| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Department of Haematology and
2
Imperial College School of Medicine, Central Middlesex Hospital, Acton Lane, London NW10 7NS, UK.
a Author for correspondence. Fax 0181 965 1115; e-mail joan{at}dcs.bbk.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: Over a 15-month period, 25 750 blood samples, collected by heel prick onto filter paper, were tested using HPLC, and the results were compared with those obtained with IEF.
Results: HPLC and IEF each identified 568 patients with FAS, 151 with FAC, 49 with FAD-Punjab, 23 with FS, 3 with FC, 6 with FSC, 5 with FE, and 1 with FD. IEF detected 62 patients with FAE, whereas HPLC detected 63. This additional FAE was observed on repeat IEF. One additional heterozygote detected by HPLC was initially not observed by IEF, but was detected on repeat IEF. HPLC detected all but six cases of Hb Barts observed by IEF. One double heterozygote and four heterozygotes were detected by IEF, but not by HPLC. The detection of hemoglobin variants expressed at low concentrations was comparable for the two methods, and carryover was not observed in routine analysis on HPLC.
Conclusions: HPLC is a sensitive, efficient, and time-saving alternative to IEF for the neonatal screening of common hemoglobinopathies.© 1999 American Association for Clinical Chemistry
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Universal neonatal hemoglobinopathy screening programs in the United Kingdom are recommended where the minority ethnic population exceeds 15% (3). Central Middlesex Hospital (CMH)1 in northwest London tests all babies born in the North Thames (West) health region, ~50 000 births per year. Maternal consent is sought before the taking of samples. The population of region is 19% non-Caucasian, with individual districts ranging from 3% to 61%. For the past 9 years, this laboratory has used isoelectric focusing (IEF; Isolab) to screen neonates for hemoglobin disorders. Although the results have been found to be satisfactory, this is a labor-intensive manual technique, and the need has arisen to identify a suitable automated method capable of replacing the existing system. For this purpose, we chose the Bio-Rad Variant HPLC system (Bio-Rad Laboratories).
There are several programs available for this system. One program is the Sickle Cell Short Program, which is a rapid 3-min assay and uses either filter paper blood spots or whole blood samples. This program is specifically designed to provide a qualitative result for hemoglobins (Hbs) A, F, S, C, D, and E in the neonate.
Another program is the ß-Thalassemia Short Program, which is a 6.5-min assay designed to quantify Hbs A2 and F. The performance of this program for the quantification of these Hbs and the identification of other Hbs, including A, S, C, D, and E, in adult samples has been described elsewhere (4)(5)(6).
For the purposes of neonatal screening as described here, a qualitative result is all that is required initially. Because of this, and the desire for a more rapid assay, the Sickle Cell Short Program was chosen as the test method.
In this evaluation, we compared "back-to-back" the efficiency and suitability of HPLC for neonatal hemoglobinopathy screening against the established technique of IEF. This included a direct comparison of results, an estimation of the lower limit of detection of both methods, and observation of carryover and the effects of dried blood spot degradation in the HPLC method.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
samples
Heel prick samples taken 7 days after birth onto Schleicher &
Schuell grade 903 filter paper were used. The 3-mm spots used for both
assays yielded a mean Hb concentration of 18.0 g/L (range, 7.9–20.5
g/L). Spots were tested with both methods on the day of receipt
or at the latest on the next working day. Before and after testing,
specimens were stored between 2 and 8 °C.
cation-exchange hplc
The Bio-Rad Variant Hemoglobin Testing System is an automated HPLC
instrument that uses a minicartridge filled with a cation-exchange
polymer material. Following the manufacturer's guidelines, we punched
3-mm blood spots from the filter paper and eluted them with 0.5 mL of
distilled water for 20 min. The system uses two phosphate elution
buffers, of differing concentrations, that are introduced into the
flow-stream at a controlled rate of 2 mL/min. Each sample was injected
into the flow stream via an automatic probe. As the concentration of
the mixture increased, the more strongly retained hemoglobins eluted
from the cartridge. Changes in absorbance were measured using a
dual-wavelength filter photometer, and these changes were plotted
against time. Each Hb has a characteristic retention time.
