Validated LC-MS/MS Method for Simultaneous Quantitation of Enasidenib and its Active Metabolite, AGI-16903 in Small Volume Mice Plasma: Application to a Pharmacokinetic Study

ABSTR ACT
Enasidenib is a selective mutant isocitrate dehydrogenase 2 inhibitor approved for the treatment of relapsed and refrac- tory acute myeloid leukemia patients. A sensitive and rapid method has been developed and validated as per regulatory guideline for the simultaneous quantitation of enasidenib and its active metabolite, AGI-16903 in mice plasma using an LC- MS/MS. Enasidenib and AGI-16903 along with internal standard were extracted from mice plasma using simple protein pre- cipitation method. Chromatographic resolution of enasidenib, AGI-16903 and the internal standard (close analogue of AGI- 16903) was achieved on a Chromolith RP-18e column using 0.2 % formic acid:acetonitrile (15:85, v/v) as an eluent, which was delivered at a flow-rate of 1.2 mL/min. The MS/MS ion tran- sitions monitored were m/z 474.1→267.2, 402.1→188.1 and 421.0→146.1 for enasidenib, AGI-16903 and the internal standard, respectively. The linearity range was 1.01–3023 ng/ mL for both enasidenib and AGI-16903. The within-run and between-run accuracy and within-run and between-run preci- sion were in the range of − 2.29 to 2.72 (as one value is in negative side). and 4.65–9.82 %, respectively for enasidenib; 0.19–10.3 and 3.22–9.22 %, respectively for AGI-16903. Both enasidenib and AGI-16903 were found to be stable in stability (up to three freeze-thaw cycles and for long-term at − 80 °C for 30 days) and processed (bench-top for 6 h and in in-injector for 24 h) samples. Application of the validated method was shown in a pharmacokinetic study in mice.

Introduction
Acute myeloid leukaemia (AML) is most commonly occurring type of leukaemia [1]. Recently it was discovered that 20 % of AML pop- ulation shown mutations on isocitrate dehydrogenase 1 or 2 (IDH1/2) [2]. This finding has driven efforts towards the discovery of novel chemical entities, which can target the inhibition of mu- tated IDH1/2 to treat AML [3]. Targeted IDH1/2 inhibitors used as a single agent drug therapy or in combination with other antican- cer drugs. Enasidenib (AG-221, CC-9000, Idhifa®; Fig. 1), is a novel first-generation selective mIDH2 inhibitor for relapsed/re- fractory AML, with 40-fold higher selectivity against mIDH2 inhi- bition over wild-type enzyme [4]. Enasidenib has shown dose-de- pendent survival advantage in aggressive human AML xenograftmodel [5]. In humans, as a monotherapy, it is administered at a daily dose of 100 mg. It has shown approximately dose proportion- al increase in area under the concentration time curve (AUC: 1970 to 8035 ng × h/mL) from 50–450 mg in patients. Following 100 mg oral administration maximum concentration in plasma (Cmax:1.3 µg/mL) attained at ~4 h (Tmax). The metabolism enasidenib is mediated by several CYPs to produce a circulatory N-dealkylated active metabolite (AGI-16903). Protein binding of enasidenib and AGI-16903 was 98.5 and 96.6 %, respectively in human plasma. Enasidenib is majorly eliminated through feces (89 %). The oral bi- oavailability of enasidenib was found to be 57 % in humans [4].Until date, two LC-MS/MS methods were reported for quantita- tion of enasidenib [6, 7] and a single LC-MS/MS method for the si-multaneous quantitation of enasidenib and AGI-16903 [8]. The first LC-MS/MS method reported by Pang et al. (2018) was in rat plasma with a linearity range of 2–500 ng/mL. Extraction of enasidenib and imatinib (internal standard, IS) from rat plasma was achieved using simple protein precipitation. The chromatographic resolution of enasidenib and the IS was achieved on BEH C18 column using a bi- nary gradient program with a total run time of 2.5 min [6].

