HEMATOPATHOLOGY
Original Article
Interference with Glycated Hemoglobin
by Hemoglobin F May Be Greater
Than Is Generally Assumed
TINA COX, B.S.,' P. PATRICK HESS, PH.D., 2 GERALD D. THOMPSON, MT(ASCP),3
AND STANLEY S. LEVINSON, P H . D . 2 3
An automated electrophoretic method to measure glycated hemoglobin (Hb Al) was compared with manual affinity methods.
Good correlation between methods was found. The electrophoretic method showed good run-to-run precision, good linearity,
was free from interference by the labile aldimine fraction, and
required less time and considerably less consumable expense
than the affinity methods. However, as previously reported, Hb
F comigrating with Hb Al caused spurious increases in glycated
Hb levels as compared with the affinity methods. This effect
was linearly dependent on the Hb F concentration. Using the
discrepancy between concentrations of Hb Al by electrophoresis
and glycated hemoglobin by affinity methods, 330 patients were
screened in two hospitals for Hb F. A 12% frequency of elevated
Hb F (defined as a level that is more than 2%) was found in
patients from a community-tertiary care hospital, which is significantly greater than the 1.5% frequency commonly thought to
occur in the adult population at large, whereas patients from a
Veterans Administration Hospital showed an elevated frequency
of 2.2%. Based on this and other studies, it is concluded that the
frequency of elevated Hb F in adults may vary substantially
among different medical centers. The authors recommend against
using this method and suggest that laboratories that persist in
using it should periodically assess the frequency in patients with
Hb F levels greater than 2% of total hemoglobin, replacing the
method if an unacceptably high frequency is found. (Key words:
Glycated hemoglobins; Hemoglobin Al; Hemoglobin F; Affinity
chromatography; Agar electrophoresis) Am J Clin Pathol 1993;
99:137-141
In 1958, D. W. Allen1 and associates used cation exchange
chromatography to separate hemoglobin (Hb A) from a
heterogeneous fast minor fraction, subsequently shown
to consist of the glycated hemoglobins Hb Ala, Hb Alb,
and Hb Ale. 2 Hemoglobin Ale has been shown to be a
valuable indicator of diabetic control.3"6 Although Hb
Ala, which contains phosphorylated sugars, and Hb Alb
are not elevated in diabetes,7 the measurement of total
glycated Hb or total Hb A1 correlates well with the measurement of Hb Ale alone.
A variety of methods for clinical laboratories are available to measure glycated hemoglobins. These include
conventional ion-exchange chromatography, high-performance liquid chromatography (HPLC), colorimetric
methods, 7 affinity separation employing boronic acid
bound to solid-phase support,8 and electrophoresis.
R. C. Allen and associates described a method to measure Hb Al using citrate agar gel electrophoresis, pH
6.1. 9 " 1 ' Electrophoresis possessed certain advantages over
the then widely employed cation-exchange chromatography methods because, unlike ion-exchange, the Hb Al
assay by electrophoresis showed temperature stability and
was unaffected by lactescence.12 However, like cation-exchange chromatography, levels of Hb A1 can be falsely
elevated by Hb F.'° In addition, electrophoresis suffers
from interference by a reversible aldimine (labile) glycated
fraction, which is not indicative of diabetic control,'' 3
from abnormal hemoglobin variants that may demonstrate interfering peaks, and is labor-intensive to perform.
As a result, with the introduction of commercial manual
affinity methods, which are not subjected to any of these
From the 'Department of Medicine, and the2 Department of Pathology.
University of Louisville, and the,} Veteran Affairs Medical Center, Louiscauses of interference, the use of electrophoresis to meaville. Kentucky.
sure glycated hemoglobin decreased.
Supported by the Department of Veteran Affairs.
Recent advances in automation have led to the develReceived November 22, 1991; revised manuscript accepted for pubopment of electrophoretic equipment that greatly reduces
lication April 29, 1992.