Chromatograms showing the results graphically and as numerical data
were printed automatically for each sample within 3 min of the
injection. The maximum batch size was 100 samples, with a run time of
5 h and a preparation time of ~20 min. The manufacturer provides
Hb FAES and Hb FADC controls, artificial mixtures of hemoglobin
containing the approximate concentrations expected in neonatal samples,
and these were tested at the beginning and end of each analytical
run.
The researcher examined each chromatogram and recorded the results on a worksheet, later transcribing the IEF results onto the same sheet. In all cases of discrepancy between the two techniques, the sample cards were repunched and retested using both methods.
ief
The IEF analysis used a precast agarose gel placed on a horizontal
water-cooled tank unit. Blood spots (3 mm) were punched from the filter
paper sample and eluted with 50 µL of elution solution for a minimum
of 30 min. A template was used to apply the eluate, and 50 W was
applied until discrete bands were visible (usually ~2 h). The plate
was then fixed with 100 g/L trichloroacetic acid, rinsed with
water, and stained with 2 g/L o-dianisidine in methanol
(7). A total of 72 samples and 4 FASC controls were
tested on each plate. Identification in the first instance was visual,
using control positions as guides, with each plate being independently
read by three different readers, of which two were experienced senior
staff.
All samples with abnormalities were repunched and repeated on subsequent IEF plates, using appropriate controls. Samples that appeared inadequately focused or that had merged with adjacent samples on the gel were repeated. Babies without Hb A or with a variant other than Hb S or Hb C were recalled for retest at 6 weeks of age.
establishing the limitations of the method
Each chromatogram that followed a positive result was examined for
carryover from the previous sample. Carbon tetrachloride hemolysates
prepared from known Hb SS and CC patients whose blood contained <2%
Hb F (quantification determined with Shimadzu IE-HPLC; Dyson
Instruments) were added to a known Hb FA cord blood sample to achieve
percentages of Hb S or C serially from 0.2% to 2%. To ascertain the
lower limits of detection, subsamples at each stage of dilution were
tested with both IEF and HPLC.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. The most common phenotype observed in this study was sickle
cell trait (60%). The Hb C trait, prevalent in peoples of West African
ancestry, was the next most common abnormality detected (16%).
Thirty-nine infants (4%) were confirmed to have hemoglobin disorders
of clinical relevance.
|
The HPLC and IEF results are summarized in Table 1
. HPLC detected all of the diseases detected by IEF except for
one double heterozygote, in which the printed chromatogram was entirely
normal (Table 2
). When detecting sickle cell trait (FAS) and Hb C trait (FAC)
present in the neonate, HPLC was successful in 100% of cases that were
detected by IEF.
|
|
There were 57 cases of FAD reported at the initial screen by IEF, 53 of which were confirmed by repeat testing at 6 weeks: 49 as AD-Punjab; 3 as AD-Iran, and 1 as AD-Korle Bu. Follow-up was not possible in four neonates. In all cases in which Hb D was detected using IEF, an abnormality was detected by HPLC. However, only cases of D-Punjab (the most prevalent Hb D type in the region) were detected in the window for Hb D in the HPLC method. Hb D-Iran was observed in, or very near, the window for Hb E in the HPLC method. Hb D-Korle Bu moved marginally slower (0.1 mm) than D-Punjab and D-Iran on IEF. In the one case of Hb D-Korle Bu detected by IEF, the Hb was in the window for Hb E.
FAE, or Hb E trait, was detected by IEF in 69 cases, 63 of which were
confirmed at 6 weeks. Six cases were lost to follow-up. HPLC detected
all of the cases of Hb E that IEF detected plus one additional. In this
case, upon reexamination of the original IEF plate, a very faint band
was observed (Table 2
).
