Simple and high sample throughput LC/ESI-MS/MS method for bioequivalence study of prazosin, a drug with risk of orthostatic hypotension

Abstract Objective The study aimed to develop a rapid, simple and sensitive LC/ESI-MS/MS method to measure prazosin concentration in human plasma and apply bedside sampling in bioequivalence study of two prazosin tablets to resolve the adverse effect of orthostatic hypotension. Significance The LC/ESI-MS/MS prazosin method was highly sensitive and selective. Bedside sampling reduced the orthostatic hypotension incidence and subject dropout rate. Methods After sample preparation, prazosin and terazosin (IS) were detected on mass spectrometer operating in multiple reaction monitoring mode using positive ionization. Mobile phase flow rate was set at 0.40 mL/min with sample run time of 1.75 min. The bioanalytical method was validated as per EMEA and FDA guidelines. Bedside sampling was performed in bioequivalence study for the first 4 h after dosing. The three primary pharmacokinetic parameters, C max, AUC0- t and AUC0-∞ and 90% confidence interval were determined. Results The small injection volume of 1 μL minimized instrumentation contamination and prolonged the analytical column lifespan. Linearity was obtained between 0.5 and 30.0 ng/mL, with coefficient of determination, r 2 ≥ 0.99. The mean extraction recovery of prazosin and IS was >92%, with precision value (CV, %) ≤ 10.3%. Only two orthostatic hypotension adverse events were reported. The two prazosin formulations were found to be bioequivalent. Conclusion The LC/ESI-MS/MS method has shown robustness and reliability exemplified by the incurred sample re-analysis result. Bedside sampling should be proposed for bioequivalence or pharmacokinetic studies of drugs demonstrating adverse event of orthostatic hypotension.

One of the serious common adverse effects of prazosin was orthostatic hypotension or postural hypotension [6][7][8]. The adverse effect is caused by the rapid increase in hydrostatic pressure in the arterial vessels of the lower body region when standing up, causing dizziness, lightheadedness or faint [8]. Orthostatic hypotension has been identified as one of the risk factors for falls [9][10][11][12], fractures [13], myocardial ischemia [10,14], cognitive impairment [15] and mortality [10].
The study aims to report a simple, sensitive and high sample throughput by LC/ESI-MS/MS method for determination of prazosin concentration in plasma samples, a drug with common adverse effect of orthostatic hypotension, in Malaysian population. The small 1 lL injection volume minimized the instrumentation contamination and prolonged the analytical column lifespan. The LLOQ of 0.5 ng/mL was below 5% of the mean C max , meeting the guideline requirement. provided by Y.S.P. Industries (Selangor, Malaysia). HPLC grade formic acid, acetonitrile and methanol were supplied by Merck (Darmstadt, Germany). Blank plasma with anticoagulant of dipotassium ethylenediaminetetraacetic acid (K 2 EDTA) was obtained from i-DNA Biotechnology (Kuala Lumpur, Malaysia). Purified water was taken from Thermo Fisher Scientific Ultrapure water system (Pittsburgh, PA, USA).
Chromatographic separation was carried out using an Poroshell 120 EC-C18 -Fast LC (Agilent, USA) analytical column (100 Â 2.1 mm; 2.7 lm particle size) fitted with a guard (UHPLC Guard Poroshell 120 EC-C18, 5 Â 2.1 mm ID, 2.7 lm particle size). The mobile phase was consisted of 0.1% formic acid and acetonitrile (35:65, v/v) run at a flow rate of 0.40 ml/min. The column oven and autosampler temperatures were 30 C and 15 C with the injection volume of 1 lL.
A tandem mass spectrometer with electrospray ionization (ESI) as ion source operated in positive ionization mode was used to detect prazosin and IS. The detector energy, conversion dynode voltage and interface bias/capillary voltage were À2120, 6000 and 4500 V, respectively. The temperature of desolvation line (DL) and heating block were 250 C and 400 C. Nitrogen gas was utilized as drying gas and nebulizing gas with gas flows of 15 and 3 L/min, respectively. Collision argon gas was set at 230 kPa. Two multiple reaction monitoring (MRM) transitions for quantification and confirmation were used for determination of prazosin and IS. The optimized MRM transitions for prazosin and IS were tabulated in Supplementary data S1.

