Chemicals and instrumentation

Standard rivastigmine tartrate {3-[(1S)-1-(Dimethylamino) ethyl] phenyl N-ethyl-N-methylcarbamate tartrate, purity ≥ 98%}, NAP 226-90 {3-[(1S)-1-(Dimethylamino) ethyl] phenol, purity ≥ 98%}, and the isotopically labeled internal standard (IS) (rivastigmine-d₄, purity ≥ 98%) were purchased from Toronto Research Chemicals (Toronto, Ontario, Canada). Acetonitrile, methanol, and formic acid (99%) were purchased from Wako Pure Chemical Ind. Ltd. (Osaka, Japan). Acetonitrile, methanol, ultrapure water, and formic acid were of LC-MS grade, and all other reagents were of analytical grade. Samples were analyzed using an Alliance 2695 System and a triple-stage quadrupole mass spectrometer (Qattro microAPI Tandem Quadrupole System). The mass spectrometer was tuned automatically to rivastigmine, NAP 226-90, and the IS using the MassLynx V4.0 software package (Waters).

Stock and working solutions

Stock solutions of both rivastigmine and NAP 226-90, at concentrations of 0.5 mg mL⁻¹, were prepared by weighing them in separate measuring flasks and dissolving them in methanol. The quality control (QC) stock solution was prepared separately from the solution intended for the calibration curve. Stock solutions were stored at −20 °C. The IS stock solution, at 0.5 mg mL⁻¹, was also prepared in methanol and stored at −20 °C. It was then diluted with acetonitrile to give a single IS working solution of 100 ng mL⁻¹. Calibration solutions at concentrations of 1–100 ng mL⁻¹ and 0.5–50 ng mL⁻¹ for measuring rivastigmine and NAP 226-90, respectively, were prepared by diluting the calibration stock solution in 0.01 N hydrochloride (HCl). For both rivastigmine and NAP 226-90, QC solutions representing their lower limits of quantification (LLOQs), as well as low, medium, and high QC solutions, were prepared by diluting the QC stock solution with 0.01 N HCl to reach concentrations of 1, 3, 15, and 75 ng mL⁻¹ for rivastigmine and 0.5, 1.5, 7.5, and 37.5 ng mL⁻¹ for NAP 226-90.

Sample preparation

The plasma samples from healthy male rats were prepared via protein precipitation with acetonitrile. To measure rivastigmine and NAP 226-90 concentrations, the QC stock solution and blank plasma for calibration (90 µL total) were placed in a 1.5 mL polypropylene tube. Then, this mixture was added to 10 µL each of eight calibrating solutions – the LLOQ, low, medium, and high QC solutions for both rivastigmine and NAP 226-90, prepared as described in previous section followed by 300 µL of the IS (100 ng mL⁻¹) prepared in acetonitrile. Afterwards, the samples were vortex-mixed and centrifuged for 10 min at 14,000×g in a 0.5 mL polypropylene tube. The supernatants were transferred into 1.5 mL polypropylene tubes and evaporated to dryness (dinitrogen [N₂] flow, 40 °C). The solid residues were reconstituted in 100 µL of the mobile phase at initial conditions (0.1% formic acid–acetonitrile 80:20, v/v), vortex-mixed, and again centrifuged for 10 min at 14,000×g. Finally, the supernatants (90 µL) were transferred into glass vials prior to being injected into the HPLC-MS/MS system.

Liquid chromatography

The analytes were chromatographically separated via an InertSustain C18 column (3.0 µm, 1.5 × 50 mm). The mobile phase used for this chromatographic separation consisted of 0.1% aqueous formic acid (solvent A) and acetonitrile. The total analysis time was 5 min per injection, and the flow rate was 0.2 mL min⁻¹. The gradient program of mobile phase in the LC is shown in Table 1. The column and sample temperatures were kept at 20 °C throughout all experiments. The injection volume for each sample was 10 μL.

