An evaluation of the Diamat HPLC analyser for simultaneous determination of haemoglobins A2 and F

The authors describe a modification of the instrumental parameters of the Diamat fully automated HPLC system for Hb A2 assay (Bio-Rad Laboratories, Milan, Italy) in order to obtain simultaneous determination of Hb A2 and Hb F. Hb A2 and Hb F measurements are reproducible (within-run CV 2.6%, with Hb A22.7%; 5.1%, with Hb F 1.3%) and accurate (from a comparison with two microchromatographic techniques for Hb A2: r = 0.9639 and 0.9755; with two alkali denaturation procedures for Hb F: r = 0.9990 and 0.9952; with radial immunodiffusion, r = 0.9877). Assay linearity has been confirmed for Hb A2 concentrations between 0 and 6.0%, and for Hb F between 0 and 60%. The data obtained from the analysis of some pathological samples for Hb Bart's, Hb H, Hb J Sardegna, Hb Lepore and Hb S are in agreement with cellulose acetate electrophoresis analysis. The Hb A2 reference intervals for normals (N = 597) and Beta-thalassemia carriers (N = 200) are respectively (95% limits) 2.02-3.27 and 3.92-5.90 in % units. Hb F values measured in normals (N = 968), in β-thal carriers (N = 302) and in δβ-thal carriers (N =3) have been found to be consistent with the usual diagnostic parameters. Some minor limitations emerged: the most relevant concerns Hb A1c, which is overestimated with respect to a reference method (y = 1.217x + 0.16; N = 79; r = 0.9235). A probable interference from labile fractions is responsible for this Hb A1c inaccuracy.


Introduction
The determination of the minor haemoglobins, Hb A2 and Hb F, in whole blood are of considerable diagnostic use for the characterization of several thalassaemic syndromes and for the correct diagnosis of several haemoglobinopathies [1]. It is also known that elevated Correspondence to Dr Andrea Mosca, Dip [2]. Therefore, a variety of laboratory techniques have been developed to accurately measure these minor haemoglobins. The most commonly used are ionexchange chromatography [3], on microcolumns or with HPLC [4][5][6][7], for Hb A2, and radial immunodiffusion [8] and alkali denaturation [9][10][11] for Hb F. This report presents the authors' experience with a new HPLC method performing a simultaneous determination of Hb A and Hb F. This is obtained by using a fully automated HPLC analyser (Diamat, Bio-Rad Laboratories, Segrate, Milan, Italy), originally dedicated to the analysis of glycated haemoglobins [12][13][14], and later adapted to Hb A determination [15].
By introducing a slight modification to the buffer elution times recommended by the manufacturer for Hb A determination, the method was adapted for measuring Hb A2 and Hb F simultaneously.
The data reported here confirm the utility of the technique to haematological practice. Some minor limitations are also discussed.

Samples
Whole blood samples (anticoagulated with potassium EDTA, g/l) were selected from routine laboratory samples for most of the analytical tests. Informed consent was obtained from those subjects studied for the evaluation of reference values.
Stabilized Hb A and Hb F liquid control materials were obtained by adding ethylene glycol (0"35 volume fraction) to haemolysates prepared by a standard tetrachloromthane lysis procedure 16]. These solutions total Hb 100 + 20 g/l) were stored at -20 C and used for longterm precision estimates. This was carried out by assaying two haemolysates, with low and high Hb A2 and Hb F concentrations, over periods of about one month (the Hb A test) and three months (the Hb F test).

Procedures
Diamat HPLC analysis This method basically consists of a cation-exchange liquid chromatography developed with three trisphosphate buffers of increasing ionic strength at a 0142-0453/89 $3.00 Taylor & Francis Ltd. controlled temperature of 23 C. The eluate is read at 415 and 690 nm. For the analysis five tl of whole blood are diluted, by means of the dilutor supplied with the apparatus, with ml ofhaemolizing reagent. The haemolysate is centrifuged at 10 000 g for min and is then loaded into the sample compartment, which is kept at between 2 and 8C. Twentytl are injected into the column for each run.
Figure shows the modifications made to the instrumental parameters recommended by the manufacturer in order to obtain a resolution of Hb F from Hb Alc. The best Hb F isolation is achieved by prolonging the buffer elution to 2 min (figure l[c]); there is no effect if the elution of buffer 2 is prolonged from 6"4 to 9"4 min. Therefore, after these tests, it was decided to adopt the following instrumental parameters: injection interval 170; injection volume 2; stepwise time 20; stepwise time 2 77; stepwise time 3 140; stop time 9900; off time 800; response time 1"00 (for all the minor fractions).
The flow is critical for optimal resolution of the Hb F peak: a flow of ml/min is obtained by setting the internal knob to a position between 25 and 30. The operating pressure, under such conditions, is between 55 and 60 kg/cm2. By using these parameters the haemaglobins are eluted in the following order (retention times in parentheses, and given in min): Hb Ala and Alb (1"7), Alc (3"1), F (3"9), A0 (7"5), A2 (9"9); each run takes approximately 16"5 min. Columns and reagents used within this evaluation were from Bio-Rad HF and KK batches. All the chromatograms in figures 1-3 have been redrawn from the original paper print-out.
Reference methods Alternative chromatographic procedures for Hb A2 have been performed either according to ICSH recommendations [3,17] or by using disposable microcolumns (Beta-Thal Quick Column, Helena Laboratories, Milan, Italy).
(c) Radial immunodiffusion for samples with Hb F lower than 12% (d) Diamat analysis for Hb Alc (see later).
The within-run imprecisions of these methods (as CV) for Hb F concentrations between 0"8 and 2"2% are: 7"3% (method a), 8"0% (method b), 4"3% (method c) and 3"3% (method d). As Hb Alc reference method another Diamat instrument was used, which was dedicated to Hb Ale measurement for the diabetic centre (later called the x-method). This instrument is equipped with a different column and buffer elution system; and its analytical characteristics have previously been evaluated [12][13][14]. Each blood sample was therefore treated independently for each analysis and loaded on the two instruments almost simultaneously (or at least no longer than 12h between the two determinations  reported when using a similar cation-exchange HPLC procedure [20]. Hb Lepore ( figure 3[c]) is slightly underestimated with respect to densitometric analysis (10"3% versus 15"0%, respectively).