Preparation of standard and sample solutions
Prazosin and IS stock standard solutions at concentration of 20.0 lg/mL were prepared in water and acetonitrile (1:1, v/v). Stock or other working standard solutions were diluted with water and acetonitrile (1:1, v/v) to prepare prazosin ( The plasma calibration standards of prazosin in the range of 0.5 À 30.0 ng/mL were prepared in blank plasma sample using appropriate working standard solutions. Another freshly prepared prazosin stock standard solution was used to prepare QC working standard solutions at 5.0, 50.0 and 250.0 ng/mL, for preparation of QC plasma samples, comprising of lower limit of quantification (LLOQ), low QC (LQC), medium QC (MQC) and high QC (HQC), at 0.5, 1.5, 15.0 and 22.5 ng/mL, respectively.

Sample preparation
An aliquot of 250.0 lL of plasma sample was measured accurately into a 2-mL microcentrifuge tube, followed by the addition of 50 lL of terazosin IS working standard solution (200.1 ng/mL) and 750.0 lL of acetonitrile. The sample was vortexed (Heidolph REAX 200, Schwabach, Germany) for 30 s and centrifuged (Eppendorf AG Mini Spin Plus, Stevenage, UK) at 9676.8 g for 5 min. A nylon membrane syringe filter (0.2 lm, 17 mm filter, Thermo Scientific, USA) was used to filter the supernatant.

System suitability
The drift of LC/ESI-MS/MS system for all the method validation runs was monitored using system suitability testing. System suitability samples were made up of five plasma samples at 21.0 ng/ mL, a solvent and an LLOQ. The conduct and acceptance criteria of system suitability test were reported in earlier studies [26,27]. Initially, the five replicates of plasma samples at 21.0 ng/mL were injected into the system to obtain a CV (%) of <6%. A solvent and LLOQ were injected right after the five replicates of these samples to assess the carryover of the method where the carryover should be less than 20% and 5% of the peak response of analyte and IS, respectively. A set of 21.0 ng/mL plasma samples were injected after every forty injections and at the end of the run. The peak area ratio of each set of these samples should be within ± 15% of the mean peak area ratio of the first five plasma samples at 21.0 ng/mL. Solvent and LLOQ were injected prior to the end of each run to ensure there was no carryover during the run.

Method validation
Method validation of prazosin was conducted as per European Medicines Agency Guideline on Bioanalytical Method Validation [28] except recovery which was performed as per Food and Drug Administration Bioanalytical Method Validation Guidance for Industry [29]. Specificity and selectivity parameter were evaluated using blank human plasma samples (without prazosin and IS) from six different subjects to check for interferences at the retention times of prazosin and IS. The sensitivity was determined at LLOQ level using six different subjects' blank human plasma. Calibration curve of prazosin was constructed with eight non-zero calibration standards at 0.5, 1.0, 5.0, 10.0, 16.0, 20.0, 26.0 and 30.0 ng/mL, calculated using least square linear regression with a weighting factor of 1/x 2 . Residual effect was assessed by analyzing a processed blank human plasma sample after injection of upper limit of quantification (ULOQ) in each bioanalytical method validation run. The accuracy and precision of within-run and between-run were assessed in a single run and three different runs on three separate days using six sets of LLOQ and QC samples (0.5, 1.5, 15.0 and 22.5 ng/mL). One set of system suitability, one set of plasma standard calibration curve and 33 replicates of QC samples were injected in one single run to determine the accuracy and precision of QC samples in a size ! study samples in bioanalytical run, which was known as extended-run precision and accuracy. Matrix effect was compared as the peak area ratio of prazosin spiked after plasma deproteinization with acetonitrile to those prepared in water and acetonitrile (1:1, v/v) using LQC and HQC samples. For dilution integrity, prazosin plasma samples at concentrations of 25.0 ng/mL (two-fold) and 5.0 ng/mL (10-fold) were used after dilution of prazosin plasma sample at 50.0 ng/mL with blank human plasma. The recovery parameter was evaluated by comparing the peak areas of analytes and IS spiked after plasma extraction to those spiked after plasma extraction using six determinations at LLOQ and QC samples.

Stability studies
Prazosin plasma sample stability was evaluated at three replicates for LQC and HQC. The short-term/bench-top stability was determined at room temperature (25 ± 4 C) for a period up to 24 h. Post-preparative stability in auto-sampler at 15 ± 3 C was evaluated until 48 h. Freeze and thaw stability was assessed after seven freeze and thaw cycles. Long-term stability in freezer was determined at À20 ± 10 C for 30 days. A freshly prepared plasma calibration curve was constructed to quantitate the concentration of QC stability samples and the calculated concentrations were compared against the nominal concentration of the QC samples.
The stability of prazosin and terazosin IS stock standard solutions was determined after storing at room-temperature (25 ± 4 C) and in chiller (5 ± 3 C) for 30 days. The stock solution of prazosin was appropriately diluted to LQC and HQC concentrations and assessed against new calibration curve from new stock standard solutions. IS stock solution stability was determined at 200.1 ng/mL concentration against a freshly prepared IS at the same concentration. The peak area responses of both samples were compared.

