with marked elevations of the BCAAs and severe ketoacidosis (19). The thiamine-responsive variant is a mild form with a biochemical phenotype between the intermediate and intermittent variants. The major clinical feature is recurrent ataxia (139, 157) although acute ketoacidotic crises may also occur (HL Levy, unpublished data). Pharmacologic doses of thiamine alone may be effective in treating these patients. The very rare E3-deficient variant of MSD is associated with deficient activities of pyruvate dehydroxygenase, α-ketoglutarate dehydrogenase, and BCKAD (18). The clinical phenotype is similar to intermediate MSD but may be accompanied by severe lactic acidosis. There is no truly effective treatment for this variant disorder.

Sickle Cell Anemia

Sickle cell anemia is a hemoglobinopathy that primarily affects African-Americans. In this ethnic group it has the extraordinarily high frequency of 1:400 (192). It is caused by a genetic mutation expressed as the substitution of valine for glutamic acid on the β-chain of hemoglobin. The sickle or S hemoglobin produced by this change causes the red blood cell to assume distorted forms that tend to block capillaries, resulting in infarcts of bones, spleen, and other organs (188). Thus, children with sickle cell anemia suffer from recurrent vaso-occlusive crises and are susceptible to bacterial infections, particularly those due to Streptococcus pneumoniae (43). In addition, increased fragility of the red blood cells leads to hemolytic anemia, requiring frequent blood transfusions. Mortality is high in sickle cell anemia, often as a result of bacterial infection and acute splenic sequestration, with the highest risk at 6 to 12 months of age (44, 84, 179).

Newborn screening for sickle cell anemia is included in the screening profile of most states, but is essentially limited to the United States. The screening techniques in widest use are “twin-tier” electrophoresis (cellulose acetate electrophoresis followed by citrate agar electrophoresis) (42) or thin-layer isoelectric focusing (79). Screening identifies not only sickle cell anemia but also benign entities such as sickle cell trait and several other hemoglobinopathies (77).

Presymptomatic identification of sickle cell anemia through newborn screening provides the opportunity for prophylactic antibiotic therapy to prevent potentially fatal bacterial infection. In addition to antibiotics, identified infants also receive immunization against Streptococcus pneumoniae and Haemophilus influenzae. Presymptomatic intervention clearly reduces mortality among identified patients (43, 45, 179). There is evidence that it also reduces disease morbidity, particularly splenic sequestration and infection (43).

The need for universal, mandatory screening for sickle cell disease has been questioned, particularly for states with relatively small African-American populations. However, accurate ethnic identification of at-risk newborns has been problematic. Consequently, newborn screening is now recommended in the United States for all newborns regardless of their ethnic background (21).

Congenital Adrenal Hyperplasia (CAH)

Congenital adrenal hyperplasia (CAH) is an autosomal recessive endocrine disorder with an incidence of about 1:15,000 (131). A low cortisol level results in reduced feedback inhibition of ACTH secretion from the pituitary gland, causing excessive release of ACTH and overstimulation of the adrenal cortex. A major consequence of this overstimulation is increased production of androgens (Figure 6), causing virilization (131).

More than 90% of CAH cases are due to 21-hydroxylase deficiency. In its most severe clinical picture, known as salt-wasting CAH, aldosterone production is decreased and life-threatening adrenal crises with hyponatremic hyperkalemic dehydration can present in the newborn period, often at 3 weeks of age (131). Female newborns have ambiguous external genitalia and may be incorrectly identified as males. In the simple virilizing form of CAH, adequate aldosterone production is retained, and salt-wasting does not occur. These children are often not diagnosed until early or middle childhood, when excess adrenal androgen secretion causes precocious puberty (133). In the late-onset form, hirsutism, acne, or amenorrhea may appear during or after adolescence (131).

The gene for steroid 21-hydroxylase, CYP21, has been cloned and mapped to the short arm of chromosome 6 in the HLA complex (33, 88). The different CYP21 mutations identified so far have shown a close correlation to the severity of disease (171).

