Evaluation of 6-year application of the enzymatic colorimetric

phenylalanine assay in the setting of neonatal screening

for phenylketonuria1

Andreas Schulze *, Ertan Mayatepek, Georg F. Hoffmann

Division of Metabolic and Endocrine Diseases, University Children’s Hospital, Im Neuenheimer Feld 150, 69120 Heidelberg, Germany

Received 19 June 2001; accepted 5 September 2001

Abstract

Background: Most reports on phenylketonuria (PKU) screening focused solely on the result of the initial investigation of

the neonatal screening sample. The aim of this study was to evaluate an enzymatic phenylalanine (Phe) determination in the

whole context spanning from the initial investigation over the recall period, up to the confirmation or exclusion of the disease.

Methods: Phe of dried blood spot specimens was analysed colorimetrically in a microtitre-plate assay based on the L-

phenylalanine dehydrogenase reaction coupled with an intermediate electron acceptor system. This assay was evaluated for

analytical variables and for neonatal PKU screening in a total number of 423,773 neonates during a 6-year period. Results:

Method validation with respect to linearity, precision (within-run CVs 3.4–4.2%, between-run CVs 6.2–10.4%), and accuracy

fulfilled all requirements for a screening method. Mean Phe (F SD) of 130,000 healthy neonates was 84 (F 22) mmol/l with a

cut-off point (mean + 3 SD) of 150 mmol/l. From 423,773 neonates, hyperphenylalaninemia was confirmed in 155 cases and

further differentiated into PKU (41 cases, 27%), BH4 deficiency (3, 2%), non-PKU HPA (67, 43%), transient neonatal HPA

(28, 18%), and secondary HPA (16, 10%). The number of false-positives (recall-rate) was 0.23%, and no false-negatives were

noted. Conclusions: Detailed studies over a period of 6 years including more than 400,000 neonates clearly show that the

enzymatic assay is a reliable and sensitive method for neonatal screening of PKU. The proven prevalence of non-PKU HPA in

the German population disclosed by the assay was twice as high as compared to the ‘‘Guthrie test’’ used previously. The

growing use and application of tandem mass spectrometry in neonatal screening will not derogate the usefulness of the

enzymatic assay in PKU screening in the foreseeable future. Careful analysis of our screening results and monitoring of all

pathological samples resulted in an evidence-based flow chart for a rational PKU screening. D 2002 Elsevier Science B.V. All

rights reserved.

Keywords: Phenylketonuria; Phenylalanine determination; Screening

1. Introduction

More than 30 years after the inauguration of the

bacteriological inhibition assay (BIA) for phenylala-

nine (Phe) measurements by Guthrie and Susi [1],

0009-8981/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.

PII: S0009-8981 (01 )00736 -7

* Corresponding author. Fax: +49-6221-563-714.

E-mail address: andreas�schulze@med.uni-heidelberg.de

(A. Schulze).

www.elsevier.com/locate/clinchim

1 In memoriam Horst Bickel.

Clinica Chimica Acta 317 (2002) 27–37

the starting point for neonatal screening of phenyl-

ketonuria (PKU) and other inherited metabolic dis-

eases, different methods for detecting PKU are used

in the screening laboratories throughout the world.

All those methods overcame the main disadvantage

of the BIA, its visual interpretation, and were devel-

oped in order to improve precision, sensitivity, prac-

ticability, and running time of screening assays.

These different methods include fluorometric [2,3],

HPLC [4], enzymatic colorimetric [5], and more re-

cently tandem mass spectrometric [6] applications for

the determination of Phe in dried blood spot speci-

mens (DBS).

The quality of the screening process mainly de-

pends on the methodology and consequent tracking,

which means the individual follow-up of each ab-

normal result. However, the latter is hard to achieve

by the screening laboratory because of insufficient

legal regularisation for feedback from the responsi-

ble physicians, thus often leading to loss of informa-

tion on the further fate of patients. Therefore, most

reports about PKU screening methods were focused

solely on the result of the initial investigation of the

neonatal screening sample. The aim of this study

therefore was to evaluate an enzymatic colorimetric

method for Phe determination in the whole context

spanning from the initial investigation over the recall

period, up to the confirmation or exclusion of the

disease.

