Use of the phenylalanine:tyrosine ratio to test newborns for phenylketonuria in a large public health screening programme

Transcription

1 J Med Screen 2000;7: Use of the phenylalanine:tyrosine ratio to test newborns for phenylketonuria in a large public health screening programme J W Eastman, J E Sherwin, R Wong, C L Liao, R J Currier, F Lorey, G Cunningham California Department of Health Services, 2151 Berkeley Way, Annex 9, Berkeley, CA 94704, USA J W Eastman J E Sherwin R Wong C L Liao R J Currier F Lorey G Cunningham Correspondence to: Dr J W Eastman, Genetic Disease Laboratory, California Department of Health Services, 700 Heinz Avenue, Suite 100, Berkeley, CA 94710, USA. jeastman@dhs.ca.gov Abstract Objective To assess the benefits of using the phenylalanine:tyrosine ratio to screen newborns for phenylketonuria (PKU). Setting Data were collected from all newborns in California during a ten month period (n = ). Methods Dried blood spot specimens were analysed at nine laboratories. To assure that the results reported from multiple sites were matched accurately, an automated methodology was chosen that included sample processing, analysis, telecommunications, reporting, and information technology. Phenylalanine and tyrosine concentrations were measured independently by continuous flow fluorometry, for which precision, recovery, detection limits, carryover, chemical specificity, reportable range, and number of repeats are reported. Results In this study, 37% of the newborns were tested at less than 24 hours of age. For this population, using a phenylalanine only cut ov of 200 µmol/l, there were 48 recalled infants per case of classic PKU. Using the phenylalanine:tyrosine ratio with a cut ov of 1.50, screen positives could be reported with phenylalanine as low as 150 µmol/l and with only 12 recalls per case. Conclusions The phenylalanine:tyrosine ratio can be measured accurately at multiple laboratories using two channel chemical analyses. Having applied the methods to the routine clinical screening of a large population, it was confirmed that the clinical sensitivity and specificity of the PKU screening test are higher when the phenylalanine:tyrosine ratio is incorporated into the cut ov than when the cut ov is based on the phenylalanine concentration alone. (J Med Screen 2000;7: ) Keywords: neonatal testing; automated assays; phenylketonuria Phenylketonuria (PKU) is an autosomal recessive disorder caused by a primary deficiency of phenylalanine hydroxylase or impaired metabolism of its tetrahydrobiopterin cofactor. 1 To prevent severe, irreversible mental retardation, avected infants are fed a diet that is low in phenylalanine. The treatment must be initiated soon after birth. Neonatal screening to determine blood phenylalanine concentrations is the best method for early detection. The concentration increases with age as the newborn is fed. If screening levels exceed a cut ov of approximately 200 µmol/l, the infants are referred for diagnosis. The early discharge of newborns, with specimens collected before 24 hours of age, has complicated the setting of cut ovs, 2 3 because feeding has not yet caused the blood phenylalanine concentration to increase. Although the phenylalanine cut ov can be reduced to minimise the risk for missing a case, this is done at the expense of specificity. 3 4 The measurement of tyrosine, calculation of the phenylalanine:tyrosine ratio, and use of this ratio as a cut ov can help to optimise the clinical sensitivity and specificity of the test In the past, the phenylalanine:tyrosine ratio has been used to identify PKU carriers, diverentiate between PKU and non-pku hyperphenylalanemia, 15 and monitor methotrexate cancer treatment. 16 Materials and methods The equipment and reagents are supplied to the California state testing laboratories under a comprehensive reagent instrument service contract. In response to an oycial request for information, two vendors one using high performance liquid chromatography (HPLC) and the other fluorometric analysis were able to supply automated test systems for the distributed network of satellite laboratories. In a side by side comparison, a two channel fluorometric system 17 demonstrated superior automation, analytical precision, and information technology, and this method was chosen for the screening programme. Phenylalanine is determined by an adaptation of the McCaman Robins procedure. Tyrosine is determined by a nitroso-naphthol method. 20 The analyses are performed on 16 instruments at nine test sites (eight screening laboratories and one central quality control laboratory). Specimens are barcoded, and positive sample identification is maintained from the punching of the dried blood spot specimens to the printing of the physicians reports. The reports are programmed to include appropriate interpretation and recommendations for follow up. For determinations of phenylalanine and tyrosine, the reagents and instrumentation were provided by Astoria-Pacific, Inc, Clackamas, Oregon, USA (API) via subcontract to EG&G Wallac Inc, San Ramon, California, USA (Wallac). The reagent kits were the API Spotcheck Phenylalanine Kit and API Spotcheck Tyrosine Kit We used broad calibration ranges (phenylalanine

