An evaluation of the Technicon DPA*1 specific protein analyser

The DPA*I analyser, a fully automated, discrete, random-access specijqc protein analyser has been evaluated. Eight, out ofa current menu of 17 analytes, were assessed for precision, accuracy and calibration stability. Within-batch precision was excellent, generally being between 1% and3%; between-batch precision was also good (3-6%). Method comparisons showed good agreement; the best correlations being obtained when the comparison methodology was also nephelometric.

The analytical performance of the DPA*I was assessed for eight different analytes out of a current repertoire of 17. The aspects evaluated included precision, linearity, calibration stability, relative accuracy, carry-over, throughput and the validity of the antigen excess checking procedures.

The instrument
The Technicon DPA*I (supplied by Technicon UK, Basingstoke, Hampshire, UK) is a bench-top unit, comprising an analytical unit with built-in keyboard, a separate VDU and keyboard, and a printer. A single 13 amp power supply is required. Up to 12 analytes may be measured at any one time, the antiserum reagent being supplied, ready-to-use in bar-coded vials, each sufficient for 100 analyses. A single probe, with level detector, performs all liquid handling and mixing functions. A sample is initially diluted in 3"4% polyethylene glycol (PEG) causing the precipitation of any serum components that might interfere in the subsequent measure-Current address: Department of Clinical Chemistry, Barking Hospital, Upney Lane, Barking, Essex IGll 9LX, UK.
ment. An aliquot of this diluted sample is mixed with antiserum and further PEG; the reaction is then monitored every 10 s for a period of 7 min. Measurement is by nephelometry using a 600 nm pulsed LED light source and a detection angle of 30. Results are produced at the rate of three tests/min, irrespective of the number of samples or tests/sample. A single point calibrator is used, the standard curve parameters being encoded on the reagent bar-code label. Samples above and below the working range of the assay are automatically rediluted and reanalysed.
There are three methods of checking for antigen excess. Firstly, the time course of the reaction is automatically monitored, checking the length of lag phase and time to peak rate. Secondly, a second aliquot of sample can be added to the reaction mixture 7 min after completion of analysis-if no further reaction occurs thesystem is in antigen excess. Thirdly, if a rediluted sample is still below the lower limit of the assay, this could be due to a very low sample concentration; or to the presence of extreme antigen excess. To distinguish between the two, an aliquot of control sample is added to the reaction mixture-if no reaction takes place extreme antigen excess is indicated.

Analytical methods and reagents
The instrument was used in accordance with the manufacturer's instructions; all calibrators and antiserum being supplied by Technicon UK. The comparison methodologies are shown in table 1, as are the sample volumes used by the DPA*I. Other standards and controls were obtained from Atlantic Antibodies (Incstar,

Results
Within-batch and between-batch precision data are shown in tables 2 and 3, respectively. The results for within-batch precision show excellent performance in all cases-the majority of the coefficients of variation being well below 3%. Inter-assay precision was good, generally Precision Within-batch precision was assessed by analysing three pools of serum at high, medium and low concentrations. The pools were arranged randomly on the 45 place sample tray and random cups were assayed in duplicate to give a total of 20 determinations at each level from the 45 samples.
Between-day precision was assayed using a mixture of frozen pools and commercial quality-control material. The study covered a single calibration period for alpha-1antitrypsin and the apolipoproteins, but more than one period for the other analytes.
Calibration stability Each assay was calibrated on day 0, no further calibration being performed for a period of at least 35 days. The calibrator and a series ofpools covering a wide concentration range were assayed as samples during the period to assess the stability of calibration.

Linearity
To determine the quality of the calibration produced using a single point calibrator a series of calibrators, in saline, of a high standard (SPS-02, Protein Reference Unit, Sheffield, UK) were made. The samples were assayed two weeks after a single point calibration. Table 2. Within-batch precision for eight analytes on the DPA *I (N 20for each sample).

Accuracy
At least 60 patient samples were assayed by the DPA* and by the comparison methodologies shown in table 1. In addition, a series of commercial calibrators and quality materials with quoted values and external quality assurance samples were analysed by the DPA* 1.
Carry-over Carry-over was estimated by running four sequences of a high pool (x 3) and a low pool (x 3) for immunoglobulin G.

Throughput
The time taken from commencement of the run to the final result printout was measured for a variety of workload patterns.  being between 3% and 5%. This probably could be improved upon in routine use as the data included a short time of poor precision, which was probably caused by a slightly blocked probe.

Calibration stability
The measured concentrations of various pools of the different analytes over a minimum of 35 days are shown in figure 1. In all cases, calibration appeared stable for this period of time, i.e. greater than the 2-week period claimed by the suppliers.
Linearity All methods were linear over the range of the assay, as shown in figure 2, confirming the suitability of the curve-fit parameters encoded on the reagent bar code.   4[a]) were lower than for the comparison method (slope 0"85), but the concentration measured showed slightly better agreement with results produced by serum electrophoresis (agarose) and densitometric scanning ( figure 4[b]), whereas agreement between the scanning and comparison method was not as good (data not shown).

Carry-over
The sequence H1HH3L1L2L3 was repeated four times for two immunoglobulin G pools and gave the following data: L1 3-796 g/1 L3 3"800 g/1 H 28"0 g/1 SD 0"051 SD 0-034 L1 and L 3 are not significantly different, indicating that carry-over was negligible.

Throughput
The throughput of the machine is dependent on the number of tests, not samples, and also on whether redilutions and automated antigen excess checking are undertaken. Figure 5 shows the time taken for a variety of workload patterns; thus, 60 tests can be analysed in 30 min if there are no redilutions etc. In the extreme cases using only known myelomas and antigen excess checking on all samples and many redilutions being performed, 100 tests can take approximately 140 min. Overall, a realistic analysis rate appears to be of the order of 100 tests/h. Within-day precision was excellent and between-day precision was good and could probably be improved upon in routine use.

Discussion
The software, the bar-coded calibrators and the ready-touse reagents all contributed to the ease of use of the  Calibration appeared stable throughout a 35 day period in all cases, indicating that monthly calibration should be sufficient. All methods were linear over the assay range, indicating the validity of the curve parameters and single point calibration.
Comparison with other methods was generally good, although the correlation coefficient for apolipoprotein A1 was less good. The significant slopes obtained for some assays is partly due to the difference in assigned values of the relative calibrants, though for alpha-l-antitrypsin and haptoglobin the slope of correlation was greater than the relative concentrations of the Technicon standard in the two measurement systems.  The difference in assigned calibration values is reflected in the concentrations measured for different calibrators for the analytes. The DPA*I is calibrated against the American College of Pathologists' reference preparation, although the software allows for the reassignment of the stated values for the calibrator to be adjusted so as to give results calibrated against a different reference preparation.
The correlation data for transferrin showed a significant intercept on the y-axis, although the straight line of the dilution curve indicates that the DPA*I is probably correct. There was no evidence of carry-over within the system, the washing procedure being satisfactory.
The antigen excess checking procedures are generally satisfactory, although two errors occurred. It is suggested that it is still advisable to run a total protein or serum protein electrophoresis if undertaking immunoglobulin assays for the first time on a patient this would indicate any samples likely to produce antigen excess in the assay system. For known paraproteins the analyser will give consistent results and such a procedure would be unnecessary [4].