ORIGINAL ARTICLE
INTERNATIONAL JOURNAL OF LA BO RATO RY HEMATOLOGY
Evaluating five dedicated automatic devices for haemoglobinopathy diagnostics in multi-ethnic populations P. VAN DELFT*, E. LENTERS †, M. BAKKER-VERWEIJ*, M. P. C. GIORDANO*
*
The Haemoglobinopathies Laboratory, Department of Human and Clinical Genetics, Leiden University Medical Centre (LUMC), Leiden, The Netherlands † Department of Clinical Chemistry, Isala Clinics, Zwolle, The Netherlands Correspondence: Dr Piero C. Giordano, Center for Human and Clinical Genetics, Hemoglobinopathies Laboratory, PO Box 9600, 2300 RC Leiden, The Netherlands. Tel.: 31 71 5269800; Fax: 31 71 5268276; E-mail:
[email protected] doi:10.1111/j.1751-553X.2009.01158.x
DE
KORTE*, U. BAYLAN*, C. L. HARTEVELD*,
SUMMARY
We have tested five haemoglobin (Hb) separation apparatuses, dedicated to haemoglobinopathy diagnostics. These are the four high performance liquid chromatography devices: VARIANT IITM, HA 8160, G7, Ultra2 and the Capillary Electrophoresis apparatus from Sebia. In the first place, we focussed on the capacity of all apparatuses to detect the most common structural variants relevant for public health, these being HbS, HbC, HbE, HbD-Punjab and HbO-Arab. We then compared how the high HbA2 b-thalassaemia carriers were identified. All apparatuses were able to identify carriers of these traits with the expected sensitivity and specificity. With the primary goal of a high degree of conformity in basic diagnostics of haemoglobinopathies, we present the interpretation and the significance of the results on all apparatuses, and we comment on the unavoidable problems and solutions.
Received 20 March 2009; accepted for publication 22 March 2009 Keywords HPLC, capillary electrophoresis, thalassaemia, sickle cell disease, diagnosis
INTRODUCTION The first challenge, encountered implementing prevention strategies for the severe forms of haemoglobinopathies in the multi-ethnic Dutch population, was to improve basic carrier diagnostics all over the country. In 2001, a working party of the Dutch Association for Hematological Laboratory Research (VHL) enquired among all laboratories, registering the diagnostic potential and providing protocols and advice 484
(Giordano et al., 2006a). During this time, we were confronted with practical problems, such as professional education and organization and also with the fact that different laboratories were using different technologies and different devices. This problem was well known to us, because in our Reference Laboratory, involved in diagnostics from the early 1960s, all possible separation and measurement methods have been used. When, in the 1990s, dedicated high performance liquid chromatography (HPLC) devices became 2009 Blackwell Publishing Ltd, Int. Jnl. Lab. Hem. 2009, 31, 484–495
P. VAN DELFT ET AL.
HAEMOGLOBINOPATHY DIAGNOSTICS ON DEDICATED DEVICES 485
available, we purchased the VARIANTTM HPLC system, a sturdy machine from a market leader in the field (Bio-Rad Laboratories), that had been previously tested by Aulesa et al. (1995) and Waters et al. (1998). In 2003, Bio-Rad Laboratories proposed that we evaluate the new VARIANT IITM version on the multiethnic Dutch population. Shortly thereafter we were approached by Menarini, Tosoh and the Primus Corporation with the same proposition. We evaluated the first three devices in our laboratory and, because of the lack of space, the Ultra2 from the Primus Corporation was placed at the Isala Clinics in Zwolle, The Netherlands. Finally, because of the good results obtained during a short test period, we decided to include in our study a relatively new capillary electrophoresis (CE) apparatus, the Capillarys from Sebia. In this way, we had an ideal chance to test four popular HPLC apparatuses dedicated to haemoglobinopathy diagnostics, in Dutch laboratories and to look into the most recent developments in dedicated CE. The aim was to reach, if not uniformity, at least conformity, using reliable hardware that could produce different, but from the diagnostic point of view, comparable and valid data. We report the results of our study and we hope to be of aid to colleagues involved in basic haemoglobinopathy diagnostics and prevention.
