Screening Newborns For Hemoglobinopathies By Hplc

Clinical Chemistiy 704-710

42:5

(1996)

Automated HPLC screening of newborns for sickle cell anemia and other hemoglobinopathies JOHN

W. EASTMAN,*

RUTH

WONG,

CATHERINE

Automated HPLC is used to test dried blood-spot specimens from newborns for hemoglobms (ITh) F, A, S, C, E, and D. We present the method and report on its performance determined during >4 years of testing 2.5 x 106 newborns. The method features automated derivation of presumptive phenotypes; quantitative quality control and proficiency testing; throughput of one specimen per minute; small sample volume; hemoglobin concentrations quantified with an mterlaboratory CV of 14-18%; retention times with interlaboratory CV of <2% and matching, within ± 0.03 miii, of laboratories and reagent lots; control of peak resolution; 0.5% detection limit for Hb S and C, and 1.0% for Hb F, A, E, and D; few interferences; and negligible background and carryover. Shortcomings of the method are the absence of microplate barcode identification and the need for manually pipetting the sample eluate into the microplate. mie1s: chromatography, INDEXING

dried blood-spot cation-exchange

specimens

#{149} Guthrie

California

Department

of Health

St., Suite 100, Berkeley, *AUthOr for correspondence.

Fax

Genetic

Disease

R.

MORALES

SYSTEM

Modular instrumentation was adapted for the California program by Bio-Rad Labs., Hercules, CA. A similar integrated instrument, ASCeNT#{174},has been marketed [14]. Each of the modular instruments used in this work consists of one to three HPLC column systems run by one Model 700 chromatography workstation. Each of the column systems consists of a Model AS-I00 HRLC automated sampling system, a 20-g.tL-loop injection valve, two Model 1350 gradient elution pumps, the cation-exchange column in a Model 1250425 35 #{176}C heater, and a dual-wavelength (415/690 nm) filter photometer (cat. no. 1961042). The 6 X 40 mm columns are packed with a nonporous 7-.tm-diameter Bio-Rad MA 7 polymeric cation-exchange material. Each column can be used for as many as 500 injections. Two sodium phosphate buffers are used as eluents, run as a gradient from -4 g/L (buffer A) to 14 g/L (buffer B) at pH 6.4. Nominal conditions of analysis are a flow rate of 2 mL/min at a pressure of 25 kg/cm2 and the following gradient (time from injection in minutes/percent of eluent that is buffer B): 0.0/0%; 0.3/10%; 0.5/24%; 1.0/52%; 1.8/100%; 1.9/100%; and 2.0/0%. Including the 1-mm wash between specimens, each chromatogram takes 3 mm. When all three columns are used, the rate of analysis is one specimen pen minute. Operation of the instrument is automated through a menudriven software program that generates the worklist, injects the sample, controls the gradient for hemoglobin separation, measures and integrates the peaks, derives the hemoglobin pattern, stores the data on electronic media, and telecommunicates the data to a remote central site. The reagent kits include whole-blood primer, wash solution, three linearity calibrators containing Hb F and A, and two lyophilized controls, one containing Hb F, A, E, and S and one Hb F, A, D, and C. To maintain the separation of hemoglobmns

cards.

Laboratory,

DANIEL

Materials and Methods EQUIPMENT/REAGENT

#{149} phenotypes

Services, CA 94710.

and

that newborns with hemoglobinopathy patterns (e.g., FS, FSC, FSA, F only) be recalled for mandatory follow-up. Also, newborns with FE patterns are recalled to differentiate EE and E//3-thalassemia. For trait patterns of FAS, FAC, and FAD, the families of the affected newborns are offered voluntary counseling.

