Validation of MCADD newborn screening

Clin Genet 2009: 76: 179 187 Printed in Singapore. All rights reserved Short Report 2009 John Wiley & Sons A/S CLINICAL GENETICS doi: 10.1111/j.1399-0004.2009.01217.x Validation of MCADD newborn screening Maier EM, Pongratz J, Muntau AC, Liebl B, Nennstiel-Ratzel U, Busch U, Fingerhut R, Olgemöller B, Roscher AA, Röschinger W. Validation of MCADD newborn screening Clin Genet 2009: 76: 179 187. John Wiley & Sons A/S, 2009 Medium-chain acyl-coa dehydrogenase deficiency (MCADD) represents a potentially fatal fatty acid β-oxidation disorder. Newborn screening (NBS) by tandem mass spectrometry (MS/MS) has been implemented worldwide, but is associated with unresolved questions regarding population heterogeneity, burden on healthy carriers, cut-off policies, false-positive and negative rates. In a retrospective case-control study, 333 NBS samples showing borderline acylcarnitine patterns but not reaching recall criteria were genotyped for the two most common mutations (c.985a>g/c.199c>t) and compared with genotypes and acylcarnitines of 333 controls, 68 false-positives, and 34 patients. c.985a>g was more frequently identified in the study group and false-positives compared to controls (1:4.3/1:2.3 vs. 1:42), whereas c.199c>t was found more frequently only within the false-positives (1:23). Biochemical criteria were devised to differentiate homozygous (c.985a>g), compound heterozygous (c.985a>g/c.199c>t), and heterozygous individuals. Four false-negatives were identified because our initial algorithm required an elevation of octanoylcarnitine (C 8 ) and three secondary markers in the initial and follow-up sample. The new approach allowed a reduction of false-positives (by defining high cut-offs: 1.4 μmol/l for C 8 ;7forC 8 /C 12 ) and false-negatives (by sequencing the ACADM gene of few suspicious samples). Our validation strategy is able to differentiate healthy carriers from patients doubling the positive predictive value (42 88%) and to target NBS to MCADD-subsets with potentially higher risk of adverse outcome. It remains controversial, if NBS programs should aim at identifying all subsets of all diseases included. Because the natural course of milder variants cannot be assessed by observational studies, our strategy could serve as a general model for evaluation of MS/MS-based NBS. EM Maier a,jpongratz a, AC Muntau a, B Liebl b, U Nennstiel-Ratzel b,ubusch b, R Fingerhut c,bolgemöller c, AA Roscher a and WRöschinger a a Research Center, Department of Biochemical Genetics and Molecular Biology, Dr von Hauner Children s Hospital, Ludwig-Maximilians-University, Munich, Germany, b Public Health Newborn Screening Center of the State of Bavaria, Oberschleißheim, Germany, and c Laboratory Becker, Olgemöller & Colleagues, Munich, Germany *These authors contributed equally to this work. Key words: acylcarnitine patterns carrier detection genotyping medium chain acyl CoA dehydrogenase deficiency mild biochemical phenotypes newborn screening population heterogeneity Corresponding author: Wulf Röschinger, Research Center, Department of Biochemical Genetics and Molecular Biology, Dr. von Haunersches Kinderspital, Ludwig-Maximilians-Universität München, Lindwurmstrasse 4, D-80337 München, Germany. Tel.: +49 89 5160 7791; fax: +49 89 5160 7792; e-mail: Wulf.Roeschinger@med.uni-muenchen.de Received 10 November 2008, revised and accepted for publication 7 April 2009 Medium-chain acyl-coenzyme A dehydrogenase deficiency (MCADD; MIM 201450) is an inherited disorder of fatty acid β-oxidation being caused by mutations in the ACADM gene (GeneBank accession no M16827.1). It has been reported to occur at an incidence of 1:8500 1:15000 (1). The clinical manifestation of MCADD, such as hypoglycemia, lethargy, or coma, is usually precipitated by catabolic stress. In the absence of screening, 17% of children with MCADD will die in childhood, and up to 10% of the survivors will develop a serious developmental disability (1). Once diagnosed, however, adverse outcome can be prevented by avoidance of fasting and a carbohydrate diet during catabolic episodes. Therefore, MCADD has been proposed to be a candidate for early detection through newborn screening (NBS) (2). The capability of tandem mass spectrometry (MS/MS) has opened the path for expanding NBS to a wider range of metabolic disorders including MCADD (3 10). Although MCADD screening 179