The purpose of newborn screening for CAH is to avoid salt-losing crises through presymptomatic therapy and assure correct gender identification of affected females. The disorder is indicated by increased 17-hydroxyprogesterone (17-OHP) determined by immunoassay in the Guthrie specimen. The immunoassays available for use in newborn screening include the original radioimmunoassay developed by Pang et al (132), the more recently developed enzyme-linked immunosorbent assay (131), and the time-resolved fluoroimmunoassay (133). Unfortunately, the false-positive rate is high (131). Increased levels of 17-OHP are very frequent in low birth weight and preterm infants (133) as well as in perinatal stress and early specimen collection (77). Identification of CYP21 mutations in the Guthrie specimen has recently been introduced as a second-tier screening [see Recent Advances/Molecular (DNA) screening, below]. This could substantially enhance the specificity of CAH screening, thereby reducing the number of false-positive results and the need to request large numbers of additional blood specimens (38).

Infants with CAH are treated with pharmacologic doses of hydroxycortisone. Immediate notification of the screening abnormality is essential to prevent salt-wasting crises and to avoid unnecessary delay in providing accurate sex assignment. Delayed turnaround time in the newborn screening process or incorrect interpretation of a positive screening result has allowed adrenal crises to occur despite newborn screening (99).

Cystic Fibrosis (CF)

Cystic fibrosis (CF) is the most common serious autosomal recessive disorder in white populations, with an estimated incidence of 1:2000 to 1:3000 (191). The defect is in a cellular Cl⁻ channel protein known as the CF transmembrane conductance regulator (CFTR), which, when defective, results in reduced permeability of Cl⁻ across the apical membranes of epithelial cells in the airways of the lungs, pancreas, intestine, and sweat glands. This secondarily affects Na⁺ transport and hydration of mucus. The thickened mucus leads to progressive lung disease and pancreatic dysfunction as well as difficulties in several other organ systems. Increased levels of Na⁺ and Cl⁻ in sweat constitute a major diagnostic sign of CF.

The gene for CFTR has been cloned, and more than 800 CFTR mutations have been identified (74, 147, 148, 151). The most common of these is ΔF508, which is present in approximately 70% of CF patients and has been associated with the most severe disease presentation, particularly in patients with a homozygous genotype (75, 118). Clinical expression of the disease varies, however, and is only partly a reflection of the different genotypes at the CFTR locus (81).

Limited newborn screening for CF has been conducted for several years (194). The rationale is that presymptomatic detection reduces the morbidity and mortality from CF (195). The indicator is increased immunoreactive trypsinogen (IRT) in the Guthrie specimen (23). As in screening for CAH, however, the false-positive rate can be high as a result of frequently increased IRT levels in normal infants. To reduce this high false-positive rate, most programs have adopted a second-tier DNA analysis for one or more of the common CF mutations in specimens with increased IRT [see Recent Advances/Molecular (DNA) Screening, below]. This two-tiered IRT/DNA approach can identify up to 95% of the CF population with a greatly increased specificity (194).

Genetic counselling with prenatal diagnosis in subsequent pregnancies of couples with an affected child who was detected by newborn screening has significantly reduced the incidence of CF in some areas that screen for CF (46). Moreover, children with CF identified by newborn screening have shown a lower percentage of early pulmonary colonization with Pseudomonas aeruginosa, less deterioration in pulmonary function, and significantly better growth than those identified clinically (26, 37).

Unfortunately, treatment of CF is still symptomatic and does not prevent the long-term complications. On that basis, newborn screening has been questioned (61). However, gene therapy has now reached the stage of phase I clinical trials in adults (69). Should these trials be successful, the advocacy for newborn screening for CF would greatly increase, since screening would enable potentially curative treatment before the onset of irreversible pulmonary damage.

RECENT ADVANCES

Tandem Mass Spectrometry

Tandem mass spectrometry (MS-MS) is a recently introduced technology that has the potential to revolutionize newborn screening. It can detect more than 25 different genetic disorders with a single assay. Given the rapidly increasing abilities to diagnose and treat genetic disease, expansion of newborn screening programs has become very important. Moreover, changes in health care, particularly discharge of mother and baby within a day or two after delivery, require more sensitive methods to detect genetic disorders at an earlier stage without increasing the number of false-positive results. Traditional screening methodologies do not completely fulfill these requirements. Furthermore, in traditional screening, adding a disorder to the screening profile requires an additional test and, consequently, extra costs (173), whereas a single MS-MS assay identifies not only the metabolic disorders already covered in newborn screening, such as PKU and MSD, but also many other aminoacidopathies, organic acidemias, and disorders of fatty acid oxidation.