2 132 Eastman, Sherwin, Wong, et al µmol/l; tyrosine µmol/l) in order to minimise the number of required dilutions. The instrumentation was Astoria- Pacific s API Spotcheck Micro Flow Analyzer. The analyser included FASPac software for interpretation of fluorescence tracings, including a partial correction for carryover. Wallac supplied the NeoGO barcode reader/punch, the Tomtec Quadra 96 automatic liquid handling instrument, as well as the specimen tracking positive identification system and the MultiCalc software for calculation of concentrations and scoring of quality control results. Wallac provided the technical service and telecommunications as a part of a comprehensive reagent instrument service purchase agreement. Newborn heelstick blood is absorbed onto S&S 903 specimen collection paper (Schleicher & Schuell, Keene, New Hampshire, USA). Two discs 6 mm in diameter are punched from the dried blood spot specimen, and the blood is eluted with 850 µl of water for 30 minutes with periodic shaking. This single eluate is used for determination of PKU, galactosaemia and haemoglobinopathies. A common aspirate of the eluate is split into separate channels for determination of the phenylalanine and tyrosine. These assays are calibrated with water solutions such that the eluate corresponds to a 38-fold dilution of whole blood applied directly to the paper from a newborn heelstick. Before the release of results to physicians, test results are telecommunicated to the central site for review. Approximately 10% of runs are repeated because either one or the other analyte has a quality control result that is beyond acceptability limits. This number of repeats is consistent with the use of 95% acceptability limits for each of the two analytes and both of the methods are under statistical control. Results LABORATORY PERFORMANCE Precision From six runs of replicate analysis at the central testing site, the following median CV results were obtained. At concentrations between 120 and 620 µmol/l phenylalanine and tyrosine in water solutions (n = 55 at each concentration), CV = 2.2 % for phenylalanine, 4.5% for tyrosine, and 4.4% for the ratio. For blood spot reference samples at concentrations between 80 and 800 µmol/l (n = 120), CV = 6.8% for phenylalanine, 8.1% for tyrosine, and 6.9% for the ratio. For pooled eluate (multiple spots eluted into one eluate) at concentrations between 80 and 800 µmol/l (n = 48), CV = 4.5% for phenylalanine, 7.0% for tyrosine, and 7.1% for the ratio. Similar CV values were observed at the remaining eight test sites when n = 1312 dried blood spot analyses were performed on the 16 instruments. Accuracy Recovery was determined at the central laboratory by measuring test results on several runs as a function of enrichment concentration at four levels. For comparison, the pools of samples were sent to a reference laboratory for determination of recovery by MS-MS (tandem mass spectrometry; Neo Gen Screening, Inc, Pittsburgh, Pennsylvania, USA). The recovery by fluorometry (r > 0.999) was phenylalanine 84% and tyrosine 89%; and by MS-MS, phenylalanine 81% and tyrosine 86%. When blood spot reference samples (n = 1312) were analysed at the satellite laboratories, there was no detectable systemwide diverence in concentration from the target values (t < 2). The maximum bias observed on any single instrument was less than ±5% for phenylalanine, tyrosine, and the ratio. Carryover No carryover was observed for phenylalanine. However, on average throughout the system of 16 instruments, a tyrosine result within the normal range was found to have an increase of +10% when the preceding sample was high ( > 500 µmol/l). Detection limit and reportable range The highest results observed on a pure water sample preceded by a variety of high and low samples (n = 246) were 20 µmol/l for phenylalanine and 34 µmol/l for tyrosine. These values also happen to equal the mean of the water data plus five