About 10 years ago, Wendel et al. [7] described a

colorimetric method for the determination of plasma

Phe using L-Phe dehydrogenase (L-PheDH; EC 1.4.1.-)

coupled with an intermediate electron acceptor system

(Fig. 1) and adapted this method as a microtitre-plate

assay for DBS [8]. Further improvement of specificity

with respect to cross reactivity towards tyrosine (Tyr)

was achieved by elevation of pH to 10.8 in the L-

PheDH reaction [9]. Upon this basis, Porton Cam-

bridge launched a commercial Phe screening kit

(QuantasekR) applicable for neonatal PKU screening

[10–12].

In 1994, after the investigation of more than 1.5

million neonates with the BIA in the past 25 years, our

laboratory which is responsible for the regional neo-

natal screening program (Baden-Wurttemberg, Ger-

many) changed their routine screening procedure from

the BIA to the enzymatic Phe determination. Our

report focuses on the evaluation of an enzymatic

colorimetric Phe determination based on the neonatal

screening results from over 400,000 investigations

during a 6-year period leading to an evidence-based

flow chart for rational PKU screening.

Fig. 1. Reaction scheme of the enzymatic Phe determination. Coupling of the L-PheDH reaction with an intermediate electron acceptor (IEA)

system for the colorimetric measurement.

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–3728

2. Materials and methods

2.1. Reagents and materials

All reagents were purchased from Sigma or

Merck. The QuantasekR Phe screening assay, the

QuantasekR Phe monitoring assay, 96-well Milli-

pore microfilter-plates (PCBH034), and the Vacuum

Manifold (PCNH001) were purchased from Quan-

tase (Perth, Scotland, UK). Filter paper, S&S 2992,

was used from Schleicher & Schuell (Dassel, Ger-

many).

2.2. Calibration materials

With the beginning of 1998, the European Work-

ing Standard [13] was used as reference material for

Phe calibrators. These blood spot calibrators were

purchased from Quantase. Before that date calibra-

tion materials were prepared by using whole blood

from healthy adult blood donors. Phe concentrations

were measured by means of the enzymatic plasma

method mentioned below and thereafter spiked with

a stock solution of 6 mmol/l to final Phe concentra-

tions of 0.120, 0.240, 0.480 and 1.2 mmol/l, respec-

tively. Thirty microliters of the calibration materials

was spotted onto filter paper, air-dried over night and

stored desiccated at 3–5 �C up to 1 month, at the

latest.

Four 4.25-mm punches of each blood spot calibra-

tor were measured in duplicate every day for calcu-

lation of the calibration curve.

2.3. Enzymatic Phe screening assay

4.25-mm discs of blood spot calibrators, of two

different internal Phe blood spot standards as well

as of the DBS from neonates, were punched out

into 96-well Millipore microfilter-plates, incubated

with 60 ml TCA 3% (w/v) and agitated for 60 min

at RT. Eluates were then transferred to microtitre-

plates using a vacuum manifold. One hundred

microliters of working enzyme (L-PheDH)/coenzyme

(NAD + ) reagent was added with a 12-channel

pipette to each well. After incubation for 30 min

at ambient temperature 100 ml colour reagent (citratebuffered solution of tetrazolium salt, intermediate

electron acceptor, and detergent) was added. Absorp-

tion at 570/690 nm was measured after 2 min on a

Multiscan MCC 349 (Labsystems, Finland) micro-

titre-plate reader.

2.4. Further methods for phenylalanine determina-

tion

In plasma, enzymatic determination of Phe was

performed with the QuantasekR Phe monitoring

assay. In this assay, the NADH production coupled

to a tetrazolium/intermediate electron acceptor detec-

tion system was measured at 570 nm with a Cobas Bio

(Hoffmann-LaRoche, Grenzach-Villen, Switzerland)

in 20 ml plasma after deproteinisation with 0.6 mol/l

perchloric acid.

Determination of plasma Phe by automated amino-

acid analysis was performed using a Biotronic 3000

analyser (Biotronic, Munich, Germany) following

standard procedures [14].