The technique of MS-MS itself is not new; it has been used for more than 15 years for the analysis of specific organic compounds in complex mixtures (112). What is new is the application of MS-MS to the Guthrie specimen, enabling its use in newborn screening (117). Accordingly, it has been applied to screening for PKU and tyrosinemia (13), maple syrup disease (11), homocystinuria (10), and MCAD deficiency (12). In a further refinement, Rashed et al (143) developed an automated method of sample application and a computerized algorithm to identify abnormal profiles of analytes measured by electrospray MS-MS.

Although the individual frequencies of the disorders identifiable by MS-MS are often quite low, the cumulative frequency is on the order of 1:4000 to 1:5000 (14, 198). This is essentially the same frequency as that of congenital hypothyroidism, which is considered to be very high by newborn screening standards. Beyond frequency, MS-MS is highly specific because it identifies disorders on the basis of metabolite patterns and ratios between metabolites rather than on a single metabolite level. Thus, the rate of false-positive results has been low (3, 198). Conversely, the high sensitivity of the method should allow for a low false-negative rate and for the measurement of metabolites at a much lower level than traditionally employed techniques. An example of the latter is the diagnosis of PKU as early as on the first day of life (15).

A triple-quadrupole mass spectrometer has two mass spectrometers linked in tandem separated by a collision chamber. A small disk is punched from a circle on the Guthrie specimen, the blood from this disk is eluted, the eluate is then esterified with butanol, and the derivatized specimen is injected into the system. These procedures are largely automated, thus greatly enhancing the speed of sample analysis and allowing its application to high-volume newborn screening. At the beginning of the system, the specimen is ionized by a “soft” ionization technique, most frequently one such as electrospray which accommodates automated sample injection (143, 144).

The first quadrupole (first mass spectrometer) determines the masses of the different metabolites in the blood specimen, separates them according to ionic charge (positive or negative ions), and transmits them to the second quadrupole (collision chamber). In the collision chamber an inert gas is introduced under high pressure, which collides with the molecular ions, producing fragmentation. The fragments are either ions or neutral (uncharged) molecular fragments. These fragments then pass into the third quadrupole (second mass spectrometer), which separates them according to mass and charge. This repeats the functioning of the first mass spectrometer but with the important exception that it also selectively scans them for characteristics associated with amino acids (ion with mass of 102 less than the original molecular ion) and acylcarnitines (ion with mass 85). Computer scanning of the two mass spectrometers synchronously with a fixed mass difference makes it possible to link the fragment results of the second spectrometer with their parent ion masses in the first spectrometer. The computerized system also measures the quantity of each amino acid or acylcarnitine (which identifies disorders of organic and fatty acids) on the basis of the ratio between the mass of the parent ion and mass of a known amount of corresponding stable isotope injected with each specimen. Excellent descriptions of this process can be found in Rashed et al (144) and Chace & Naylor (14).

To date, MS-MS has been incorporated into relatively few newborn screening programs. The largest of these programs is Neo Gen Screening Inc., a private newborn screening laboratory in Pittsburgh, which is using MS-MS to screen more than 400,000 infants per year in several American states and a few other countries (http://www.neogenscreening.com). The state of Massachusetts recently began newborn screening by MS-MS to test about 80,000 infants per year. The New South Wales program in Australia has been using MS-MS for more than a year. Among the 137,120 infants screened, 31 (1:4400) have been identified with a genetic disorder (198). In Germany (Bavaria), newborn screening with MS-MS has been conducted since January 1999. Among 108,000 infants screened, the frequency of identified disorders has been approximately 1:3850 (A Roscher, personal communication). Other programs in Japan (162) and Saudi Arabia (142) are also using MS-MS in newborn screening. It will be the challenging task for the future to assess the true frequency of the disorders identifiable by MS-MS, the sensitivity for each disorder, the rate of false-positive results, and the effects of early diagnosis and treatment on the outcomes.

MS-MS represents the possibility of a highly efficient and comprehensive approach to newborn genetic screening. It is likely that this single method will soon replace the multiple procedures currently used by most screening programs (92).