standard deviations. We have defined these levels as the detection limits. The reportable range for phenylalanine is the range between the detection limit and the concentration of the calibration range extended with dilution: µmol/l. The reportable range for tyrosine is the range between the detection limit and the concentration of the calibrator with the highest concentration: µmol/l (no dilutions made). POPULATION DATA During 10 months of pilot screening of newborns (age < 2 days), we found the following frequency distributions (median, 1st 99th percentile): phenylalanine 111, µmol/l (n = ); tyrosine 150, µmol/l (n = ); and the ratio 0.75, (n = ). (The lower n value occurs because during the pilot study the out of control tyrosine assays were not repeated.) During the pilot period there were 13 cases of classic PKU, and the results were distributed as follows (median, range): phenylalanine 401, µmol/l; tyrosine 119, µmol/l; ratio 3.49, Archived newborn specimens were retrieved from freezer storage to help establish cut ov levels. These specimens had been tested previously using the prior analysis method and stored in plastic bags for up to eight years without desiccant at nominally 15 o C. Specimens were retrieved in groups of three: one from a newborn with PKU with two additional adjacent non-pku specimens as controls. For the 162 controls the median tyrosine concentration (150 µmol/l) for the retrieved specimens was the same as that for new specimens. The median phenylalanine (115 µmol/l) and the ratio (0.79) were slightly higher (<+5%)

3 Phenylalanine:tyrosine ratio to screen PKU 133 Phenylalanine/tyrosine molar ratio PHE RTO 21 PHE RTO+ 403,601 PHE RTO 131 PHE+ RTO+ 492 PHE+ RTO 6 PHE+ RTO Phenylalanine (µmol/l) The number of screen negatives ( ) and positives (+) for newborns using cut ovs based on the phenylalanine concentration only (PHE) and the phenylalanine:tyrosine ratio (RTO). The RTO screen is negative when ratio > 1.5 and phenylalanine < 150 µmol/l, and the RTO screen is positive when ratio < 1.5 and phenylalanine > 400 µmol/l. In this population there were 13 cases of classic PKU. For phenylalanine only screening, the number of presumptive positives is 15.6 per newborns (48 recalls per case of classic PKU). When the ratio is used, the number of positives is 3.9 per (12 recalls per case). than the medians on new specimens. The diverence did not depend on the duration of storage, from one to eight years. The distributions for the data on 81 known PKU cases with specimens retrieved from freezer storage are (median, range): phenylalanine 364.9, µmol/l; tyrosine 120.4; µmol/l; and ratio 2.97, CLINICAL SENSITIVITY AND SPECIFICITY The population data from the current study are summarised in fig 1. Cut ov values were set using historical practice as well as results obtained by testing the current and freezer stored specimens. When phenylalanine alone is used to screen newborns for PKU, the cut ov is 200 µmol/l. The cut ov is 1.50 for the phenylalanine:tyrosine ratio, with two exceptions. When the ratio is less than 1.50 and the phenylalanine concentration exceeds 400 µmol/l, the newborn is recalled to obtain a second sample. Also, when the ratio is greater than 1.50 and the phenylalanine concentration is less than 150 µmol/l, the newborn is not recalled. For phenylalanine only screening, the number of presumptive positives is 15.6 per newborns (48 recalls per case of classic PKU). When the ratio is used, the number of positives is 3.9 per (12 recalls per case). Discussion RATIO SCREENING The phenylalanine:tyrosine ratio is used because interferences from non-specific physiological elevations of phenylalanine are eliminated by measurement of the substrate:product ratio of the hydroxylase reaction. Additionally, when the ratio is used, the cut ov does not depend on the number that one assumes for the volume of blood that is eluted for analysis from the blood collection card, and small variations in the collection paper are cancelled out. It is not necessary to prescribe a particular measurement process. The ratio has good predictive power when the eluate from a dried blood spot specimen is split for independent measurement of phenylalanine and tyrosine by diverent chemical methods. Ideally, one may wish to measure both analytes with the same detection technology, such as HPLC, MS-MS, or ion exchange chromatography. We used the two channel methodology because it incorporated the automation, telecommunications, and information technology required by our large, distributed public health screening system. Precise, accurate, and