Professional Documents
Culture Documents
A R T I C L E I N F O A B S T R A C T
Keywords: Background: Unrecognized hemoglobinopathies can lead to measured hemoglobin A1c (Hb A1c) concentrations
Hemoglobin variants that are erroneous or misleading. We determined the effects of rare hemoglobin variants on capillary electro-
Hemoglobin A1c phoresis (CE) and HPLC methods for measurement of Hb A1c.
Diabetes Methods: We prospectively investigated samples in which Hb A1c was measured by CE during a 14-month period.
Ion-exchange HPLC
For samples in which the electropherograms suggested the presence of rare hemoglobinopathies, hemoglobin
Capillary electrophoresis
variants were identified by molecular analysis or by comparison with electropherograms of known variants.
Boronate affinity chromatography
When sample volume permitted, Hb A1c was measured by 2 HPLC measurement procedures and by boronate
affinity HPLC.
Results: Hb A1c was measured by CE in 33,859 samples from 26,850 patients. 15 patients (0.06%) were iden-
tified as having rare hemoglobinopathies: Hbs A2 prime, Agenogi, Fannin-Lubbock I, G Philadelphia, G San Jose,
J Baltimore, La Desirade, N Baltimore, Nouakchott, and Roanne. Among 6 of these samples tested by 2 ion-
exchange HPLC methods, the rare Hb was detected by both HPLC methods in only one sample, and none were
detected by boronate affinity HPLC. The mean of the Hb A1c results of 2 HPLC methods differed from the result of
the CE method by 0.7–2.2% Hb A1c in samples with variant hemoglobins versus < 0.2% Hb A1c in samples
without variants.
Conclusion: Measurement procedures differ in the ability to detect the presence of rare Hb variants and to
quantify Hb A1c in patients who harbor such variants.
https://doi.org/10.1016/j.cca.2017.11.012
Received 29 September 2017; Received in revised form 10 November 2017; Accepted 13 November 2017
Available online 14 November 2017
0009-8981/ © 2017 Elsevier B.V. All rights reserved.
S.W. Strickland et al. Clinica Chimica Acta 476 (2018) 67–74
hemoglobin variants or the effects of such variants on the measured A, and Hb F) based primarily upon charge alone, in comparison to CE
concentration of Hb A1c [20,21]. that separates based upon size and charge. Detection is performed with
an absorbance reading at 415 nm and the software calculates Hb A1c as
Hb A1c/“other Hb species”, which can include Hb A1a, Hb A1b, Hb A0,
2. Material and methods and/or Hb F and varies based upon vendor and methodology used.
68
S.W. Strickland et al. Clinica Chimica Acta 476 (2018) 67–74
Table 1
Detection of hemoglobin variants by CE- and HPLC-based measurement procedures.
Hb variants found by Capillarys2 Mutation Number of patients Presence of variant detected by Presence of variant detected by Presence of variant detected by
identified† Tosoh G7?c Tosoh G8? Variant II Turbo?
Hb Fannin-Lubbock I β-G119D 1 No No No
Hb La Desirade β-A129V 4 No Yes/No No
Hb Roanne α-D94E 1 No No No
Hb G-San Jose β-E7G 1 No – –
Hb A2 prime δ-G16R 1 No – Yesb
Hb G-Philadelphia/Hb AS α-N268K 1 Yes Yesa Yesa
Hb J-Baltimore β-G16D 3 No Noa Yesa
Hb N-Baltimore/Hb AS β-K95E 1 Yesa Noa Yesa
Hb Agenogi β-E91K 1 – – –
Hb Nouakchott α-P114L 1 – – –
a
Indicates data from previous report [16].
b
Indicates data from previous report [19].
c
Includes data from a previous report [14].
69
S.W. Strickland et al. Clinica Chimica Acta 476 (2018) 67–74
Fig. 3. Hb A1c analysis on samples from 2 patients with confirmed Hb La Desirade. Panels A and C: Sebia Capillarys electropherograms; Panels B and D: Tosoh G8 chromatograms. Panels
A and B: Patient 1; Panels C and D, Patient 2. Arrows indicate peaks not seen in samples from patient lacking hemoglobin variants.
