ANALYST MAY 1986 VOL. 111 517 Correlation Between Fluorescent Polarisation lmmunoassay and Enzyme lmmunoassay of Anticonvulsant Drugs and Stability of Calibration Graphs* Neville Ratnaraj Valerie D. Goldberg and Peter T. Lascelles Department of Chemical Pathology The National Hospital Queen Square London WCI N 3BG UK Quality control materials and serum samples from patients on long-term drug therapy were analysed for anticonvulsant drugs by enzyme immunoassay (EMIT) and fluorescence polarisation immunoassay (TDX). The accuracy and precision of the two procedures were studied and the stability of calibration graphs was evaluated over a 30-d period. The accuracy and precision of both assays were satisfactory over the therapeutic ranges of phenobarbitone primidone phenytoin and carbamazepine and there was a good correlation between the results obtained by EMIT and TDX; for sodium valproate the accuracy and precision of the EMIT assay were poor.Calibration graphs generated by the TDX procedure were found to be stable but with the EMIT procedure calibration graphs for phenobarbitone carbamazepine and sodium valproate showed considerable drift. Keywords Enzyme immunoassa y; fluorescence polarisation immunoassa y; calibration graph stability; anticonvulsant drugs Over the last few years the concept of therapeutic drug monitoring (TDM) has become widely accepted as a valuable aid in the care and management of patients with epilepsy.' Well established techniques used for measuring concentra-tions of anticonvulsant drugs in serum include UV spectropho-tometry,2-6 gas - liquid chromatography (GLC)7-14 and high-performance liquid chromatography (HPLC) ,13-17 but as a result of the increase in the numbers of determinations carried out new methods of analysis have been developed that emphasise speed in addition to precision and accuracy.13 The introduction of radioimmunoassay~~J9 and its adaptation for the analysis of anticonvulsant drugs20 was followed by the development of the related techniques of enzyme immuno-assay fluoroimmunoassay and substrate-labelled fluores-cence immunoassay.21-29 These methods have proved to be particularly well suited for the production of commercial kits and hence for automation under microprocessor control and have enabled the drug monitoring units to keep abreast of a rapidly increasing work load.13 The technique of enzyme immunoassay (EIA) has had an enormous impact in the field of TDM,21 mainly because of the development of EMIT a homogeneous immunoassay system that was introduced by the SYVA Company in 1974. Using a discrete analyser samples can be processed at a rate of between 60 and 100 per hour and the results obtained correlate well with results from standard methods including GLC and HPLC.24JO-34 In 1981 as a further development in the field of TDM the Abbott Company introduced the TDX, which employs the principle of fluorescence polarisation.35J6 Although the principle was first described as early as 1926,37 and adapted for drug determinations in 1973,38 the TDX was the first commercial application of fluorescence polarisation immunoassay (FPIA).Results obtained using this method correlate well with those obtained using other pro~edures.39~40 Both the EMIT and TDX procedures are now widely used in the monitoring of anticonvulsant drugs in the routine situation and the accuracy and precision of both methods are acceptable in the context of routine clinical chemistry which commonly works to a 95% confidence limit. Both methods have been designed to take full advantage of modern instrument technology which is increasingly based on micro-processor control of the function and sequence of the analysis; sampling and dilution procedures can thus be automated by * Presented at Analyticon 85 London UK September 17-19th, 1985. the use of robotic arms and probes.Further introduction of EPROM (erasable program read-only memory) has the advantage that calibration graphs can be stored in memory and their stability monitored over a long period so that a