MATERIALS AND METHODS The machines We have tested the following devices: the HPLC VARIANT IITM (Bio-Rad Laboratories, Hercules, CA, USA); the Capillary Electrophoresis device Capillarys (Sebia, Lisses, France); the HA 8160 (Menarini, Florence, Italy); the G7 (Tosoh Bioscience, Tokyo, Japan); and the Ultra2 (Primus Corporation, Kansas City, KS, USA). The samples A cohort of diagnostically significant cases (749), selected from our routine service, was analysed at the Reference Laboratory, Leiden University Medical Centre, Leiden, The Netherlands, on the first three machines. A selection of relevant samples, previously analysed at Leiden, was run double blind at the Isala Hospital at Zwolle, The Netherlands, on the Ultra2. 2009 Blackwell Publishing Ltd, Int. Jnl. Lab. Hem. 2009, 31, 484–495
The methods All apparatuses were tested using columns and buffers provided by the manufacturers and following the original instructions. All samples were examined at the haematological, biochemical and molecular levels using standard methodologies. The haemoglobin (Hb) S fractions were confirmed using the classic sickle test first (Daland & Castle, 1948) and characterized by direct DNA sequencing of the b-globin genes when needed. All other variants were directly confirmed at the molecular level using previously described methods (Giordano et al., 2006b). All cases with elevated or borderline HbA2 levels showing haematological abnormalities were examined at the molecular level using the same methods (Giordano et al., 2006b). The normal and abnormal values Normal adult samples were all those presenting after the age of 2 with ±97% HbA, >1% HbF and ±2.5% HbA2. Normal neonatal samples were those presenting with ±80% HbF, ±20% HbA and a trace of HbA2 below 0.5%. We evaluated the normal HbA2 range, obtained measuring the HbA2 fraction in a large cohort (i.e. 100 individuals) of confirmed non irondepleted and non a- or b-thalassaemia (thal) carriers. The normal value obtained in this way in our laboratory was 2.9%. This figure may go with a 2 · SD from 2.32 to 3.48%. In an extended study involving 674 cases (Mosca et al., 2009), we observed that in iron deficiency, d- and a-thal, in normal HbA2 b-thal and in cases of overestimations in the presence of HbS, the HbA2 values could respectively be low, normal, barely higher or much higher than the normal ranges (Figure 1). For this reason, and because normal ranges are ‘apparatus dependent’, we considered only values higher than 4.0% to be reliable for the diagnosis of b-thal trait, while we considered normal values in the range between 2.5% and 3.5% but only when measured in combination with a normal haematological picture. We proceeded with additional (molecular) analyses when the HbA2 values were in the grey zone (3.5–4.0%) or when the haematological data were persistently abnormal or unclear. These procedures are necessary especially in multiethnic societies where many different (combinations
486
P. VAN DELFT ET AL.
HAEMOGLOBINOPATHY DIAGNOSTICS ON DEDICATED DEVICES
Specificity and overlapping of the HbA2 values measured in different cohorts of patients an artifact in HbS carriers
HbA2 elevated as
HbA2 reduced in δ-thalassemia and variants carriers In δ/β-thalassemia In β/α-thalassemia combinations In classic high HbA2 β-thalassemia carriers In normal HbA2 β-thalassemia carriers α Thalassemia
In (α-/--)
α Thalassemia
(αα/--) (α-/-α)
α Thalassemia
(αα/-α)
α Thalassemia In iron depletion
The normal
0
1.0
2.0
range is method dependent
3.0
4.0
5.0
6.0
7.0
8.0
9.0 %
Figure 1. Data from 100 non a-thalassaemic, non iron-depleted normal individuals, and from large cohorts of patients (N = 674) tested on the VARIANT IITM high performance liquid chromatography apparatus at Leiden University Medical Centre. In exceptional cases, b-thal carriers may present with 8–9% HbA2. Higher values (12–16%) usually indicate Hb Lepore, a d/b hybrid chain with a b-thal minor phenotype.
of) mutations occur. In these cases, a correct evaluation of the genotype/phenotype correlation associated with prognosis and risk prediction for the different forms of abnormal Hbs and a- or b-thal is needed. For this reason, most of the samples analysed in this study were also confirmed at the molecular level, and routinely controlled for a-thal deletions or point mutations using gap-polymerase chain reaction (gap-PCR) and direct sequencing (data not shown).
ANALYTICAL RESULTS The common abnormal haemoglobins
23). Few carriers were in the range of ‘slightly elevated HbA2’. The Ultra2 was tested on a smaller number of samples. Data showing the comparisons are summarized in Tables 2 and 3. Normal and abnormal HbA2 values Tables 2 and 3 show that elevated HbA2 levels are well measured but also that it is difficult to have perfectly matching figures on different machines. It is therefore unsafe to exclude a b-thal carrier based on a presumed cut-off value, without considering the complete haematological picture.
The large majority of the 749 cases presented no major abnormal fractions while 77 did. All machines tested were able to identify the common carriers of HbS, HbC, HbE, HbD-Punjab, HbO-Arab and HbH disease. The data are summarized in Table 1 and separation examples of the common HbS, HbC, HbE, HbD-Punjab variants on the five different devices are shown in Figure 2a–f, respectively.
Because of the different integration factors and to overlapping with the HbA1c peak, the HbF fraction is often imprecisely measured, especially in the low ranges (1–3%). However, the significance of a precise HbF measurement in haemoglobinopathy diagnostics is limited. The data are summarized in Table 4.