In addition, in the HPLC method, there were 11 cases in which an abnormality was observed in the window for Hb D and 9 cases in which an abnormality was observed in the window for Hb E for which the abnormality was not identified as Hb D or E by IEF. These cases were reported initially as FA + Band(s) by IEF, and most were subsequently confirmed and typed, and are detailed in the Discussion.
less common abnormalities
HPLC also detected various abnormalities outside its design
capabilities for Hbs A, F, S, C, D, and E. In 71 cases, the
chromatograms showed anomalies, mainly in the "Fast" region, the
"Unknown 2" region (between Hb F and Hb A), and the area between Hb
S and Hb C. In one additional case, a definite abnormality was observed
in the window for Hb E in the HPLC method, but was not detected by IEF.
This band was noted on repeat IEF in a position 1 mm slower than Hb F,
and review of the original plate demonstrated poor focusing, which
obscured detection (Table 2
).
In three other instances, a possible abnormality—a slight peak between Hb F and Hb A—was detected by HPLC, but the IEF results were normal. Two of these infants were resampled at 6 weeks of age and found to be normal (FA) by HPLC and by all other methods, including IEF. The third family refused the retest.
Thirty-four bands detected at birth by IEF were found to be much
reduced or to have disappeared at the time of the 6-week repeat
testing. These were classified as 
-chain variants and could not be
investigated further. HPLC detected 24 of these -chain variants,
whereas 10 were not observed either at the initial testing or at the
6-week repeats.
In four cases, the presence of a heterozygous variant was detected by
IEF and confirmed by the laboratory at 6 weeks, but was not detected by
HPLC (Table 2
). One of these cases was identified as Hb Chicago, and
the other three are still under investigation.
Seventeen additional cases were lost to follow-up. Five of these were
not detected by HPLC; two were thought to be the -chain variant
F-Sardinia. Table 3
gives a breakdown of the Hb types confirmed during the test
period. Confirmation methods included cellulose acetate, acid gel and
globin chain electrophoresis, urea triton, amino acid analysis,
and mass spectrometry. With the exception of Hbs Chicago and Harrow,
all were detected by HPLC. In addition to those that appeared in the
windows for Hbs D and E, the other abnormalities observed fell outside
the preset windows of the HPLC system. Examples of chromatograms are
shown in Fig. 2
.
|
|
Hb BARTS
Hb Barts (4) occurs in fetal life in 
-thalassemia when there
are reduced chains available. In the neonate, Hb Barts is
recognized as two or three very fast-running bands on IEF. Studies have
shown that the degree of Hb Barts present in the neonate is suggestive
of the number of nonfunctional or deleted genes and thus the
severity of the -thalassemia (8). At CMH, only when three
Hb Barts bands are present on IEF, with the middle band approximately
equal in appearance to the Hb A, are babies resampled at 6 weeks of
age. This is done to test for Hb H disease, which is characterized by
moderately severe anemia and three nonfunctional genes. Since 1989,
only five cases of Hb H disease have been identified by CMH, with none
observed during the test period.
HPLC was capable of detecting Hb Barts in routine analysis. Hb Barts
appears as a distinct, increased peak in the Fast region (first
peak observed) on the chromatogram (Fig. 2
). HPLC detected 208 cases of
such an increase in the Fast region (values ranging from 4% to 23%
Fast). Each case was compared with the corresponding sample tested on
IEF. Fig. 3
shows all 208 such cases observed by HPLC, and the
corresponding IEF result. A Student t-test performed between
the "observed" and "not observed" groups demonstrated a
significant difference in the means between the groups
(P = 0.0000; = 0.05). In 58 cases detected by
HPLC (varying from 3% to 9% Fast), no Hb Barts was observed on IEF.