fresh calibration graph does not have to be generated for each run. In order to establish limits for the stability of the EMIT and TDX reagents in the routine analysis of anticonvulsant drugs, we now present a comparison of the two techniques in respect of the stability of calibration graphs over a period of 30 d. We also present data on accuracy and precision obtained by analysing quality control materials and pooled patients' sera for anticonvulsant drugs by both methods. Experimental Patients' Samples Blood samples collected without anticoagulant from patients at the National Hospitals and Chalfont Centre for Epilepsy for the routine measurement of anticonvulsant drugs were used; all the patients were on two or more drugs.After centrifuga-tion the sera were stored at -20 "C. Quality Control Sera Quality control sera prepared by pooling patients' sera known by previous analyses to contain the drugs required were used to establish the within- and between-batch precision of the assays and the stability of the calibration graphs. The patients' sera were divided into two groups. One group was pooled, filtered twice with thorough mixing aliquoted into bottles and deep frozen at -20 "C (QCRa). The other group was subdivided into three batches of sera with subtherapeutic (SL) therapeutic (TL) and toxic (TO) levels of each drug, aliquoted and stored deep frozen (Table 1).In addition a commercial quality control material Ortho Bi-Level Assayed Anticonvulsant/Antiasthmatic Control Set I (Corning Medical) was used to investigate the accuracy of the assays and the stability of the calibration graphs. Methods Kits for the determination of phenobarbitone primidone, phenytoin carbamazepine and sodium valproate by the EIA (EMIT) procedure (SYVA) and FPIA (TDX) procedure (Abbott) were used according to the manufacturers' instruc 518 ANALYST MAY 1986 VOL. 111 Table 1. Therapeutic ranges of anticonvulsant drugs Therapeutic range*/ Drug pmoll-1 Phenobarbitone . . . . 20-130 Primidone . . . . . . 15-60 Phenytoin . . . . . . 28-67 Carbamazepine . . . . 12-50 Sodiumvalproate .. . . 360-600 * Values adapted from the literature and used routinely at The National Hospital. tions and the analyses carried out using an EMIT Auto-Carousel with CP5000 EMIT Clinical Processor (SYVA) and a TDX Analyser (Abbott). Calibration graphs were generated on the EMIT Auto-Carousel and Abbott TDX according to the manufacturers’ instructions. Correlation between the EIA and FPIA methods was examined by analysing randomly chosen patients’ samples for the five anticonvulsant drugs in duplicate on the same day and carrying out regression analysis and the accuracy and precision of the two procedures and the stability of the calibration graphs were assessed using the quality control sera Ortho I and QCRa and the pooled patients’ sera SL TL and TO as follows.The accuracy of the EIA and FPIA methods was investi-gated by analysing samples of the commercial quality control serum Ortho I in duplicate by both procedures. The within-batch precision of the EIA and FPIA methods was determined by analysing 40 samples of the quality control serum QCRa in duplicate by both procedures on the same day and the between-batch precision was assessed by analysing samples of QCRa in duplicate by both procedures in separate routine runs using newly generated calibration graphs for each run. The stability of the calibration graphs for the five anticon-vulsant drugs over a 30-d period and the precision of the assays under these conditions were determined by the following procedure. Using freshly prepared calibration graphs samples of QCRa and Ortho I and one sample from each of the groups of pooled patients’ sera SL TL and TO were determined by both methods.The calibration graphs for the EIA and FPIA procedures were stored in the memories of the CP5000 and TDX respectively and the analyses of the quality control samples and pooled patients’ sera were repeated at 3-d intervals for 30 d using the stored calibration graphs which were