Subsequently, Dittakavi et al. (2018) reported [7] a mice dried blood spot (DBS) method for quantitation of enasidenib on LC-MS/MS with higher sensitivity over the previously published method along with addressing the demerits of previously published method [6]. Re- cently, Li et al. [8] reported simultaneous quantitation of enasidenib and AGI-16903 in Japanese and Caucasian subjects plasma as part of Phase I study to evaluate the pharmacokinetics and safety in these subjects. The reported LLOQ was 1.0 ng/mL (both the ana- lytes) using 100 µL plasma. Both enasidenib and AGI-16903 along with corresponding deuterated internal standards were extracted from plasma and the reconstituted sample was resolved on a C18 column using a ternary gradient mobile phase with a total run time of 5 min. The demerits of this paper are (i) no details were provid- ed on method validation and linearity range (though LLOQ was given) (ii) no details given on LC-MS/MS conditions and transition pair used for enasidenib and AGI-16903 (iii) attaining 1.0 ng/mL as an LLOQ with 100 µL plasma indicates that method is less sensitive(iv) usage of complicated gradient program (v) long run time etc. The aim of the present study is to develop and validate an LC-MS/ MS method using 10 µL of mice plasma with an LLOQ of 1.01 ng/ mL and its application to a pharmacokinetic study in mice.Materials and MethodsChemicals and reagentsEnasidenib (purity: 98 %) was purchased from Aaron, Shanghai, China. AGI-16903 ( Fig. 1; purity: 99.6 %) and internal standard [IS; 4-Chloro-N,6-bis-6-(trifluoromethyl)pryidin-2-yl)-1,3,5-tria- zin-2-amine; purity: 99.5 %; Fig. 1) were synthesized by Medici- nal Chemistry Group, Jubilant Biosys using literature information[9] and characterized using chromatographic (HPLC, LC-MS/MS) and spectral techniques (IR, UV, Mass, 1H and 13C-NMR) by the An- alytical Research Group, Jubilant Biosys. HPLC-grade acetonitrile, methanol and ethyl acetate were purchased from Rankem, New Delhi, India.

Dimethylsulfoxide was purchased from Sigma-Aldrich, Bangalore, India. HPLC-grade formic acid was purchased from Sig- ma-Aldrich, St. Louis, USA.Chromatography and MS/MS conditionsSciex 6500 triple quadrupole (Sciex, Redwood City, CA, USA) mass spectrometer was coupled to a HPLC (UFLC Prominence, Shimad- zu, Japan) was used for the analysis of enasidenib, AGI-16903 and the IS in the present study. The instrument was controlled using Analyst software (version 1.6.2). The HPLC was equipped with a degasser, binary pump, auto-sampler and a Chromolith RP-18e col- umn (100 × 4.6 mm), which was maintained at 40 ± 1 °C. Mobile phase is a mixture of 0.2 % formic acid and acetonitrile (15:85, v/v) was delivered at a flow rate of 1.2 mL/min into the mass spectrom- eter electro spray ionization chamber to resolve enasidenib, AGI- 16903 and the IS from the endogenous components of mice plas- ma within a total run time of 3.0 min. Quantitation of the analytes was achieved by MS/MS detection in positive ion mode by monitoring the transitions pair (Q1→Q3) of m/z 474.1→267.2, 402.1→188.1 and 420.1→146.1 for enasidenib, AGI-16903 and the IS, respectively. The interface temperature and ion spray voltage was 500 °C and 5500V, respectively. Quadrupole Q1 and Q3 were set on unit resolution. The dwell time was 100 msec. The source and compound param- eters for enasidenib, AGI-16903 and the IS are shown in Table 1.Preparation of stock solutionsIn order to prepare calibration curve (CC) and quality control (QC) samples two separate stock of enasidenib (1454 µg/mL) and AGI- 16903 (1069 µg/mL) were prepared in methanol:water (80:20, v/v). Primary stock solution of IS (100 µg/mL) was prepared in DMSO, which was subsequently used for prepare the working stock solu- tion. The prepared primary stock solutions of enasidenib, AGI-16903 and the IS were stable for 50 days at − 20 °C. The composite stock solution of enasidenib and AGI-16903 (made from the first primary stock solutions) was diluted appropriately with methanol:water (80:20, v/v) to prepare CC standards. From primary stock solution of the IS, working stock IS solution (50 ng/mL) was prepared in DMSO.