This may be because IEF is not a sensitive test for Hb Barts, and it is
possible that many of the cases of increased Fast observed by HPLC, but
not by IEF, are in fact true cases of single nonfunctional or deleted
genes. There were six instances in which Hb Barts was observed on
IEF but there was no increased peak observed on the corresponding
chromatogram. Without genetic analysis, which was not within the scope
of this study, no further explanation can be given with regard to these
anomalies. Hb H disease was not apparent in any of these cases.
|
carryover and limits of detection
No carryover was observed at any time during the test period
during routine neonatal analysis using HPLC. The lower limits of
detection are shown in Table 4
. The IEF results were determined blind on day 0 from
hemolysates poured onto filter paper cards. HPLC results were
determined from fresh hemolysates diluted on day 0.
|
The IEF results were very similar to those reported previously for the detection of Hb S and Hb C (9), although we were able to produce a detection limit of <0.5% for Hb C and <0.75% for Hb S, in contrast to the previous report in which detection by HPLC was successful at 1% Hb abnormality or higher.
degradation
The effect of aging of the dried blood spots on the HPLC analysis
was tested without back-to-back IEF testing because the minimal effects
of storage on results from IEF have been well established
(10)(11). Wajcman et al. (10)
successfully performed IEF analyses on samples that had been stored at
4 °C for up to 18 months.
Between 2 and 4 months after the initial testing, 20 heel prick samples
(randomly chosen from a larger sample representing many of the various
abnormal Hbs) were retested using HPLC. Although high baselines,
increased noise in the Fast region, and wider areas for both normal and
abnormal Hbs were noted, the results were satisfactory for Hbs A, F, S,
and C except from premature infants born at 35 weeks gestation or
less. However, ambiguous results were observed in cases of other
abnormalities (especially -chain variants that occur in small
percentages in the neonate) and Hb Barts.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Costing data from our laboratory reported elsewhere (12) demonstrated only marginal differences between techniques for programs screening >25 000 births annually. The Bio-Rad Variant has interchangeable programs for quantitative analysis of Hbs A2 and F in nonneonates and for Hb A1c in diabetics.
The capacity of a single HPLC system using this type of rapid program is ~230 patient samples per 8-h working day. One hundred samples can be analyzed during a 5-h period essentially unaided, and 20–30 samples can be analyzed before the last batch of 100, which can continue after regular work hours. The system shuts down automatically at the end of the run. However, some time must be allowed for routine maintenance and running retention markers; therefore, 200 patient samples per day is more realistic. This means that one system should be adequate for a laboratory testing of up to 50 000 samples per year. When IEF is used, 72 samples can be tested in 2 h, giving a daily workload of 432 samples for one tank. Each power pack and cooler can run two tanks; therefore, the capacity can be doubled for little extra capital cost.
IEF plates cannot be read until the plate is dry, usually overnight, whereas results from HPLC are available 3 min after sample injection. In addition, the STAT option on the Bio-Rad Variant allows urgent queries to be met. In our experience, however, such urgent results are rarely requested (less than one per year).
When performed correctly, IEF produces excellent hemoglobin separation, with very little band overlap when bands are measured to 0.1 mm against controls. However, results obtained with IEF can be inconsistent because of variable application and staining techniques. Sample merging and insufficient focusing occur even with experienced staff, leading to a substantial number of otherwise unnecessary repeats. The manufacturer's guidelines for sample preparation for the HPLC method are simple and lead to very few technical repeats.
When IEF is used, samples are first punched and eluted in microtiter plates and then transferred by pipette to the appropriate position on the gel template. Not only is there the possibility of a punching error, there is the risk of a pipetting error at the sample application stage. The biggest single advantage of automated HPLC is that the sample is punched directly into the tray that is used for analysis. Thus, although the possibility of an initial punching error remains the same, the risk of a pipetting error is eliminated.
The interpretation of results obtained with IEF requires much
experience. Chromatograms obtained with HPLC are less subjective than
IEF for common abnormalities because the results are observed in preset
windows, although rarer variants are found outside these windows and
can be harder to recognize. Caution is needed in interpreting Hbs D and
E as detected by HPLC because many other variants are observed in these
windows. For example, Hb D-Iran, D-Korle Bu, Abruzzo, and several
-chain variants elute in or near the window for Hb E, whereas Hb
G-Philadelphia, G-Norfolk, Matsue-Oki, Winnipeg, Stanleyville II, and
certain -chain variants overlap in the window for Hb D. Hb Memphis
was observed in the window for Hb S, but at a much lower concentration
than would be expected in a true sickle cell trait. The overlap does
not significantly affect the screening outcome, providing the
laboratory has a suitable protocol for confirmatory testing and that
the counseling strategy is planned with this in mind.