retrieved from memory as required. Results and Discussion Figs. 1-5 show the results obtained from the analysis of random patients’ samples by EIA and FPIA. It can be seen that the correlation between the two methods is acceptable over a wide range of concentrations; the correlation coeffi-cient lies between 0.98 and 0.99. Results of the within-batch analysis of samples of Ortho I by the two immunoassay procedures are shown in Table 2.It can be seen that the accuracy of the two immunoassay procedures is acceptable for phenobarbitone primidone phenytoin and carbamazepine although with sodium valproate the mean result of 40 analyses of Ortho I using the EMIT procedure barely falls within the range for the analysis quoted by the manufacturer. On the other hand it can be seen from Table 3 that the accuracy of the EMIT assay is unacceptable for phenobarbitone and sodium valproate with the mean result of 40 analyses of Ortho I falling well outside the manufacturer’s quoted range when between-batch analyses were carried out using the same calibration graphs for a period of 30 d. The -0 5 200 2 n I- > 160 II J I I I I I 0 40 80 120 160 200 240 Phenobarbitone level by EMlTiymol I-1 Fig.1. Correlation of TDX with EMIT assay of phenobarbitone. Analysis of randomly collected patients’ samples n = 40; r = 0.99; slope = 1.07; and intercept = -5.27 70 c I1 60 > 401 Q) -.E 20 a‘ 301 l o t //. I / I I I 1 I 0 10 20 30 40 50 60 70 80 Primidone level by EMIT/ymol I - 1 Fig. 2. Correlation of TDX with EMIT assay of primidone. Analysis of randomly collected patients’ samples n = 40; r = 0.98; slope = 1.01; and intercept = -0.99 100 r / 90 -80 -r I -Z 7 0 -8 60-I-9” 50-U 40-5 30 5. -0) C 0 C .--E 20-10 -0 10 20 30 40 50 60 70 80 90 100 Phenytoin level by EMITiymol 1-1 Fig. 3. Correlation of TDX with EMIT assay of phenytoin.Analysis of randomly collected patients’ samples n = 50; r = 0.99; slope = 0.99; and intercept = 0.7 ANALYST MAY 1986 VOL. 111 accuracy of the TDX procedure is acceptable under these conditions (Table 3). Table 4 shows the results of the within-batch analysis of QCRa by both immunoassay procedures. It can be seen that the within-batch precision of both immunoassay procedures measured by the analysis of QCRa is acceptable for phenobar-bitone primidone phenytoin and carbamazepine with a standard deviation of less than 2.5 pmol l-1 and a coefficient of variation of less than 5%. For sodium valproate however, whereas the coefficient of variation for the analysis of 40 samples by both procedures is less than 570 the standard deviation for the EMIT assay is as high as 15.37 pmol l-1 and for the TDX assay the standard deviation is 9.13 pmol 1-’ 900 7 800 0 5 700 600 - -2 > a 500 a -2 400 2 Tz > 7ij 300 .- 5 200 -u v 100 519 ---------I I 1 I 1 I I a I / 0 10 20 30 40 50 60 70 Carbamazepine level b y EMlTiprnol I-’ Fig.4. Correlation of TDX with EMIT assay of carbamazepine. Analysis of randomly collected patients’ samples y1 = 40; r = 0.98; slope = 0.96; and intercept = -1.41 Only for phenobarbitone and primidone in this experiment are the standard deviation and coefficient of variation lower for the EMIT assay than for TDX. Similar results were obtained from the between-batch analysis of the five drugs by the TDX and EMIT procedures as shown in Table 5 .The performance of both assays is satisfactory for phenobarbitone, primidone phenytoin and carbamazepine but less so for sodium valproate with a standard deviation of 6.35 pmol 1-1 for the TDX assay and 12.64 pmol 1-1 for EMIT under the conditions of the experiment. In the EMIT assay these results are in marked contrast to the results shown in Table 6, obtained from the between-batch analysis of QCRa for all five drugs using the same calibration graphs throughout the 30-d Table 2. Results of within-batch analysis of Ortho Control Set I by TDX and EMIT Drug Method Phenobarbitone . . . . . . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate . . . . TDX EMIT EMIT EMIT EMIT EMIT No.of samples 40 40 40 40 40 40 40 40 40 40 Mean/ pmoll-1 65.24 76.88 33.82 32.24 28.24 29.45 17.28 19.16 330.42 420.87 Standard deviation/ pmol 1-1.10 2.86 0.98 1.89 0.99 1.24 0.78 1.38 10.24 20.38 Coefficient of variation % 1.38 4.28 2.64 4.68 2.86 4.28 2.69 4.88 4.08 9.85 Mean concentration quoted by manufactured pmoll-1 59 t- 12 65 k 13 33 f 6 33 k 7 25 f 5 28 k 6 17 k 3 18 f 3 326 f 62 354 f 69 Table 3. Results of between-batch analysis of Ortho Control Set I by TDX and EMIT using the same calibration graph over a 30-d period Mean concentration Standard quoted by No. of Mean/ deviation/ Coefficient of manufactured Drug Method samples pmol 1- 1 pmoll-1 variation % pmol 1-Phenobarbitone .. . . . . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate . . . . TDX EMIT EMIT EMIT EMIT EMIT 11 11 11 11 11 11 11 11 11 11 62.09 80.73 33.64 30.64 26.64 27.45 16.73 20.64 321.55 547.64 1.58 5.00 1.03 2.11 1.03 1.44 0.65 1.43 11.41 113.08 2.45 6.20 3.05 6.89 3.86 5.24 3.87 6.95 3.55 20.65 59 k 12 65 f 13 33 f 6 33 f 7 25 k 5 28 f 6 1 7 f 3 1 8 k 3 326 f 62 354 k 6 520 ANALYST MAY 1986 VOL. 131 ~~~~ ~ ~ Table 4. Results of within-batch analysis of quality control serum QCRa by TDX and EMIT Drug Method Phenobarbitone . . . .. . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate . . . . TDX EMIT EMIT EMIT EMIT EMIT No. of samples 40 40 40 40 40 40 40 40 40 40 Mean/ pmol I-' 77.63 76.32 22.03 22.20 44.00 40.82 34.40 36.97 375.88 360.35 ~ Standard deviation/ pmoll-1 2.00 1.83 1.07 0.41 0.99 1.80 0.50 1.58 9.13 15.37 ~ Coefficient of variation o/o 2.67 2.40 4.88 1.82 2.24 4.40 1.44 4.26 2.43 4.27 Table 5. Results of between-batch analysis of quality control serum QCRa by TDX and EMIT Drug Method Phenobarbitone . . . . . . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate .. . . TDX EMIT EMIT EMIT EMIT EMIT No. of samples 40 40 40 40 40 40 40 40 40 40 Mean/ pmol I-' 75.15 78.70 22.03 21.38 43.38 42.38 33.65 36.70 368.80 378.40 Standard deviation/ pmol l-1 2.55 3.60 0.62 0.74 1.76 1.27 1.12 1.11 6.35 12.64 Coefficient of variation Yo 3.44 4.57 4.81 3.46 4.07 4.01 3.30 3.04 1.75 4.62 Table 6. Results of between-batch analysis of quality control serum QCRa by TDX and EMIT using the same calibration graph over a 30-d period Drug Method Phenobarbitone . . . . . . TDX Primidone . . . . . . . . TDX Phenytoin . . . . . . . . TDX Carbamazepine . . . . . . TDX Sodiumvalproate . . . . TDX EMIT EMIT EMIT EMIT EMIT No.of samples 11 11 11 11 11 11 11 11 11 11 Mean/ pmol - * 71.27 86.18 21 .oo 19.18 42.09 40.36 33.55 40.91 350.73 557.91 Standard deviation/ pmol 1- 1 2.15 7.85 1 .oo 2.32 1.22 3.98 0.82 4.68 10.46 111.36 Coefficient of variation O/O 3.02 9.10 4.76 12.07 2.90 9.86 2.44 11.44 2.98 19.96 period. Whereas the standard deviation for the EMIT procedure is satisfactory for primidone phenytoin and carba-mazepine under these conditions (less than 5 pmol 1-1 in all instances) it is 7.85 pmol 1-1 for phenobarbitone and 111.36 pmol 1-1 for sodium valproate; the range of the coefficient of variation lies between 9.10% for phenobarbitone and 19.96% for sodium valproate which is unsatisfactory.On the other hand it can be seen that the results obtained from the between-batch analysis of QCRa for all five anticonvulsant drugs by the TDX assay under these conditions (shown in Table 6) are comparable to those obtained from the within-batch analyses (Table 4) and the between-batch analyses where a fresh calibration graph was generated for each run (Table 5). Tables 7-11 show the results of the analyses of the pooled patients' sera SL TL and TO for the five drugs by TDX and EMIT using the same calibration graph over a 30-d period. It can be seen that the variation in the results for the pooled sera containing drugs below the therapeutic range (SL) within the therapeutic range (TL) and above the therapeutic range (TO) is random for primidone and phenytoin (Tables 8 and 9).For phenobarbitone and carbamazepine the results of the analysis of the sera TL and TO by EMIT unlike the results obtained by TDX show a pronounced high bias particularly