Molecular (DNA) Screening

In 1973, one of us (HLL) wrote, in reference to newborn screening for genetic disorders, “Since the direct analysis of genes, however desirable, is still not possible, genetic screening usually refers to testing relatively large numbers of individuals for gene products or resulting metabolites with the aim of identifying disorders due to mutant genes” (89). Less than two decades later, analysis for genetic mutations in newborn screening had become possible. In 1987, McCabe et al (109) extracted DNA from the Guthrie card, and 3 years later Schwartz et al (153) used the polymerase chain reaction (PCR) to amplify a targeted gene sequence directly from this specimen. Today, most mutations can be detected in the newborn screening specimen.

The hope that mutation analysis would become the primary method of newborn screening, however, has been mitigated by several realities, particularly the existence of multiple mutations in each of the genetic disorders for which screening is conducted (see Screened Genetic Disorders, above). Analyzing each newborn specimen for all of these mutations would require an enormous and expensive effort. Although several of these disorders have predominant mutations, testing for only these mutations would leave a substantial number of affected infants undetected. Only in sickle cell anemia is there a single mutation that would readily accommodate primary molecular screening. However, simple and inexpensive separation of the hemoglobins by electrophoresis or isoelectric focusing allows for specific identification of these mutant β-globin gene products (192), obviating any advantage of molecular screening for the sickling disorders. Nevertheless, it is possible that in the future primary molecular screening will be applied to the newborn specimen for a limited number of genetic abnormalities not otherwise identifiable. One example is the detection of gene fusions such as the MLL-F4 and TEL-AML1 sequences for presymptomatic detection of acute lymphoblastic leukemia in childhood (40, 193).

Mutation analysis is, however, very important as secondary or “second-tier” testing in newborn screening. Table 1 lists the disorders for which this testing is conducted and the general molecular methods employed. Almost all programs that include screening for cystic fibrosis add secondary molecular screening for the CFTR gene mutations to increase the positive predictive value of the primary IRT screening result (29, 47, 141, 196). Specifically, a cutoff level for IRT at the 95th or 99th percentile is selected to include essentially all affected infants. Most of the infants in this selected pool, however, do not have cystic fibrosis. To identify the infants who have cystic fibrosis and further select those who might have cystic fibrosis while eliminating those who are unaffected (false positives), the newborn specimens in this pool are examined for the CFTR mutations most prevalent in the screened population. This molecular battery always includes the prevalent ΔF508 mutation but varies from only this mutation in New South Wales, Australia (196), to 11 additional CFTR mutations in Wisconsin (47). Over 90% of these infants have no identified mutation and are considered to be false positives. Approximately 2% are homozygous or compound heterozygous for these mutations, thus confirmed as having cystic fibrosis and referred to a clinical program for treatment. The remaining infants with only one identified mutant allele are considered to possibly have cystic fibrosis and are referred for confirmatory sweat testing. Most of these infants are carriers with a normal sweat test result, whereas the sweat test in a few is positive, indicating that they have cystic fibrosis and a second mutant allele not identified by the molecular battery.

TABLE 1

Second-tier molecular analysis in newborn screening

Molecular testing for the prevalent A985G mutation in medium-chain acyl-CoA dehydrogenase deficiency (MCADD) and the predominant E474Q mutation in long-chain 3-hydroxyacyl-CoA dehydrogenase deficiency (LCHADD) is performed for second-tier testing of newborn screening specimens identified by MS-MS (EW Naylor, personal communication). Second-tier molecular analysis has also been used for the sickling hemoglobinopathies (206) and for galactosemia (177).

It is likely that the use of molecular analysis for second-tier testing will expand with advances in DNA technology. Neo Gen Screening Inc. in Pittsburgh is examining secondary molecular testing in newborn screening for congenital adrenal hyperplasia, selecting specimens with increased 17-OHP (EW Naylor, personal communication). This screening generates very high false-positive rates (see Screened Genetic Disorders, above), and increasing its positive predictive value by second-tier testing would be very beneficial. Developments in DNA chips could allow for second-tier testing with efficient detection of hundreds of mutations in a single analysis (105). If so, confirmatory molecular testing for frequently screened genetic disorders such as PKU and maple syrup disease could be accommodated, substantially reducing the number of false-positive results and the need to retest normal infants.