Working stock solutions of the analytes and the IS were stable for 15 days at 4 °C.Preparation of quality control samplesSecond primary stock solutions of enasidenib and AGI-16903 were used to prepare QCs for the determination of within-run and be- tween-run precision and accuracy for enasidenib and AGI-16903. The QCs were 1.01, 3.02, 1574 and 2267 ng/mL at LLOQ QC (lower limit of quantitation quality control), LQC (low quality control), MQC (medium quality control) and HQC (high quality control). Pre- pared QCs were stored at − 80 ± 10 °C until analysis.Sample preparationTo an aliquot of 10 µL plasma, 400 μL of methanol enriched with IS (100 ng/mL IS) was added and allowed to vortex mix for a period of 3 min on a Thermomixer (Eppendorf). Following this the contents in the micro-centrifuge tubes were centrifuged for 5 min at 14 000 rpm in a refrigerated centrifuge (Eppendorf 5424R) maintained at5 °C to remove the particulated matter and debris. Post centrifu- gation step from the clear supernatant 150 µL was pipetted out into a HPLC vial and 5.0 µL was used for analysis on LC-MS/MS.Method validation parameters were performed to meet the re- quirements based on FDA guideline [10]. Various parameters cov- ered under validation are – selectivity, carry over, recovery, matrix effect, linearity, precision, accuracy, stability studies (in-injector, bench-top, freeze/thaw and 30-day long-term at − 80 ± 10 °C), di- lution integrity and incurred samples reanalysis.Pharmacokinetic studyTwelve male Balb/C mice (weigh range: 27–31 g) were procured from Vivo Biotech, Hyderabad, India and housed at Jubilant Animal House facility (having 12/12 h light/dark cycles with controlled hu- midity and temperature) for a period of seven days (during this pe- riod mice had free access to feed and water) before performing pharmacokinetic study [approved by Institutional Animal Ethics Committee (IAEC/JDC/2018/158)]. Mice were fasted ~4 h on the day of pharmacokinetic study and before dosing. Mice received enasidenib orally [strength: 1.0 mg/mL (suspension formulation prepared using methyl cellulose and Tween-80); dose volume: 10 mL/Kg] at a dose of 10 mg/kg. Blood samples (25 μL) were col- lected at pre-determined time points (0.5, 1, 2, 4, 8, 10, 12, 24, 36, 48, 54 and 60 h) through tail vein (using Micropipettes) into poly- propylene tubes (having dipotassium EDTA as an anti-coagulant). Sparse sampling technique (n = 3 mice at each time point) was adopted during blood collection so that blood loss from each mouse was kept less than 10 %. Feed was provided to all mice 2 h post-dosing of enasidenib.

Results and Discussion
During initial chromatography optimization, we used the previous- ly optimized chromatographic conditions reported by us [7] with slight alteration in mobile phase composition [0.2 % formic acid:acetonitrile (15:85, v/v)] at a flow-rate of 1.0 mL/min for the elution of AGI-16903 along with enasidenib, but we found that the peak shape was very broad for AGI-16903 on Atlantis dC18 column (100 × 4.6 mm, 3 µM). As the IS was structurally very close to AGI- 16903, the peak shape for the IS was also broad. Subsequently, we tried several commercially available columns [Chromolith RP-18e (100 × 4.0 mm), X-Terra Phenyl (150 × 3.9 mm, 5 µM) and Kinetex HILIC (50 × 4.6 mm, 2.6 µM) and] and found that Chromolith RP 18e column was best suited to provide the base-line separation and symmetric peaks for enasidenib, AGI-16903 and the IS.As a next step, to obtain optimum ionization and sensitivity 100 ng/mL AGI-16903 and the IS solutions were individually inject- ed into mass spectrometer and electro-spray ionization (ESI) full scans were carried out in positive ion detection mode. Like enasidenib, both AGI-16903 and the IS gave very good intensity signals and formed protonated [M + H] + at m/z 402.1 and 421.0, respectively. Surprisingly, during fragmentation pattern in MS/MSmode we did not see the intense and consistent product ion i.e., Q3: m/z 267.2, which was selected as Q3 for enasidenib for AGI- 16903 and the IS ( Fig. 2a, b). Eventually, the most intense ion m/z 188.1 and 146.1 was picked up as Q3 for AGI-16903 and the IS, respectively for quantitation purpose. The postulated fragmen- tation pattern of AGI-16903 and the IS are shown in Fig. 2a, b. In this validation, we have used structurally closer analogue of AGI- 16903 as an IS, which helped us in overcoming the routine issues encountered in chromatographic (elution on column; reproducible recovery etc) and mass spectrometry parameters (ionization and matrix effect issues) optimization. Examination of the fragmenta- tion pattern of enasidenib and AGI-16903 suggested that there may be opportunity to further increase the sensitivity of the quan- titation by employing ion summation approaches which have been used for many drugs to achieve better sensitivity [11, 12].