Mass spectrometry

Samples were ionized using the following parameters: electrospray voltage: 3.5 keV; cone voltage: 25 eV; source temperature: 120 °C; cone gas flow (N₂): 50 L h; desolvation gas flow (N₂): 600 L h; and desolvation temperature: 350 °C. Selective reaction monitoring (SRM) was carried out using N₂ as a collision gas, and the MS/MS transitions monitored in the positive ion mode for rivastigmine, NAP 226-90, and the IS were m/z 251.0 → m/z 206.0 at 15 eV; m/z 166.0 → m/z 121.0 at 15 eV; and m/z 255.0 → m/z 206.0 at 15 eV, respectively. A dwell time of 200 msec was given for each transition.

Validation of the method

The validation procedure was carried out to determine whether the proposed method met the acceptance criteria for bioanalytical methods from the FDA [6] and the EMA [7].

Accuracy and precision

The LLOQs for rivastigmine and NAP 226-90 were the lowest analyte concentrations on the calibration curve that could be determined with precision < 20%, and accuracy < 40%. The response of the blank plasma sample should be at least five times lower than that of a spiked sample at the LLOQ. Three validation batches, each containing eight calibration and 24 QC samples at different concentrations (LLOQ, low, medium, and high QCs in sextuplicate preparation), were analyzed. Precision and accuracy were calculated and expressed as coefficient of variation (CV%), relative error (RE%), and total analytical error (%) for each analytical batch (within-run) and across the three validation batches (between-run).

Recovery rate

Blank plasma was spiked at three different QC concentrations and extracted (A). Additionally, three QCs were prepared in the extracted blank plasma (B). Theses samples were prepared in triplicate and subjected to HPLC-MS/MS. The peak areas of samples A and B were obtained, and the extraction recovery rate (%) was calculated as A/B × 100.

Matrix effect

The corresponding peak areas of the three QCs prepared in extracted blank plasma (B) and three other QCs diluted in matrix-free LC eluent (C) were obtained in triplicate. Then, the matrix effect (%) on ionization efficacy was calculated as B/C × 100.

Selectivity

Selectivity was evaluated by comparing chromatograms of the plasma samples obtained from six healthy male rats, without adding rivastigmine, NAP 226-90, or the IS. The baseline signals at the expected analyte retention times were evaluated as interfering peaks.

Linearity

The calibration samples and analyte-specific SRM quantifier transitions were used to construct calibration curves for the rivastigmine and NAP 226-90 measurements. Peak areas of the respective analytes were divided by those of the IS. Weighted linear regression (1/x) was performed for each analytical batch. Method linearity was validated by performing simple linear regression between the introduced concentrations and the back-calculated concentrations via the selected calibration curve.

Application of the HPLC-MS/MS method

The developed HPLC-MS/MS method was successfully applied to determine the plasma concentrations of rivastigmine and NAP 226-90, following intravenous rivastigmine administration, in healthy male Sprague Dawley rats (n = 6) (CLEA Japan, Inc., Japan). The animals were housed in temperature- and humidity-controlled rooms with a 12 h/12 h light/dark cycle. They were kept on a standard chow diet and were given free access to water. All animal experiments were carried out according to the guidelines of the Institutional Animal Ethical Committee (approval no: 2021-A-18) at Fukuyama University. Rivastigmine (1 mg kg⁻¹) was intraperitoneally administered as a suspension in saline to three of the sampled rats. This rivastigmine dose was equivalent to the clinical dose, calculated according to FDA guidelines for human-to-animal dose extrapolation. Blood samples were drawn from the jugular vein and collected in heparinized polypropylene tubes at 30 min, 60 min, 90 min, 180 min, 6 h, and 12 h post-dose under light pentobarbital anesthesia. After each blood sample was taken, the plasma was separated by centrifuging the blood at 2,000×g for 20 min. The extracted plasma was kept frozen at −80 °C until analysis.

Plasma concentration-time profile and pharmacokinetics parameters of rivastigmine and NAP 226-90 were calculated by moment analysis. The pharmacokinetic parameters includes the area under the plasma concentration-time curve from zero to infinity (AUC_0-∞); the mean residence time (MRT); the terminal elimination half-life (t _1/2); the distribution volume on steady state (Vd_ss) and total clearance (CL).