Address reprint requests to Dr. Levinson: Laboratory Service, Veteran labor-intensity and permits the simultaneous assay of
Affairs Medical Center, 800 Zorn Ave., Louisville, Kentucky 40206.
multiple samples. Laboratories have purchased this
137
138
HEMATOPATHOLOGY
Article
equipment to perform cardiac isoenzyme assays. This
equipment offers an apparently simple, rapid, low-costper-test alternative to assay glycated hemoglobin. As a
result, the use of electrophoresis for this assay appears to
be increasing: the first 1990 College of American Pathologists Survey (EC-B) showed that 12% of 675 reporting
laboratories use this method, whereas the first Survey for
1991 (E-A) showed that 17% of 614 laboratories use this
method.
Rapid electrophoresis (REP) is an automated electrophoresis system (Helena Laboratories, Beaumont, TX).
In this report, we describe our evaluation of the REP for
measuring Hb A1 in comparison with affinity methods.
Indeed, the REP offers a less tedious, reproducible, and
inexpensive alternative to assay Hb A1 in appropriate patient populations. However, comparative data on methods
indicate that the frequency of elevated Hb F is higher in
some adult populations than is commonly believed, and
therefore presents a greater potential for interference than
is generally assumed.
MATERIALS AND METHODS
Rapid Electrophoresis
Lysis of cells and rapid electrophoresis was performed
with kits from the manufacturer according to instructions.
Erythrocytes were lysed by the addition of 25 fiL whole
blood to 75 nL saponin in 1 mol/L borate buffer, pH 6.1
(Lysing Reagent, formulated to remove the labile aldimine
fraction), mixed, and incubated for 10 minutes. The lysis
was the only manual step. Lysates were automatically pipetted, electrophoresed on a citrate agar gel (pH 6.1), and
dried. Up to 30 samples can be electrophoresed simultaneously. After electrophoresis, the gels were scanned at
415 nm in an automated densitometer, and results were
printed as a percentage of fast-migrating peak area to the
total peak area. Samples were assayed in less than 2 hours
with approximately 20 minutes of hands-on time.
Affinity Separation Methods Using Solid-Phase
M-Aminophenylboronic Acid
Glycated hemoglobin was assayed by two affinity
methods, Glycoscreen (Pacific Hemostasis, Ventura, CA),
and Glycoglobin (Endocrine Sciences Products, Tarzana,
CA), which is a conventional affinity chromatography
method. The two assays yielded similar results. The affinity column method for assaying glycated hemoglobin is
a multistep procedure that has been well described in other
reports. 7-814 It took approximately 4 hours to assay 30
samples by the column technique.
Patient samples were selected randomly from those with
routine requests for glycated Hb. Bloods for glycated Hb
A.J.C.P.-
determinations were collected in EDTA-containing tubes
in the Teaching Hospitals of the University of Louisville
by standard methods. One-hundred eighty blood samples
were obtained from the Veterans Administration Hospital
(VAMC) and 150 from Humana Hospital, University of
Louisville (HHUL). All assays were performed individually, as suggested by the manufacturers.
Hemoglobin F from cord blood erythrocyte lysates was
estimated by electrophoresis with the REP, after assay of
total hemoglobin by standard hematologic methods. Glycated hemoglobin in cord blood was determined by the
affinity column chromatography method, and shown to
be 5.6 g/L, or 4.7% of the fast-migrating Hb fraction. This
is a very small component of the total fast-migrating hemoglobin. Therefore, the amount of fast-migrating hemoglobin in cord blood was considered to be Hb F. Hemoglobin F in patients' samples from both institutions
was assayed using alkaline denaturation. 15
To assess interference of irreversibly bound glucose by
reversibly bound glucose, erythrocytes from two different
patient samples were diluted with phosphate-buffered saline containing 0 and 50 mmol/L D-glucose and incubated
for 3 hours at 37 °C, as previously described.16 The
amount of glycated hemoglobin was assayed in triplicate
by electrophoresis and by affinity separation (using Glycoscreen).