Phe, Tyr, and the Phe/Tyr-ratio in DBS were

measured by means of electrospray tandem mass

spectrometry (ESI-MS/MS) as already described in

detail [6].

2.5. Calculations and statistics

Blood Phe concentration was calculated from the

calibration curve obtained by linear regression analy-

sis (least-square method). For that purpose, the on-line

data were transferred in the Porton Cambridge Data

Management SystemR producing the report as a

Microsoft ExcelR file.

Data are expressed as the meanF SD. Statistical

significance was determined using Student’s t-test.

Differences with P values < 0.01 were considered

significant.

2.6. Dried blood spot specimens

DBS of all neonates (n= 423,773) born between

1994 and 1999 were investigated within the scope of

the regional neonatal screening program (Baden-

Wurttemberg, Germany). The blood regularly

obtained by a heelstick on the fifth day of life (range

1–10, for detailed distribution see Fig. 4) was spotted

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–37 29

onto filter paper, allowed to dry and sent per post mail

to our laboratory.

3. Results and discussion

3.1. Method validation (analytical variables)

3.1.1. Calibration curves, linearity, and limit of detec-

tion

For the study of the calibration curve, DBS of 98,

282, 568, and 1077 mmol/l Phe were measured four-

fold on 20 days. Results are shown in Fig. 2. Analysis

of the daily obtained calibration curve over a period of

4 years revealed a slope of 0.144F 0.053 and inter-

cept of 0.014F 0.003 (meanF SD, respectively).

The detected absorbance for the calibrators was

linear in the range of 98–1077 mmol/l (r2 = 0.974).

Limit of detection was determined as the concen-

tration corresponding to a signal 3 SD above the

mean for a calibrator free of analyte. This limit of

detection for Phe in DBS was 43 mmol/l, highlight-

ing the limitation of the method in the lower Phe

range. However, this is not of any relevance for

neonatal screening, but must be taken into account

for potential use in the monitoring of long-term PKU

treatment.

3.1.2. Precision

We assessed the precision of the whole Phe screen-

ing assay according to NCCLS EP5-T2 (Wayne, PA:

National Committee for Clinical Laboratory Stand-

ards, June 1984). Estimates of within-run and bet-

ween-run SD were obtained by determination of DBS

with four different Phe concentrations. Two repli-

cates per specimen per run and two runs per day for

20 days were performed. The results are presented in

Table 1.

The CV for the determination of the within-run

precision was between 3.4% and 4.2%. The CV for

the determination of the between-run precision was

between 6.2% and 10.4%.

3.1.3. Analytical recovery and comparison-of-meth-

ods study

For the determination of the accuracy of the Phe

screening assay, Phe recovery in the DBS was ana-

lysed by spiking whole blood with Phe amounts

reaching final Phe concentrations of 121, 485, and

1212 mmol/l of pooled blood, respectively. The

pooled blood was spotted onto filter paper and, in

addition, deproteinised plasma was prepared. Subse-

Fig. 2. Calibration curve of the enzymatic Phe screening assay.

Dried blood spot Phe (DBS) calibrators were measured fourfold on

20 consecutive days. Phe concentrations are plotted as meanF SD.

Calibration curve was obtained by linear regression analysis. Slope

and intercept are expressed as meanF SD. SEE, standard error of

estimate (standard deviation about the regression line).

Table 1

Precision of the enzymatic phenylalanine screening assay

DBS phenylalanine Within-run Between-run

concentration (mmol/l)Concentration (meanF SD),

(mmol/l)

CV

(%)

Concentration (meanF SD),

(mmol/l)

CV

(%)

98 95.5F 3.7 3.8 95.5F 10.0 10.4

282 260.3F 9.0 3.4 260.3F 23.7 9.1

568 600.5F 21.7 3.6 600.5F 44.8 7.5

1077 1065.8F 44.8 4.2 1065.8F 66.2 6.2

DBS: dried blood spot.

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–3730

quently, Phe concentrations were measured in DBS

with the Phe screening assay and by ESI-MS/MS

15 times in the same batch, and in plasma by the

QuantasekR Phe monitoring assay on Cobas Bio

and by automated amino-acid analysis 12 times in the

same batch (Table 2).

