Computer evaluation of the EMIT assays carbamazepine, ethosuximide, phenobarbital, phenytoin, quinidine and theophylline on the Gemsaec centrifugal fast analyser

ACKNOWLEDGEMENTS We thank the American Hospital Supply (U.K.) Ltd., for the loan of the instrument and the Department of Health for the financial support to enable us to carry out this evaluation. We are also grateful to Mr. J. Korten (Electronucleonics,Breda, Holland) and Mr. I. West (American Hospital Supply Company) for advice and assistance throughout the evaluation and to Mr. P. Brooks (Mayday Hospital, Croydon) for providing patient samples assayed for total protein.


Introduction
The adaptation of EMIT antiepileptic drug assays to the Aminco Roto Chem II centrigufal fast analyser has been described by Finley et al [1,2]. The analysis procedure was very convenient and accurate; however, most of the data handling had to be done with the aid of a HP 9815A desktop computer. Using Finley's modification of the EMIT assays the authors, wished to adapt EMIT similarly to the Gemsaec centrifugal fast analyser and perform all the data processing with the Gemsaec computer. As no EMIT-program was available it was decided to develop one that would operate for routine clinical chemistry work. The program was written in FOCAL 8

(the version used by Electro Nucleonics
Inc., the manufacturer of the Gemsaec) computer language.
The Gemsaec transfer disc has only 16 positions, so it was considered important not to occupy a large part of the disc with standards when analysing unknown samples. Apparatus A Gemsaec centrifugal fast analyser attached to a PDP 8/e computer, with magnetic tape (dectape) as the storage device, was used for the evaluation. The Rotor temperature was kept within 37.0 + 0.1C. Electro Nucleonics Inc.'s loader for the Gemini analyser was used for the preparation of transfer discs for Gemsaec. The loader was prepared for the automatic delivering of two reagents and the sample flushed with buffer solution into the transfer disc.

Reagents
Reagent kits from the Syva Corporation were used through out the evaluation.

Data handling
Five absorbance readings were taken on each of the rotor positions. The delta absorbance, i.e. the last reading minus the first reading, is a measure of the enzyme activity of the sample in its cuvet. The enzyme activity is correlated to the concentration of the drug in the sample. Similarly to Finley et al the measured delta absorbances were treated as free counts of radioactivity in the radioimmunoassay. Applying the logit-log transformation used by Finley et al, in the computation of the standard curve, a second order, unweighted, non-linear curve provided a better fit for the experimental data than a linear curve did.
The following formulas were used as the basis of calculation of the FOCAL program developed here: (2) ln(conc) K-logit 2 + L.logit + M K,L and M factors were evaluated using the following matrix equation: The curve fitting procedure was carefully inspected for falsely computed results from sera containing very low drug concentration. The logit-log parabola may reverse "direction" if B is close to BO. K,L and M factors were stored on dectape in order to check the stability of the calibration between runs. During the first part of the program the calculation is carried out using the stored K,L and M factors in equation 4. If there is an unacceptable deviation from target values, e.g. when only standards are analysed, the curve fitting factors have to be recalculated. Three different modes of recalibration may be performed.  5" 33.6 CUV. 6

Results and discussion
The pipetting station, the loader, was carefully optimised before setting up EMIT runs. It was extremely important to prime the pump syringes prior to analysis. Table shows the precision data obtained when five standards on each disc were used to calculate the standard curve. It is apparent that the coefficients of variation for the EMIT assays studied in this work were low. Unfortunately only nine positions were then left free on the disc for the evaluation of unknown samples. As the first section of the program makes no correction of the stored curve fitting factors, it was possible to check the agreement of standard curves between runs. However, results obtained from several runs of standard curves showed that the agreement was poor. Either a correction of the stored "standard curve" or a recalibration using a full set of standards (= five standards) must be performed. As stated in the data handling section, the facilities to correct a stored "standard curve" using measured data of either one or two standards on the disc were included in the program. Using just one standard for recalculation, the parallel shift correction, resulted in unreliable performance of the assay (results are not shown in this paper). However, recalibration using two standards was acceptable, see Table 2. Using this procedure an additional three samples per disc could be analysed, providing a more economical analysis than that performed with a full set of standards on each disc, although the latter was always superior to other modes of calibration. Two long-term quality control studies were made, see Table 3; they demonstrate the routine applicability and reliability of our computerised EMIT procedure.   Figure 3. A carbamazepine assay performed with a "full set" of standards on the disc. In this case the stored calibration factors are acceptable. However, the standard curve was recalibrated and the outprint from the curve fitting procedure is shown in the second part. By use of the specially designed calculation program for the Gemsaec computer, the manual plotting and evaluation is totally eliminated. No additional computer is required to perform calculation of EMIT results. Standard curves can be stored on magnetic tape and may be corrected by use of measured data of two standards in new runs. Three more samples per disc could therefore be analysed.
A great advantage of the EMIT assays is the enormous time saving as compared to other methods of analysis (GC or HPLC). EMIT enhances the possibility of producing fast reports both during the night and the day for the clinician.
In an emergency situation this is a great asset for guiding the treatment of drug-intoxicated patients or underdosed patients. Table 3. Long-term quality control