between day 18 and day 30 (Tables 7 and 10). For sodium valproate (Table 11) the variation in the results obtained for the sera SL TL and TO by TDX is apparently random but it can be seen that the results obtained for SL using the EMIT assay increase by a factor of 480 pmol l-1 between day 1 and day 30 and that the test ceases to give a result for serum TL after day 24 and for serum TO after day 3 because the apparent increase goes above the upper limit for the test quoted by the manufacturer (1145 pmol 1-1). The results in Tables 7-10 show that the between-batch precision for the TDX method is acceptable for phenobarbi-tone primidone phenytoin and carbamazepine for the sera TL and TO with a standard deviation of less than 5.5 pmol l-1 and a coefficient of variation of less than 7%.For the EMIT assay the corresponding values in particular for the coeffi-cient of variation are poor except for phenytoin (Table 9). In the sub-therapeutic range it can be seen from Tables 7-10 that both assays yield comparable results for the determination of phenobarbitone primidone phenytoin and carbam-azepine but that the mean of the 11 results from the determination of phenobarbitone by EMIT is considerably higher than the mean of the 11 results from the TDX assay (Table 7). For sodium valproate the between-batch precision of the TDX assay is less satisfactory in terms of standard deviation than for the other drugs analysed under these conditions.Although the coefficient of variation is less tha ANALYST MAY 1986 VOL. 111 521 ~ ~~ Table 7. Analysis of pooled patients’ sera for phenobarbitone by TDX and EMIT using the same calibration graph over a 30-d period Phenobarbitone/pmol 1 - 1 * Sub-therapeutic level (SL) Therapeutic level (TL) Toxic level (TO) Day 1 3 6 9 12 15 18 21 24 27 30 Mean . . . . . . . . . . Standarddeviation . . . . . . Coefficient of variation Yo . . * Mean of three determinations. TDX 16 15 13 13 13 14 14 13 13 14 15 13.91 1.04 7.51 EMIT 18 24 21 19 19 21 21 21 20 23 20 20.64 1.75 8.47 TDX 107 105 104 100 100 99 103 102 102 103 103 102.55 2.34 6.89 EMIT 105 116 121 119 112 115 128 130 127 119 132 120.36 8.30 6.89 TDX 224 227 221 222 215 233 225 229 219 228 226 224.45 5.03 2.24 EMIT 216 232 242 254 244 245 269 265 263 258 291 252.64 20.11 7.96 Table 8.Analysis of pooled patients’ sera for primidone UY TDX and EMIT using the same calibration graph over a 30-d period Primidonelpmoll-l* Sub-therapeutic level (SL) Therapeutic level (TL) Toxic level (TO) Day 1 6 9 12 15 18 21 24 27 30 Mean . . . . . . . . . . Standarddeviation . . . . . . Coefficient of variation Yo . . * Mean of three determinations. TDX 10 10 11 11 9 11 11 11 12 10 12 10.73 0.90 8.43 EMIT 13 12 11 9 10 10 10 10 9 11 10 10.45 1.21 11.61 TDX 47 47 48 48 46 47 46 46 48 44 48 46.82 1.25 2.67 EMIT 50 48 46 40 43 41 41 44 38 41 40 42.91 3.73 8.69 TDX 71 70 70 68 65 65 68 67 70 69 67 68.09 1.92 2.82 EMIT 69 68 66 55 56 56 57 59 54 58 55 59.36 5.55 9.36 ~ Table 9.Analysis of pooled patients’ sera for phenytoin by TDX and EMIT using the same calibration graph over a 30-d period Phenytoin/pmol 1 - 1 * Sub-therapeutic level (SL) Day TDX EMIT 1 19 20 3 16 20 6 19 20 9 18 22 12 18 19 15 18 17 18 16 18 21 16 18 24 16 19 27 17 20 30 16 17 Mean .. . . . . . . . . 17.18 19.09 Standarddeviation . . . . . . 1.25 1.51 Coefficient of variation Yo . . 7.28 7.93 * Mean of three determinations. Therapeutic level (TD) Toxic level (TO) TDX 49 44 48 48 46 48 47 47 47 46 43 46.64 1.80 3.87 EMIT 48 48 48 51 43 44 44 45 48 51 45 46.82 2.79 5.95 TDX 96 96 85 93 95 95 91 93 91 92 90 92.45 3.24 3.50 EMIT 96 96 89 97 82 89 85 85 90 93 85 89.73 5.20 5.79 5% for the analysis of all three sera the standard deviation ranges from 17.25 to 26.66 pmoll-1 (Table 11). However the performance of the EMIT assay in the analysis of serum SL is extremely poor with a standard deviation of 132.72 pmol 1-l and a coefficient of variation of 22.15% and no calculations of any kind could be made on the results of the analysis of the sera TL and TO by this