CONSIDERATIONS

Criteria

The contribution of Guthrie in founding newborn screening lies more in the specimen he originated than in the test for phenylketonuria that he developed. While his bacterial assays (“Guthrie tests”) are being replaced by much better techniques, the Guthrie specimen is here to stay. Among the reasons it continues to be such a valuable specimen is that it can be reliably tested for a virtually endless number of factors present in blood. This possibility has led to increasing opportunities for the expansion of newborn screening which, in turn, has led to a new examination of the criteria for newborn screening. Although these criteria have traditionally followed the 10 criteria suggested by Wilson and Jungner for screening in general (200) (Table 2), newborn screening has never been an entirely comfortable fit for them (16). Following is an examination of the Wilson-Jungner criteria considered to be most important for newborn screening and the problems of adhering to them as new information and new technologies emerge.

TABLE 2

Summary statements of the Wilson-Jungner criteria for screening (200)

The first rule of screening is that the condition should be an important health problem. This usually means that the disorder has both the capacity to produce illness and a relatively high frequency in the population. There should be no argument that a genetic disorder considered for screening must produce illness. The issue is the extent of the potential disability required to justify screening. While some genetic disorders can produce severe disability, such as death in infancy or mental retardation, others may produce mild cognitive impairment or learning disabilities only in certain settings. Histidinemia and Hartnup disorder, for example, may be among the latter (96, 156, 158). Moreover, most genetic disorders that produce severe disability also have variant forms that result in little or no disease (169, 204) or can have marked clinical variability even in the biochemically (and genotypically) classic forms (140). The question of whether to screen has usually been resolved on the side of newborn screening when it seems that most of those affected will have substantial clinical problems without screening and early treatment. However, the frequency of such problems may not be known with certainty when screening is considered. A current example is the question of adding screening for medium-chain acyl-CoA dehydrogenase deficiency (MCADD). There is no doubt that this disorder can result in sudden death (66). There are also affected adults who have remained asymptomatic into middle-age (57). The percentage of those with MCADD in either category is yet unknown.

Aside from clinical effect, frequency of the disorder per se also bears on the perception of importance to health. This is, of course, irrelevant to the affected individual and family but assumes great importance when the screening program utilizes methods such as bacterial or specific fluorometric assays, which require a separate test for each disorder. With newer technologies such as MS-MS, wherein many genetic disorders are covered by a single analysis (see Recent Advances/Tandem Mass Spectrometry, above), frequency becomes much less important (92). Thus, new developments in technology as well as new information about the disorders demand a fresh examination of “important health problem.”

Newborn screening began with that for PKU, based on the availability of treatment that prevents mental retardation (80). Until screening for sickle cell anemia was added, the criterion that presymptomatic treatment for the prevention of long-term complications must be available was an absolute requirement for screening. With sickle cell screening, however, the objective was not to prevent the long-term complications, for which there was no therapy, but to provide penicillin prophylaxis to all identified infants in order to protect the 10% who might die from overwhelming bacterial sepsis. Screening for cystic fibrosis is based on a similar justification.

The potential benefit of early diagnosis to the family is another advantage of newborn screening, rarely considered when there is little or no treatment for the affected infant. It is quite common for diagnosis and supportive treatment in genetic disorders to be delayed for years after clinical complications appear (24). In such situations, screening detection could have led to the avoidance of unnecessary medical procedures, inaccurate diagnoses, and continued ignorance of the cause of the difficulties.

The issue here is to what extent must the natural history of a disorder be understood before newborn screening is acceptable. In fact, newborn screening itself has, albeit inadvertently, led to a much greater understanding of the natural history of every disorder screened. It was newborn screening for PKU and the occurrence of mental retardation in some identified children despite early dietary treatment that led to the recognition of the pterin defects as causes of hyperphenylalaninemia (168). Newborn screening has shown that congenital adrenal hyperplasia clearly follows an autosomal recessive pattern of inheritance, with males as frequently affected as females, unlike the previous belief based on clinical recognition of ambiguous genitalia that females are much more often affected than males (131). Newborn screening is expanding our knowledge of biotinidase deficiency, a treatable disorder that can produce severe disability but for which asymptomatic adults have been reported (32). No disorder has been known in its full natural history prior to screening. The challenge is to learn the full extent of the phenotype in every screened disorder as the screening continues and to identify indicators that can discriminate infants who will remain asymptomatic so that unnecessary treatment is avoided.