method for Hb A1c quantification. As previously described [14], the The CE electropherograms for samples from 2 patients with Hb La
HPLC method reported Hb A1c results for 10 of the 11 patients (91%) Desirade (Fig. 3A and C) showed peak doublets at Hb A0, Hb A1c and
with no indication that a hemoglobin variant was present. The sample “Other Hb A” (arrows). The patterns were identified by the analyzer's
that the Tosoh G7 HPLC method identified as containing an Hb variant software as atypical and no results were reported for Hb A1c. On HPLC
was recognized as Hb G-Philadelphia/Hb AS (Table 1). analyses (Tosoh G8), extra peaks near the Hb A0 and Hb A1c peaks were
Samples from 13 of the 17 patients with suspected hemoglobino- seen in the sample from patient #1 (Fig. 3B; arrows), but the sample
pathies were available for molecular testing. Sequencing identified from patient #2 had no HPLC doublet for the Hb A0 peak and only a
hemoglobinopathies in all 13 (Table 1). Two of the 17 patients with small doublet for the Hb A1c peak (Fig. 3D, arrow). The Tosoh G8 (and
suspected hemoglobinopathies were lost to follow-up, but for another 2 the Variant II Turbo) gave no indication that this was an atypical
of the 17 samples, the original capillary electropherograms matched the chromatogram, and the %Hb A1c was calculated.
capillary electropherogram of one of the confirmed hemoglobino- The CE profile for Hb Roanne showed a variant peak eluting in the
pathies for a total of 15 patients with confirmed or presumptively- same location as Hb F (Fig. 4A; arrow). No extra peak was seen in the
identified hemoglobinopathies. Among the 15 patients, 10 distinct Tosoh G8 chromatogram (Fig. 4B) nor the Variant II Turbo chromato-
molecular variants were identified (Table 1). gram, and this sample would have been reported as having a normal
profile.
3.1. Resolution of hemoglobins by CE and HPLC measurement procedures Hb Fannin-Lubbock I showed an atypical profile on CE analysis with
a variant peak eluting between the Hb A1c and Hb A0 peaks (Fig. 4C;
The hemoglobinopathies that were detected by CE were incon- arrow); this peak hampered the software's ability to calculate Hb A1c.
sistently recognized by other methods. CE electropherograms and the Neither the Tosoh G8 HPLC nor the Variant II Turbo chromatograms
Tosoh G8 HPLC elution patterns are shown in Figs. 2–4: Fig. 2 shows revealed the presence of a hemoglobin variant (Fig. 4D).
results for a sample without a Hb variant, and Figs. 3 and 4 show results Blood containing Hb N-Baltimore/Hb AS was not identified as
for samples with Hb variants. Biorad Variant II Turbo HPLC chroma- having a variant by the Tosoh G8, but was identified as non-reportable
tograms are shown in Fig. 5. For samples without Hb variants, CE by the Variant II Turbo (Table 1).
achieved better baseline resolution of the hemoglobin peaks (Fig. 2A) In summary, among samples from the studied patient population in
than was seen with the HPLC methods (e.g., Fig. 2B). Samples that which a variant had been detected by the CE method, the 3 con-
contained Hbs La Desirade and Roanne were also tested by a gel-elec- temporary HPLC measurement procedures for Hb A1c detected the
trophoresis method for detection of hemoglobinopathies; it failed to presence of hemoglobin variants in only 2 of 17 chromatograms (1 of
identify their presence (not shown). 11 for Tosoh G7, 1 of 3 for Tosoh G8 and 0 of 3 for Variant II Turbo).
70
S.W. Strickland et al. Clinica Chimica Acta 476 (2018) 67–74
Fig. 4. Hb A1c analysis on samples from patients with confirmed Roanne and Hb Fannin-Lubbock I. A: Hb Roanne, Sebia Capillarys electropherogram; B: Hb Roanne sample, Tosoh G8
chromatogram; C: Hb Fannin-Lubbock I, Sebia Capillarys electropherogram; D: Hb Fannin-Lubbock I, Tosoh G8 chromatogram. Arrows in panels A and C indicate peaks not seen in
samples from patient lacking hemoglobin variants.
The boronate method does not recognize the presence of hemoglobi- and provided Hb A1c results for the samples, including the samples from
nopathies since separation of glycated Hb is based on the structure of 2 patients with Hb La Desirade (shown in Table 1).
the glucose bound to hemoglobin and not based on the properties of the For the sample that contained Hb Roanne, the CE result was higher
hemoglobin itself. by 0.8% Hb A1c than the results of the ultra2 (Table 2). The Hb A1c
Other published reports are available on HPLC results for 3 of the result for the sample containing Hb Fannin-Lubbock I was flagged by
hemoglobin variants [20,22]: Hbs J-Baltimore, G-Philadelphia/Hb AS, the CE analyzer, but the quantitative result that it calculated was 1.4%
and N-Baltimore/Hb AS. The published results are shown in par- Hb A1c higher than results from the ultra2 (See Discussion). Results also
entheses in Table 1 with citation of the published reports. All 3 have differed among the G8, ultra2, and Variant II Turbo when a hemoglobin
been characterized by the Variant II Turbo which detected the presence variant was present (Table 2).
of all 3, and 2 were studied by the Tosoh G8 which detected 1 of the 2.