Bioequivalence study
A two-treatment, two-period, two-sequence, randomized, open label, single-dose, two-way crossover oral bioequivalence study of two prazosin products, Minison tablet (1 mg prazosin, Y.S.P. Industries, Malaysia) and Minipress tablet (Pfizer Australia Pty. Ltd., Australia) in 30 healthy subjects under fasting conditions was conducted, with a washout period of 7 days. The study protocol was approved by the Medical Research and Ethics Committee (MREC) Ministry of Health, Malaysia. The subjects were informed about the possible risks and benefits of joining the study and signed written informed consent form before participating in the study. Each subject was dosed with two tablets (2 Â 1 mg) of either test or reference product.
Male subjects between 18 and 55 years old, Body Mass Index (BMI) from 18.5 to 30.0 kg/m 2 , in good health at screening, no history of hypersensitivity or allergy to prazosin or any other related drugs, willing to give written informed consent were the inclusion criteria of the study. The subjects were refrained from taking vitamin supplementation, herbal remedies, natural products or medications 7-day before admission, hospitalized 30 days before consent taking, drinking alcoholic beverages 24 h or red wine 7-day before product dosing, blood donation ! 500 ml 90-day before consent taking, involvement in any other clinical trial and the last drawn blood sample 60-day before taking the first dose of studied product.
The blood samples of the study were drawn before dosing (0 h) and at 0.25, 0.50, 0.75, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 5.0, 6.0, 8.0, 10.0, 12.0 and 16.0 h after product dosing. Blood samples were drawn at the bedside for the first 4 h. The plasma samples were stored in freezer at À20 ± 10 C until analysis. Adverse events were observed and recorded in the case report form throughout the study.

Pharmacokinetic parameters and analysis
The values of C max (peak plasma concentration) and T max (time to achieve peak plasma concentration) were obtained from the prazosin plasma concentration versus time profiles for each subject. AUC 0-t (area under the plasma concentration-time curve) and AUC 0-1 (area under the curve from time zero to infinity) were calculated from prazosin plasma concentration versus time profiles. A linear-logarithmic trapezoidal method was used to calculate AUC 0-t . The last measurable prazosin plasma concentration was divided by elimination rate constant (k e ) to calculate the AUC t-1 . The k e was calculated from the slope of the prazosin plasma concentration-time profile after logarithmic transformed. The elimination half-life was calculated using 0.693/k e [30][31][32]. Values below LLOQ were put as zero (0) for calculation purposes.
Incurred sample re-analysis (ISR) was performed by choosing at least 10% of the total number of analyzed samples. A total of four points (two points per period) were selected from each subject, plasma samples at C max and elimination phase (within four times of LLOQ level). The ISR results were compared with the data obtained during the first analysis.

Statistical analysis
The primary pharmacokinetic parameters, C max , AUC 0-t and AUC 0-1 , were transformed to natural logarithm scale. An analysis of variance (ANOVA) with geometric least square mean procedure was used to distinguish effects due to subject within sequence, period and treatment for the ln-transformed values of C max , AUC 0-t and AUC 0-1 of the two products [30,33]. Bioequivalence was concluded if the 90% confidence interval (CI) of the geometric mean ratios of C max , AUC 0-t and AUC 0-1 of test/reference product fell between 0.80 and 1.25 (80.00 À 125.00%). Two one-sided t-tests were used to test for bioequivalence.