Mass spectrometric and chromatographic characteristics

During MS tuning, [M+H]⁺ signals for rivastigmine and NAP 226-90 were more intensely detected in the positive ion mode. Figure 1 shows the precursor and product ion mass spectra of the [M+H]⁺ ion for rivastigmine and NAP 226-90. Sensitive detection via SRM was used to choose identical fragment ions for rivastigmine, NAP 226-90, and the IS. The identical fragment ions of rivastigmine and the IS were m/z 206.0, and NAP 226-90 was m/z 121.0, respectively.

View Full Size

Full-scan product ion mass spectra in positive mode of (a) rivastigmine, (b) NAP 226-90, (c) rivastigmine-d₄ (IS)

Citation: Acta Chromatographica 34, 4; 10.1556/1326.2021.00989

• Download Figure
• Download figure as PowerPoint slide

The chromatographic conditions were optimized to rapidly and efficiently separate rivastigmine, NAP 226-90, and the IS from the plasma components. Figure 2 shows the chromatographic characteristics of rivastigmine, NAP 226-90, and the IS for blank plasma and for the LLOQs. Under optimized chromatographic conditions, the retention times of rivastigmine, NAP 226-90, and the IS were all about 1.3 min.

View Full Size

Selective ion chromatograms of (a) Blank plasma; (b) Plasma spiked with rivastigmine at concentration of LLOQ (1 ng mL⁻¹); (c) Plasma spiked with NAP 226-90 LLOQ at concentration of LLOQ (0.5 ng mL⁻¹); (d) Plasma spiked with IS at concentration of 100 ng mL⁻¹

Citation: Acta Chromatographica 34, 4; 10.1556/1326.2021.00989

• Download Figure
• Download figure as PowerPoint slide

Method validation

No interfering peaks derived from the plasma matrix were observed in the plasma samples collected from the six rats, which indicated that the method had acceptable selectivity (Fig. 2). For the calibration curves, the regression coefficient (r ²) of the rivastigmine concentrations from 1 ng mL⁻¹ to 100 ng mL⁻¹ was ≥0.9982, and that for the NAP 226-90 concentrations from 0.5 ng mL⁻¹ to 50 ng mL⁻¹ was ≥0.9983, confirming the linearity of the method’s eight-point calibration curves. The LLOQs determined for rivastigmine and NAP 226-90 fulfilled the FDA and EMA criteria for both within-run and between-run accuracy (88.8% and 106.6%, respectively, for rivastigmine; 98.1% and 98.1%, respectively, for NAP 226-90) and precision (≤14.60% and 7.53%, respectively, for rivastigmine; ≤ 11.31% and 4.53%, respectively for NAP 226-90) (Tables 2 and 3). All analyte signals of the LLOQs were at least five times higher than the signals detected in plasma samples without rivastigmine and NAP 226-90. Tables 2 and 3 present within-run and between-run accuracy and precision data for rivastigmine and NAP 226-90 measurement. In the rivastigmine measurements, the within-run the accuracy for the three QCs was 101.1–107.8%; and the precision was 4.34–11.27%. The between-run the accuracy was 103.2–114.9%; and the precision was 1.66–7.56%. For the NAP 226-90 measurements, the within-batch the accuracy for the three QCs was 99.1–105.8%; and the precision was 99.1–105.8%. The between-run the accuracy was 96.7–102.1%; and the precision was 3.44–7.78%.

The rivastigmine extraction recovery rates from the rat plasma in the low, medium, and high QCs were 107.4% ± 12.03%, 101.2% ± 9.82%, and 103.6% ± 3.47% (mean ± standard deviation), respectively. The NAP 226-90 extraction recovery rates from the rat plasma in the low, medium, and high QCs were 109.8% ± 8.05%, 102.1% ± 10.23%, and 91.9% ± 11.62%, respectively. The matrix effects of rat plasma for rivastigmine in the low, medium, and high QCs were 87.3% ± 12.04%, 95.4% ± 8.88%, and 103.8% ± 3.47%, respectively. The matrix effects of rat plasma for NAP 226-90 in the low, medium, and high QCs were 106.8% ± 8.05%, 102.3% ± 10.21%, and 99.1% ± 11.62%, respectively. There were no significant differences in matrix effect among the three QCs for rivastigmine and NAP 226-90 measurements (Table 4).