Reproducibility studies for the different methods were
conducted with controls provided by the manufacturer of
the kit. The level of glycated hemoglobin in normal and
high controls, respectively, averaged 7.7% and 16.3%.
Stability of storage for glycated hemoglobin was assessed
by lysing whole-blood cells as for electrophoresis, and
storing the lysates, along with duplicate unlysed wholeblood cells, at 4 °C, - 2 0 °C, and - 7 0 °C for varying
amounts of time, after which electrophoretic analysis was
performed.
Linearity and sensitivity of fast-migrating hemoglobin
was assessed by electrophoresis of lysates obtained from
diluted cell suspensions of adult or cord blood. The dilutions were made with phosphate-buffered saline, pH 7.4
(Beckman Instruments, Brea, CA; product number
449690), and the cells lysed according to the protocol for
the electrophoretic method.
RESULTS
Coefficients of variability for 13 between-day assays for
normal and high controls by electrophoresis were 5.2%
and 3.3%, respectively. The within-run coefficients of
variations were 3.8% for the normal and 2.1% for the
elevated samples (n = 13).
After incubation, with and without glucose, the mean
levels of glycated hemoglobin from a sample containing
ruary 1993
COX ET AL.
Interference with Glycated Hemoglobin by Hemoglobin F
no glucose were 10% and 7% by the electrophoretic and
affinity methods, respectively. The mean levels in the same
sample incubated with glucose were 10.2% and 7.2%, respectively. A second sample, containing no glucose,
showed mean levels of 17.8% by the electrophoretic
method and 14.7% by the affinity method, and 17.5% and
15.2%, respectively, when incubated with glucose. We
concluded that irreversibly bound glucose does not significantly interfere with the assay.
Linearity and sensitivity for the fast-migrating hemoglobin fraction by electrophoresis is depicted in Figure 1.
Measurement of patients' blood (shown in the figure with
circles, triangles, and squares) indicates that the assay is
linear over the physiologic range of from 2 g/L to at least
24 g/L (equivalent to 17% at a total Hb of 140 g/L). Measurement of Hb F from cord blood (diamonds) indicates
that linearity extends to at least 120 g/L (linearity greater
than 30 g/L is not shown). Hemoglobin F contributes
linearly to the fast-migrating fraction and cannot be distinguished from Hb Al.
The concentration of glycated Hb from two normal
and two elevated samples, stored as cell suspensions or
as hemolysates for as long as 28 days, showed no significant change. Interestingly, although the fast-migrating
fraction remained unaffected relative to the total, some
samples showed a split of the Hb A fraction into a minor,
faster migrating fraction when stored (Fig. 2). This was
observed with samples stored as frozen cells or as hemolysates, except that a split in the Hb A fraction occurred
after 3 days of storage as cells but not until about 14 days
as hemolysates. Storage at —20 °C produced the greatest
splitting.
A normal reference range of 4% to 7.2% for the affinity
method was previously established in our laboratory
(HHUL). The normal reference limits for the electrophoretic method was determined to be between 6.5% and
9.8% using regression line analysis and clinical data from
100 patients from the VAMC.
The regression line between the electrophoresis (y) and
the column method (x) for 150 patients from HHUL was
y = 0.66 X + 5, r = 0.86, P = 0.0001; yielding a normal
range of 7.8% to 9.8% for electrophoresis, which correlated
well with the established range.