Validation results of the methodSelectivityAs shown in Fig. 3a, b no significant interferences were found from endogenous components in drug-free mice plasma at the re- tention times of enasidenib and AGI-16903 at LLOQ concentration and the IS (50 ng/mL) indicating that the method is selective.Sensitivity and carry overThe lowest limit of reliable quantification for the analyte was set at the concentration of the LLOQ. The precision and accuracy at LLOQ concentration were found to be 10.3 and 103 % for enasidenib and8.60 and 102 % for AGI-16903. After the injection of highest cali- bration sample for either enasidenib or AGI-16903 no peak re- sponse was observed in the immediate blank plasma sample indi- cating that there was no carry-over.RecoverySelection of a proper extraction solvent plays very important role in getting the consistent and reproducible results for the analyte and the IS. By having cleaner samples, the life of the analytical col- umn will be extended and it will also help in achieving higher sen- sitivity on mass spectrometer. We tried both protein precipitation and liquid-liquid extraction trials for the recovery of enasidenib and AGI-16903 from mice plasma. In liquid-liquid extraction trials (ethyl acetate, dichloromethane, n-hexane and tert-methyl butyl ether) recovery of AGI-16903 was very low (ranged between 23.1–28.5 %) though enasidenib recovery was satisfactory (ranged between 43.7–60.3 %). Using protein precipitation with acetonitrile the re-covery for enasidenib and AGI-16903 was > 80 %. The recovery data for enasidenib and AGI-16903 are presented in Table 2.Matrix is a common barrier in the development of LC-MS/MS based assays, which causes either ion suppression/enhancement if it is not properly evaluated during method development [13, 14]. Six different lots of plasma samples, spiked with enasidenib and AGI- 16903 at LQC and HQC levels were analyzed to evaluate the matrix effect. The results have shown that the precision and accuracy for the analyzed samples were within acceptance range ( Table 2). These results indicate that neither ion suppression nor ion enhance- ment was observed for enasidenib, AGI-16903 and the IS.Calibration curveLinearity was evaluated by analyzing eight non-zero calibration standards: 1.01, 2.02, 20.2, 101, 504, 1008, 2016 and 3023 ng/mL.Calibration plot (analyte/IS peak area ratio vs the nominal concen- tration) data were subjected to statistical analysis using a linear re-gression model, showed the regression equation of y = 0.000494x + 0.00066 and y = 0.000494x + 0.00066 for enasidenib and AGI-16903, respectively. The average regression (n = 4) was found to be ≥ 0.997 for both the analytes. The accuracy ( %) and precision ( %RSD) observed for the mean of back-calculated concentrations for four calibration curves were within 94.1–110 % and 0.64–2.74 % for enasidenib and 94.1–110 % and 0.64–2.74 % for AGI-16903.A summary of accuracy and precision data for within- and between- run for enasidenib and AGI-16903 is listed in Table 2.