Capillary electropherograms for additional hemoglobin variants 4. Discussion
found in this study are shown in Fig. 6 and were also analyzed in a
second laboratory giving the same electropherogram results. According to the WHO, approximately 7% of the worldwide popu-
lation is affected by a hemoglobin variant [23]. Many of these variants
3.2. Quantitative results reported for Hb A1c% by CE and compared are clinically silent and are detected only fortuitously. Identification of
methods variants is critical for interpretation of Hb A1c results for several rea-
sons: First, Hb variants may produce altered red cell survival [2], thus
The Hb A1c results for samples that contained Hb variants varied making even analytically accurate Hb A1c results clinically misleading.
among methods. Quantitative Hb A1c results for samples containing Hb Second, unresolved peaks can contribute to falsely increased or de-
Roanne and Hb Fannin-Lubbock I, are presented in Table 2 along with creased Hb A1c results in both HPLC and CE methodologies: Peaks that
results for a sample without a variant hemoglobin. Results for the overlap with the Hb A0 peak or with the Hb A1c peak prevent proper
sample without an identified Hb variant differed by a maximum of integration of the peaks of interest and compromise calculations of the
0.3% among the four methods. For these hemoglobinopathy samples peak area, thereby affecting the reported Hb A1c%. Third, hemoglobin
(and for a third Hb, Hb La Desirade), all CE electropherograms were variants may affect hemoglobin glycation [24], further increasing the
identified by the CE procedure's software as “Atypical Profile” and, in need for identification of the presence of these variants.
some cases, the software provided no Hb A1c result. By contrast, the This study shows that the presence of Hb variants is not reliably
HPLC instruments flagged none of the HPLC chromatograms as atypical detected by contemporary ion-exchange HPLC-based (or boronate-
71
S.W. Strickland et al. Clinica Chimica Acta 476 (2018) 67–74
Fig. 5. Biorad Variant II HPLC Chromatogram with Hb La Desirade and Hb Fannin Lubbock. A: Expected pattern of Biorad Variant II chromatogram. B: Hb Fannin-Lubbokc sample Biorad
Variant II chromatogram; C: Hb La Desirade sample Biorad Variant II chromatogram.
based) methods for measuring Hb A1c. The analytical resolution of CE data for these 2 hemoglobins is that the calculation of Hb A1c% by the
allowed identification of 10 patients who had hemoglobin variants that CE (Sebia) software included glycated Hb variant (not visible on the
had gone undetected by an HPLC method, and Hb variants were not electropherogram; presumably integrated with the Hb A1c peak) with
reliably detected by 2 additional HPLC-based measurement procedures the Hb A1c while the software fully resolved the nonglycated Hb variant
used during this study. The HPLC methods were unable to efficiently as part of the Hb A0, thus the result was an overestimate of Hb A1c% by
resolve the Hb variants; and, despite potential interference from the CE method.