We screened bloods for Hb F by dividing values obtained by the electrophoretic method by the corresponding
value obtained using the affinity method. Blood samples
having more than a 50% discrepancy (equivalent to 3%
Hb F at 6% glycated Hb) were examined to determine if
the absolute difference was 3% or more of total hemoglobin. We used this screening approach to avoid the need
to analyze samples with very high concentrations of glycated Hb for Hb F because such samples were substantially
in agreement by both methods clinically. Nineteen of 150
0
0.1
0.2
0.4
139
0.6
DILUTION
FIG. 1. Linearity of fast migrating hemoglobin with cell dilution. Glycated
hemoglobin was assayed in three patient samples (D), (A), and (O), and
Hb F in cord blood (0). The dashed lines (—) indicate the percentage
of total hemoglobin in the fast migrating fraction (right vertical axis),
and the solid lines (
) represent the absolute level of hemoglobin in
g/L (left vertical axis). The top dashed line (—) represents percentage of
Hb F in the cord blood specimen and reflects the four points to the right,
but excludes the lowest dilution that deviated markedly.
values obtained from HHUL exhibited a difference of 3%
or more of the total hemoglobin, and 4 of 180 bloods
from the VAMC showed a deviation of 3% or more of
the total. One of the 4 VAMC bloods and 3 of the 19
HHUL bloods showed spurious elevations when the
comparative normal reference ranges for the assays were
compared. Table 1 shows the levels of glycated Hb and
Hb F for these samples. Eleven unhemolyzed specimens
of the 19 blood samples from HHUL and 2 unhemolyzed
of the 4 samples from the VAMC were assayed for Hb F
(also shown in Table 1). We assumed that the normal
range for hemoglobin F in adults was less than 2%.' 718
All of these blood samples showed values of 2% or more
Hb F, although the two samples from the VAMC were
barely increased at 2%. On the other hand, all 11 blood
samples from the HHUL were well above the 2% level,
with the lowest being 2.5%.
DISCUSSION
The REP is the first automated electrophoretic system
available for use in the clinical laboratory. In our laboratory, it has proved to be easy to use for quantitation of
Hb A1 levels, showing good precision (coefficients of variation, 5.2% and 3.3% at levels of Hb Al of 7.3 and 16.3,
respectively), and requiring less hands-on time than the
manual affinity column method. Many laboratories have
purchased the REP to assay cardiac isoenzymes. When
Vol. 99 • No. 2
140
HEMATOPATHOLOGY
Article
HgbA1
HgbA1
!
HgbA(Fr1)
Fr2
HgbA
Fr 1
FIG. 2. Splitting of Hb A into two fractions with storage. Electrophoretic
densitometer profile of hemoglobin using fresh blood (upper) and blood
stored as cells for 14 days at -20°C (lower). The concentration in percentage for Hb Al in the fresh sample was 15.3% and for the frozen
sample it was 16.1%. By 14 days, the Hb A in the frozen sample (lower)
split into two fractions, Fr 1 and Fr 2. Migration is in the direction of
HbAl.
the REP is already in the laboratory, its use is substantially
less expensive for assaying glycated Hb than affinity
methods (approximately $0.30 vs. $1.50 per sample).
The labile aldimine fraction of Hb A, formed before
the irreversible Amadori rearrangement, does not appear
to interfere significantly with the assay. Several approaches
have been used to reduce this interference: a 22-hour incubation in saline,16 a 30-minute incubation with carbazide and aniline, pH 5.0,19 and simply incubation at
pH 5-6 for 30 minutes. 20 As formulated by the manu-
A.J.C.P. •
facturer, incubation at pH 6.1 for 10 minutes in 1 mol/
L borate eliminates this interference.
The assay showed good linearity for fast-migrating hemoglobin from 2 g/L to more than 120 g/L (shown to 30
g/L; Fig. 1). The lower level of linearity is well below the
lower limit of the reference range. It is also apparent (Fig.
1) that Hb F behaves in a manner similar to Hb A1 with
dilution, and thus the two cannot be differentiated.
Boyer and associates21 reported a normal range of 0 to
3% for Hb F, with most adults showing levels of less than
2%. The normal range for Hb F in the adult population
is generally assumed to be less than 2% of the total hemoglobin. 1718 Hb F concentrations below this level do
not significantly interfere with Hb Al analyzed by electrophoresis because a bias for this small amount has been
incorporated into the normal reference range. Using a
normal range of less than 2%, Krause17 found an elevated
frequency of about 1.5% for Hb F, which is, in our experience, a value commonly quoted by representatives of
manufacturers producing these electrophoretic methods.