Estimation of the HbA2 fraction
Sensitivity and specificity
We compared the estimated HbA2 value on the five machines, running cohorts of b-thal carriers (N = 57–
The degree of sensitivity for the structural mutations, for the carriers of high HbA2 b-thal and HbH
Estimation of the HbF fraction
2009 Blackwell Publishing Ltd, Int. Jnl. Lab. Hem. 2009, 31, 484–495
P. VAN DELFT ET AL.
HAEMOGLOBINOPATHY DIAGNOSTICS ON DEDICATED DEVICES 487
Table 1. Degree of sensitivity for common Hb S, C, E, D-Punjab, HbO-Arab and HbH genotypes on the different devices VARIANT II
Capillarys
HA 8160
G7
UItra2
Name Putative genotype N/score peak
Name N/score peak
Name N/score peak
Name N/score peak
Name N/score peak
HbA/S HbS/S HbA/C HbC/C HbA/E
30/30 6/6 6/6 2/2 6/6
S window S window C window C window A2 window
30/30 6/6 6/6 2/2 6/6
S-zone S-zone C-zone C-zone E-zone
30/30 6/6 4/4 1/1 6/6
30/30 6/6 5/5 2/2 6/6
S+ S+ C+ C+ P08
3/3 2/2 2/2 NT 2/2
S* S* C* ?*
HbE/E
3/3
A2 window
3/3
E-zone
3/3
2/2
P08
NT
-
HbA/DPunjab HbA/OArab HbH disease
19/19
D window
19/19
D-zone
16/16
18/18
D+
7/7
?*
1/1
Unknown
l/l
Unknown l/l
S/C window S/C window S/C window S/C window Abnormal separation Abnormal separation Divided in 2A2 and P8 S/C window
1/1
P06
1/1
C*
4/4
Peak seen, 4/4 not named
H-zone
P1 peak
4/4
Peak seen, 2/2 not named
?*
4/4
NT, not tested. *Eventually commented as ‘consisted with’ or unspecified.
disease was 100% on all devices. Only a few cases of borderline HbA2 b-thal could have been missed without haematological indication and molecular analysis. All common abnormal separations observed during this survey as well as all high HbA2 b-thal traits matched the characterization at the DNA level. The molecular confirmation might suggest that specific identification of the common variants, separated on HPLC or CE, can be considered absolute. However, the full match is rather due to the high prevalence of the mutants in a particular population than to the precision of the retention time (RT).
EVALUATION OF THE DIFFERENT DEVICES General considerations Dedicated HPLC and CE machines designed to diagnose haemoglobinopathies have many advantages and few disadvantages. The great advantage is that all these devices are fast, automatic and offer an interpretation of the diagram and a quantitative measurement of the fractions. However, these advantages may lead to a diagnostic misinterpretation when the results are taken too easily for granted, without cautious examination. 2009 Blackwell Publishing Ltd, Int. Jnl. Lab. Hem. 2009, 31, 484–495
THE VARIANT IITM APPARATUS (BIO-RAD LABORATORIES) The machine The VARIANT IITM is a computerized apparatus provided with data registration software and based on the same technology of the previous VARIANTTM model, consisting of a gradient elution system driven by two peristaltic pumps. The common and rare variants The VARIANT IITM separates the normal or the common abnormal Hb fractions at a specific RT and at a high degree of specificity. The apparatus comes with an online database (Clinical System Network) that enables the registration of abnormal separations by many users worldwide. The database allows ‘putative identification’ for the common and a ‘possible indication’ for the rare variants. HbA2 and HbF The HbA2 levels are quite accurate and correlate well with those measured with other apparatuses. High
488
P. VAN DELFT ET AL.
HAEMOGLOBINOPATHY DIAGNOSTICS ON DEDICATED DEVICES
(a)
(b)
(c)
(d)
(e)
(f)
Figure 2. From left to right: Capillarys, VARIANT II, HA 8160, G7 and Ultra2. (a) Normal; (b) HbA/S; (c) HbA/C; (d) HbA/E; (e) HbA/D; (f) high HbA2 herozygous b-thalassaemia. HbO-Arab is not shown in this figure.
HbA2 cases are well identified; borderline cases will need confirmation at the molecular level. If when changing the kit the fraction does not elute within
the expected RT, elution should be adjusted by regulating the temperature of the column as indicated by the manufacturer. 2009 Blackwell Publishing Ltd, Int. Jnl. Lab. Hem. 2009, 31, 484–495
P. VAN DELFT ET AL.
HAEMOGLOBINOPATHY DIAGNOSTICS ON DEDICATED DEVICES 489
Table 2. Showing the number of cases diagnosed as b-thalassaemia trait
Table 4. Results from samples with HbF range between