Method validation
The coefficient of variation (CV, %) of first five replicates of high system suitability samples was 4.3%, indicating that the system reached sufficient equilibration before the start of method validation runs. The percentage of deviation was within the acceptance criteria with the deviation value 13.6%. Carry over was not found in neat solvent after injecting high system suitability samples.
Six subjects' blank plasma showed absence of interfering peak at or near the retention times of prazosin and IS (Figure 1). The signal to noise (S/N) ratios of prazosin and IS at LLOQ level (0.5 ng/mL) were more than 8 and 70 in the study. Prazosin and IS retention times ranged from 0.52 to 0.57 min, respectively.
Prazosin calibration curve was linear for concentration ranged from 0.5 to 30.0 ng/mL. The weighting factor of 1/[prazosin concentration] 2 , was selected because it provided the most suitable approximation of variance and the highest accuracy for all nonzero calibration points. The accuracy (RE, %) of the back-calculated concentration for each plasma calibration point of prazosin varied from À9.0% to 14.8%. All calibration curves reported r 2 (coefficient of determination) > 0.99. Figure 2 shows the mean equation of calibration curves was y ¼ 0.004703 x (± 0.002381) þ 0.010700 (± 0.006265). After injection of ULOQ calibrator, no peak was detected in blank sample at prazosin and IS retention times indicating that no residual effect was observed in the method. Table 2 shows the precision and accuracy results. The prazosin within-run accuracy (RE, %) and precision (CV, %) of LLOQ and QC samples ranged from À4.0% to 8.0% and 3.5% to 12.5%, respectively. The prazosin between-run accuracy (RE, %) and precision (CV, %) values ranged from À4.0 to 4.0 and 5.2 to 10.9, respectively. The extended run precision (CV, %) of 2.6 À 5.9% and accuracy (RE, %) results of 2.0 À 10.0%, indicated that the method would be able to run a total of $91 plasma samples with system suitability samples (n ¼ 7), plasma sample calibration curve (n ¼ 10), two sets of volunteers' samples (n ¼ 68) and two sets of QC samples (n ¼ 6), with high degree of confidence.
The precision (CV, %) of IS normalized matrix factors of prazosin at two QC levels using six different human plasma lots were 6.8% and 8.5%, indicating that current extraction method was free of matrix effect. The mean extraction recovery value of prazosin and IS was more than 92% with precision (CV, %) value 10.3%.
The 2-fold and 10-fold dilution gave mean accuracy of 108.6% and 104.2% with precision (CV, %) values 3.0% and 5.2%. Accuracy and precision values of the dilution integrity test were within the guidelines requirements. The result of dilution integrity is shown in Table 3.

Stability studies
Prazosin stability in plasma sample under different studied conditions are shown in Table 4. The plasma samples of prazosin were stable under all tested conditions (short-term/bench-top, post-preparative in autosampler, freeze-thaw cycles and long-term stability) with reported accuracy values ranged from 92.3% to 113.3%. The stock standard solution of prazosin was stable for a month at room temperature (104.7 À 107.3%) and in the chiller (96.2 À 112.7%). IS stock solution was stable for 1 month at room temperature and in chiller, with accuracy value >92%.

Application to bioequivalence study and ISR
After validation, the method was applied to quantify prazosin concentrations in plasma samples from a bioequivalence study of two prazosin formulations. A total of 30 healthy subjects were recruited in the study. Four subjects were withdrawn from the study (one in first period and three in second period) and 26 subjects completed the study. Two out of the four subjects were withdrawn due to the side effect of orthostatic hypotension, while the other two subjects dropped out voluntarily. Figure 3 shows the mean prazosin plasma concentration-time curves. Table 5 displays the summary of the main prazosin pharmacokinetic parameters of the bioequivalence study. The 90% CI of the geometric mean ratio of test over reference were 0.91 À 1.10 for C max , 0.93 À 1.10 for AUC 0-16 and 0.92 À 1.10 for AUC 0-1 , which were within the bioequivalence limit of 0.80 À 1.25. The ANOVA results showed that there was no statistically significant difference (p > 0.05) between Minison and Minipress tablets in the ln-transformed C max , AUC 0-t and AUC 0-1 values for the treatment effect. The intra-subject coefficient of variation (CV%) values obtained from the ln-transformed AUC 0-t , AUC 0-1 and C max values were 18.20, 17.96 and 20.49%, respectively. The inter-subject coefficient of variation (CV%) values obtained from the ln-transformed AUC 0--t , AUC 0-1 and C max values were 22.91%, 22.34% and 18.01%, respectively. The power of the test in detecting a difference of 20% between test and reference products based on significance level of 0.05, using 26 healthy subjects, was more than 80% for AUC 0-t and AUC 0-1 values.
ISR study was done to verify the reliability of the first analyzed prazosin concentrations in subjects' plasma samples. More than ten percent of all samples were chosen and analyzed in separate runs on different days. The ISR results were within 20% from the initial value for 92.0% of prazosin samples, proving the reproducibility of the method where the value must be 20% for at least of the plasma samples [28,29].