For enasidenib, the between-run accuracy ranged from −2.29 to 1.98 % and be- tween-run precision ranged from 6.49 to 9.82 %. Similarly, for AGI- 16903 the between-run accuracy ranged from 0.19–9.27 % and between-run precision ranged from 6.78–9.22 %. These results on both occasions (within- and between-run precision and accuracy) were found to be within the limits as specified in regulatory guideline indi- cating that the developed method is valid; precise and accurate for the quantification of enasidenib and AGI-16903 in mice plasma.The predicted concentrations for enasidenib at 3.02 and 2267 ng/ mL samples in processed (on bench-top for 6 h and in in-injector for 24 h) and stability studies (up to three freeze/thaw cycles and long-term for 30 days at − 80 ± 10oC) are shown in Table 3. All the results %RE were within 85-115 % and %RSD ± 15 %.The results of the dilution integrity support that the upper limit for enasidenib and AGI-16903 can be extended up to 9239 ng/mL. The precision and accuracy for 10-fold diluted samples was < 10 % for both the analytes.Incurred samples reanalysisTwelve samples selected for ISR (six samples each for enasidenib and AGI-16903) met the acceptance criteria. The back calculatedaccuracy values ranged between 89.5–106 and 92.5–108 % for enasidenib and AGI-16903, respectively from the initial assay val- ues ( Table 4) authenticating the reproducibility and ruggedness of the validated method.The sensitivity and specificity of the validated assay was found to be sufficient for accurately characterizing the pharmacokinetics of enasidenib and AGI-16903 (metabolized from enasidenib) in mice plasma following oral administration. Enasidenib and AGI-16903 (metabolized from enasidenib) were quantified up to 60 and 36 h post-dosing. The time versus plasma concentration profiles for enasidenib and AGI-16093 are presented in Fig. 4. Following oral administration of enasidenib maximum concentration (Cmax) in plasma (1.47 µg/mL) was achieved at 4.0 h (Tmax, time to reach Cmax).

The AUC0-∞ (area under curve from time zero to infinity) wasfound to be 26.4 µg × h/mL with a terminal half-life (t½) of 6.01 h. On other hand, AGI-16903 attained Tmax at 8.0 h. The AUC0-∞ and t½ for AGI-16903 were 227 ng × h/mL and 12.4 h, respectively. In order to investigate the huge difference in AUC0-∞ and Cmax be- tween enasidenib and AGI-16903, we have determined the meta- bolic stability in mice and human liver microsomes using the pro- cedure reported by Gabani et al [15]. The % remained was 94.7 and74.4 in mice liver microsomes and 97.6 and 80.7 in human liver mi- crosomes, for enasidenib and AGI-16903, respectively at the end of 30 min incubation time. These results clearly indicate that the percent remained for enasidenib was higher than AGI-16903 in both matrices and support for the higher exposure for enasidenib in mice (present study) and in humans as reported by Li et al [8]. We have also digitalized using DigitizeIt (version 2.0.0; accessed on 27 July 2019 available at https://www.digitizeit.de) the report- ed plots of enasidenib and AGI-16903 in healthy Caucasian and Jap- anese subjects post administration 100 mg oral dose of enasidenib[8] and found that enasidenib and AGI-16903 reaches the reported LLOQ (1.01 ng/mL) by 14th and 10th day, respectively which is con- trary to the reported (up to 28th day). This indicates that the meth- od used by Li et al. [8] for quantitation of enasidenib and AGI-16903 may not be sensitive enough to determine the plasma concentra- tions of enasidenib and AGI-16903 till the reported time despite using higher volume of plasma (100 µL).

Conclusion
We developed and validated an LC-MS/MS method for the deter- mination of enasidenib and AGI-16903 in mice plasma. The meth- od is selective, linear, accurate and precise in the range of 1.01– 3023 ng/mL. The developed method has several advantages viz., low plasma volume (10 µL), improved sensitivity, reduction in large number of animal use by the avoidance of sample intensive com- posite curves, allows serial blood sampling from mice, useful in tox- icokinetic studies etc. The utility of this assay was demonstrated through its successful application to a mice pharmacokinetic study.