well matched results were achieved among 16 instruments at nine test sites. METHOD PERFORMANCE The results of 4 9% CV for test precision are comparable with those reported for MS-MS 6 and ion exchange-hplc 8 (range 6 12% CV). Also, the recovery values are consistent with published data. Recovery of phenylalanine or tyrosine has been reported at 74 99% by fluorometry, 82 92% by reversed phase HPLC, % by ion exchange-hplc, 8 and % by MS-MS. 6 The calibration of the API method assumes a concentration of blood on the collection card of 0.38 µl/mm 2. We attribute the recovery of less than 100% to adiverence between the direct application of heelstick blood, relating to the cited absorbency, and the application of surrogate blood in the laboratory to produce a concentration of 0.32 µl/mm The API software adequately corrects the carryover in the fluorometer tracings for phenylalanine, but additional correction is needed for tyrosine. Carryover in both channels of analysis is observed on water samples, and the detection limit takes this into account. For both analytes, the detection limits are close to one half of the concentration of the non-zero calibrator at the low end of the calibration curve. The detection limits are significantly higher than those reported for MS-MS 6 and for ion exchange-hplc, 8 approximately three to 10 µmol/l for both analytes. Nevertheless, the detection limits for the fluorometric method are less than one half the concentrations found in newborn specimens at the lowest one percentile of the population. Population data are consistent with published averages for phenylalanine (range µmol/l) and tyrosine (range µmol/l), obtained from the fluorometric measurement of serum samples, and for phenylalanine (91 µmol/l) and tyrosine (110 µmol/l) on dried blood spot samples (For phenylalanine and tyrosine, the concentrations in the plasma or serum are comparable with the concentrations in whole blood. 26 ) (Measurement of the ratio by fluorometry has not been systematically studied before.) Chemical interferences in the fluorometric methods can be inferred from a comparison of our results with the population mean values for phenylalanine and tyrosine determined by ionexchange chromatography, HPLC, 5 31 MS-MS, 6 and enzymatic methods for plasma and dried blood spot samples. Compared with the literature data for ion-exchange

4 134 Eastman, Sherwin, Wong, et al Table 1 Birth prevalence of classic PKU chromatography and MS-MS, the population medians determined in our study are higher by approximately 60 µmol/l and 100 µmol/l respectively for phenylalanine and tyrosine. We attribute this to method interference. Ninhydrin reactive compounds such as amino acids and peptides can interfere with the phenylalanine method. Hydroxyphenyl acids and tyramine cause interference with the tyrosine method (API product labeling). The ratio distribution did not show the evects of interference. The median result in this work for the ratio is very close to values reported on control adult and newborn populations of by ion-exchange chromatography, and 0.78 by MS-MS. 6 CUT OFFS Historically, California has used a cut ov of 200 µmol/l for phenylalanine screening (scaled to the new method). The results on freezer stored specimens and the pilot population have led us to adopt 1.50 as the cut ov for the ratio. This is consistent with the published range of The numbers of screen negatives and positives found in our study are given in fig 1. The six results with the ratio < 1.50 but phenylalanine > 400 µmol/l (fig 1) were managed as presumptive positives, so the newborn would be recalled for a second heelstick. Additionally, the 130 results with the ratio > 1.50 but with phenylalanine < 150 µmol/l are classed as negatives (fig 1). This is because we have found that very low tyrosine concentrations (and ratio > 1.50) are correlated with low birth weight and intensive nursery care when phenylalanine is below 150 µmol/l, and these infants do not have PKU. When tyrosine results are > 700 µmol/l, the screening programme recommends that another dried blood spot specimen be collected, and free testing on the second specimen is provided. Years This work (July 1998 May 1999) All ethnicity Screened Cases Screened per case Cases per White only Screened ND* Cases Screened per case Cases per * ND, not determined. POPULATION SCREENING When phenylalanine only is used for PKU screening, the positive predictive value decreases with increasing numbers of newborns tested at an early age In the population reported here, 37% of the newborns were tested at