Optimization of chromatographic conditions and sample preparation
The obtained precursor molecular ions [M þ H] þ of prazosin and IS were 384.20 and 388.20 using Q1 and Q3 scans on drug standard solutions with concentration of 2000 ng/mL. Positive ionization mode was more favored by prazosin and IS due to the molecular structures of prazosin and IS having piperazine group, a six-membered ring with two nitrogen atoms at opposite positions in the ring. A proton donor was contributed with the use of formic acid in mobile phase. In the current study, two transitions were selected as quantification and confirmation product ions for prazosin and IS, m/z 95.00 and m/z 247.05 for prazosin, while m/z 71.10 and m/z 247.10 for IS. Terazosin was selected as IS as it is easily available and not costly. It showed similar chromatographic behavior as prazosin and its use has been reported in other studies [21,34].
Methanol and acetonitrile are common organic modifiers in the preparation of mobile phase. A higher ionization efficiency was observed with the use of acetonitrile compared with methanol. An increase of $40-50% in peak intensity was observed for acetonitrile due to its stronger dipole moment than methanol. The mobile phase composition with 0.1% formic acid and acetonitrile was optimized to obtain a symmetrical peak shape, sufficient sensitivity and a short sample run time. A distorted peak shape was observed when 0.1% formic acid and acetonitrile at 30:70 (v/v) was used as the mobile phase. Intensity was slightly reduced (5 À 7%) when the acetonitrile content was reduced to 60%. 0.1% formic acid and acetonitrile at 35:65 (v/v) gave the most optimum result with a short run time of 1.75 min. The short run time and the use of low flow rate reduced the consumption of organic solvents and hence the cost of analysis.     A sufficiently sensitive analytical method is essential for pharmacokinetic and bioequivalence studies to accurately determine the pharmacokinetic parameters. According to FDA and EMA guidelines, LLOQ should be set to at least 5% of the anticipated C max . As there was no published pharmacokinetic data of prazosin on Malaysia's population, pharmacokinetic data of other populations were thus used as guidance [21,25]. Gwak and Chun [21] reported a mean C max of 23.1 ng/mL after administration of 2 mg prazosin, while Zhu et al. [25] reported a C max range of 47.55 À 50.04 ng/mL after dosing with 4 mg prazosin. Based on these published C max values, 1 ng/mL was chosen as the LLOQ value for the current study.
Acetonitrile was used as a deproteinization agent in the study. Different ratios of acetonitrile to plasma at 1:1, 2:1, 3:1 and 4:1, were used. At ratios of 1:1 and 2:1, viscous and dirty samples were obtained, which were not suitable to be injected as they may damage the analytical column, clog the tubing and contaminate the system. When the acetonitrile and plasma ratios were increased to 3:1 and 4:1, clean supernatants were obtained after centrifugation. These two ratios were compared in terms of matrix effect and sensitivity at HQC level (22.5 ng/mL). Matrix factors of 0.92 and 0.93 w obtained for ratio of 3:1 and 4:1, respectively, which were close to each other. It was found that acetonitrile and plasma of ratio 3:1 gave a higher sensitivity than that obtained at a ratio of 4:1. LLOQ of 0.5 ng/mL at S/N ratio of above 5 was achieved. Therefore, acetonitrile and plasma at a ratio of 3:1 was selected for sample preparation. Based on the LLOQ of 0.5 ng/mL with 1 lL injection volume, the present analytical method is more sensitive than earlier published methods (Table 1) for the determination of prazosin in plasma samples.
In a previously published terazosin study [34], method validation was performed on terazosin with prazosin as an IS. No method validation was carried out on prazosin. In the present study, method validation was conducted on prazosin as an analyte to evaluate the reliability of the bioanalytical method [28,29]. Modification was made to the composition of mobile phase, volume of acetonitrile used as deproteinization agent and sample injection volume. Acetonitrile was reduced from 1000.0 lL to 750.0 lL as a deproteinization agent. Furthermore, a lower sample injection volume (1 lL instead of 2 lL) was used. A smaller injection volume helps to enhance peak symmetry by reducing the solvent diffusion effect and reduce any clogging issue, hence prolonging the shelf life of the analytical column.