Contrary to these findings, Wood and associates22 reported a higher normal range of up to 4.8% in normal
adults for Hb F, with a substantial number of patients
showing values greater than the 2% level. The present
study suggests that one reason for the discrepancy between
studies is because the number of adults with Hb F levels
greater than 2% may vary between populations of various
medical centers. Accordingly, the patients at the VAMC
exhibited an elevated frequency consistent with the lower
estimate (4 of 180, or 2.2%), whereas a much higher frequency was observed in samples from HHUL (19 of 150,
or 12.7%). This difference in the frequency of elevated
Hb F between the two hospitals was statistically significant
(chi-squared test = 10.45, P = 0.001, DF = 1). Furthermore, It is noteworthy that a conservative approach
(screening with a 50% difference) was used to identify
discrepancies.
It is known that Hb F commonly accompanies chronic
anemias, such as hereditary spherocytosis, myeloproliferative, and myelodysplastic disorders. As indicated in
Table 1, patients with increased levels of Hb F from
HHUL included men and women, whose ages varied
widely. No specific features that would better characterize
the bias in terms of disease or racial category were noted
in this small sample. The reason for the difference in frequency of elevated Hb F among patients from the HHUL
and VAMC is unknown at this time.
In light of current pressures to reduce costs in clinical
laboratories, given the good analytical performance along
with cost and time savings over manual affinity methods,
there appears to be an increasing use of electrophoretic
methods to assay glycated Hb. Despite these advantages,
we recommend against using these methods because our
jruary 1993
COX ET AL.
Interference with Glycated Hemoglobin by Hemoglobin F
141
TABLE 1. HEMOGLOBIN F SAMPLES WITH 2:3% DIFFERENCE (ABSOLUTE) BETWEEN METHODS
HHUL*
Sample and Method
Glycated Hb (%) by Affinity
Method
Hemoglobin A1 (%)
Electrophoresis
Hemoglobin F (%) by Alkaline
Denaturation
Age of patients^ assayed for
HbF
Mean
SD
5.9%
1.2
10%
4%
49 years
VAMC*
n
Mean
SD
Range
n
4.4-8.7%
19f
5.9%
0.4
5.2-6.1%
4t
1.8
8.4-14.4%
19t
9.4%
0.9
8.2-10.3%
4t
0.82
2.5-5%
11
2.0
0
2.0-2.0%
2t
11
62.5 y
59-63 y
2
—
Range
8-84 years
* One hundred fifty HHUL blood samples and 180 VAMC blood samples were screened.
t Four of the HHUL blood samples showed a discrepancy by electrophoresis on the basis of
normal reference ranges for the methods (2.6%). and I of the VAMC bloods showed a reference
range discrepancy (0.5%).
data indicate that elevated Hb F levels in patients may be
more common than generally assumed. We suggest that
laboratories that persist in using this method should periodically assess the frequency of patients with Hb F levels
greater than 2% in their populations and replace the
method if an inordinately high frequency is found.
% HHUL: median age = 57 years: Sex of 11 HHUL patients = 6 women and 5 men; VAMC:
Sex = both male,
10.
11.
12.
REFERENCES
13.
1. Allen DW, Schroeder WA, Balog J. Observations on the chromatographic heterogeneity of normal adult and fetal hemoglobin: A
study of the effects of crystallization and chromatography in the
heterogeneity and the isoleucine content. J Am Chem Soc
1958;80:1628-1632.
2. Schneck AG, Schroeder WA. The relationship between the minor
components of whole normal human adult hemoglobin as isolated
by chromatography and starch block electrophoresis. J Am Chem
Soc 1961;83:1472-1478.
3. Rahbar S, Blumenfield O, Ranney HM. Studies of an unusual hemoglobin in patients with diabetes mellitus. Biochem Biophys
ResCommun 1969;36:838-843.