Maier et al. has been introduced in many NBS programs, it is still associated with unresolved questions (11, 12). It has revealed a wider population heterogeneity than anticipated including genotypes with as yet unpredictable clinical expressivity. Because the outcome of individuals with MCADD still remains unclear, the attempt to define biochemical phenotypes with clinical relevance is challenging (13). The genotypes of patients identified by NBS differ from those identified after metabolic decompensation (5, 14). The mutation c.985a>g is known to be most prevalent in patients with clinically manifested MCADD (80% homozygous) (15) and may thus be considered as a genotype of high-penetrance susceptibility (16). This genotype, however, was shown to be considerably less frequent in individuals identified by NBS (63%/47% homozygous) (5, 14). In NBS, numerous novel mutations including a second prevalent mutation, c.199t>c, were identified (5, 14). Patients carrying these mutations often express patterns with significantly lower concentrations of acylcarnitines (5, 17). In addition, it has been shown that c.985a>g carriers also display mildly pathological acylcarnitine concentrations (18, 19). Healthy carriers, however, should not be burdened by screening programs. As a consequence, strategies to dissect patients with mild MCADD from carriers need to be developed. Furthermore, it remains unclear whether all individuals with MCADD are recognized by the various cut-off policies. We performed a retrospective study of our NBS database comprising samples obtained within 4 years. Because both false-positives and falsenegatives can be expected to reveal borderline acylcarnitine patterns, the molecular bases of these samples were assessed with respect to the two most frequent mutations c.985a>g and c.199t>c. Our specific aims were (i) to decrease the falsepositive rate by providing biochemical features that allow to discriminate between patients carrying ACADM mutations on both alleles and healthy carriers, and (ii) to define criteria that differentiate genotypes with presumably high clinical penetrance (homozygous c.985a>g) from those with milder biochemical expression (compound heterozygous c.985a>g/c.199t>c). Patients and methods MCADD NBS and population data Acylcarnitines in dried blood spots were usually collected between 36 and 72 h after birth and were analyzed by MS/MS (20, 21). Data interpretation was performed applying a multiparametric rating system combining octanoylcarnitine (C 8 ) concentrations and secondary markers (C 6, C 10, C 10:1 ), as well as analyte ratios (C 8 /C 6,C 8 /C 10, C 8 /C 12 ). Parameters were considered increased when their value exceeded the 99.5 th percentile. Elevations of both the primary marker C 8 and at least three secondary markers in the initial screening card and the follow-up sample were regarded as being suspicious (5). In addition, single cards not fulfilling recall criteria were reinvestigated due to excessively elevated C 8. These individuals were referred for complete clinical evaluation and diagnostic workup including plasma acylcarnitines, urinary acylglycines, and mutational analysis of the complete coding region of the ACADM gene. Within 4 years, 470,247 newborns were screened. Fifty-eight MCADD patients were identified giving a frequency of 1:8108 (95% CI: 1:6277 1:10684). Thirty-one of these patients were found to be either homozygous for c.985a>g (n = 24) or compound heterozygous for c.985a>g and c.199t>c (n = 7). Three compound heterozygous (c.985a>g/c.199t>c) neonates not born in Bavaria were additionally included (Table 3, group 1b). The mutation data provided the basis to calculate expected allelic frequencies for carriers of c.985a>g (1:70) and c.199t>c (1:480), following the Hardy Weinberg equilibrium. Selection of study group and matched controls To identify mild biochemical phenotypes, the initial NBS database was queried for samples with four medium-chain acylcarnitine concentrations 95 th percentile (C 6, C 8, C 10, and C 10:1 ), that did not reach recall criteria for a suspicion of MCADD defined in our NBS program (>99.5 th percentile). Threshold concentrations were: C 6 0.61 μmol/l, C 8 0.23 μmol/l, C 10 0.12 μmol/l, and C 10:1 0.13 μmol/l. Confirmed patients were excluded. A total of 938 samples were retrieved with borderline MCADD patterns in comparison with 81 false-positive samples generated within the same time period by routine NBS. Because our program requires anonymization of filter cards at regular intervals, 333 stored blood spots were available for anonymized mutational analysis, 6.3% of these neonates had a birth weight below 1500 g. An identical number of controls with normal medium-chain acylcarnitines were collected after close matching for birth weight (maximum difference 410 g) and age of blood 180