In keeping with national recommendations, California legislation mandates screening all newborns for sickle cell disease [1-3].This legislation was implemented for the State of California by the Department of Health Services’ Genetic Disease Laboratory (GDL) and Genetic Disease Branch, Berkeley, CA [4]. Because of its potential as a quantitative method of analysis amenable to automation, GDL chose cation-exchange HPLC [5-12] as the screening method. Cation-exchange HPLC has been used to screen cord blood for hemoglobinopathies [13]. Here, we report on the application of this technique to screen newborns by assaying their dried blood-spot (DBS) specimens. The method reported here is designed to resolve hemoglobins (Hb) F, A, 5, C, E, and D. The California program requires

Heinz

L. Lao,

700

510-540-2228.

abbreviations: GDL, Genetic Disease Laboratory; DBS, dried blood spot; NB, newborn; Hb, hemoglobin(s); and AU, hemoglobin concentration in area units, the area of the chromatographic peak. Received October 4, 1995; accepted January 25, 1996. ‘Nonstandard

among

704

all lots of the cation-exchange

resin,

the software

that

Clinical Chemistry

controls the gradient is modified when needed to accommodate any differences in the performance of the different lots. A key feature of the test system is its quantification of the concentration of the hemoglobin variants. Chromatographic peaks are reported with heights in microvolts and with areas in relative response units (AU, area units). With the integration settings used for this screening method, 1 AU is approximately three times the area in p.V-min. The three linearity calibrators are used to monitor the dose-response curve for photometer readings vs hemoglobin concentration. The reagent-instrument system must maintain the photometer readings for the linearity calibrators (-0.2, 0.4, and 0.6 g/L Hb F) within ±20% of a stated nominal value. HPLC

SCREENING

METHOD

Specimen collection and preparation.Blood from a subject is absorbed into S&S 903 specimen collection paper (Schleicher & Schuell, Keene, N}-l). A disc 0.95 cm (3/8 in.) in diameter is punched from the DBS specimen, and the blood is eluted into 1.00 mL of water for 30 mm with periodic shaking (equivalent to a 36-fold dilution of the whole-blood specimen). The eluate is further diluted 1:6 with water and dispensed into a 96-well microplate, which is loaded onto the autosampler for injection. Worklistgenerationand reports. A worklist

is generated

automat-

ically when the analyst sets up the run file information. At the end of each run, a Neonatal Hemoglobin Summary Report is printed out. The summary lists the identification number of each specimen and the percentages of all the hemoglobin variants found in the specimen. A Pattern Report is also printed out, listing the hemoglobin pattern (presumptive phenotype) for each specimen. This report lists, in order of decreasing concentration, the letter designation of each of the hemoglobins found in the specimen. Quality control. For each column system, the FAES and FADC liquid controls are run at the beginning and end of each run. At the beginning and end of each tray (microplate) of newborn specimens, eluates of blood spots containing Hb S are run as tray controls. The mean and SD values for retention time are fixed as quality-control action limits in the controller software by specification. Mean and SD values for hemoglobin concentration are established by replicate analysis of the controls at GDL before they are put into use. Trays of newborn results are flagged for review by a quality-control officer when (a) the retention time of Hb S in the tray control exceeds the ±3SD limit; (b) the concentration of Hb S in the tray control exceeds the ± 3 SD limit; or (c) the blank water injections give a chromatographic peak height >5000 j.tV. Because the microplates are not barcoded for identification, a positional control is also included on each tray. This is an eluate of a blank disc of unique position such that it sequence of analysis (e.g., position 2). The positional on each tray.

blood-collection paper placed in a documents each tray’s number in the tray 2 must have a blank result at control also serves as a water blank

42, No. 5, 1996

DERIVATION

705

OF PATTERNS

For each newborn tested, a hemoglobin pattern presents the observed hemoglobins in order of relative concentration from the highest to lowest. GDL developed algorithms (see Appendix) for use on the chromatography workstation software so that chromatographic peaks from noise and minor hemoglobins are not included in the hemoglobin patterns. The retention time of Hb A, is similar to that of Hb E. However, Hb A2 is never expressed in a hemoglobin pattern derived for a newborn specimen because its concentration is low, and the interpretation algorithms remove such low peaks from the pattern (Appendix). Satellite hemoglobins (e.g., E1, 5,, C,) produce small chromatographic peaks that elute -0.17 mm faster than the peak of the corresponding major hemoglobin [5,7]. GDL introduced a 1:4 rule to eliminate the satellite hemoglobmns from the reported patterns. Also, based on published information on the relative percentage concentrations of Hb A in thalassemia cases [8, iS], GDL developed a 1:2 rule to differentiate patterns FAS (sickle cell trait) from FSA (S/f3thalassemia) (see Appendix). Hb (1), an unidentified hemoglobin eluting between Hb F and Hb A, appears in the HPLC chromatograms for the linearity calibrators, the liquid controls, surrogate in-house DBS samples, and in most DBS specimens from newborns. The retention time for Hb (1) corresponds to that expected for Hb A,d [16]. The program used to derive the hemoglobin pattern adds the concentration of Hb (1) to the concentration of Hb A and reports the total concentration as Hb A (Appendix). SPECIMENS