The enzymatic Phe determination in DBS re-

vealed a recovery of 89–101% and CVs of 6.8–

11.2%, thus fulfilling the requirements for a screen-

ing method.

Phe determination in DBS by ESI-MS/MS (recov-

ery 106–125%, CVs 6.5–16.9%) revealed increased

recoveries and CVs comparable to the enzymatic

method.

The enzymatic Phe measurement in plasma

showed the best accuracy (recovery 92–109%, CVs

1.1–2.9%), whereas automated amino-acid analysis,

an established method for quantitative plasma amino

acid analysis, was less good as expected (recovery

74–120%, CVs 3.1–14.1%). This was especially true

for the low (121 mmol/l) and medium (485 mmol/l)

Phe concentrations which yielded recoveries of about

80–120% and 74–108% as well as CVs of 10.2% and

14.1%, respectively (Table 2).

3.1.4. Reference interval (normal range) of neonates

Data of a total number of 130,000 healthy neo-

nates were analysed for the determination of the Phe

reference interval, distribution analysis, as well as

Phe concentration in relation to age of sampling.

Mean (F SD) Phe concentration was 84 (F 22)

mmol/l with a median of 83 mmol/l. Cut-off point cal-

culated by mean + 3 SD was 150 mmol/l. Frequency

distribution of Phe values followed a nearly Gaussian

distribution (Fig. 3). The curve only slightly flattens

on the right side.

The Phe concentration related to age of the neo-

nates at the time of blood sampling was investigated

in order to estimate the need of changing the cut-off

value in cases which deviate from the time of

sampling regularly performed on the fifth day of life.

That is of special importance for early discharge of

neonates, because of the well-known elevations of

Phe concentrations immediately post-natally [15]. In

only 2% of our investigated population of 130,000

neonates blood sampling was performed before the

fourth day of life, whereas 5% were investigated after

the sixth day of life (Fig. 4). The mean (F SD) Phe

concentrations on the first and second days of life

were 96 (F 27) and 93 (F 24) mmol/l, respectively,

thus, significantly higher than on the fifth day. From

the third to the 10th day of life concentrations did not

differ significantly from the latter (Fig. 4). Higher

Phe levels on the first and second days resulted in

Fig. 3. Frequency distribution of Phe concentration in DBS from

130,000 neonates (circles). The lined curve represents the ideal

Gaussian distribution.

Table 2

Recoveries of phenylalanine spiked in whole blood. Comparison between different methods of measurement in dried blood spots (DBS) and

plasma

Method Specimen N Recovery mean (range) (%)

121 mmol/l

adjusted (%)

CV

(%)

485 mmol/l

adjusted (%)

CV

(%)

1212 mmol/l

adjusted (%)

CV

(%)

Screening assay DBS 15 94 (75–107) 10.4 89 (69–108) 11.2 101 (90–111) 6.8

ESI-MS/MS DBS 15 106 (87–160) 16.9 125 (102–143) 10.5 125 (108–136) 6.5

Enzymatic, Cobas Bio Plasma 12 104 (95–109) 2.9 94 (92–96) 1.1 100 (96–102) 1.7

Amino Acid Analyzer Plasma 12 99 (80–120) 10.2 83 (74–108) 14.1 99 (94–104) 3.1

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–37 31

2.5% and 0.8% of samples over the cut-off (150

mmol/l), respectively, in contrast to 0.2% on the fifth

day of life.

3.2. Diagnostic outcome of neonatal screening for

PKU

DBS of a total number of 423,773 neonates which

were investigated from 1994 to 1999 in the context of

the neonatal screening program of Baden-Wurttem-

berg, Germany, were included in the study. 93% of

samples were taken on the fourth to the sixth day of life

(Fig. 4). Measurement of Phe from the same DBS was

repeated if Phe concentration was found to be over the

cut-off of 150 mmol/l. This was necessary in 0.78% of

all samples. In case that Phe elevation was confirmed,

either a written recall was sent out (0.26% of all

samples) or when Phe levels were > 360 mmol/l the

sender of the DBS was contacted by phone and asked

for transferal of the neonate to a specialised hospital

(60 neonates, 0.01% of all samples) (Table 3).