Validation of MCADD newborn screening sampling (maximum difference 33 h). Of the 81, 68 false-positive samples were available for analysis. Genotyping of c.985a>g and c.199t>c and mutation analysis of the entire coding region of the ACADM gene Genomic DNA was extracted from dried blood spots. Briefly, blood spots were washed twice in 1.5 ml 10 mmol/l NaCl/EDTA (20 min), boiled in 150 μl 50 mmol/l NaOH (10 min), neutralized by adding 30 μl 1 mmol/l Tris-HCl (ph 7.5), and brought to a total volume of 200 μl with water. Mutations c.199t>c (exon 3) and c.985a>g (exon 11) were tested by restriction assays. Exons 3 and 4 were amplified using sense primer 5 - GTCTGAGTTTCTGATAATCAAGG-3 and antisense primer 5 -CTTACTCATATGCATTCCAG- 3. Mutation c.199t>c generates a restriction site for NlaIII (New England Biolabs). Exon 11 was amplified using a mutagenic sense primer, which introduces an NcoI restriction site in the variant c.985a>g sequence (22). Mutation analysis of the entire coding region was performed as described (5). Identification of c.985a>g and c.199t>c Three hundred and thirty-three samples fulfilled the criteria of the study group. The frequency of carriers of c.985a>g in the study group (1:4.3) was significantly higher than in the group of matched controls (1:42; p < 0.001) (Table 1). No difference in the frequency of carriers of the mutation c.199t>c (1:333) was observed. Sixty-eight newborns showed elevated acylcarnitine markers consistent with MCADD in the initial NBS, but normal patterns in the recall samples (false-positives). Thirty of these were heterozygous for c.985a>g and three for c.199t>c corresponding to a frequency of about 1:2.3 and 1:23, respectively (Table 1). These frequencies significantly differ from the expected number of heterozygous individuals for either mutation in the control group (p < 0.001). Among the newborns with borderline acylcarnitine patterns, no individuals were identified being homozygous for c.985a>g or c.199t>c or compound heterozygous for these two mutations. It has to be noted, however, that a few of the carriers might be compound heterozygous with other untested mild mutations. Results NBS samples showing borderline acylcarnitine patterns were genotyped with respect to the two most common mutations c.985a>g and c.199t>c. In Fig. 1, the subgroups are depicted in relation to increasing medium-chain acylcarnitine concentrations. They comprised confirmed MCADD patients identified by NBS (n = 34), the study group (n = 333), matched controls (n = 333), and false-positives (initial and recall samples; n = 68). Discrimination of ACADM genotypes by acylcarnitine patterns Within the study group, the carrier frequencies for c.985a>g and c.199t>c were related to the biochemical phenotype. The rate of carriers declined with decreasing C 8 concentrations (50% 15%), whereas the rate of non-carriers increased (50% 85%) (Table 2). Acylcarnitine concentrations and ratios were related to ACADM genotypes (Table 3). Patients Fig. 1. Studied subgroups in relation to increasing medium-chain acylcarnitine concentrations: matched controls and study group [based on the initial newborn screening (NBS) sample], medium-chain acyl-coa dehydrogenase deficiency (MCADD) patients and false-positives (based on the initial NBS and the recall sample). 181

Maier et al. Table 1. Carrier frequencies of the ACADM mutations c.985a>g and c.199t>c in the study group, matched controls, and false-positives Carriers of c.985a>g Carriers of c.199t>c Non-carriers of c.985a>g or c.199t>c Total Study group Number expected a 4.76 0.69 Number found 77 1 255 333 Matched controls Number expected a 4.76 0.69 Number found 8 1 324 333 False-positives Number expected a 0.97 0.14 Number found 30 3 35 68 Frequencies Carriers of Study group Matched controls False-positives c.985a>g 1:4.3 (1:3.6 1:5.4) b 1:42 (1:25 1:132) b 1:2.3 (1:1.6 1:3.4) b c.199t>c 1:333 (1:60 1:13, 162) b 1:333 (1:60 1:13, 162) b 1:23 (1:7.8 1:110) b a derived from newborn screening population data. b 95% confidence intervals. Table 2. Carrier frequencies of the ACADM mutations c.985a>g and c.199t>c related to octanoylcarnitine (C 8 ) concentrations in dried blood spots (expressed as μmol/l) Study group C 8 1.0 1.0 > C 8 > 0.5 0.5 C 8 0.23 Total Number of carriers of c.985a>g or c.199t>c 11 (50%) 36 (36%) 31 (15%) 78 (23%) Number of non-carriers of c.985a>g or c.199t>c 11 (50%) 64 (64%) 180 (85%) 255 (77%) Total 22 100 211 333 homozygous for c.985a>g (group 1a) were significantly different from patients compound heterozygous for c.985a>g and c.199t>c (group 1b) with respect to C 6,C 8,C 10:1,C 8 /C 6,C 8 /C 10, and C 8 /C 12. Group 1b was well distinguished from carriers with borderline acylcarnitine patterns (group 2) applying the markers C 6,C 8,C 10:1, C 8 /C 10,andC 8 /C 12. Within controls (group 3), the trend to higher metabolic markers in carriers with normal acylcarnitine patterns compared to noncarriers was not statistically significant. Compound heterozygous patients (group 1b) significantly differed from non-carriers with high discriminatory power as to all metabolites and ratios analyzed. Receiver-operated-curve (ROC) analysis revealed that the ratios C 8 /C 10 and C 8 /C 12 were best to discriminate c.985a>g homozygotes (group 1a) and c.985a>g/c.199t>c compound heterozygotes (group 1b) with areas under the curve (AUC) of 0.976 and 0.924, respectively (data not shown). Possible cut-off values for discrimination were 5 for C 8 /C 10 and 27 for C 8 /C 12 to achieve a sensitivity of 96% and 88%, respectively. Compound heterozygous patients (group 1b) were best discriminated from carriers with borderline acylcarnitine patterns (group 2) by the concentration of C 8 and the ratio C 8 /C 12 (AUC C 8 : 0.990; C 8 /C 12 : 0.995). The AUC for the ratio C 8 /C 10 was slightly lower (0.960). To achieve a theoretical sensitivity of 100%, cutoff values for all samples were set to 1.4 μmol/l for C 8 and 7 for C 8 /C 12, respectively and depicted as a scatter plot (Fig. 2a). Patients (group 1a+b) and carriers with borderline acylcarnitine patterns (group 2) were clearly distinguished except for four carriers (Fig. 2a, No. 1 4). To identify a potential mutation on the second allele, the entire coding regions of these specimens were sequenced. In individuals No. 1 3, mutations were identified proving that these individuals were finally MCADD patients. One mutation is novel with unknown functional consequences, the others have 182

Validation of MCADD newborn screening Table 3. Acylcarnitine concentrations (expressed as μmol/l) and acylcarnitine ratios related to various ACADM genotypes a Group n Genotype C 6 C 8 C 10 C 10:1 C 8 /C 6 C 8 /C 10 C 8 /C 12 1a b 24 c.985a>g / 2.41 ± 1.85 e 11.59 ± 5.71 e 0.77 ± 0.46 0.98 ± 0.46 e 5.66 ± 2.17 e 18.56 ± 11.98 e 70.11 ± 31.38 e c.985a>g 1b b 10 c.985a>g / c.199t>c 1.02 ± 0.50 f 2.68 ± 1.05 f 0.90 ± 0.37 0.63 ± 0.33 f 2.78 ± 0.78 3.56 ± 2.08 f 17.07 ± 6.25 f 2 c 78 Carriers for c.985a>g or c.199t>c 0.75 ± 0.24 0.66 ± 0.37 0.67 ± 0.40 0.29 ± 0.12 0.92 ± 0.45 1.12 ± 0.56 2.47 ± 1.65 3 d 333 Matched controls 0.09 ± 0.05 0.08 ± 0.04 0.09 ± 0.05 0.18 ± 0.09 1.08 ± 0.70 1.08 ± 0.73 0.80 ± 0.70 MCADD, medium-chain acyl-coa dehydrogenase deficiency; NBS, newborn screening. a All values are expressed as mean ± SD. b MCADD patients identified in NBS (1a: homozygous for c.985a>g; 1b: compound heterozygous for c.985a>g / c.199t>c). c Carriers with borderline acylcarnitine patterns of the study group. Due to the small sample size of carriers at position c.199 (n = 1), no separation of the carriers (c.985 vs. c.199) was performed. d Matched controls with normal acylcarnitine patterns including nine carriers at position c.985 or c.199. Significant differences (p < 0.05) between groups (Mann Whitney U test): e 1a and 1b. f 1b and 2. been described in NBS earlier (5, 23). In individual No. 4, no second mutation could be identified. However, cryptic splice site mutations, mutations within the promoter region and large deletions cannot be ruled out. A scatter plot was also drawn for the remaining individuals analyzed being non-carriers for both c.985a>g and c.199t>c (Fig. 2b). Five individuals were found above both cut-off values. Sequencing of the ACADM gene of these specimens revealed no variations. Technical performance of NBS for MCADD We retrospectively compared our initially utilized multiparametric rating system (5) with the use of optimized cut-off values defined in this study (1.4 μmol/l for C 8 and 7 for C 8 /C 12 ). The changes (multiparametric rating/optimized cut-off values), provoked by inclusion of the