UPLC-Tandem Mass Spectrometry for Quantification of Busulfan in Human Plasma: Application to Therapeutic Drug Monitoring

Busulfan (Bu) is an alkylating agent commonly used in preparative regimens for hematologic malignant and non-malignant patients undergoing hematopoietic stem cell transplantation (HSCT). The objective of the present study was to develop an UPLC-MS/MS method for quantification of Bu in human plasma. A total of 55 patients with hematologic malignancies (n = 34) and non- malignancies (n = 21) received myeloablative Bu therapy prior to HSCT. A tandem mass spectrometric method was developed and validated to quantify Bu levels in these patients. The method was fully validated over the concentration range of 25–2000 ng/mL (r > 0.99). The assay method demonstrated good precision and accuracy. Stability studies indicated that the drug was stable in various conditions. Incurred sample reanalysis findings were within acceptable ranges (<15% of the nominal concentration). Based on the 1st dose AUC results, one third of hematologic malignant patients and half of non-malignant patients needed dose adjustment. However, in subsequent doses (5th, 9th, and 13th), 77%, 82% and 82%, respectively, of hematologic malignant patients and 71%, 67% and 86%, respectively, of non-malignant patients achieved the target range of Bu AUC. The suitability of the developed method for routine TDM of Bu in HSCT patients was demonstrated. The study suggests that the pharmacokinetic profile of Bu varies in both groups.

Instrumentation. Acquity UPLC system (Waters Assoc., Milford, MA, USA) coupled to a triple quadruple mass detector (TQD) was used for analysis of Bu plasma samples. The analytes were separated with Acquity UPLC BEH C 18 column (2.1 × 50 mm, 1.7 μm) at ambient temperature. The analytical column was equipped with VanGuard pre-column filter of the same packing material (Waters Assoc., Milford, MA, USA). The mobile phase consisted of methanol: 20 mM ammonium acetate buffer (90:10, v/v) and delivered at a flow rate of 0.3 mL/ min to a positive electrospray ionization interface (ESI + ) of TQD (Waters Assoc., Milford, MA, USA). Tuning parameters of MS and MS/MS were adjusted by directly infusing solutions of Bu and IS (prepared in the mobile phase) into the ionization probe at a flow rate of 10 μL/min using Hamilton syringe. The ion source and desolvation temperatures were fixed at 150 °C and 350 °C, respectively. The capillary voltage was set at 3.08 kV, cone Linearity. The linearity of the present assay method was assessed by spiking Bu in blank human plasma at six non-zero calibrators covering the range of 25 to 2000 ng/mL and then analyzed in replicates of twelve over several days. Bu concentrations were plotted versus the detector responses to obtain the slope, intercept, and correlation coefficient (r) of the linear line employing the least squares' linear regression method using MassLynx software. The precision (determined as relative standard deviation; RSD, %) of the calibration standards should not deviate www.nature.com/scientificreports www.nature.com/scientificreports/ by values of more than 15% except the lower limit of quantification (LLOQ) which should not exceed 20% of the nominal concentration 33 . Accuracy (bias) and precision. Quality control (QC) samples of Bu were prepared in blank plasma at concentrations of 50, 500, 1250, and 1750 ng/mL (covering low, medium and high ranges of the calibration standards) and were measured in ten replicates to evaluate the intra-and inter-run precision and accuracy. The intra-run precision was determined from ten replicate analyses of QC samples from one calibration curve batch in one day 34 . However, the inter-run precision was determined over a period of four weeks 34 . The RSD, % was considered as a  www.nature.com/scientificreports www.nature.com/scientificreports/ measure of precision whereas, accuracy (or bias, %) was the percent of deviation from the nominal concentration. The precision (RSD, %) determined at each concentration should be ≤15% 33 .
