Synthesis, Screening and Pharmacokinetic Evaluation of Potential Prodrugs of Bupropion. Part One: In Vitro Development

In general, prodrugs are developed to circumvent deficiencies associated with the absorption, distribution, metabolism, excretion or toxicological (ADMET) profile associated with the active drug. In our study, we select bupropion, a drug with broad pharmacology incorporating dopaminergic, noradrenergic, nicotinic and cytokine modulation properties, but which is rapidly metabolized in vivo. We exploited its carbonyl and secondary amine functionality to facilitate the synthesis of bioprecursor prodrug forms with the sole objective of identifying analogues with enhanced properties over bupropion. A range of analogues were synthesized, ranging from N-methyl, N-benzyl, oximes, enol acetate and ether forms to examples where both functional groups were utilized to form oxadiazine, oxadiazinone, oxazolone and acetylated derivatives. We then developed an in vitro metabolic screen to simulate the human oral delivery route for these analogues. The selection of media in the screens contained a variety of pH, enzymatic and co-factor systems which mimic metabolic in vivo environments that drugs encounter when delivered orally. By coupling our in vitro screening tool to a selective hyphenated technique such as LC-MS, we were able to quickly select potential prodrugs for further in vitro and in vivo development. From the data generated, the N-alkylated bupropion analogues were shown to have the highest potential to act as bioprecursor prodrugs of bupropion.


Plasma LLOQ/Precision
Human plasma was spiked with N-methylbupropion, bupropion and metabolites at a concentration of 10 ng/mL. Aliquots (6 × 50 µL) were transferred to 1.5 mL microcentrifuge tubes. ACN (200 µL) was added and the solution was vortex mixed. The remaining mixture was centrifuged at 10,000 rpm for 10 min. An aliquot (100 µL) of supernatant was removed and added to 0.001 M HCl (100 µL). The solution was vortex mixed again and 100 µL was placed in a micro-insert for HPLC analysis. The final concentration was 1 ng/mL. All LLOQ solutions had an analyte peak signal to noise ratio greater than 10.

Plasma LOD
The limit of detection of the method was determined by injection of serially diluted standards from highest to lowest level. The limit of detection was determined as the lowest peak seen with a minimum signal to noise ratio of three.

Brain LLOQ Accuracy/Precision
Rat brain homogenate (165 mg brain per mL 0.01 M HCl) was spiked with N-methylbupropion, bupropion and metabolites at two concentrations, 250 and 2500 ng/mL. Aliquots (5 × 200 µL) were transferred to 1.5 mL microcentrifuge tubes. The mixture was centrifuged at 15,000 rpm for 10 min. Supernatant (100 µL) was transferred to 1.5 mL microcentrifuge tubes. ACN (300 µL) was added and the solution was vortex mixed. The remaining mixture was centrifuged at 15,000 rpm for 10 min. 200 µL of supernatant was removed and added to 800 µL of water. The solution was vortex mixed again and analysed by HPLC. Final concentrations were 12.5, 125 ng/mL.

Specificity
The bioanalytical LCMS method was shown to be specific by injection of all standards, plasma and diluent solutions separately. No interference was observed with mass values similar to the analytes of interest. The ability of the Orbitrap to extract ions ± 2.0 ppm routinely during the course of the analysis significantly improved the signal to noise ratio affording excellent specificity and sensitivity. The diastereomeric pair rac-erythro-hydrobupropion and rac-threo-hydrobupropion which have identical mass (242.1306 amu) required baseline resolution for accurate quantitation. Resolution between these two compounds over the linear range was >1.5 which was considered acceptable. The same resolution was observed for the diastereomeric pair of N-methylaminoalcohol metabolites. For use of this method for determination of bupropion and metabolites in guinea-pig plasma and brain, selectivity is not an issue for rac-erythro-hydrobupropion as this is not generated in guinea pigs in significant amount [13].

Linearity
Linearity was assessed by injection of a range of standards from 0.1 ng/mL to 1,000 ng/mL. The linear part of the calibration curve was evaluated by least squares-linear regression analysis where the linear range was calculated by the range of points with a correlation coefficient of greater than 0.995. The method was found to be linear for all analytes for at least three orders of magnitude, with a minimum correlation coefficient of r 2 ≥ 0.995. Table S1 presents the correlation coefficient values for bupropion, N-methyl bupropion and their metabolites. The linear range for rac-N-methyl aminoalcohols was lower as this standard was 2:1 mixture of the two diastereomers. N-methylbupropion had the largest range of quantitation, 1-500 ng/mL.

Carryover
Carryover was assessed by injection of a blank sample following analysis of the highest concentration linearity sample. Initially during the method development there was carryover (>2%) of bupropion but after careful adjustment of needle flushing and washing between injections, this was eliminated. The needle wash solution was optimized as 0.5% formic acid in MeOH. The acidified needle wash improved system suitability, reproducibility and eliminated carryover (<0.1%).

Limit of Detection
The limit of detection was determined as the lowest peak detected with a minimum signal to noise ratio of three. The LOD of N-methylbupropion, bupropion and metabolites was shown to be in the pg/mL range (Table S2).

Accuracy and Precision
The accuracy/recovery and precision of the method was evaluated using human plasma. Six replicate plasma samples were analysed at three different levels across the linear range. Each was spiked with N-methyl bupropion, bupropion and metabolites in the expected pharmacokinetic concentration range. All accuracy results were within 85%-115%,% RSD ≤ 15% (Table S3).

LLOQ-Accuracy and Precision
Plasma lower limit of quantitation-accuracy and precision was established at the 1 ng/mL level for N-methylbupropion, bupropion and metabolites. At 1 ng/mL each analyte had a signal to noise ratio of greater than 10. The results are tabulated in Table S4. Brain LLOQ was determined at 12.5 ng/mL and accuracy/recovery at 125 ng/mL. The CV of six replicates at each level was higher than that of plasma samples. This was attributed to dissolved protein precipitating during sample preparation. This displaces the solution volume and can have an effect of concentrating the solutions. Results are tabulated in Table S5.

Sample and Standard Stability
Sample and standard stability was evaluated over 24 h. Samples and standards were shown to be >90% assay after 24 h when stored at 5 °C. Medium and long term stability was not established. Sample and standard stability in plasma is already documented in literature for bupropion and metabolites [14,15]. Bupropion plasma samples are stable at low pH and metabolite samples are stable in the pH range 2.5-10.0. Therefore our samples and standards were prepared in 0.001 M HCl. This is the standard sample preparation procedure in the current USP monograph for preparation of assay samples for bupropion hydrochloride extended release tablets [16]. Standard stability was demonstrated throughout the run and at the end of the run by calculating the %RSD of a repeat injection standard. This verified standard stability over the course of the analysis.