identified false-negative MCADD variants and by reduction of false-positives, were evident in terms of MCADD birth prevalence (1:8108/1:7464), positive predictive value (42%/88%), and falsepositive rate (0.017%/0.016%). Age-dependency of octanoylcarnitine concentrations in MCADD patients We addressed the question whether the biochemical identification of MCADD patients is influenced by age of sampling, as acylcarnitine concentrations increase after birth and decline subsequently (24). C 8 concentrations of 470 initial and recall samples of MCADD patients were analyzed. We observed an increase of C 8 concentrations between days 1 and 2 3 followed by a highly significant decrease (p < 0.001) between days 2 3 and 4 5, and days 4 5 and 6 7, respectively (Fig. 3). In contrast, the 99.5 th percentile of C 8 concentrations (0.4 μmol/l) of 376,657 healthy individuals did not change significantly. Discussion The spectrum of ACADM genotypes identified by NBS is heterogeneous and includes novel variants showing milder biochemical phenotypes (5, 14, 17). First outcome data revealed that MCADD patients, and in particular patients homozygous for c.985a>g, benefit from early detection by showing a reduced incidence of metabolic crisis or death compared with unscreened patients (13, 25). However, the natural course of the disease in newborns with novel mutations is unknown. Many of these individuals may remain asymptomatic even without detection by NBS. Nevertheless, the so far available evidence is not conclusive to safely propose that these genotypes are conferring an asymptomatic MCADD condition. Consequently, most current NBS programs still aim at detecting the full spectrum of MCADD including the milder variants. Thus, the clear distinction of carriers with mild biochemical abnormalities (18, 19) from genotypes with variants on both alleles and milder biochemical expression is of high practical importance. 183

Maier et al. (a) (b) Fig. 2. (a) Scatter plot of the best discriminating parameters [C 8 (μmol/l) and C 8 /C 12 ] to distinguish 34 medium-chain acyl-coa dehydrogenase deficiency (MCADD) patients from carriers with borderline acylcarnitines in the initial newborn screening (NBS) sample. Circles symbolize MCADD patients (open circles with genotype c.985a>g/c.985a>g n = 24; closed circles with genotype c.985a>g/c.199t>c, n = 10), whereas triangles symbolize carriers. Only four presumed carriers showed metabolic markers consistent with MCADD (No. 1 4); three samples (stars, No. 2 4) were retrospectively proven to be MCADD patients (for details, see Table 4). (b) Scatter plot for all the remaining individuals being non-carriers for both c.985a>g and c.199t>c symbolized by open squares (study group and matched controls) and closed squares (false-positives). Only five individuals were found above both defined cut-off values. All five specimens were sequenced and revealed no variations. Two individuals showed C 8 /C 12 values below 0.1 (not shown). In a retrospective, cross-sectional case control study, we searched for the two most prevalent ACADM mutations in individuals that had been designated as MCADD negative in routine NBS, but showed acylcarnitine markers 95.0 th percentile. They were labeled as borderline MCADD pattern and considered to have an increased likelihood to comprise both false-negatives and false-positives. Recalled samples from NBS that had been designated upon follow-up as falsepositives were additionally genotyped. The rate of carriers for c.985a>g in the study group was about 10-fold higher than in matched controls. The high carrier frequency found for c.985a>g 184

Validation of MCADD newborn screening Fig. 3. Blood spot octanoylcarnitine concentrations [C 8 (μmol/l)] in 470 initial NBS and recall samples of confirmed MCADD patients. C 8 concentrations increase (p = 0.09) between day 1 and days 2 3 and decline with increasing age showing a highly significant change between days 2 3 and 4 5, and days 4 5 and 6 7, respectively (p< 0.001). The 99.5 th percentile (0.4 μmol/l; horizontal line) of C 8 concentrations in healthy newborns did not change significantly. Box plots display the 25 th and 75 th percentiles as lower and upper boundaries of the boxes. The line within the boxes represents the median (50 th percentile). Whiskers above and below the boxes indicate the 95% confidence interval of the mean. Highly significant changes are indicated ( p < 0.001). of 1:4.3 compares to a rate of 1:70 we