Result of all positive tested screening samples,

number and prevalence of confirmed diagnoses as

well as rates of false positives are summarised in

Table 3. A more detailed consideration of the different

diagnosis groups is the content of the following

paragraphs. Quantitative data for all different groups

are shown in (Table 4).

3.3. Phenylketonuria

Phenylketonuria is caused by phenylalanine

hydroxylase (PAH; EC 1.14.16.1) deficiency (McKu-

sick 261600). Following the German recommenda-

tions of PKU treatment [16], patients with Phe >600

mmol/l while on normal diet were classified as PKU.

We detected a total number of 41 neonates suffering

from PKU. The resultant prevalence in our population

was estimated to be 1:10,336. In general, confirma-

tion of PKU was made on the 10th day of life. At that

time, the Phe concentration was twofold increased in

comparison to the screening value (Fig. 5). In 39/41

cases, the initial screening Phe levels were >360

mmol/l and in 35/41 >600 mmol/l. From the six

neonates with Phe < 600 mmol/l, Phe restriction was

Table 3

Results of neonatal screening in 423,773 neonates

True positive False positive False negative

Phe 150–360 mmol/l 95 (8.2%) 995 (86.2%) 0 1090 (94.4%)

Phe>360 mmol/l 60 (5.2%) 4 (0.3%) 0 64 (5.5%)

155 (13.4%) 999 (86.5%) 0 1154 (100%)

Confirmed diagnoses in true

positives (n= 155)

Number Prevalence

PKU 41 1:10,336

BH4 Deficiency 3 1:141,258

Non-PKU HPA 67 1:6325

Transient neonatal HPA 28 1:15,135

Secondary HPA 16 1:26,486

HPA: hyperphenylalaninemia.

Fig. 4. Phe concentration in DBS from 130,000 neonates related to

age of blood sampling. The line plot shows the mean (open circles)

and meanF 3 SD (filled circles) of Phe concentration. ** marks

significant ( p< 0.01) difference of the mean of Phe concentration

compared to that obtained on the fifth day of life. The number of

investigated DBS on the respective days is plotted as columns.

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–3732

started in two of them before screening was taken

because of already known older siblings affected with

PKU. In one other case (screening Phe 480 mmol/l),

DBS was taken on the second day of life, and in two

neonates which were primarily classified as non-PKU

hyperphenylalaninemia (HPA) (screening Phe 310

and 260 mmol/l), a follow-up revealed a ‘‘mild’’

PKU with delayed increase of Phe exceeding 600

mmol/l after 6 weeks and 10 months, respectively.

Only in one neonate (Phe 490 mmol/l on fifth day of

life) did we not find a lucid explanation.

3.4. Tetrahydrobiopterin deficiency (BH4 deficiency)

BH4 is the cofactor of PAH. HPA in the inves-

tigated neonates was caused in a total number of

three patients by a defect in BH4 metabolism. Two of

them suffered from 6-pyrovoyl-tetrahydropterin syn-

thase deficiency (McKusick 261640), and one from

dihydropteridine reductase deficiency (McKusick

261630). Their initial screening Phe levels ranged

between 479 and 596 mmol/l. BH4 deficiency was

suspected because of decrease of Phe levels after oral

BH4 administration. Final diagnoses were made by

enzyme measurements. The proportion of BH4 defi-

ciencies of the proven HPAs accounted for 2.7%,

reflecting a prevalence of about 1:140,000 in our

population.

3.5. Non-PKU hyperphenylalaninemia (mild PKU)

Non-PKU HPA (defined as 150 mmol/l < Phe < 600

mmol/l while on normal diet) due to mild PAH

deficiency was confirmed in 67 cases. The resultant

prevalence of 1:6325 was markedly higher as those

obtained by the BIA (1:14,000, former results of our

screening centre). This is of special importance for

females with regard to the risk of maternal PKU,

which can be prevented by achieving Phe levels < 360

mmol/l preconceptually and during pregnancy [17].

Therefore, it is important to identify women at risk.