The question this criterion poses is which costs should be considered in such calculations. Screening for a disorder that can cause death in infancy, such as maple syrup disease, may at face value not be economically cost effective, since the costs of screening and lifelong therapy for an identified case may be high, whereas the infant who dies in the absence of screening may not incur high costs. However, if the loss of lifetime productivity by the death of an infant is considered, the cost factor assumes a different face (2). Moreover, the costs of diagnostic procedures and lifelong medical care for those unscreened who do not die in infancy may further offset the costs of screening and therapy. Consequently, all potential averted costs should be included when the cost of screening is weighed against the cost of not screening.

Missed Cases

Most newborn screening programs have experienced missed cases. Holtzman et al (60) documented 76 such cases in screening for phenylketonuria and congenital hypothyroidism. The reasons included failure to collect a newborn specimen in the hospital nursery or collection of a specimen that was inadequate for testing, errors in laboratory procedures, errors in follow-up of abnormal results, and falsely reporting abnormal screening results as normal. In some instances the error could be documented, while in other instances the error was assumed on the basis of circumstantial evidence.

The information cited above, obtained by surveying directors of public health laboratories, represents only a fraction of the missed cases. Many missed cases are not called to the attention of the screening laboratory, because the physician making the diagnosis either is unaware of newborn screening or does not consider it important to notify the laboratory of the diagnosis. Moreover, the survey only included the two universally screened disorders whereas many programs include other disorders in the screening profile.

Most cases missed in newborn screening are laboratory errors (60). This can be established by testing the stored newborn specimen (see Considerations/Stored Specimens, below). For phenylketonuria screening, specimen retesting has shown distinctly increased levels of phenylalanine in specimens originally reported as normal when the children were found to have PKU (114; HL Levy and JR Simmons, unpublished data). Low thyroxine and increased TSH have been found upon retesting newborn specimens reported as normal in children with congenital hypothyroidism (128).

False-negative results can also result from mistakes in procedures or policy within the screening program. One notable error is in setting a cutoff level so high (or low) that affected infants with relatively mild abnormalities are not identified. Mild abnormalities in the presence of a disorder are increasingly frequent with early hospital discharge (136). Adjusting cutoff levels closer to normal limits can address this problem (30, 136). Collection of the newborn blood specimen after rather than before a blood transfusion is another program error. This results in the presence of normal donor red blood cells masking the abnormality present in the red blood cells of the infant. The most frequent example of this is missing galactosemia when detection requires the absence of GALT activity (170, 190). This type of program error can also mask the presence of sickle cell anemia. In these instances, the newborn screening result is reported as normal.

Biological variability virtually assures the likelihood that some infants with disorders for which screening is conducted will have normal findings when the newborn specimen is collected. These are disorders in which successful screening requires postnatal accumulation of a metabolite (or reduction of an analyte, such as thyroxine for congenital hypothyroidism). When the specimen is collected within the first 24 hours, as occurs with early hospital discharge (25), the reliability of newborn screening may be a concern.

In some of the screened disorders, however, an occasional infant will not have the screening abnormality, even when the specimen is collected at several days of age. This has been documented for homocystinuria (182), cystic fibrosis (196), tyrosinemia (48), and congenital adrenal hyperplasia (164) and undoubtedly accounts for most of the infants with undetected non-PKU hyperphenylalaninemia. One potential method for overcoming this problem is to change the primary screening indicator to one that has greater sensitivity. This has been done in the screening for tyrosinemia type I in Quebec, where the frequency of this disorder is high (83). It was recognized that some affected infants were unidentified by newborn screening because their level of tyrosine was normal when the specimen was collected before 48 hours of age. However, succinylacetone was known to always be markedly increased. Consequently, the screening program added an assay for activity of δ-aminolevulinic acid dehydratase, an enzyme inhibited by succinylacetone (48). This has resulted in the identification of all affected infants (A Grenier, personal communication).