extrapolate from birth incidences and allelic frequencies of confirmed MCADD in our NBS population. As expected, the carrier frequency for c.985a>g was even higher (1:2.3) in false-positives and was related to the extent of C 8 elevation we observed in the initial NBS. The higher the level of C 8 above the 95.0 th percentile was, the higher the likelihood of ACADM heterozygosity. In contrast, c.199c>t carriers were not found more frequently in the study group when compared to controls (1:333). In both groups, the expected number of carriers was observed indicating that carriers of this mutation do not persistently contribute to borderline screening results. This finding is consistent with the hypothesis of c.199c>t being a mild mutation (14). However, a 14-fold higher frequency of c.199c>t carriers in the group of false-positives was identified indicating that c.199c>t carriers can show transient abnormalities. The mutation has been shown to be a temperature sensitive folding variant, which is usually mild but may have an impact on protein function at increased temperatures (26). We subsequently analyzed the borderline acylcarnitine markers of the study group in relation to those measured in patients unraveled by routine NBS (5). A tendency of differing acylcarnitine profiles between various ACADM genotypes has been reported (5, 14, 17, 27). Our approach, however, revealed for the first time, that compound heterozygotes (c.985a>g/c.199t>c, group 1b) can biochemically be discriminated with almost negligible overlap from healthy carriers (unwanted information in NBS) by the markers C 8 and the ratio C 8 /C 12. c.985a>g homozygotes (group 1a) with severe biochemical phenotypes served as control cohort. This group was best differentiated from the compound heterozygotes by the ratios C 8 /C 10 and C 8 /C 12. It is noteworthy, however, that the defined cut-offs are difficult to translate to other NBS programs unless the methodology is exactly comparable. Table 4. Identified mutations and biochemical phenotypes [initial newborn screening (NBS) and recall sample] of four retrospectively analyzed newborns showing metabolic markers consistent with medium-chain acyl-coa dehydrogenase deficiency (MCADD) a Case-no. Genotype Age b C 0 (μmol/l) C 6 (μmol/l) C 8 (μmol/l) C 10 (μmol/l) C 10:1 (μmol/l) C 8 /C 6 C 8 /C 10 C 8 /C 12 51.30 c 0.83 c 0.39 c 0.50 c 0.41 c 4.50 c 10.00 c 4.00 c 1 c.50g>a / 3 20.59 1.23 2.68 1.86 0.93 2.18 1.44 8.38 c.985a>g 8 16.25 0.45 0.31 0.18 0.13 0.69 1.94 2.82 2 c.698t>c / 4 18.08 0.73 1.64 0.80 0.43 2.24 2.05 8.63 c.985a>g 9 21.40 0.71 0.60 0.27 0.20 0.84 2.22 6.00 3 c.797a>g / 3 47.10 0.83 2.32 0.65 0.41 2.79 3.57 7.25 c.985a>g 10 27.10 0.18 0.57 0.10 0.12 3.17 5.70 6.33 4 d n.i. e / 3 30.76 1.84 3.44 0.41 0.16 1.87 8.39 11.86 c.199t>c 12 26.25 0.85 1.03 0.16 0.24 1.21 6.44 20.60 a Markers exceeding the defined cut-offs are shown bold. b Age at blood sample collection in days. c 99.5 th percentile. d Not fulfilling recall criteria, but reinvestigated due to excessively elevated C 8. e Not identified. 185

Maier et al. Four designated carriers showed biochemical features consistent with MCADD (Fig. 2a). Subsequently, we sequenced their entire ACADM coding regions and identified mutations on the second allele in three of four cases. From a biochemical and genetic point of view, these individuals are to be designated retrospectively as false-negatives. They escaped detection as our initial interpretation algorithm required both an elevation of C 8 and at least three secondary markers above defined thresholds in both the initial and the followup sample. The individuals were recalled due to clearly abnormal acylcarnitine profiles in their initial samples, but the follow-up samples showed a marked decline not fulfilling our criteria any more. The initial algorithm was clearly too strict and individuals with comparable acylcarnitine patterns would not have escaped detection in most routine NBS programs. Acylcarnitine concentrations have been shown to increase after birth and decline subsequently (24). To consider this highly significant decrease of C 8 over time is crucial, as many screening protocols rely primarily on C 8 concentrations (Fig. 3). This phenomenon led to missing the diagnosis of glutaric acidemia, type I in a screened infant (28). We early redefined our interpretation algorithm and performed a complete clinical evaluation and diagnostic workup including plasma acylcarnitines, urinary