ANALYZED

GDL collected Bio-Rad FAES

data on three types of samples: the liquid and FADC controls, which are not processed

through the punching, elution, and dilution steps of the method; surrogate DBS specimens prepared in house (GDL-DBS samples); and newborns’ specimens collected on paper (NB-DBS specimens). The liquid controls contain Hb F, A, 5, C, E, D in concentrations similar to those found in actual newborn specimens, i.e., mostly Hb F with 10-20% Hb A and 3-10% of the other variants. The GDL-DBS samples, which contain Hb F, A, and 5, were prepared by mixing commercial bulk AS blood or cord blood with adult outdated bank blood and spotting the mixture onto specimen collection paper. These DBS sample pools contain various concentrations of the hemoglobin variants for use as controls and proficiency-test samples. For GDL-DBS pools containing Hb S, the concentration of Hb A is large (-75%), and the ratio of [Hb A]/[Hb S] does not match the ratio found in newborn specimens. Nevertheless, these samples yield satisfactory chromatograms. EVALUATION

OF PERFORMANCE

CHARACTERISTICS

The precision and accuracy of retention times, as well as the precision of the variant quantification, were determined from data collected with 23 column systems at nine laboratories using one lot of cation-exchange resin. We also evaluated the accuracy and precision of data collected with four different lots of resin by one laboratory using three column systems. The other perfor-

706

Eastman

et al.: Screening

mance characteristics were determined column systems and four lots of resin.

newborns

for the aggregate

for hemoglobinopathies

of 23

Results PERFORMANCE

by HPLC

Chromatograms. Fig. 1 is a chromatogram obtained from a liquid sample made by combining controls with Hb F, A, 5, C, E, and D. Fig. 2 is a chromatogram of an eluate of a DBS specimen obtained from a newborn with sickle cell disease. Precision of retention times.Table 1 compares our results with the instrument specifications. The SD specification is one-sixth the range of the retention time identification window (window ± 3SD). The observed SD for Hb S exceeds the limit somewhat; nonetheless, at this precision, all cases of sickle cell disease have been correctly identified. Accuracy of retention times. The observed mean retention time (Table 1) for each hemoglobin is compared with the retention time for the center of the hemoglobin identification window in the integration software. In most cases the observed mean values are within 0.01 mm of the specified value, and in no case does the bias exceed 0.03 mm. Precision of quantification of hemoglobins. According to the specifications (Table 2, last column), with a one-column system (with one photometer) the interrun CV for Hb F in the middleconcentration linearity calibrator should be no more than 5%. Also, among all systems the mean should always lie between 80 000 and 120 000 AU (±20% range for matching the 23 systems). The results are in acceptable agreement with the specifications. The CV of 10.9% over all 23 column systems is equivalent to a ±2SD range of 21.8%, which is close to the ±20% range limit. On multiple column systems the observed CVs for liquid controls and DBS samples span a range from 14% at 30 000

blood

spots

AU to 18% at 5000 AU. The 5000 AU (18% CV) is to a NIB-DBS specimen containing a hemoglobin 2.5% relative concentration (in a total area of 200 000 30 000 AU (14% CV) is equivalent to a NB-DBS containing

ASSESSMENT

of dried

a hemoglobin

variant

at 15% relative

equivalent variant at AU). The specimen

concentration.