False-Positive Results

The bane of newborn screening is false-positive results (92). The cutoff result in screening must be set so as to avoid missing affected cases, but this inevitably identifies unaffected infants with transiently abnormal values (Figure 7). Every such abnormal result on the primary screen requires a number of individualized additional procedures. These include a search for the specimen among the hundreds or thousands of specimens tested that day and repeat testing of the specimen, perhaps even multiple (second-tier) testing. Should this testing confirm the abnormal result, the attending physician or health facility must be notified and a repeat specimen requested. This notification may also go to the parents, accompanied by an explanation that includes the possible implications of the result. If these implications include a disorder that requires immediate attention (e.g. maple syrup disease, galactosemia, or congenital adrenal hyperplasia), the notification is by telephone. This may result in an emergency visit for evaluation of the infant. Inevitably, anxiety is generated in medical personnel and the family, and many questions come up that the physician or the newborn screening program is required to answer. The repeat specimen must be collected and submitted to the screening program. In the screening laboratory the repeat specimen must be separately processed and tested, with the results recorded and transmitted to the physician or health facility.

Most often, the repeat specimen is normal, indicating that the original result was a false positive. Table 3 lists the approximate rates of false-positive results for the newborn screening tests. The rates are highest when the indicator is an abnormal metabolite level, such as thyroxine in congenital hypothyroidism, 17-OHP in congenital adrenal hyperplasia, and galactose in galactosemia. False-positive rates are also high when the indicator is low enzyme activity, such as that of GALT in galactosemia. When the gene product can be directly examined, such as in the hemoglobinopathies, the false-positive rate is low.

TABLE 3

False-positive rates in newborn screening

The challenge is to adopt primary tests that reduce false-positive results. The most successful example of meeting this challenge is the substitution of MS-MS for the traditional bacterial or other specific assays (see Recent Advances/Tandem Mass Spectrometry, above). By determining metabolite profiles instead of individual metabolite levels, MS-MS usually allows immediate recognition of the disorder. A notable example is PKU. In a group of specimens from California, Chace et al (15) found that only one of the 91 false-positive results generated by fluorometric measurement of phenylalanine was positive when the phenylalanine:tyrosine ratio was determined by MS-MS. Similarly, MS-MS testing, by determining the levels of the branched-chain amino acids and allowing for determination of the leucine/alanine ratio, is a more specific indicator of maple syrup disease than the bacterial assay for leucine (119).

When highly specific primary screening is unavailable, second-tier testing of specimens is important [see Recent Advances/Molecular (DNA) Screening, above]. While secondary testing does not eliminate the load imposed on screening programs by the need to recover and perform additional tests on the Guthrie specimen and can be misleading, as occasionally occurs in screening for cystic fibrosis (17, 186), it does reduce the number of notifications for false-positive results. Second-tier DNA testing has improved screening for cystic fibrosis and galactosemia [see Recent Advances/Molecular (DNA) Screening]. It has traditionally been used to render screening for congenital hypothyroidism more specific by secondarily measuring TSH levels in those specimens that have low thyroxine (see Screened Genetic Disorders, above). Second-tier testing after ether extraction in screening for congenital adrenal hyperplasia has also markedly reduced the rate of false-positive results by eliminating those specimens in which the measurement of 17-OHP by immunoassay was falsely elevated by interfering steroids (108).

Stored Specimens

Most newborn screening programs store the Guthrie specimen for a period of time after the screening is completed (176). This practice began in Massachusetts in the early 1960s to enable retesting of a newborn specimen should a child who was not detected by newborn screening for PKU be later diagnosed with PKU. This would be an important means of quality control, i.e. determining the reliability of newborn screening and specifying readjustments.

Testing stored specimens has yielded valuable information in assessing the quality of newborn screening performance. Infants with PKU not identified by newborn screening have been shown to have had increased levels of phenylalanine in the newborn specimen (false-negative) and were missed for programmatical, not biological, reasons (114). Conversely, retesting the stored specimen has excluded laboratory error in failure to detect other disorders, such as a defect in pterin metabolism with hyperphenylalaninemia (9) and Duchenne muscular dystrophy (68). In addition, the stored newborn specimen has been used to obtain clinical information not otherwise obtainable. For instance, Zellweger syndrome was confirmed in a deceased infant by determining increased very long-chain fatty acids in the stored specimen after the autopsy suggested the possibility of this disorder (70) and genetic information for family counseling was obtained by determining the cystic fibrosis–related haplotype in the stored newborn specimen from a deceased child who had CF (199). Recovery of stored specimens with testing by MS-MS has disclosed the presence of medium-chain acyl-CoA dehydrogenase deficiency (MCADD) in several infants who died with the presumptive diagnosis of sudden infant death syndrome (SIDS), and in two infants with long-chain 3-hydroxyacyl-CoA dehydrogenase deficiency (LCHADD) who also were presumed to have had SIDS (www.neogenscreening.com).