acylglycines, and mutational analysis of any individual with an abnormal acylcarnitine pattern in the initial blood spot. Currently, elevated concentrations of C 8 and the two ratios C 8 /C 10 and C 8 /C 12 are regarded as being suspicious. In addition, C 8 concentrations above 1 μmol/l result in further diagnostic workup. From our data, an improved screening strategy might be delineated that allows a reduction of both false-positive and false-negative rates. The clear distinction of affected individuals from carriers decreases the false-positive rate and can be achieved by defining high biochemical cut-offs (C 8 and C 8 /C 12 ). The reduction in the false-negative rate can be achieved by sequencing the entire coding region of all initially suspicious samples (Fig. 2a,b). Due to the high specificity, this would involve only about two samples per 100,000 newborns. This strategy minimizes the potential burden and health care costs arising from follow-ups. It will remain a matter of debate, whether it is worthwhile to identify all ACADM variants in NBS including those where adverse outcome is not yet documented. As long as the outcome cannot be predicted, one might argue that MCADD is defined only on biochemical and molecular grounds. Findings in clinically diagnosed patients including residual enzyme activity might help to separate milder conditions (29). With gaining experience, the diagnosis of MCADD should be further refined (13). This issue applies to all diseases included in NBS programs. Our data devise biochemical features that allow to safely differentiate (i) healthy carriers from patients and (ii) ACADM genotypes with presumably high clinical penetrance (homozygous c.985a>g) from those with milder biochemical expression (compound heterozygous c.985a>g/c.199t>c) that have not shown to manifest clinically so far. In theory, this provides a technical tool to specifically target NBS for MCADD to only those subsets with a higher risk for adverse outcome. In addition, the burden of suffering from unwanted information (i.e. detection of carriers) or from equivocal results can be kept very low. Because the natural course of mild disease variants detectable by NBS cannot reliably be assessed by observational studies, the described validation strategy could serve as a general model for evaluation of MS/MS-based NBS. Acknowledgements This work was supported by the Bavarian Genome Research Network (BayGene) and SHS International (Gesellschaft für klinische Ernährung mbh). We are grateful to Prof. Müller- Mysok and Dr. Müller for helpful statistical discussions and the team at the Public Health Newborn Screening Center of the State of Bavaria for technical assistance. This article is part of a thesis by Julia Pongratz to fulfill the requirements for a medical degree at the Ludwig-Maximilians-University of Munich. References 1. Grosse SD, Khoury MJ, Greene CL et al. The epidemiology of medium chain acyl-coa dehydrogenase deficiency: an update. Genet Med 2006: 8 (4): 205 212. 2. Wilson CJ, Champion MP, Collins JE et al. Outcome of medium chain acyl-coa dehydrogenase deficiency after diagnosis. Arch Dis Child 1999: 80 (5): 459 462. 3. Chace DH, Hillman SL, Van Hove JL et al. Rapid diagnosis of MCAD deficiency: quantitative analysis of octanoylcarnitine and other acylcarnitines in newborn blood spots by tandem mass spectrometry. Clin Chem 1997: 43 (11): 2106 2113. 4. Clayton PT, Doig M, Ghafari S et al. Screening for medium chain acyl-coa dehydrogenase deficiency using electrospray ionisation tandem mass spectrometry. Arch Dis Child 1998: 79 (2): 109 115. 5. Maier EM, Liebl B, Röschinger W et al. Population spectrum of ACADM genotypes correlated to biochemical phenotypes in newborn screening for medium-chain acyl-coa dehydrogenase deficiency. Hum Mutat 2005: 25: 443 452. 186