Discussion
Accuracy was tested by making a comparison of the Diamat proposed procedure against several reference methods, as reported in table 2. The correlations between these methods were excellent.
Finally, the procedure has no carry-over and is insensitive to changes in sample total haemoglobin, in the range 4"4-20"4 g/dl (data not reported in detail).
With regards to column life, we found that Hb F can be properly resolved for up to 750-800 runs. Up to 1100 runs of Hb A can be made.

Reference values
Hb A2 reference values were evaluated for normal adults and -thal carriers. From the statistical analysis of elementary data the following values were obtained: The authors first experiences with the Diamat system applied to the determination of haemoglobin species in blood are reported. It is clear that this apparatus can improve routine haematological management in the clinical laboratory. The system offers the following advantages: (1) Hb A is measured accurately and reproducibly. This is relevant because Hb A accounts for only a small amount of total haemoglobin (Hb A concentrations higher than 6% are rarely found) and because the separation limit between normals and [3-thalassaemics is very narrow. A similar reproducibility is rarely obtained with manual chromatographic methods.
The reference intervals found for normals and [3thalassaemia carriers are in agreement with those already reported [3].  (2) () Simultaneous Hb F determination eliminates the need for separate analysis; this kind of measurement is, for most Hb F assay protocols, timeconsuming, poorly reproducible and seldom linear to high concentrations. Chromatographic analysis is able to discriminate between several haemoglobin variants; their quantification is, in the majority of cases, in agreement with the electrophoretic pattern.
Sampler capacity (48 tubes) and the system's automation allow up to 90 samples per day to be analysed; no special skills are required to operate the instrument.
A cation exchange HPLC analysis recently proposed by Bisse and Wieland [20] seems to offer similar advantages.
The system optimized by these authors is probably superior, in terms of resolution, but compared to the Diamat procedure, has the following drawbacks: The analysis is more time-consuming (each run takes more than 60 rain, while the Diamat run is only one-third as long).
No information is available on accuracy in Hb A2 measurement.
Hb A lc accuracy has been evaluated on a small number of subjects (N 25), with only three of these subjects out of normal range.
No other fast high resolution HPLC procedure for the separation of Hb F, Hb A0, Hb A2 and other Hb variants seems to have been presented.
The Diamat system has some minor problems. For example Hb Alc quantification seems to be overestimated with respect to a reference method of proved accuracy [12]. A possible explanation for this could be the influence of labile aldimine forms which, if not removed during the haemolysis step, can co-migrate with the Hb Alc component (stable ketoamine form). The importance of removing these labile forms has been stressed by several investigators [23][24][25]. It has also been demonstrated that the Diamat Alc analysis (the x-method in figure 4) is specific for the stable forms [25]. In fact, a significative difference has been found by analysing the haemolysing agent used for Hb Ale determination (xmethod) and the one used with the test method (ymethod in figure 4); the two agents differ in pH (5"95 versus 7"20, respectively) and in conductivity (246 versus 132tS, respectively). In order to confirm that the haemolysing agent has a powerful effect on Hb Ale concentration, Hb Alc was measured using the proposed method, in a sample treated with the two haemolysing agents separately. The following values were found: 5"3% (x-method haemolysing solution) and 7"3% (y-method haemolysing solution). In conclusion, the Hb Alc measurement on Diamat equipped with reagents and column for Hb A and Hb F is probably inaccurate, because the labile aldimine form is not removed during the haemolysis step.
A second limitation of the technique concerns peak identification. This is easily done for normal samples, but, when dealing with abnormal samples (for example those reported in figures 2 and 3) peak identification should be performed by an experienced haematologist. It is also evident that, if a pathological result is obtained, further analyses, such as DNA analysis and globin chain synthesis, have to be made in order to obtain a final diagnosis.
Apart from these limitations, this kind of HPLC analysis is useful for diagnosis of haemoglobin-related disorders.
Improvements, such as a further reduction of analysis times for a correct estimation of lib Alc, would be useful.

NOTES FOR AUTHORS
Journal of Automatic Chemistry covers all aspects of automation and mechanization in analytical, clinical and industrial environments. The Journal publishes original research papers; short communications on innovations, techniques and instrumentation, or current research in progress; reports on recent commercial developments; and meeting reports, book reviews and information on forthcoming events. All research papers are refereed.

Manuscripts
Two copies of articles should be submitted. All articles should be typed in double spacing with ample margins, on one side of the paper only. The following items should be sent: (1)