Stored dried blood specimens have been used for research in biochemical genetic disorders. Hostetter et al (64) determined that liver disease in tyrosinemia type I is present before birth by demonstrating markedly increased levels of α-fetoprotein in stored cord blood specimens obtained for routine screening. Naughten et al (126) also used these stored specimens to establish the enzymatic expression of prolidase deficiency at birth.

More recently, the effectiveness of MS-MS as a new technology for expansion of newborn screening has been investigated in stored newborn specimens. This has been performed by testing recovered newborn specimens from children clinically identified with disorders assumed to be detectable in the newborn period. The results have indicated this capability of MS-MS for a number of the disorders (39, 106, 166). Similar studies have been performed in patients with peroxisomal disorders, the results suggesting that measurement of phytanic and pristanic acids in the newborn specimen could be used in newborn screening for these defects (174).

One of the most important features of these stored specimens, however, is the availability of DNA in the specimen, allowing for the use of this specimen in many areas of genetic research. The DNA can be extracted or used without extraction in the PCR amplification of gene segments that, in turn, can be examined for perhaps any mutation [see Recent Advances/Molecular (DNA) Screening, above]. This capability has been exploited to determine prevalences of the predominant mutation in MCADD (22, 107, 159), the frequency of a novel PAH mutation in PKU (101), and in searching for a polymorphism in the transforming growth factor α-locus among infants with cleft palate to examine the possibility of a genetic basis for this birth defect (65).

It is this feature of the stored Guthrie specimen that has generated the interest and concern about the storage of newborn screening specimens since it represents essentially the creation of “DNA banks” of a substantial portion of the population (111). A survey of all newborn screening programs in the United States disclosed that 75% of the programs store all of the newborn specimens for at least several months following completion of the tests, and most retain the specimens for a year or longer, some as long as 25 years (111). The storage conditions are usually at room temperature and humidity, although on occasion the specimens are refrigerated or even frozen. The conditions of storage may be unimportant for some metabolites which, like the amino acids phenylalanine and leucine, can be almost fully recovered from newborn blood specimens stored at room temperature for up to 20 years (98) but may be critical for the recovery of other analytes used in newborn screening, especially enzyme activities (176). The conditions of storage are probably irrelevant for the recovery of DNA, however, given its remarkable stability (111).

The key issue in the testing of stored newborn specimens is whether the infant from whom the specimen was obtained can be identified after the test result becomes known. This issue is usually framed in terms of whether an identifier will or will not be retained with the tested specimen. For some purposes, especially those of epidemiological importance such as determining gene prevalences or the frequency and distribution of infectious disease markers in populations, anonymous testing (no identifiers) is sufficient (134). For other purposes, such as retesting specimens in affected children whose newborn screening result was reported as normal, testing stored specimens to determine whether a new screening procedure reliably identifies a disorder, or retrospectively examining the effects of a disorder so as to decide whether it might be a valid candidate for newborn screening, identifiers are essential.

The retention of an identifier introduces the question of informed consent or legal mandate. It is generally agreed that informed consent should be required when retrieval of a specimen is requested by a physician or the family to establish diagnosis in a deceased child. A subpoena should be required to retrieve a specimen for legal purposes (e.g. to establish identity of a kidnapped child or for evidence in a malpractice action). Whether stored specimens could be used for non-anonymous retrospective studies in the absence of informed consent is an intensely contested issue. Some believe that this should only be allowed if informed consent for this use in the future is obtained from a parent at the time the newborn specimen is collected. However, the difficulty of obtaining truly informed consent in this context might severely limit such potentially very valuable use of the specimen. For instance, the informed consent would have to be nonspecific, and many parents might view this as providing consent for future inappropriate use. Newborn screening programs and public agencies are now actively addressing this issue. Hopefully, it will be resolved in a manner that will allow this unique specimen to be used for a better understanding of human genetics as well as the proper role of newborn screening.

ACKNOWLEDGMENTS

This review was supported by the Deutsche Forschungsgemeinschaft (DFG), the Metabolic Research Fund at Children's Hospital, and the Meister Family Homocystinuria Research Fund. Donald Chace and Edwin Naylor of Neo Gen Screen, Inc, kindly provided valuable information.