Validation of MCADD newborn screening 6. Röschinger W, Olgemöller B, Fingerhut R et al. Advances in analytical mass spectrometry to improve screening for inherited metabolic diseases. Eur J Pediatr 2003: 162 (Suppl. 1): 67 76. 7. Carpenter K, Wiley V, Sim KG et al. Evaluation of newborn screening for medium chain acyl-coa dehydrogenase deficiency in 275 000 babies. Arch Dis Child Fetal Neonatal Ed 2001: 85 (2): F105 F109. 8. Zytkovicz TH, Fitzgerald EF, Marsden D et al. Tandem mass spectrometric analysis for amino, organic, and fatty acid disorders in newborn dried blood spots: a two-year summary from the New England Newborn Screening Program. Clin Chem 2001: 47 (11): 1945 1955. 9. Wilcken B, Wiley V, Hammond J et al. Screening newborns for inborn errors of metabolism by tandem mass spectrometry. N Engl J Med 2003: 348 (23): 2304 2312. 10. Schulze A, Lindner M, Kohlmüller D et al. Expanded newborn screening for inborn errors of metabolism by electrospray ionization-tandem mass spectrometry: results, outcome, and implications. Pediatrics 2003: 111 (6 Pt 1): 1399 1406. 11. Leonard JV, Dezateux C. Screening for inherited metabolic disease in newborn infants using tandem mass spectrometry. BMJ 2002: 324 (7328): 4 5. 12. Dezateux C. Newborn screening for medium chain acyl-coa dehydrogenase deficiency: evaluating the effects on outcome. Eur J Pediatr 2003: 162 (Suppl 1): S25 S28. 13. Wilcken B, Haas M, Joy P et al. Outcome of neonatal screening for medium-chain acyl-coa dehydrogenase deficiency in Australia: a cohort study. Lancet 2007: 369 (9555): 37 42. 14. Andresen BS, Dobrowolski SF, O Reilly L et al. Mediumchain acyl-coa dehydrogenase (MCAD) mutations identified by MS/MS-based prospective screening of newborns differ from those observed in patients with clinical symptoms: identification and characterization of a new, prevalent mutation that results in mild MCAD deficiency. Am J Hum Genet 2001: 68 (6): 1408 1418. 15. Yokota I, Coates PM, Hale DE et al. Molecular survey of a prevalent mutation, 985A-to-G transition, and identification of five infrequent mutations in the medium-chain Acyl-CoA dehydrogenase (MCAD) gene in 55 patients with MCAD deficiency. Am J Hum Genet 1991: 49 (6): 1280 1291. 16. Zlotogora J. Penetrance and expressivity in the molecular age. Genet Med 2003: 5 (5): 347 352. 17. Waddell L, Wiley V, Carpenter K et al. Medium-chain acyl-coa dehydrogenase deficiency: genotype-biochemical phenotype correlations. Mol Genet Metab 2006: 87 (1): 32 39. 18. Lehotay DC, LePage J, Thompson JR et al. Blood acylcarnitine levels in normal newborns and heterozygotes for medium-chain acyl-coa dehydrogenase deficiency: a relationship between genotype and biochemical phenotype? J Inherit Metab Dis 2004: 27 (1): 81 88. 19. Blois B, Riddell C, Dooley K et al. Newborns with C8- acylcarnitine level over the 90th centile have an increased frequency of the common MCAD 985A >G mutation. J Inherit Metab Dis 2005: 28 (4): 551 556. 20. Fingerhut R, Röschinger W, Muntau AC et al. Hepatic carnitine palmitoyltransferase I deficiency: acylcarnitine profiles in blood spots are highly specific. Clin Chem 2001: 47 (10): 1763 1768. 21. Liebl B, Nennstiel-Ratzel U, von Kries R et al. Expanded newborn screening in Bavaria: tracking to achieve requested repeat testing. Prev Med 2002: 34 (2): 132 137. 22. Yokota I, Indo Y, Coates PM et al. Molecular basis of medium chain acyl-coenzyme A dehydrogenase deficiency. An A to G transition at position 985 that causes a lysine- 304 to glutamate substitution in the mature protein is the single prevalent mutation. J Clin Invest 1990: 86 (3): 1000 1003. 23. Rhead WJ. Newborn screening for medium-chain acyl-coa dehydrogenase deficiency: a global perspective. J Inherit Metab Dis 2006: 29 (2 3): 370 377. 24. Meyburg J, Schulze A, Kohlmüller D et al. Postnatal changes in neonatal acylcarnitine profile. Pediatr Res 2001: 49 (1): 125 129. 25. Nennstiel-Ratzel U, Arenz S, Maier EM et al. Reduced incidence of severe metabolic crisis or death in children with medium chain acyl-coa dehydrogenase deficiency homozygous for c.985a >G identified by neonatal screening. Mol Genet Metab 2005: 85 (2): 157 159. 26. O Reilly L, Bross P, Corydon TJ et al. The Y42H mutation in medium-chain acyl-coa dehydrogenase, which is prevalent in babies identified by MS/MS-based newborn screening, is temperature sensitive. Eur J Biochem 2004: 271 (20): 4053 4063. 27. Nichols MJ, Saavedra-Matiz CA, Pass KA et al. Novel mutations causing medium chain acyl-coa dehydrogenase deficiency: under-representation of the common c.985 A > G mutation in the New York state population. Am J Med Genet A 2008: 146A (5): 610 619. 28. Smith WE, Millington DS, Koeberl DD et al. Glutaric acidemia, type I, missed by newborn screening in an infant with dystonia following promethazine administration. Pediatrics 2001: 107 (5): 1184 1187. 29. Derks TG, Boer TS, van Assen A et al. Neonatal screening for medium-chain acyl-coa dehydrogenase (MCAD) deficiency in The Netherlands: the importance of enzyme analysis to ascertain true MCAD deficiency. J Inherit Metab Dis 2008: 31 (1): 88 96. 187