Selectivity. The selectivity of the present method was assessed by analyzing six independent sources of blank human plasma samples for possible interferences with endogenous components. In this regard, heparinized, hemolyzed and lipemic plasma samples were assessed 33 . On the other hand, some of exogenous compounds at concentrations of 1.0 µg/mL were tested for potential interferences with Bu. The investigated drugs were phenytoin, ethosuximide, oxcarbazepine, lamotrigine, acetaminophen, metoclopramide, domperidone, metronidazole, diclofenac, nicotine, levetiracetam, zonisamide, chloramphenicol, mefenamic acid, and lormetazepam. The mass  www.nature.com/scientificreports www.nature.com/scientificreports/ detector response (peak area) of various plasma samples' extracts as well as exogenous compounds (at the retention times of Bu and IS) were compared to that of the spiked blank human plasma samples at the LLOQ 33 . Stability. The stability of Bu in human plasma sample was evaluated by several studies 33,34 . Freeze-thaw stability was evaluated employing five freeze-thaw cycles from −80 °C to room temperature. At each cycle, QC plasma samples were stored frozen for at least 12 h before they were taken out for thawing at the bench top. The QC samples were kept at room temperature for 2 h to allow for complete thawing before analysis 33,34 . On the other hand, Bu stability in the autosampler was evaluated by storing the processed Bu samples in the autosampler and then the samples were injected at designated time intervals for up to 24 h 33 . In addition, bench-top stability was evaluated by keeping the processed QC samples at room temperature (25 °C) for up to 24 h and the samples were analyzed at various times over that period of time. Moreover, long-term stability assessment was performed by storing the QC samples at −80 °C for 1 month. The samples were analyzed at various times during that period and compared with freshly prepared standard curves and QC samples. The mean value results of stability investigations were calculated and compared to the nominal concentrations 33,34 . Recovery and matrix effect. The matrix effect (ME) was evaluated by post-column infusion method 35    www.nature.com/scientificreports www.nature.com/scientificreports/ Bu and IS was evaluated by comparing the peak areas obtained from blank human plasma samples spiked with the analytes (using the four QC samples) before extraction to those spiked after extraction 33 .
Incurred sample reanalysis. Incurred sample reanalysis (ISR) was evaluated by re-analysis of some patients' samples 36 . The original and repeated analyses of patients' samples were performed employing the present method procedures. The % difference of the original and repeated Bu concentration results was determined as follows 33 : = * -%Difference 100 (Repeated Original)/Mean Injection carryover. This test was assessed by injecting the upper limit of quantification (ULOQ) of the calibration curve (2,000 ng/mL) followed by blank human plasma sample (0.0 ng/mL). The potential peak areas at the retention times of Bu and IS were compared with the corresponding peak areas of the ULOQ 33 .
Dilution effect. For Bu plasma samples above the ULOQ, dilution of samples is necessary to provide the concentrations within the calibration curve range and hence the dilution effect is necessary to be assessed. The  www.nature.com/scientificreports www.nature.com/scientificreports/ dilution integrity was evaluated by spiking blank human plasma samples with Bu at concentrations above the ULOQ (e.g., 3, 000 ng/mL) and then the samples were diluted by blank human plasma samples at various ratios involving 1:1, 1:2, and 1:3. Five replicates for each dilution ratio were assessed. The accuracy (bias) and precision (RSD, %) of the diluted plasma samples were determined and compared to the nominal concentrations 33 .
Clinical application. The present assay method was applied in clinical settings by analysis of Bu plasma samples refereed to our TDM laboratory for routine monitoring of patients taking Bu.

Patients' characteristics.
A total of 55 patients with hematological (n = 34) and non-hematological (n = 21) malignancies received myeloablative Bu therapy prior to HSCT at Kuwait Cancer Control Center (KCCC). The characteristics and demographics of the patients are presented in Table 1. The present study was approved by Ethical Committees (Ministry of Health and Health Science Center-Kuwait University) as well as approval of KCCC hospital. Informed consent form was not required from the patients because Bu samples were analyzed for routine monitoring prior to HSCT therapy as well as it is a retrospective study (data collected from the patients' files). None of the research team members had an access to identifying patient information when analyzing the samples or processing the data. All the methods described have been conducted according to Good Clinical Practice and Good Clinical Laboratory Practice guidelines.