Protein precipitation was chosen as the method for extracting rivastigmine and NAP 226-90 from the rat plasma. However, previous reports have employed protein precipitation extraction via SPE. SPE and LLE typically provide cleaner extracts with fewer matrix effects, minimum ion suppression, and a low propensity for building backpressure in the chromatographic column. Nevertheless, in our preliminary investigation using SPE or LLE as the extraction procedure, extracts with satisfactory analyte recoveries were not obtained. Therefore, in this study, the procedure was simplified, and protein precipitation with acetonitrile was used, as it costs less and allows for faster sample preparation. In addition, the amount of required plasma was reduced from 100 µL [5] to 90 µL. The extraction recoveries ranged between 91.9% and 109.8%. The repeatability was also good for rivastigmine and NAP 226-90, when an analyte CV < 12.03% was considered.

Plasma concentrations of rats receiving rivastigmine administration

Figure 3 shows the profiles of rivastigmine and NAP 226-90 concentrations, as measured by the HPLC-MS/MS method developed in this study. All measured drug concentrations fell within the range (rivastigmine; 2.1–80.2 ng mL⁻¹, NAP 226-90; 1.0–11.2 ng mL⁻¹). The corresponding values of the pharmacokinetic parameters by non-compartment model are shown in Table 5.

View Full Size

Comparison of plasma concentration profiles for the determination of rivastigmine (a) (n = 6) and NAP 226-90 (b) (n = 6) by the developed HPLC–MS/MS in rats

Citation: Acta Chromatographica 34, 4; 10.1556/1326.2021.00989

• Download Figure
• Download figure as PowerPoint slide

Regarding rivastigmine measurement in rat plasma, a previous report using LC-MS [5] has found a range from 10–100 pmol mL⁻¹ (2.6–26 ng mL⁻¹). In another report, the range is 0.1–100 ng mL⁻¹ [8] The concentration range of the calibration curve in the current research stretched from 1–100 ng mL⁻¹, and in vivo pharmacokinetic study has suggested that plasma concentrations of rivastigmine are dose proportional. Karasova et al. [8] summarizes the results of previously published investigations, reporting the values of pharmacokinetic parameters and the plasma and brain concentrations of rivastigmine after single intramuscular injections. According to that report, the C _max of rivastigmine, after a single injection of 137 μg kg⁻¹ in healthy rats was 4.96 ± 0.67 ng mL⁻¹. Recently, many reports have demonstrated the efficacy and safety of maintenance doses (9 or 18 mg/body) of rivastigmine for dementia [9, 10]. In the current study, rivastigmine doses were calculated according to recommended human doses and were given to each rat via injections of 1 mg kg⁻¹ or 2 mg kg⁻¹. Taking these findings into consideration, the calibration curve range for rivastigmine determination might be set with an upper limit of 100 ng mL⁻¹. The current method of rivastigmine measurement, therefore, has the advantage of a wider calibration range compared to the methods applied in previous research.

A study using LC-MS to measure NAP 226-90 in rat plasma has found a range from 10–100 pmol mL⁻¹ [2.6–26 ng mL⁻¹] [5]. However, pharmacokinetics data on NAP 226-90, especially in terms of rat plasma targeting, have not been previously reported. Preclinical and clinical data about NAP 226-90 are limited to the results of acetylcholinesterase/butyrylcholinesterase inhibition studies, which do not report the pharmacokinetic profile of NAP 226-90. The LLOQ of our method was 0.5 ng mL⁻¹, which was smaller than that of a previous report [5]. In our pharmacokinetics study, C _max and C _min of NAP 226-90 was 8.8–11.1 ng mL⁻¹ and 1.0–3.2 ng mL⁻¹, especially when a rat is given rivastigmine at higher doses of more than 1 mg kg⁻¹. The range of the calibration curve for the current method of NAP 226-90 measurement is 0.5–50 ng mL⁻¹, which fulfills the recommended range, including the LLOQ and the upper limit.