Comparison of prazosin pharmacokinetics with other studies
The mean C max values of Minipress tablet (reference) and Minison tablet (test) were 21.1 ± 1.1 and 21.6 ± 1.1 ng/mL, respectively. Compared with the LLOQ value of 0.5 ng/mL obtained in the current study, LLOQ values >1.0 ng/mL reported in certain published studies [19,[22][23][24] are not sufficiently sensitive for the present BE study. Quantification of the last few sampling time points at elimination phase will be very challenging with a high LLOQ value. This can affect the estimation of k e , t 1/2 and extrapolated AUC. The LLOQ value of 0.5 ng/mL in the current study was sufficiently sensitive to reliably determine prazosin concentration in plasma samples of the bioequivalence study. The present bioanalytical method not only fulfilled the requirement of the LLOQ (5% of the  expected C max of the study) but also fulfilled the percentage of extrapolated AUC (AUC t-1 ) of below 20%, with percentage of AUC 16-1/ AUC 0-1 of all subjects <10%. The results also indicate that sufficient sampling points were collected in the study. The C max and AUC values were similar to the results reported by Guelen et al. [20] and Gwak and Chun [21]. Guelen et al. [20] conducted a pharmacokinetic study in 12 healthy subjects using two formulations of 2 mg prazosin tablets. The study reported mean C max , AUC 0-10 , t max and t

Adverse events and orthostatic hypotension
Alpha adrenoceptor antagonists are known to cause orthostatic hypotension, which may contribute toward reduced vascular resistance [35][36][37]. In a study done by Graham et al. [38] in hypertensive patients, all patients (100%) showed tachycardia and developed serious orthostatic hypotension after administration of 2-mg of prazosin. The side effects disappeared on day 2 after the administration of the second 2-mg tablet. Bhanu et al. [39] and van der Worp et al. [40] found that the a1-adrenoceptor antagonists were associated with up to two-fold increased orthostatic hypotension, compared with placebo. Hiremath et al. [41] demonstrated a consistently higher risk of hypotension-related adverse events with the use of alpha blockers. The study reported the proportion of hypotension and syncopal adverse events among alpha blocker users within the first 90 days were 38.4% and 40.2%, respectively. Rivasi et al. [42] stated that orthostatic hypotension risk is higher for a1-adrenoceptor antagonists which present the lowest uroselectivity and the drop in blood pressure could be more severe after the first dose (first-does phenomenon). Therefore, bed-time administration of these alpha blockers is preferable. Other studies also reported orthostatic hypotension adverse event with the use of a1-adrenoceptor antagonist [43][44][45]. Peak concentration occurred between 1 and 4 h post administration [39]. High odds of drug-induced orthostatic hypotension was taken into safety consideration at the planning stage of the study. A precaution step was taken in the study by bedside taking of blood samples for the first 4 h of the study. During the first 4 h, subjects rested supinely in bed with a pillow placed under their head. All subjects underwent clinical assessment before allowing them to leave their bed for the first 4 h post dosing of 2-mg prazosin tablet. Even with the implementation of the safety policy, the study still reported a total of five adverse events post dosing of 2-mg prazosin tablet. Three of the five adverse events were dizziness while the other two were orthostatic hypotension (6.67%) where their systolic and diastolic blood pressures dropped more than 20 and 10 mmHg, respectively. However, the incidence of orthostatic hypotension from our studies was much lower compared with others [38,44]. The subjects recovered after a few h on the same day and withdrew from the study when subsequent blood samplings were not possible due to the orthostatic hypotension. Other subjects reported a drop in systolic and diastolic blood pressures of 20 and 10 mmHg, respectively, which should not be categorized as orthostatic hypotension [46,47]. The decrease in systolic and diastolic blood pressures was well tolerated and a total of 26 subjects completed the study. The findings suggest that bedside taking of blood samples reduces the occurrence and frequency of orthostatic hypotension and hence the dropout rate of the study. High dropout rate can lead to an underpowered study and fail to meet bioequivalence. This will require an increased investment in time, people and money to identify the cause of bioequivalence failure, as well as a re-run of the study [48]. This sampling approach can be applied to other a1-adrenoceptor antagonists such as terazosin, doxazosin or other medications that cause postural hypotension.

Conclusion
The study developed a sensitive, simple, accurate and rapid LC/ ESI-MS/MS method for the determination of prazosin in human plasma and the method was applied to a bioequivalence study of two different formulations containing prazosin. The validated prazosin bioanalytical method was sufficiently sensitive to accurately determine the pharmacokinetic parameters. Bedside taking of blood samples is critical to reduce the occurrence of orthostatic hypotension and subject dropout rate. This practice should be recommended when conducting bioequivalence study of drugs with orthostatic hypotension risk.