method (Table 11).The results obtained from the within and between-batch analysis of quality control samples with freshly generated calibration graphs show that the performance of both immuno 522 ANALYST MAY 1986 VOL. 111 ~ Table 10. Analysis of pooled patients’ sera for carbamazepine by TDX and EMIT using the same calibration graph over a 30-d period Carbamazepinelymol 1-~~ ~ ~~ Sub-therapeutic level (SL) Therapeutic level (TL) Toxic level (TO) Day 1 3 6 9 12 15 18 21 24 27 30 Mean . . . . . . . . . . Standarddeviation . . . . . . Coefficient of variation % . . * Mean of three determinations.TDX 7 7 7 7 5 5 5 6 6 6 6 6.09 0.83 13.65 EMIT 8 8 7 7 8 8 7 7 7 8 8 7.55 0.52 6.92 TDX 22 21 19 19 19 19 18 19 19 19 19 19.27 0.90 4.69 EMIT 22 22 22 22 24 23 23 22 24 26 27 23.36 1.75 7.48 TDX 50 50 50 49 52 49 48 48 53 49 50 49.82 1.54 3.09 EMIT 55 55 54 50 69 75 60 63 75 89 93 67.09 14.50 21.61 Table 11. Analysis of pooled patients’ sera for sodium valproate by TDX and EMIT using the same calibration graph over a 30-d period Sodium valproate/ymol ] - I * Sub-therapeutic level (SL) Day 1 3 6 9 12 15 18 21 24 27 30 Mean . . . . . . . Standarddeviation .. . . . . Coefficient of variation % . . * Mean of three determinations. TDX 354 354 389 368 368 361 382 382 410 354 382 382.18 17.80 4.78 EMIT 397 474 516 54 1 541 62 1 643 643 569 733 877 599.18 132.72 22.15 Therapeutic level (TL) EMIT TDX -590 643 604 732 590 748 569 805 569 805 618 922 590 679 597 878 625 937 604 >1145 625 >1145 600.18 17.25 2.87 Toxic level (TO) TDX 923 923 923 854 854 902 95 1 902 902 923 909 914.82 26.66 2.91 EMIT 937 1139 >1145 >1145 >1145 >1145 > 1145 >1145 >1145 > 1145 >1145 assay procedures is satisfactory in the context of a clinical chemistry laboratory where the standard of a 95% confidence limit is widely accepted.The results of the analysis of the pooled patients’ sera however show that the calibration graphs generated by the TDX procedure largely remain stable over that period whereas the calibration graphs for EMIT exhibit a considerable drift in the determination of sodium valproate and to a lesser extent phenobarbitone and carba-mazepine marked by a high bias in the results for the determination of these drugs. The fact that the standard deviation and coefficient of variation for the between-batch analysis of quality control sera by the EMIT assay are acceptable when a fresh calibration graph is generated for each run is a further indication that the results shown in Tables 7 10 and 11 are primarily a function of the drift of the EMIT calibration graph under the conditions of the experiment, rather than cross-reactivity with other substances present in patients’ sera.On the other hand the relatively unsatisfactory performance of the TDX assay of sodium valproate com-pared with the performance of the other TDX assays examined in this study may be a function of antibody cross-reactivity with derivatives of fatty acids present in patients’ sera as the fluctuations in the results of the analysis of pooled patients’ sera by TDX shown in Table 11 are purely random in character. The two immunoassay procedures under investigation here gave comparable results for all five anticonvulsant drugs when used for the analysis of random patients’ samples over a wide range of concentration.Nevertheless recent reports of interferences in immunoassay procedures from substances present in patients’ sera have given rise to concern over the specificity of commercial kits for the assay of anticonvulsant drugs.41342 A major problem has been the report of interfer-ence with the EMIT assay of phenytoin by a metabolite of phenytoin or its derivatives.43.44 Problems have also been described although to a lesser extent with the TDX assay for phenytoin .41,45 The increasing development and use of mono-clonal antibodies will undoubtedly improve the specificity of assays for anticonvulsant drugs based on immune reactions in the future although there may be problems still to overcome. The qualities of simplicity and speed already make immuno-assays indispensable in the situation of ever-expanding need and increase in workload for the analysis of anticonvulsant drugs.References 1. 