Detection limits. HPLC peak criteria in the integration parameters are set so that Hb S and C are expressed in the hemoglobin pattern when the relative concentration of each exceeds 0.5%. For Fib F, A, E, or D the detection limit is 1.0%. To achieve these detection limits requires that the height threshold used to eliminate background noise be reduced to 500 jV. To test the practicality of this threshold, GDL submitted two GDL-DBS samples containing 1.1% and 0.8% Hb S for analysis at the satellite laboratories. In 12 weekly shipments (one sample per week) to nine laboratories (n = 106 analyses after excluding 2 results that were invalid for other reasons), there were no missed Hb S peaks. By now, >450 000 newborns have been tested with use of the 500-.tV height threshold, and there have been no known missed cases of Hb S or other clinically significant variant exceeding 0.5% in the newborns’ specimens. That a 500 .tV threshold is required is shown by results obtained during the first 3.5 years of HPLC screening, during which time we used a threshold of either 3000 or 2000 j.tV. At 3000 j.V not all satellite laboratories reported the presence of Hb 5, even though it made up 1.0% of the total hemoglobin in the GDL-DBS proficiency-test samples. However, retrospective examination of the chromatograms, which are stored on magnetic tape, and reprocessing at a lower threshold detected all of the Fib S peaks. Also, during that period, -2.1 million newborns were screened. In rare instances Hb S, A, E, and D that were not reported in the newborn screening pattern were found in children at an older age (Table 3). In all cases, when the chromatographic raw data were electronically reprocessed at a lower threshold (e.g., 500 .tV), we found a well-resolved bellshaped peak at the retention time of the missing variant.

300

bOO

‘20O E a)

400

a)

(I)

Cl)

0

0 a-

100

C’)

a)

Li

0 0

2

0 Time

(mm)

Fig. 1. Chromatogram of a 1:1 mixture of containing Hb FAES and Hb FADC.

two

Bio-Rad controls

(from left to right): Hb FAST, Fl, F, (1), A, E, D, S, and C. (See Appendix for hemoglobin nomenclature.) Peaks

2

0 Time (mm)

FIg. 2. Chromatogram of a newborn dried blood-spot specimen with pattern ES. Peaks (from left to right): void volume and Hb FAST, Fl, F, (1), and S.

Clinical Chemistry

Table 1. PrecisIon

42, No. 5, 1996

707

and accuracy of retention times. Retention

time, mm x 100

Mean Hb

Liquid samples F A E D S C Dried blood spots F A S

n

Mean

422

64.08 83.29 98.14 107.93

422 211 211 211

172.60

90 97 204

84.29

63.37

-2.37

+

61.00 83.00 118.00

117.57

0.97 1.00

+

83.00

98.00 107.00 118.00 172.00

118.25

211

3.08 0.29 +0.14 +0.93 +0.25 +0.60

61.00

+

SD specification

SD

BIas

specification

1.39 1.39 1.59

2.33 1.67 1.33 1.67 1.33 2.00

0.9

1.17

2.33

1.8

1.02

1.67 1.33

1.2

1.26

1.29

1.41

-0.43

CV, %

1.5 1.2

1.3 1.4 1.2

1.2

Tables 1 and 2: Data collected from nine sites using resin lot 4 on 23 column systems.

The concentrations determined after reprocessing are given in Table 3. Theoretically, the detection limit of the method might be decreased by a further reduction in the threshold setting. However, all cases of sickle cell disease are detected at the current value. Specificity. In the past 4 years, GDL has been notified of 12 instances in which diagnostic follow-up results did not agree with the newborn screening hemoglobin patterns, and for which the discrepancy could be attributed to the specificity of the HPLC method (Table 4). Also, we have found that -1% of the chromatographic peaks at the retention time for Hb S have an atypical concentration ratio for [Hb A}:[Hb S] of -6:1 (usually the ratio is -1:1). The screening patterns for these newborns are FAS (also in Table 4). Hb G, an a-chain variant with four chromatographic peaks, is readily identified by visual inspection of the chromatogram. According to the rules given in the Appendix for derivation of patterns, and depending on the lot of cation-exchange resin in use, Hb G has been reported in newborn screening as Hb E, Hb D, or combinations of Hb E and D with Hb (2).

particular

Background and carryover. Water blanks are included on the calibration tray and on each tray of newborns’ samples for analyses. GDL staff review all chromatograms that show a peak with height >5000 .tV. These situations occur infrequently (9 times per 100 000 newborns) and do not affect newborns’ hemoglobin patterns. In some cases small spikes are observed, usually at the void volume (retention time 0.2 mm). However, the retention times are too fast, the peak widths too narrow, and the heights too low for these spikes to affect the hemoglobin patterns generated by the method algorithms (Appendix). Peak shape. Laboratory staff monitor the Hb F peak shapes for the controls and for 5% of all newborns’ specimens. By remov-