4. Koenig RS, Peterson CM, Jones RL, et al. Correlation of glucose
regulation and hemoglobin Ale in diabetes mellitus. N Engl J
Med 1976;295:417-420.
5. Gabbay KH, Hasty K, Breslow JL, et al. glycated hemoglobin and
long term blood glucose control in diabetes mellitus. J Clin Endocrin Metab 1977;44:854-864.
6. Bunn HF, Gabby KH, Gallop PM. The glycosylation of hemoglobin:
Relevance to diabetes mellitus. Science 1978;200:21-27.
7. Mayer TK, Freedman ZR. Critical Review. Protein glycosylation in
diabetes mellitus: A review of laboratory measurements and their
clinical utility. Clin Chim Acta 1983; 127:147-184.
8. Mallia AK, Hermanson GT, Krohn RI, et al. Preparation of a boronic
acid support for separation and quantitation of glycated hemoglobins. Anal Lett 1981; 14:649-661.
9. Allen RC, Stastny M, Hallett D, Simmons MAA. Comparison of
isoelectric focusing and electochromatography for the separation
—
14.
15.
16.
17.
18.
19.
20.
21.
22.
Vol. 99 • No. 2
and quantitation of hemoglobin Ale. In: Radola BJ, ed. Electrophoresis, New York: Walter deGruyter, 1979, pp 663-670.
Menard L, Dempsey ME, Blankstein LA, et al. Quantitative determination of glycated hemoglobin Al by agar gel electrophoresis.
Clin Chem 1980;26:1598-1602.
Hayes EJ, Gleason RE, Soeldner JS, et al. Measurement of hemoglobin A1 by liquid chromatography and by gel agar electrophoresis compared. Clin Chem 1981;27:476-479.
Aleyassine H, Gardiner RJ, Blankstein LA, Dempsey ME. Agar gel
electrophoretic determination of glycated hemoglobin: Effect of
variant hemoglobins, hyperlipidemia, and temperature. Clin
Chem 1981;27:472-475.
;
KJenk DC, Hermanson GT, Krohn RI, et al. Determination of glycated hemoglobin by affinity chromatography: Comparison with
colorimetric and ion-exchange methods, and effects of common
interferences. Clin Chem 1982;28:2088-2094.
Abraham EC, Perry RE, Stallings M. Application of affinity chromatography for separation and quantitation of glycated hemoglobins. J Lab Clin Med 1983;102:187-196.
Betke K, Marti HR, Schlicht I. Estimation of small percentages of
fetal hemoglobin. Nature 1958; 184:1877-1878.
Nathan DM. Labile glycated hemoglobin contributes to hemoglobin
A1 as measured by liquid chromatography or electrophoresis.
Clin Chem 1981;27:1261-1263.
Krause JR, Viktor S, Cambell E. The effect of hemoglobin F upon
glycated hemoglobin determinations. Am J Clin Pathol 1982;78:
767-769.
Fairbanks VF, Klee GG. Biochemical aspects of hematology. In:
Tietz NW, ed. Fundamentals of Clinical Chemistry. Philadelphia:
WB Saunders, 1987, p 812.
Nathan DM, Avezzano E, Palmer JL. Rapid method for eliminating
labile glycated hemoglobin from the assay for hemoglobin Al.
Clin Chem 1982;28:512-15.
Bannon P. Effect of pH on the elimination of the labile fraction of
glycated hemoglobin. Clin Chem 1982; 28:2183.
Boyer SH, Belding TK, Margolet L, et al. Variations in the frequency
of fetal hemoglobin-bearing erythrocytes (F-Cells) in well adults,
pregnant women, and adult leukemics. Johns Hopkins Med J
1975;137:105-115.
Wood WG, Stamatoyannopoulos G, Lim G, Nute PE. F-cells in the
adult: Normal values and levels in individuals with heredity and
acquired elevations of Hb F. Blood 1972;46:671-682.