2. 3. 4. Pippenger C. E. Ther. Drug. Monit. 1979 1 3. Broughton P. M. G. Biochem. I . 1956 63 207. Wallace J. E. Biggs J. D. and Dahl E. V. Anal. Chem., 1965 37 10. Fuhr J. Arzneim-Forsch. 1964 14 74 ANALYST MAY 1986 VOL. 111 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Hamilton H. E. and Wallace J. E. in Pippenger C . E., Penry J. K. and Kutt H. Editors “Antiepileptic Drugs: Quantitative Analysis and Interpretation,” Raven Press New York 1978 pp. 175-183. Kupferberg H. J. in Pippenger C. E. Penry J. K. and Kutt, H. Editors “Antiepileptic Drugs Quantitative Analysis and Interpretation,” Raven Press New York 1978 pp.9-17. Grimmer G. Jacob J. and Schafer H. Arzneim.-Forsch., 1969 19 1287. MacGee J. Anal. Chem. 1970 42 421. Toseland P. A. Grove J. and Berry D. J. Clin. Chim. Acta, 1972 38 321. Chambers R. E. and Cooke M. J. Chromatogr. 1977 144, 257. Rambeck B. and Meijer J. W. A. Ther. Drug Monit. 1980, 2 385. Toseland P. A. and Wicks J. F. C. in Richens A. and Marks V. Editors “Therapeutic Drug Monitoring,” Churchill Livingstone London 1981 pp. 85-109. Goldberg V. D. Ratnaraj N. Elyas A. and Lascelles P. T., in Rose F. C. Editor “Research Progress in Epilepsy,” Pitman Press Bath 1983 pp. 462-483. Burke J . T. andThenot J. P. J. Chromatogr. 1985,340,199. Huisman J. W. Clin. Chim.Acta 1966 13 323. Atwell S. H. Green V. A. and Haney W. G. J. Pharm. Sci. 1975 64 806. Soldin S. J. and Hill J. G. Clin. Chem. 1976 22 856: Yalow R. S. and Berson S. A. Nature (London) 1959,184, 1648. Yalow R. S. and Berson S. A. J. Clin. Invest. 1960 39, 1157. Cook C. E. Christensen H. D. Amerson E. W. Kepler, J. A. Tallent C. R. and Taylor G. F. in Kellaway P. and Petersen I. S. Editors “Quantitative Analytic Studies in Epilepsy,” Raven Press New York 1976 pp. 39-58. Marks V. in Richens A. and Marks V. Editors “Thera-peutic Drug Monitoring,” Churchill Livingstone London, Rubenstein K. E. Schneider R. S. and Ullman E. F., Biochem. Biophys. Res. Commun. 1972 47 846. Berchou R. C. Lodi R. A. and Sherman J. A. Ann. Clin. Biochem. 1979 16 205. 1981 pp.155-182. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 523 Goldberg V. Ratnaraj N. Elyas A. and Lascelles P. T., Anal. Proc. 1981 18 313. McGregor A. R. Crookall-Greening J. O. Landon J. and Smith D. S. Clin. Chim. Acta 1978 83 161. Kamel R. S. McGregor A. R. Landon J. and Smith D. S., Clin. Chim. Acta 1978 89 93. Kamel R. S. Landon J. andsmith D. S. Clin. Chem. 1980, 26 1281. Wong R. C. Burd J. F. Carrico R. J. Buckler R. T., Thoma J. and Boguslaski R. C. Clin. Chem. 1979,25,686. Li T. M. Robertson S. P. Crouch T. H. Pahuski E. E., Bush G. A. and Hydo S. J. Clin. Chem. 1983 29 1628. Blake C. and Gould B. J. Analyst 1984 109 533. Hemmila I . Clin. Chem. 1985 31 359. Shaw W. and McHan J. Ther. Drug Monit. 1981 3 185. Kulpmann W. R . and Oellerich M. J. Clin. Chem. Clin. Biochem. 1981 19 249. Jolley M. E. J. Anal. Toxicol. 1981 5 236. Popelka S. R. Miller D. M. Holen J. T. and Kelso D. M., Clin. Chem. 1981 27 1198. Jolley M. E. Stroupe S. D. Schwenzer K. S. Wang C. J., Lu-Steffes M. Hill H. D. Popelka S. R. Holen J. T. and Kelso D. M. Clin. Chem. 1981 27 1575. Perrin F. J . Phys. Radium 1926 7 390. Dandliker W. B. Kelly R. J. Dandliker J. Farquhar J., and Levin J. Zmmunochemistry 1973 10 219. Lu-Steffes M. Pittluck G. W. Jolley M. E. Panas H. N., Olive D. L. Wang C.-H. J. Nystrom D. D. Keegan C. L., David T.-P. and Stroupe S. D. Clin. Chem. 1982,28,2278. Wang S. T. and Peter F. Clin. Chem. 1985 31 493. Haughey D. B. Matzke G. R. Halstenson C. E. and Keane W. F. J. Anal. Toxicol. 1984 8 106. Walson P. D. Brain Dev. 1985 7 116. Flachs H. and Rasmussen J. M. Clin. Chem. 1980,26,361. Aldwin L. and Kabakoff D. S. Clin. Chem. 1981 27 770. Oeltgen P. R. Clark T. Blouin R. A. and Shank W. A., Jr. Clin. Chem. 1984 30 1032. Paper A51430 Received November 20th 1985 AcceDt-ed December 12th. 198