Table 3. ComparIson the result

No. of Hb

Liquid samples F F A S Dried blood spots F A S

systems

samples

18

40

Mean

SO

103 834 101 679 29 057 7 338

4621

4.4

11 112 4 295 1 002

10.9 14.8 13.6

23

105 213

23

213

23

27

37838

23

97

5284

23

206

23b

8CV <5% specified. Rangeof ±20% specified.

5 113

5 147 918 894

CV, %

13.6 17.4 17.5

after reprocessing data, % of total Hb

Resuft Newborn

Older

1

FS

FSaC

HbA, 0.9

2

FS

FSa

HbA, <0.5

Case no.

AU No. of

result with

Pattern

Table 2. PrecIsion of quantification. column

of the newborn screening

obtaIned at an older age.a

8

b

3 ES FSa HbA, <0.5 4 ES FSAC HbA, 0.8 5 ES FAS HbA, <1.0 6 FAd FAS HbS, 0.5 7 FA#{176} FAS FIbS, 0.88 8 F only FAS HbS, 1.08 9 FE FEa HbA, <0.5 10 FA’ FAE HbE, 0.8 11 F only HbA, 0.9, HbD, 0.8 Differencesattributable to prior use of higherthreshold (>500 MV). Original newborn screening data reprocessed to reflect current 500 V

threshold. r FSApatterns are reported as trait unless(Hb SJ >2 lHb Al,inwhichcase the pattern isreported as FSa forS/p.thalassernia (see Appendix). d Follow-up initiated from familystudies. 8 HbS >0.5%.

‘Reanalysis ofNB-DBS specimen gave 1.1% HD E and a patternofFAE. v represents an unidentifiedvariant.

g

Eastman

708

et al.: Screening

newborns

for hemoglobmnopathies

by HPLC

of dried

blood

spots

results dIfferIng from clinIcal foilow-up results because other Variants eluted at retention tImes for Hb S, E, and D. Screening pattern Follow-up pattern Follow.up methodb Comment Hb variant elutes on HPLC as FIb S. Incidence FAS FAV IEF 1/10 000 newborns (1% of FAS patterns). [FIb A1/[Hb V] - 6/1 FA IEF Hb variant,not resolved by IEF, elutes on FAE

Table 4. Newborn screening Case no.

See comment

1

HPLC as Hb E.

2 3

FAE FAE

FA FAV

IEF IEF

4 5 6

FAE FAE FAE

FAV FAC FAD/G

IEF Mother CAE

7 8 9 10

FAE

FAD/G FAG

FDE FAS

FETak

FAV

CAE Mother MS IEF

11

FAS

FA

IEF

Same as case 1. Hb variant

(not Hb C) elutes

on HPLC as

Hb E. Same as case 3. Same as case 3. Hb G elutes on HPLC as Hb E. (Hb D ruled out by the HPLC method.)

FDA

Same as case 6. FIb G elutes

on I-IPLC as FIb D.

Hb Tak elutes on HPLC as Hb D. elutes on HPLC as Kb S. [Hb Al/ [Hb V] - 1. Kb variant, not resolved by IEF, elutes on F$PLCas HbS. [Hb A]/[Hb V] - 1. Hb variant

V represents an unidentified variant. 8 Not the same as the cases in Table 3. IEF,isoelectric focusing; CAE,cellulose acetate electrophoresis;MS, mass spectrometry;Mother, the newborn’smother was tested for hemoglobinopathies.

ing defective columns from use in testing specimens, the staff maintain the height/area ratio for Hb F in liquid controls within a specification of 3.5 MV/AU. During a recent 12-month period when two lots of resin were in use, GDL rejected (6%) of 1769 columns because of broad Hb F peaks. Resolution. Hb F, A, E, D, 5, and C are well resolved

102

(Fig. 1).

Effects ofchangingthereagentlot. In 4 years we have used four lots of resin. The CV for retention times on the different lots was comparable with that given in Table 1. The accuracy of retention times for four lots of resin was comparable with that reported in Table 1 for lot 4. In two-thirds of the measurements, the observed mean values for retention times were within 0.01 mm of the specified center of the hemoglobin identification window. The maximum observed bias was 0.03 mm. The observed mean values for all hemoglobins were the same for liquid controls and eluates from DBS samples (no matrix effect). The precision of quantification was determined by measuring Hb F in the linearity calibrators with three column systems at one laboratory; the CVs were <6% for each of the four lots of resin. Photometer readings for the three linearity calibrators were within acceptable limits (nominal value ±20%). CVs for measuring Hb F, A, and S in DBS samples were <16% for all lots, which is similar to the results obtained on one lot (Table 2). Linearity. For four lots of resin and the corresponding lots of linearity calibrators, the dose-response curves for the three relative concentrations gave the following multiple R2 values: lot 1, 0.977; lot 2, 0.975; lot 3, 0.976; and lot 4, 0.985.

FREQUENCY

DISTRIBUTION

OF TOTAL

HEMOGLOBINS

When applied to the California newborn population, the HPLC screening method gave the following distribution for the total area for all hemoglobins: n = 151 000; mean = 208 000 AU; SD = 43 400 AU; skewness, 0.283; 1st percentile, 110 000 AU; 50th percentile, 207 000 AU; 99th percentile, 320 000 AU. Within the overall CV of 21% for the frequency distribution, the variance (SD)2 is estimated to be distributed among its components as follows: 14%, and HPLC

physiological methodology

14%, 72%.

DBS

sample

collection

Discussion The HPLC screening method quantifies the relative concentrations of hemoglobin variants and has good reproducibility with singleton determination. Quantitative ratio rules are invoked to derive automatically the presumptive phenotype for each newborn. Setting quantitative limits allows application of routine quality-control rules. Proficiency tests are scored with the use of quantitative acceptability limits. Analyte contents measured in newborns’ DBS specimens are dependent on the adequacy of the specimen. A hemoglobinopathies screening test result of an extremely low or high concentration of hemoglobin reveals specimens that are not suitable for determinations of any of the newborn screening analytes (phenylalanine, thyroxine, thyrotropin, uridyl transferase, etc.). In such cases a second blood specimen must be obtained from the newborn. The HPLC screening method requires only a small sample. One punch of a 0.95-cm-diameter disc from a blood-collection card is eluted in water to separate the hemoglobmns. This same

Clinical Chemistry

eluate is used for the determinations of two other newborn screening analytes, phenylalanmne and uridyl transferase. According to Bio-Rad, the rapid separation of hemoglobins is possible because proteins do not penetrate the resin. Also, any degraded hemoglobins and other proteins are removed from the cation-exchange resin before the hemoglobmns of interest are eluted. The interferences from variant hemoglobmns that have retention times similar to Hb S, C, E, and D (Table 4) are relatively few and do not compromise the detection of newborns with sickle cell disease. Also, degradation (if any) of hemoglobmns in a DBS sample does not interfere with the reporting of an accurate phenotype. Disadvantages of the method include the requirement for manual aliquoting and dilution of the specimen eluate into the microplate, which is subject to specimen identification error, given that a specimen may be pipetted into the wrong well of the microplate. Also, because the microplates have no barcode identification, a positional control is needed to maintain the sequence of analysis. Although the California program does not screen for Hb Barts, this variant is measurable with the HPLC screening method. The chromatographic peak for Hb Barts elutes with a retention time close to that of the void volume and can be seen, for example, in some DBS specimens from newborns with Hb E, giving a chromatogram characteristic of EJa-thalassemia. Other methods used to screen newborns for hemoglobinopathies are cellulose acetate (basic) and citrate agar (acidic) electrophoresis and isoelectric focusing. In a large-scale screening program, these methods do not compare favorably with HPLC screening, because they are not automated and quantitative. When electrophoresis is used, the presumptive phenotypes are derived by visual inspection, consensus decisionmaking, and manual data entry, all of which are subject to human error and judgment. With HPLC screening, presumptive phenotypes are derived automatically. Quality control and proficiency testing are quantitative. Compared with other HPLC techniques such as anion-exchange chromatography, the cation-exchange chromatography used here has the advantage that hemoglobin degradation products are eluted rapidly from the column and do not interfere with quantification of the principal hemoglobmns.