shoulder peaks in some cases (e.g., Fig. 3), the HPLC measurement This study has several strengths, notably that each of the 10 rare
procedures nonetheless reported results for Hb A1c. The HPLC chro- hemoglobin variants was definitively identified by sequencing. By
matograms of other Hb variants (e.g., Hb Roanne and Hb Fannin-Lub- analyzing samples both by CE and by representative comparison
bock I) showed no sign of the variant, most likely due to lack of re- methods, the study revealed that CE recognized the presence of he-
solution of these Hbs from Hb A0. moglobinopathies that the other studied methods did not recognize as
The quantitative results of the various methods deserve further at- present. A limitation of the study is that it did not prospectively com-
tention. The Hb A1c result in the CE method derives from the following pare the frequencies of identification of hemoglobin variants by CE and
ratio: Hb A1c peak/(Hb A1c peak + Hb A0 peak). In the Fannin-Lubbock by other methods in a series of patients that were all tested by both
I patient sample, this calculation resulted in a reported Hb A1c% of methods to ascertain whether CE is more sensitive than other methods
9.3%, whereas results of the other methods ranged from 7.1–7.9%Hb in detection of hemoglobin variants. The observed frequency of detec-
A1c (Table 2). However, if the CE calculation had included the aberrant tion of rare hemoglobinopathies, however, appears to be near the ex-
peak for Hb Fannin-Lubbock I (which was resolved from the Hb A0 pected prevalence of rare hemoglobinopathies in this population. As
peak as indicated by the arrow in Fig. 4C) as part of the Hb A0 peak, the reported by Huisman in a study of hemoglobinopathies in the
Hb A1c% result would have been 7.2%, a result consistent with the Southeastern United States, the prevalence of rare hemoglobin variants
other methods. This could suggest that the Hb Fannin-Lubbock I peak with known phenotypes was 0.012% [26], similar to the prevalence of
on the HPLC chromatogram is hidden within the large A0 peak 0.06% in our study population [14] in the same part of the U.S. Al-
(Fig. 4D) and is contributing to a decreased Hb A1c% by increasing the though this difference can be expected to occur by chance, a possible
area of the “A0” peak. Moreover, it suggests that neither method se- explanation for the higher prevalence in the present study than in the
parated Hb A1c from a presumed Hb Fannin-Lubbock I Hb that is gly- previous study [14] is that the high resolution of CE allowed identifi-
cated on its N-terminal valine. A similar situation was observed for Hb cation of hemoglobin variants that were not recognized as being present
Roanne, which showed a slightly lower Hb A1c% for the HPLC-based by the screening gel-electrophoresis method used in the earlier (1980)
method than for the CE method. Another possible explanation of the study. This is supported by our observation that two Hb variants that
72
S.W. Strickland et al. Clinica Chimica Acta 476 (2018) 67–74
Fig. 6. Rare Hemoglobin variants identified by CE A: Capillary electropherogram of sample with Hb Nouakchott. Arrows represent extra peaks present in the patient sample. B: Capillary
electropherogram of sample with Hb A2 prime. Arrows represent aberrant peaks present in the patient sample. Representative of either a homozygous A2 prime mutation or a patient that
is heterozygous for A2 prime mutation and delta thalassemia. C: Capillary electropherogram of sample with Hb G Philadelphia/AS. Arrows represent extra peaks present in the patient
sample [18]. D: Capillary electropherogram of sample with Hb N-Baltimore/Hb AS. Arrows represent aberrant peaks present in the patient sample. Peak 1: Hb-N Baltimore, Peak 2: other
HbA, Peak 3: unknown, Peak 4: Hb AS, Peak 5: putative A2.
were resolved by the CE measurement procedure were not detected by a Rare Hb variants occur with considerable frequency and their de-
gel-electrophoresis-based method for hemoglobinopathy detection. tection by contemporary HPLC measurement procedures for Hb A1c is
Moreover, the most common of the rare hemoglobinopathies that were unreliable. These variants can interfere analytically in measurements of
identified in the earlier study were Hb A2 prime, G-Philadelphia and N- Hb A1c and may produce misleading results if associated with decreased
Baltimore, all of which were also identified in our study in the South- red cell survival or decreased rates of glycation. The resolving power of
eastern United States. These observations suggest that measurement of the studied CE measurement procedure for quantification of Hb A1c
Hb A1c by the CE method used here identifies a high proportion of the appears sufficient to recognize the presence of a range of rare he-
patients with the rare hemoglobinopathies that are encountered in the moglobinopathies in addition to the more common Hbs C, D, E, F, and
geographic area studied. Thus, other methods, if tested in parallel with S.
the CE method, would be unlikely to identify appreciably more patients
who have variant Hbs than were found by the CE method. References
Rare hemoglobinopathies are unlikely to be diagnosed clinically. In
the present study, the presence of a hemoglobin variant was unknown [1] H.F. Bunn, R. Shapiro, M. McManus, L. Garrick, M.J. McDonald, P.M. Gallop,
K.H. Gabbay, Structural heterogeneity of human hemoglobin A due to none-
in all patients prior to testing, despite the fact that most had a long nzymatic glycosylation, J. Biol. Chem. 254 (10) (1979) 3892–3898.
association with medical facilities. Three of the hemoglobin variants [2] D.B. Sacks, W.G. John, Interpretation of hemoglobin A1c values, JAMA 311 (22)
identified by CE in the present study — Hbs Roanne, La Desirade, and (2014) 2271–2272.
[3] S. Rahbar, O. Blumenfeld, H.M. Ranney, Studies of an unusual hemoglobin in pa-
Nouakchott — are typically silent clinically, and are associated only
tients with diabetes mellitus, Biochem. Biophys. Res. Commun. 36 (5) (1969)
with mild anemia [27–29]. These 3 variants accounted for 5 of our
73
S.W. Strickland et al. Clinica Chimica Acta 476 (2018) 67–74
74