an age <24 hours, and with the phenylalanine only cut ov, there were 48 recalls per case of classic PKU. When the ratio was used, there were 12 recalls per case. For other populations, the utility of the ratio for routine screening of newborns will depend on the prevalence, the population characteristics, and the cut ovs chosen. For example, during the period , an average 50% of newborns in California were tested before 24 hours of age. For this population, we have estimated the performance characteristics for ratio screening. With a phenylalanine cut ov of 200 µmol/l for phenylalanine only screening, and a ratio cut ov of 1.58 for ratio screening (normalised to the same test sensitivity), the number of recalls per case would have been 58 for phenylalanine screening and 5.2 for ratio screening. In a review of California screening data reported during the period 1980 to the present, we have identified three cases of classic PKU for which the phenylalanine concentrations in the screening specimens (130, 163, and 190 µmol/l when scaled to the current method) were below cut ov (200 µmol/ l). Although contrary to programme policy, the specimens had been collected at 2 4 hours after birth. Two of the these specimens are no longer available, but analysis of the third using the new methods has confirmed the low phenylalanine (initial 190, repeat 181 µmol/l). In this case the phenylalanine:tyrosine ratio of 1.84 was screen positive (cut ov 1.50). This result, and the capability to report positives at concentrations of phenylalanine as low as 150 µmol/l, underscore the improved sensitivity of ratio screening. For diagnosis of classic PKU or variant hyperphenylalaninaemia, newborns are referred to a metabolic specialist. First, each newborn with a presumptive positive screening result is recalled to obtain a second dried blood spot specimen. The same screening methodology and cut ovs are applied to the confirmatory specimen. When the second result is positive, the newborn is referred to the specialist. The patients are classified according to the concentration of phenylalanine and other metabolic criteria. Although the focus of the programme is to screen for classic PKU, we have examined the registry of outcomes to determine if the implementation of ratio screening has avected the number of cases classified as variant. Before implementation, an average 21 variants were logged per year (range 16 31). In the 12 month period following implementation of ratio screening, 12 variants were logged. With ratio screening, it appears that fewer newborns are identified with non- PKU hyperphenylalaninaemia. The birth prevalence for classic PKU in the California population is presented in table 1. During this pilot phase of screening, the prevalence in Caucasians was not determined, because the breakdown of results by ethnicity was not readily available. The birth prevalence of PKU in California is consistent with published information concerning this condition, 1 and with the percentage of Caucasians in the state population. Conclusions Previous reports had demonstrated the advantages of ratio screening. 5 9 We show that the ratio can be applied successfully to the routine clinical testing in a large scale public health screening programme. Also, we show that the advantages of ratio screening do not require

5 Phenylalanine:tyrosine ratio to screen PKU 135 that both analytes be measured with the same method of detection. Previous authors used a common detection system, such as HPLC, 5 7 MS-MS 6 9 or ion exchange chromatography. 8 We show that the use of the phenylalanine: tyrosine ratio is a powerful screening approach, even when the eluate from a dried blood spot is split into two channels for independent chemical analysis of the analytes. This underscores the fundamental rationale for ratio screening, namely, the enzyme is characterised best by measuring both the substrate and the product of the reaction. Also, any variations in calibration for the amount of blood eluted from the collection card are cancelled when the ratio is used. 1 Scriver CF, Kaufman S, Eisensmith RC, et al. The hyperphenylalaninemias. In: Scriver CR, Beaudet AL, Sly WS, et al, eds.the metabolic and molecular bases of inherited disease. Vol. 1, 7th ed. New York: McGraw Hill, 1995: Jew K, Kan K, Koch R,et al. Validity of screening early collected newborn specimens for phenylketonuria using a fluorometric method. Screening 1994;3: McCabe ERB, McCabe L, Mosher GA, et al. Newborn screening for phenylketonuria: predictive validity as a function of age. Pediatrics 1983;72: Arnopp