Certain -chain variants are only expressed neonatally in very low
concentrations and can, therefore, be hard to detect. Hb Winnipeg was
present in the sample we tested at only 0.7% in the window for Hb D
despite being from a full-term baby, and it could be overlooked by
inexperienced operators. We therefore suggest that any anomalous
chromatogram be treated as a possible hemoglobin abnormality and
repeated and/or recalled as appropriate.
Of the five abnormalities detected with IEF, but not with HPLC, three
cases are still under investigation, and their clinical significance is
not known. Two are ß-chain variants that could be significant if
present in the homozygous or double heterozygous state. The -chain
variant Hb Chicago has no known clinical significance. The new
hemoglobin (Hb Harrow, ß118 Phe-Cys) eluted in the window for Hb A on
HPLC, giving the impression of a normal neonate. This baby had no Hb A,
and we do not yet know the full clinical effects. At 1 year of age,
however, the baby remains well and is maintaining his hemoglobin at 94
g/L. We do not think he will need a lifelong transfusion regime.
This case does highlight the problems of identifying abnormal
hemoglobins that fall in the window for Hb A.
The two heterozygotes initially missed by the IEF method (Hbs E and Abruzzo) were attributed to technical problems in focusing and staining the gels.
The performance of HPLC was comparable to IEF with few exceptions. The
sensitivity, or proportion of positives correctly identified by HPLC,
as calculated with the aid of a "truth table" (13), is
0.9945 and reflects the five cases not detected with this method. True
positives were based on the confirmed results detected with the
reference method and, therefore, do not include those cases lost to
follow-up. Because of the subjective detection of Hb Barts by IEF,
these numbers have not been included in the truth table analysis.
Because of the lack of clinical significance of -chain variants and
because their inclusion in the truth table analysis would give an
erroneous impression of the real performance of HPLC, these numbers
were also excluded. The positive predictive value is the
proportion of patients with a positive result who are correctly
identified, and is established at 0.9978. We have defined the two cases
that were recalled for repeat testing for abnormalities on the basis of
the HPLC analysis alone as false positives. The proportion of negatives
correctly identified by HPLC, the specificity, was 0.9999, whereas the
proportion of neonates with negative results who were correctly
identified, the negative predictive value, was 0.9998. For a
screening program such as this, it is most important that the
specificity and negative predictive value are high. Any false positives
will be eliminated by the use of new samples and the inclusion of
different methods on follow-up.
HPLC detected fewer -chain variants than IEF during the test period.
Because these are generally clinically insignificant and disappear
during the first few months of life, this could be considered an
advantage because there would be fewer repeat samples requested and
thus less anxiety and burden on the families involved.
Although dried blood spots are not the preferred medium for the detection of Hb Barts, we found that HPLC performed as well as IEF.
Hemoglobin carryover from one sample to the next was not observed during routine analysis using HPLC.
In summary, the HPLC method performed satisfactorily throughout the evaluation, and the results compared well with our existing IEF method for all common and clinically significant Hb abnormalities. The use of an automated system will produce substantial time saving in our laboratory.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  L. B. Thomas, S. J. Agosti, M. A. Man, and S. M. Mastorides Screening for Hemoglobinopathies During Routine Hemoglobin A1c Testing Using the Tosoh G7 Glycohemoglobin Analyzer Ann. Clin. Lab. Sci., January 1, 2007; 37(3): 251 - 255. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Joutovsky, J. Hadzi-Nesic, and M. A. Nardi HPLC Retention Time as a Diagnostic Tool for Hemoglobin Variants and Hemoglobinopathies: A Study of 60000 Samples in a Clinical Diagnostic Laboratory Clin. Chem., October 1, 2004; 50(10): 1736 - 1747. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. M. Clarke and T. N. Higgins Laboratory Investigation of Hemoglobinopathies and Thalassemias: Review and Update Clin. Chem., August 1, 2000; 46(8): 1284 - 1290. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