Screening Phe levels of the non-PKU HPA group

ranged between 158 and 636 mmol/l (median 246

mmol/l) and remained normally unchanged during

the recalls (Fig. 6). If Phe exceeds levels >240

mmol/l in the first recall, we recommend initiation of

further confirmation tests including exclusion of

BH4 deficiency, instead of requesting further recalls

(Fig. 9).

In three cases with positive first and second recalls,

further confirmation analyses revealed Phe levels of

about 120–150 mmol/l. Long-term Phe-monitoring

performed in monthly intervals, however, confirmed

the diagnosis of non-PKU HPA.

Two neonates, both breast fed, suspected of suf-

fering from PKU because of their screening Phe

levels of 600 and 636 mmol/l, respectively, showed

a decline of their Phe levels already at the time of

their hospitalisation. Further work-up also revealed

non-PKU HPA.

Table 4

Phenylalanine concentration in the different HPA groups at neonatal

screening, during the recall period, and at time of confirmation

Diagnosis Number Age of Sampling Phenylalanine

Median

(range) day

Median (range)

mmol/ l

PKU (n= 41)

Screening 41 5 (2–7) 970 (261–2024)

Confirmation 35a 10 (5–18) 1697 (448–3212)

BH4 Deficiency (n = 3)

Screening 3 5 (5–6) 564 (479–596)

Confirmation 3 17 (14–17) 970 (824–1515)

Non PKU-HPA (n= 67)

Screening 67 5 (1–10) 246 (158–636)

1st Recall 61 15 (7–122) 261 (164–588)

2nd Recall 40 29 (15–180) 245 (170–570)

Confirmation 44a 41 (10–330) 255 (121–564)

Transient neonatal-HPA (n= 28)

Screening 28 5 (1–21) 172 (152–521)

1st Recall 28 17 (5–42) 182 (121–309)

2nd Recall 27 36 (12–90) 133 (61–267)

3rd Recall 13 49 (20–270) 109 (36–145)

Secondary-HPA (n= 16)

Screening 16 6 (5–13) 355 (170–1357)

1st Recall 7b 13 (8–18) 218 (79–539)

2nd Recall 4 29 (23–39) 173 (79–261)

3rd Recall 1 65 145

Healthy

neonates

130,000 5 (1–10) 84.0 (F 22.1)c

aDifference of cases and number of confirmations result from

failed quantitative reports in a part of confirmed cases.b In 9 cases no recall was done because of medical reports

explaining secondary Phe elevation.cMean (F SD).

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–37 33

3.6. Transient neonatal HPA

Increased Phe concentration in 28 DBS was caused

by transient HPA of the neonate. In 27/28 cases, the

screening Phe was 152–194 mmol/l (median 172

mmol/l). Phe in the first recall samples (taken on

17th day of life on average) tended to result in higher

levels, but declined in the further recall samples. After

7 weeks, the Phe levels in all infants were found to be

< 150 mmol/l (Fig. 7). In order to omit unnecessary

Fig. 5. Phe concentration of the PKU group in neonatal screening (DBS) and in confirmation test (plasma). Plasma Phe was measured by means

of different methods used in the respective metabolic centres where the confirmation analyses were made.

Fig. 6. Phe concentration of the non-PKU HPA group in neonatal screening, in recalls (DBS), and in confirmation test (plasma). Plasma Phe was

measured by means of different methods used in the respective metabolic centres where the confirmation analyses were made.

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–3734

recalls, we recommend to perform the first and all

succeeding recalls with an interval of 2 weeks if the

initial screening Phe levels are < 200 mmol/l (Fig. 9).

In one neonate, we measured a high Phe concen-

tration (521 mmol/l) which normalised in subsequent

samples (120 as well as 60 mmol/l, at 16th and 41st

Fig. 7. Phe concentration in DBS of the transient neonatal HPA group in neonatal screening and in recalls.

Fig. 8. Phe concentration versus molar Phe/Tyr-ratio in the different HPA groups. Phe concentrations of all neonatal screening DBS in which

the Phe/Tyr-ratio was available are plotted against the latter. The Phe concentration in transient neonatal HPA was below 210 mmol/l in 17 of

17 cases (horizontal line). Note that the Phe of 521 mmol/l from one infant with transient HPA, discussed in the text, is omitted because of the

missing Phe/Tyr-ratio. The Phe/Tyr-ratio in 16/17 DBS from transient HPA and in 7/7 DBS from secondary HPA was found below 2.7

(vertical line).

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–37 35

day of life, respectively). We could exclude a mix-up

of samples, an error of measurement, and intercurrent

illness of the child which could explain such a unusual

finding.

3.7. Secondary HPA

Increased Phe in neonatal screening not caused by

genetic PAH deficiency was detected in 16 babies. In

them, Phe levels were found to be elevated because of

liver impairment in the neonates due to galactosemia

(in 4 of 16 cases), multi-organ failure (5/16), cardio-

vascular disease (2/16), severe asphyxia (cord pH 6.9)

(1/16), chromosomal aberration (trisomy 18 as well as

21) (2/16), and extreme immaturity (25 as well as 26

weeks of gestational age) (2/16). The wide range of

Phe concentrations within these group (170–1357

mmol/l) disallowed a differential diagnosis of HPA

by the Phe levels. As recently shown by our group, the

use of the Phe/Tyr-ratio obtained by means of ESI-

MS/MS, therefore, is a good aid in the differential

diagnosis [6]. In the DBS of seven neonates with

secondary HPA, their molar Phe/Tyr-ratio could be

estimated and was found to be below 2.7 in all cases

(Fig. 8). In comparison, Phe/Tyr-ratio in PKU ranged

between 6.7 and 47.7, except for two milder cases of

PKU mentioned above, where a Phe/Tyr-ratio of 5.3

as well as 3.4 was found.

3.8. False positives and pitfalls

In 995 of 423,773 DBS, the increased Phe con-

centration (150–360 mmol/l) of the neonatal screening

was found to be < 150 mmol/l in the recall sample,

reflecting a false positive rate of about 0.23%. Phe

determination in four DBS revealed concentrations

>360 mmol/l. Two of them were caused by intra-

venous amino-acid administration (Phe 582 and 988

mmol/l; Phe/Tyr 10.2 and 38.3, respectively). In two

filter cards, a high Phe concentration (679 and 1760

mmol/l, respectively) was traceable in a circumscribed

area of the blood spot only. In both of them, measure-

ments of other punches from the same spot and of

other blood spots from the same filter card revealed

decreased or normal Phe concentrations, respectively.

Thus, the high Phe may have been caused by con-

tamination of the filter card and highlights the need

for internal confirmation tests of abnormal results

before initiating recalls or hospitalisations of neo-

nates.

Fig. 9. Evidence-based flow chart for PKU screening.

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–3736

4. Conclusion

Detailed studies of an enzymatic colorimetric Phe

determination in DBS over a period of 6 years

including more than 400,000 neonates clearly show

that this represents a reliable and sensitive method for

neonatal screening of PKU. During the whole study

period, we got no information and were not aware of

any false negative result. The overall recall rate of

0.23% was found to be in an acceptable range. More-

over, the prevalence of non-PKU HPA detected by

this method in our centre was twice as high compared

to the BIA, demonstrating the improved sensitivity of

the former.

Based on careful analyses of the screening results

and the monitoring of all abnormal samples, a new

evidence-based flow chart for neonatal PKU screen-

ing is proposed (Fig. 9). Its application should reduce

recalls without loss of sensitivity or any delay in

diagnosis and early treatment.

The observation of two PKU cases with delayed

Phe increase within the first weeks up to several

months highlights the need of continuous investiga-

tion and careful follow-up of infants with Phe levels in

the non-PKU HPA range. This is also true for those

neonates whose screening Phe levels are distinctly

elevated, because of the possible Phe decrease as we

have seen in two other babies. Furthermore, it must be

a general rule to confirm all abnormal results firstly by

internal re-measurement before initiating recalls or

hospitalisation of neonates in order to avoid unneeded

anxiety of the families.