JJ, Lorey FW, Currier RJ,et al. Results of screening for phenylketonuria using a lower cutov value in early collected specimens. Screening 1995;3: Qu Y, Miller JB, Slocum RH, et al. Rapid automated quantitation of isoleucine, leucine, tyrosine and phenylalanine from dried blood filter paper specimens. Clin Chim Acta 1991:203: Chace DH, Millington DS, Terada N,et al. Rapid diagnosis of phenylketonuria by quantitative analysis for phenylalanine and tyrosine in neonatal blood spots by tandem mass spectrometry. Clin Chem 1993;39: Levy PA, Miller JB, Shapira E. The advantage of phenylalanine to tyrosine ratio for the early detection of phenylketonuria. Clin Chim Acta 1998;270: Reilly AA, Bellisario R, Pass KA. Multivariate discrimination for phenylketonuria (PKU) and non-pku hyperphenylalaninemia after analysis of newborns dried blood-spot specimens for six amino acids by ion-exchange chromatography. Clin Chem 1998;44: Chace DH, Sherwin JE, Hillman SL, et al. Use of the phenylalanine-to-tyrosine ratio determined by tandem mass spectrometry to improve newborn screening for phenylketonuria of early discharge specimens collected in the first 24 hours. Clin Chem 1998;44: Perry TL, Tischler B, Hansen S, et al. A simple test for heterozygosity for phenylketonuria. Clin Chim Acta 1967;15: Perry TL, Hansen S, Tischler B, et al. Determination of heterozygosity for phenylketonuria on the amino acid analyzer. Clin Chim Acta 1967;18: Rosenblatt D, Scriver CF. Heterogeneity in genetic control on phenylalanine metabolism in man. Nature 1968;218: Jackson SH, Hanley WB, Gero T, et al. Detection of phenylketonuric heterozygotes. Clin Chem 1971;17: GriYn RF, Elsas LJ. Classic phenylketonuria: diagnosis through heterozygote detection. J Pediatr 1975;86: Blitzer MG, Bailey-Wilson JE, Shapira E. Discrimination of heterozygotes for phenylketonuria, persistent hyperphenylalaninemia and controls by phenylalanine loading. Clin Chim Acta 1986;161: Hilton MA, Patel CC, Bertolone SJ, et al. Plasma phenylalanine:tyrosine ratios during high-dose methotrexate-citrovorum rescue. Med Pediatr Oncol 1982;10: Mönch E, Winkelmann M, Stäblein W. Screeningtests auf hyperphenylalaninämien und tyrosinämien: fluorometrische mikromethoden zur quantitativen bestimmung von phenylalanin und tyrosin mit dem gleichen Auto Analyzer-System. Monat Kinderheilkiunde 1977;125: McCaman MW, Robins E. Fluorimetric method for the determination of phenylalanine in serum. J Lab Clin Med 1962;59: Hill JB, Summer GK, Pender MW, et al. An automated procedure for blood phenylalanine. Clin Chem 1965;11: Hochella NJ. Automated fluorometric determination of tyrosine in blood. Anal Biochem 1967;21: Phillips RE, de Boton R. A simplified determination of blood phenylalanine using samples collected on paper. Palo Alto: GK Turner Associates, Inc, Rudy JL, Rutledge JC, Lewis SL. Phenylalanine and tyrosine in serum and eluates from dried blood spots as determined by reversed-phase liquid chromatography. Clin Chem 1987;33: Hill JB, Palmer P. Filter paper blood collection and punching as a means of quantification. Clin Chem 1969;15: Hsia DYY, Litwack M, O Flynn M, et al. Serum phenylalanine and tyrosine levels in the newborn infant. N Engl J Med 1962;267: Wong PWK, O Flynn ME, Inouye T. Micromethods for measuring phenylalanine and tyrosine in serum. Clin Chem 1964;10: Hsia DYY, Berman JL, Slatis HM. Screening newborn infants for phenylketonuria. JAMA 1964;188: Levy H L, Barkin E. Comparison of amino acid concentrations between plasma and erythrocytes. Studies in normal human subjects and those with metabolic disorders. JLab Clin Med 1971;78: Ghadimi H, Pecora P. Plasma amino acids after birth. Pediatrics 1964;34: Dickinson JC, Rosenblum H, Hamilton PB. Ion exchange chromatography of the free amino acids in the plasma of the newborn infant. Pediatrics 1965;36: Lepage N, McDonald N, Dallaire L, et al. Age-specific distribution of plasma amino acid concentrations in a healthy pediatric population. Clin Chem 1997;43: Godel H, Graser T, Földi P, et al. Measurement of free amino acids in human biological fluids by highperformance liquid chromatography. J Chromatogr 1984; 297: Shen R, Richardson CJ, Rouse BM, et al. An enzymatic assay of plasma phenylalanine and tyrosine for the detection and management of phenylketonuria. Biochem Med 1981;26: Dhondt JL, Paux E. Evaluation of an enzymatic, colorimetric method for the neonatal screening of phenylketonuria. Screening 1993;2: Ambrose JA, Ingerson A, Garrettson LG, et al. A study of the fluorometric method for phenylalanine in serum samples. Clin Chim Acta 1967;15: