Comparison between the immunoassay and high performance liquid chromatography for therapeutic monitoring of carbamazepine and phenytoine

Objective: To investigate the correlation of the immunoassay and chromatography method for quantitative measurement of two antiepileptic drugs (AED), carbamazepine (CBZ) and phenytoin (PHT) and determination of relation between the CBZ and it’s metabolite carbamazepine 10,11epoxide (CBZ-E). Additionally we investigated whether there is a di4 erence in the determination of serum concentration of CBZ and PHT when measured in two di4 erent labs by high performance liquid chromatography (HPLC). Materials and methods: This study was carried out on 102 blood samples (72 CBZ and 30 PHT) collected from epileptic outpatients. Plasma concentrations of CBZ and PHT were determined by validated HPLC (Shimadzu and Agilent) and the CEDIA-immunoassay method. Results: The correlations of serum concentrations of CBZ between CEDIA and HPLC1 and between CEDIA and HPLC2 were good (R = 0.97 for both techniques). Even better correlation was found between concentrations of CBZ measured by the two HPLC systems (R = 0.99). Similar, for PHT, we found good correlation between CEDIA and the two systems of HPLC (HPLC1 and HPLC2, R = 0.98) and between the two systems of HPLC of R =0.98. The moderate correlation coe> cient was found between serum concentrations of CBZ and its metabolite CBZ-E, measured in two labs by di4 erent HPLC (R = 0.49 and 0.43, respectively; P < 0.001). Conclusion: We observed good correlation for estimation of CBZ and PHT concentration obtained by means the immunoassay and two di4 erent HPLC. The possibility of measurement of CBZ-E could be advantage of chromatography in comparison with immunoassay.


Introduction
Antiepileptic drugs are characterized with important inter-and/or intra-individual variation in pharmacokinetics and di erent susceptibility to adverse reaction (1).The clinical e ect of many antiepileptic drugs correlates better with blood levels than with doses (2,3).
Up until approximately 20 years ago, anti-epileptic therapeutics were restricted to very few drugs that were developed in the rst half of the 20th century, so called old antiepileptic drugs.A relatively small therapeutic index and the desirability to guide the therapy with therapeutic drug monitoring (TDM) characterize the antiepileptics of the old, but also of the newer generation (4,5).
The old antiepileptic drugs (AED) still remain valuable in the epilepsy treatment, and the optimiza-tion of therapy with these drugs through TDM is imperative.
Carbamazepine (CBZ) is considered the rst-line anticonvulsant for the treatment of generalized tonic-clonic and complex partial seizures.Its use is complicated by variable absorption, auto-induction, hetero-induction of the CYP3A4 isoenzyme and its active metabolite carbamazepine-10,11epoxide (EPO-CBZ) (6).The pharmacologic activity of the metabolite CBZ-E is similar to that of CBZ (7).
PHT is a primary anticonvulsant drug for the prophylaxis and treatment of generalized tonic-clonic and partial seizures.PHT exhibits nonlinear pharmacokinetics in the therapeutic range.All these properties of antiepileptic drugs argue the use of TDM (8,9), as the interpretation of plasma levels of AED in the light of the clinical situation of epileptic patients can markedly in uence dose optimisation of these drugs (10).
Selecting the most appropriate analytical methods is often not easy and the choice depends on the availability of sta , expertise and equipment, the nature of the service to be provided and the range of drugs to be assayed.
The major advantage of chromatographic techniques is their exibility and adaptability to a wide range of drugs.However, in comparison to immunoassays they are slow, labour-intensive techniques which usually demand a high degree of technical expertise.Immunoassays have higher development cots than chromatography, narrow range of drugs to be analysed and are generally more susceptible to interference to endogenous components (cross-reactivity) (11).
The objective of our study was to investigate the interchangeability of immunoassay and chromatography for the quantitative measurement of CBZ and PHT and to investigate whether measurements in two labs using di erent HPLC systems could results in di erences in the determination of CBZ and PHT in same serum samples.Additionally we investigated the relation between the serum concentrations of CBZ and its metabolite, CBZ-E.

Materials and methods
This retrospective study was carried out on 102 blood samples collected from epileptic outpatients who were referred for routine TDM in the IMD lab-Berlin (Institute for Medical Diagnosis, Berlin, Germany).Serum concentrations of CBZ were determined in 72 samples, while the concentrations of PHT were determined in 30 serum samples.Blood samples were taken at steady state concentration of these drugs.The serum concentrations of CBZ and its metabolite CBZ-E and PHT were determined comparatively by two HPLC systems, HPLC 1 and HPLC 2 (HPLC 1: model Shimadzu LC-9A in the Institute of Clinical Pharmacology of University Clinical Center "Charite", Berlin; HPLC 2: model Agilent 1050 HPLC-System in the Institute for Medical Diagnosis-IMD Berlin).In both institutions measurements of AED are established for routine TDM.
In the HPLC 1 we used the commercial available kit for determination of AED (Chromsystems Instruments & Chemicals GmbH, München, Germany).We used other components of kit: Mobile Phase, high resolution, 1000 mL; AED Serum calibration standard; Internal standard, 15 mL and Precipitation reagent, 5 mL.We performed a sample preparation as was described in the manual of manufacturer.The HPLC 1 parameters were: We used a Carbamazepine kit (CEDIA) and Phenytoine kit (CEDIA) and Biorad controls (Immunoassay Plus Control -Lyphocheck level 1, and Lyphocheck level 2).The calibration stability (Cedia® Core TDM Multi calibrator) was controlled by daily running of controls to check the intra assay specity.Between run coe cients of variation were < 8% for all assays in two HPLC analytical systems.Serum samples were stored at -20 °C.The statistical analysis of relations between HPLC and immunoassay results was performed with Microsoft EXCEL® software (Analyse-it®), using the linear regression analysis and Passing-Bablok method and Bland-Altman plots (12).
For presentation of relation between the CBZ and CBZ-E we used parametric tests when the data had a Gaussian distribution (Student-t test and Pearson correlation coe cient) (13).

Results
Analysis of the results revealed that the compliance of serum concentrations of CBZ and PHT between immunoassay and HPLC and between two di erent HPLC (HPLC 1 and 2) is very good.We observed excellent correlation (r = 0.97, P < 0.001) between CBZ concentrations obtained by HPLC 1 and CEDIA (Figure 1a).The same correlation (r = 0.97, P < 0.001) of CBZ concentrations were found between HPLC 2 and CEDIA (Figure 1b).In the analysis of compliance between the two HPLC systems in determination of CBZ concentrations, we obtained even better correlation, r = 0.99, P < 0.001 (Figure 1c).   between HPLC 1 and CEDIA (r = 0.98, P < 0.001) and between HPLC 2 and CEDIA (r = 0.98, P < 0.001) is presented in gure 2a and 2b, respectively.Comparison of PHT serum concentrations obtained by two di erent HPLC (HPLC 1 and 2) in two di erent labs also showed very good correlation (r = 0.98, P < 0.001) (Figure 2c).
In addition possible di erences between two different HPLC systems were further evaluated by comparison of correlations between CBZ and its metabolite, EPO-CBZ.The correlation coe cients between CBZ and EPO-CBZ measured by HPLC in two di erent labs are shown in the Figure 3a and 3b.The correlations between CBZ and EPO-CBZ in both HPLC systems were moderate, i.e. r = 0.486 (P < 0.001) for HPLC 1 and r = 0.427 (P < 0.001) for HPLC 2. The correlation between HPLC 1 and HPLC 2 in measurement of EPO-CBZ was good (R = 0.88, P < 0.001).
In order to compare two analytical methods and two HPLC using di erent commercial kit we applied the comparison by Passing-Bablok method.
The detail results between di erent HPLC (HPLC 1 and HPLC 2) and between HPLC and immunoassay we have presented in table 1.
The agreement between HPLC 1 and HPLC 2 in the determination of CBZ, PHT and CBZ-E we have evaluated through Bland-Altman agreement.This method we used to compare the immunoassay and chromatographic method (HPLC 1 and HPLC 2) and to present di erence vs. mean average (Figure 4a-c; 5a-c and 6a).The results of Passing Bablok comparison are presented in the table 1.

Discussion
TDM as routine service includes the measurement of serum concentrations of numerous drugs presenting widely varying chemical structures.Thus, in the TDM service a compromise has to be made between using the best method for each individual analyte and using the techniques that allow quantization of wide range of substances.
In our study besides the correlation between two analytical methods (immunoassay-CEDIA and HPLC), we attempt to present the impact of di erent HPLC systems with di erent procedures of validation.The correlation between two HPLC systems and correlation between two HPLC methods and immunoassay method (CEDIA) observed in the present study proved very satisfactory for CBZ and PHT.
The correlation coe cient of serum concentration of CBZ and PHT between the HPLC 1 and CEDIA in our study were the same (each r = 0.974), while the correlations of serum concentrations of CBZ and PHT between the HPLC 2 and CEDIA in our study were slightly di erent (r = 0.969 and r = 0.990).The above ndings almost match results of Rambeck et al., who found that in the case of PHT there was a highly linear correlation (r = 0.985, y = 1.113x -0.589) between HPLC and the Biotrack system, while in the case of CBZ, the correlation between HPLC and Biotrack system was somewhat lower (r = 0.931, y = 1.29x -0.136) (14).Similar results of correlation have found others authors, too (15,16).
In contrast to immunoassay, the TDM of CBZ using the chromatography system enables to measure the concentrations of CBZ-E.The determination of CBZ-E presents an advantage in TDM concerning the fact that this metabolite has similar pharmacological activity to the parent drug (17,18), and this should be considered in the optimization of CBZ therapy (19).In our study we noted the moderate correlation between CBZ and CBZ-E level determined by HPLC 1 (r = 0.486) and HPLC 2 (r = 0.427).Previous authors con ned the variability of correlation between the CBZ and its metabolite, CBZ-E.Mihaly et al. presented moderate correlation coefcient of plasma concentrations of CBZ and its metabolite, CBZ-E (r = 0.495) (20).Mihaly suggested that there was a considerable inter-subject variation in the metabolism of CBZ to CBZ-E, without presenting any other possible factor.Semah et al. observed correlation between CBZ and CBZ-E plasma levels found in the mono-therapy group, but not in the poly-therapy group (21).Further we found a good correlation in determination of CBZ-E serum concentration between HPLC 1 and HPLC 2.
Unfortunately, in our study were not able to determine which serum samples present mono-therapy or poly-therapy of CBZ.
Pasing-Bablok regression and Altman -Bland plots supported the comparison between two di erent HPLC and between chromatography and immunoassay.The Bland-Altman scatter diagram showed the di erences plotted against the averages of the two measurements, drawing the mean di erence, and the limits of agreement, de ned as the mean di erence plus and minus 1.96 times the standard deviation of the di erences.The results showed the good agreement between di erent measurement techniques.

Conclusion
The correlation between 2 analytical methods and between 2 HPLC systems is very good indicating the complementarities of results of TDM in both institutions.
The moderate correlation between CBZ and CBZ-E obtained in our study by both HPLC systems show the possible implication of other pharmacokinetic factors and possible concomitant AED that was not subject of investigation in this study.The analytical determination of CBZ-E could be advantage of chromatography in comparison with CEDIA immunoassay.

FIGURE 1A .
FIGURE 1A.The comparison of serum concentration of CBZ between CEDIA immunoassay and HPLC 1

FIGURE 6A .
FIGURE 6A.Bland-Altman plot: Di erence HPLC1 vs. HPLC2 (EPO-CBZ) In the HPLC 2 we used a ClinRep® Complete Kit for AED in serum (order no.15000; RECIPE Chemicals + Instruments GmbH, München, Germany).The Kit components of ClinRep® were: Mobile phase 1000 mL; Standard solution 3 mL; P Precipitant with internal standard IS 15 mL; Serum calibrator, lyophylisate 3 mL; Sample preparation vials with reagent (lyophilisate) and D diluting solution.The procedure of sample preparation was described in the manual of ClinRep® Complete Kit.