42, No. 5, 1996

709

2. Sickle Cell Disease Guideline Panel. Sickle cell disease: screening, diagnosis, management, and counseling in newborns and infants. Clinical Practice Guideline No. 6. AHCPR Pub. No. 930562. Rockville, MD: Agency for Health Care Policy and Research, Public Health Service, US Department of Health and Human Services, 1993. 3. California Health and Safety Code §309.5 (West 1990 & Suppl 1995). 4. Lorey F, Cunningham GC, Shafer F, Lubin B, Vichinsky E. Universal screening for hemoglobinopathies using high performance liquid chromatography: clinical results of 2.2 million screens. Eur J Hum Genet 1994;2:262-71. 5. Wilson JB, Headlee ME, Huisman THJ. A new high-performance liquid chromatographic procedure for the separation and quantitation of various hemoglobin variants in adults and newborn babies. J Lab Clin Med 1983:102:174-86. 6. Ou C-N, Buffone GJ, Reimer GL. High-performance liquid chromatography of human hemoglobins on a new cation exchanger. J Chromatogr 1983:266:197-205. 7. Huisman THJ. Percentages of abnormal hemoglobins in adults with a heterozygosity for an a-chain and/or a chain variant. Am J Hematol 1983;14:393-404. 8. Kutlar A, Kutlar F, Wilson JB, Headlee MG, Huisman THJ. Quantitation of hemoglobin components by high-performance cationexchange liquid chromatography. Am J Hematol 1984;17:39-53. 9. Rogers BB, Wessels RA, Ou C-N, Buffone GJ. High performance liquid chromatography in the diagnosis of hemoglobinopathies and thalassemias; report of three cases. Am J Clin Pathol 1985;84: 67 1-4. 10. Wilson JB, Wrightstone RN, Huisman THJ. Rapid cation-exchange high-performance liquid chromatographic procedure for the separation and quantitation of hemoglobins S, C, and 0 Arab in cord blood samples. J Lab Clin Med 1986:108:138-41. 11. Huisman THJ. Separation of hemoglobins and hemoglobin chains by high-performance liquid chromatography. J Chromatogr 1987; 418:277-304. 12. Ou C-N, Rognerud CL. Rapid analysis of hemoglobin variants by cation-exchange HPLC. Clin Chem 1993;39:820-4. 13. van der Dijs FPL, van den Berg GA, Schermer JG, Muskiet FD, Landman H, Muskiet FAJ. Screening cord blood for hernogmobinopathies and thalassernia by HPLC. Clin Chem 1992;38:1864-9. 14. Loomis SJ, Go M, Kupeli L, Bartling DJ, Binder SR. An automated system for sickle cell screening. Am Clin Lab 1990;Oct.:33-4O. 15. Weatherall DJ. The thalassaemia syndromes. Oxford, UK: Blackwell Scientific Publications, 1965:268 pp. 16. Bisse E, Wieland H. High-performance liquid chromatographic separation of human haemoglobins, simultaneous quantitation of foetal and glycated haemoglobins. J Chromatogr 1988;434:95110.

Results of the laboratory analyses in clinical follow-up presented in Tables 3 and 4 were determined under contract to the California Department of Health Services by the Children’s Hospital Oakland Research Institute, directed by F. Shafer, B. Lubin, and E. Vichinsky.

References 1. Office of Medical Applications

of Research. Newborn screening for sickle cell disease and other hemoglobinopathies. National Institutes of Health consensus development conference statement, Vol. 6, No. 9. Bethesda, MD: National Institutes of Health, Public

Health Service, US Department of Health and Human Services, 1987.

Appendix:

Rules for Derivation of Hemoglobin

Pattenis

A list of the hemoglobmns with their retention times is given below. Numbers in parentheses are used for unidentified species. For example, Hb (1) elutes between Hb F and Hb A. Noise and minor hemoglobins. When present, Hb A is used internal standard. Any chromatographic peak with an area the area of the Hb A peak is not included in the pattern. A is not present, the other adult hemoglobin (e.g., Hb S in cell disease) is used as the internal standard.