Results and discussion
UPLC analysis. For obtaining optimized conditions for quantification of Bu in human plasma samples, appropriate tuning parameters were utilized to determine the precursor and product ions of Bu and IS using ESI + mode. The precursor/product ions were determined at m/z 263.7 > 151.1 and 272.2 > 159.1 for Bu and IS, respectively. Several endeavors comprising the use of various mobile phase compositions were accomplished to achieve the optimal conditions concerning good resolution and symmetrical peak shapes. The mobile phase consisting of methanol: 20 mM ammonium acetate buffer (90:10, v/v) was found optimal for the formation of the precursor and product ions of both Bu and IS. In addition, Acquity UPLC BEH C 18 column (2.1 × 50 mm, 1.7 μm) equipped with a VanGuard pre-column filter (employing a flow rate of 0.3 ml/min) was found to be the best column for obtaining good peak shape and well separation of both Bu and IS. Using a simple ether extraction step, the total sample pretreatment procedure was very short (approximately 20 min) as well as providing good chromatograms (free of endogenous components) as shown in Fig. 1. The use of an appropriate IS is mandatory for achieving satisfactory method performance. In this regard, deuterated Bu (Bu-d 8 ) was utilized for quantification of Bu in human plasma samples. The advantage of this IS is that it has similar physicochemical properties as Bu and hence it compensates for matrix effects since Bu and IS have the same retention times (Fig. 2).
The selectivity of the present method was assessed by investigating six independent lots of blank human plasma samples involving heparinized, hemolyzed, and lipemic samples. No interferences from endogenous components were observed at the retention times of Bu or IS in blank human plasma samples. In addition, evaluation of interferences from various exogenous compounds demonstrated lack of interferences from the assessed compounds. The current method however, verified good selectivity as shown by lack of interfering peaks from both endogenous and exogenous compounds at the retention times of the drug and IS. Figure 2 depicts typical MRM chromatograms of blank human plasma sample spiked with Bu at the LLOQ (25 ng/mL) and IS. In addition, Fig. 3 demonstrates typical MRM chromatograms of a plasma sample taken from thalassemic child patient who was on Bu therapy.
The present assay method was linear for Bu concentration in the range of 25 to 2000 ng/mL (mean slope = 0.0012, RSD% = 8%, N = 12). Linear correlations (R 2 > 0.999) were obtained using least squares linear regression method using peak area ratios with the LLOQ of 25 ng/mL (RSD% = 7.4%,). The accuracy and precision of the LLOQ were within the acceptable limits (<15%) 33 .
Intra-and inter-run precision of Bu assay method of four QC levels (50, 500, 1250 and 1750 ng/mL) ranged from 1.23 to 6.52% with accuracy (bias) ranged from −10.66 to −0.24% demonstrating adequate precision and accuracy ( Table 2) 33 . www.nature.com/scientificreports www.nature.com/scientificreports/ The findings of Bu stability at various situations involving storage of processed samples in the autosampler for up to 24 h and freeze-thaw for five cycles were presented in Table 3. The processed Bu samples were stored in the autosampler (at ambient temperature) and the results demonstrated that the processed samples can remain in the autosampler for up to 24 h without showing appreciable loss in the quantified levels. The results of freeze-thaw stability study demonstrated that the precision ranged between 2.4% to 2.8% and the accuracy ranged between 93.7% to 98.1% (Table 3). The findings exhibited that Bu in human plasma was stable for at least 5 freeze/thaw cycles from −80 °C to room temperature. On the other hand, the findings of Bu stored at −80 °C on its stability demonstrated that the drug was stable at this temperature for at least one month. In addition, storing Bu samples at room temperature indicated that Bu plasma samples were stable for a minimum period of 24 h ( Table 3).
The ME on the present assay method was assessed by post-column infusion procedure during the method development procedure and consequently the separation system was optimized 35 . The post-column infusion experiment demonstrated that the signals at the retention times of Bu and IS were unchanged when the blank plasma extract was injected into the UPLC-MS/MS system indicating a lack of ion suppression/enhancement using the current assay method.