The growing use and application of tandem mass

spectrometry in neonatal screening, based on the

advantage of this method to obtain information about

many other metabolites in addition to Phe at the same

time and within one analytical step, will not derogate

the usefulness of the enzymatic colorimetric Phe

determination in PKU screening in the foreseeable

future. This is especially true and important for

screening laboratories which are not able to afford

the greater expenses for tandem mass spectrometry.

References

[1] Guthrie R, Susi A. A simple phenylalanine method for detect-

ing phenylketonuria in large populations of newborn infants.

Pediatrics 1963;32:338–43.

[2] Blau K. Determination of phenylalanine in filter paper blood

spots by a simplified automated fluorimetric method without

dialysis. Clin Chim Acta 1983;129:197–200.

[3] Gerasimova NS, Steklova IV, Tuuminen T. Fluorometric meth-

od for phenylalanine microplate assay adapted for phenylke-

tonuria screening. Clin Chem 1989;35:2112–5.

[4] Vollmer DW, Jinks DC, Guthrie R. Isocratic reverse-phase

liquid chromatography assay for amino acid metabolic disor-

ders using eluates of dried blood spots. Anal Biochem 1990;

189:115–21.

[5] Dooley KC. Enzymatic method for phenylketonuria screening

using phenylalanine dehydrogenase. Clin Biochem 1992;25:

271–5.

[6] Schulze A, Kohlmueller D, Mayatepek E. Sensitivity of elec-

trospray-tandem mass spectrometry using the phenylalanine/

tyrosine-ratio for differential diagnosis of hyperphenylalanine-

mia in neonates. Clin Chim Acta 1999;283:15–20.

[7] Wendel U, Hummel W, Langenbeck U. Monitoring of phenyl-

ketonuria: a colorimetric method for the determination of plas-

ma phenylalanine using l-phenylalanine dehydrogenase. Anal

Biochem 1989;180:91–4.

[8] Wendel U, Koppelkamm M, Hummel W, Sander J, Langen-

beck U. A new approach to the newborn screening for hyper-

phenylalaninemias: use of l-phenylalanine dehydrogenase and

microtiter plates. Clin Chim Acta 1990;192:165–70.

[9] Campbell RS, Brearley GM, Varsani H, et al. Development

and validation of a robust specific enzyme mediated assay for

phenylalanine in serum. Clin Chim Acta 1992;210:197–210.

[10] Dhondt JL, Paux E. Evaluation of an enzymatic, colorimetric

method for the neonatal screening of phenylketonuria. Sreen-

ing 1993;2:141–7.

[11] Keffler S, Denmeade R, Green A. Neonatal screening for phe-

nylketonuria: evaluation of an automated enzymatic method.

Ann Clin Biochem 1994;31(Pt 2):134–9.

[12] Elvers LH, Diependaal GAM, Blonk HJ, Loeber JG. Phenyl-

ketonuria screening using the Quantase phenylalanine kit in

combination with a microfilter system and the dye tartrazine.

Sreening 1995;5:209–23.

[13] Dhondt JL, Loeber J, Elvers LH, Paux E. Preparation of the

first European working standard for phenylalanine determina-

tion in dried blood spots. J Med Screening 1998;5:63–6.

[14] Slocum RH, Cummings JG. Amino acid analysis of physio-

logical samples. In: Hommes FA, editor. Techniques in diag-

nostic human biochemical genetics: a laboratory manual. New

York: Wiley-Liss; 1991. p. 87–126.

[15] Chace DH, Sherwin JE, Hillman SL, Lorey F, Cunningham

GC. Use of phenylalanine-to-tyrosine ratio determined by tan-

dem mass spectrometry to improve newborn screening for

phenylketonuria of early discharge specimens collected in

the first 24 hours. Clin Chem 1998;44:2405–9.

[16] Burgard P, Bremer HJ, Buhrdel P, et al. Rationale for the

German recommendations for phenylalanine level control in

phenylketonuria 1997. Eur J Pediatr 1999;158:46–54.

[17] Recommendations on the dietary management of phenylketo-

nuria. Report of Medical Research Council Working Party on

Phenylketonuria. Arch Dis Child 1993;68:426–7.

A. Schulze et al. / Clinica Chimica Acta 317 (2002) 27–37 37