Hb (1).Hb (1) (possibly A. The

computer

program

Hb Aid) elutes between adds the concentration

as an <0.1 If Hb sickle

Hb F and Hb of Hb (1) to

710

Eastman

et al.: Screening

newborns

the concentration of Hb A. The total concentration in the pattern as Hb A.

for hemoglobinopathies

is expressed

Hb F1 The concentration to the concentration pattern as Hb F.

of Hb F1 (acetylated Hb F) is added of Hb F and the sum is expressed in the

Satellite hemoglobins. Consider Hb X as the major hemoglobin and Hb Z as a potential satellite hemoglobin [5, 7]: If the concentration of Hb Z is <0.25 the concentration of Hb X, Hb Z is deleted from the pattern. In this work the combinations of major and satellite hemoglobmns are (X/Z): E/A, D/(2), D/A, SIE, S, and Cl(s). Thalassemia flag. If Hb A and Hb Y (any variant) are both present, divide the concentration of Hb Y by the concentration of Hb A. If the quotient is >2, then change the representation of Hb A in the pattern from “A” to “a”. For example, once it has been determined that Hb A and Hb S are both present, the pattern report code of Hb A is changed to Hb a if the concentration of Hb S is more than twice that of Hb A. Thus S/-thalassemia is reported as FSa. For follow-up, FAS and FSA are both reported as trait, unless [Hb S] >2[Hb A], in which case the pattern is reported as S1f3-thalassemia [8, 15]. Inadequate specimens. When the total area is <60 000 AU, the pattern report is “not determined (low area).” Similarly, when the total area exceeds 420 000 AU, the pattern report is “not determined (high area).” When a repeat analysis confirms either low area or high area, the specimen is declared made-

Table 5. Peak criteria Peak name

Retention

F A

E

0.98

0

1.07

S

1.18

Fl

of dried

blood

spots

quate for all the newborn screening analytes, and a new specimen is requested. In the California program, the number of inadequate specimens so detected is -16 per 100 000 newborns tested. Sample degradation. A flag is used to identify specimens with excessive concentrations of hemoglobin degradation products. In the chromatography system used, these compounds are eluted before Hb F and appear in the two identification windows defined as FAST and Fl. (In most specimens, window Fl holds Hb F1, the acetylated form of Hb F, and no degradation products.) When the total relative concentrations of Hb FAST plus Fl exceed 50%, the hemoglobin pattern is reported as “not determined (FAST exceeds 50%).” In practice, after review of the chromatogram by a quality-control officer, valid Hb patterns can be reported with a total [FAST + Fl] as high as 75%. In the California program the incidence of chromatograms with [FAST + Fl] >50% is <1 per 100 000 newborns tested. Many of those found are the result of improper collection of the blood sample from an umbilical line. Such specimens are designated inadequate for determining all of the newborn screening analytes. F only. If the pattern

is F only, the result is printed

out as “not

determined (F only).” This result, which is expected only in cases of f3-thalassemia major, must be confirmed by repeat injection of the DBS eluate. Peak criteria. The peak criteria for inclusion of a species hemoglobin pattern are summarized in Table 5.

for Inclusion of a species

time, mm

0.18 0.45 0.61 0.83

FAST

by HPLC

in the

In the hemoglobin pattern report. Peak criterion for Inclusion In pattern

report

Always excluded Always added to F, so the total of [Fl] + [F] is reported in the pattern as F Always included Included, unless: (a) E is present and [A] <[E]/4, in which case A is deleted from the pattern, or (B) D is present and [Al <[D]/4, in which case A is deleted from the pattern Included, unless S is present and [E] <[S]/4, in which case E is deleted from the pattern

C (1)

1.72

(2)

0.73 0.91

(3) (4) (5)

1.13 1.33 1.55

(6)

1.85

and [Dl <[S]/4, in which case D is deleted from the pattern Always included Always included If F is present, the total of [(1)1 + [A] is reported in the pattern as A Included, unless D is present and [(2)] <[D]/4, in which case (2) is deleted from the pattern Always included Always included Included, unless C is present and [(5)] <[C]/4, in which case (5) is deleted from the pattern Always included Included,unless S is present