The findings of the mean Bu recovery from the human plasma samples ranged between 77.9 to 81.7% whereas that of the IS was 82%, indicating the suitability of liquid extraction procedure for separation of Bu from human plasma samples (Table 4).
The findings of incurred sample re-analysis (ISR) by using the present method demonstrated acceptable results which ranged between −13.6% to +13.9% ( Table 5). The findings of injection carryover demonstrated that the results were within the acceptable limits (below 15% of the LLOQ of Bu and below 3% for the IS). The dilution integrity of Bu samples above ULOQ assessment demonstrated adequate accuracy and precision (<15% of the nominal concentration).
The present method demonstrated many advantages over the reported methods such as good precision and accuracy. In addition, the ME was assessed and the results showed lack of ion suppression/enhancement indicating good reliability in Bu detection and quantitation. Moreover, ISR of Bu results were within the acceptable ranges.
Therapeutic drug monitoring of Bu. Inter-individual variability in the pharmacokinetic profile of Bu has been observed in patients on oral Bu as well as i.v. Bu injection therapy, and the inter-individual variability in metabolic drug handling may contribute to suboptimal outcome 7,37,38 . Monitoring of Bu levels in human plasma samples is necessary for patients on i.v. Bu injection, and dose adjustment based on pharmacokinetic analysis of Bu is commonly experienced. At AUC values of ≤900 μΜ.min, the incidence of graft failure increases, whereas at ≥1550 μM.min there is an increased risk of treatment-related toxicity such as hepatic veno-occlusive disease 7 .
The pharmacokinetic profile of i.v. Bu in hematologic malignant and non-malignant patients using non-compartmental methods are presented in Table 6. Figure 4 depicts individual AUC values following 1 st , 5 th , 9 th , and 13 th Bu doses for both groups. Following the 1 st Bu dose, the median AUC values in malignant and non-malignant patients were 1181 and 1117 μM.min, respectively, and the median CL values were 2.46 and 3.45 mL/min/kg for malignant and non-malignant patients, respectively. Furthermore, the median Vd values were 0.54 and 0.61 L/kg for malignant and non-malignant patients, respectively.
It has been observed that after the 1 st Bu dose to the hematologic malignant patients, 5 patients (14.7%) needed dose increase whereas 6 patients (17.6%) needed dose reduction to achieve an optimal Bu target range of 950-1500 μM.min. However, for the successive dose adjustments, the patients were solely required Bu dose reduction. Following the 5 th , 9 th , and 13 th doses, Bu dose was reduced in 20.6%, 17.6%, and 17.6% of patients, respectively. On the other hand, following the 5 th dose, 2.9% of the patients needed dose increase. However, no increase in Bu dose was needed after the 9 th and 13 th doses. Similarly, Bu doses were also adjusted for the non-malignant patients. In this regard, after the 1 st Bu dose, 5 patients (23.8%) needed dose increase whereas 6 patients (28.6%) needed dose reduction to attain Bu target range of 900-1350 μM.min.
After the initial Bu dose, the median CL in hematologic malignant patients was 2.46 mL/min/kg which was significantly lower (P = 0.0055) than that in non-malignant patients (3.45 mL/min/kg). The same trend of significantly lower CL values was seen in both groups in subsequent Bu doses (Table 5 and Fig. 5). Moreover, the 1 st dose median CL was significantly high in both groups whereas no significant changes were observed in the CL values in subsequent doses (Fig. 5). This finding concurs with a previous observation in non-malignant children 39 .
The relationship between age and Bu clearance is depicted in Fig. 6. As shown in the figure, non-malignant patients (mainly pediatrics) demonstrated high Bu CL values in contrast to hematologic malignant patients (mainly adults) indicating that non-malignant patients may require high Bu dose (mg/kg) than hematologic malignant patients to achieve an optimal clinical response. However, this is not the case because the therapeutic range of Bu in non-hematologic patients is lower than that of hematologic malignant patients.

Conclusions
An accurate, precise, reliable, specific, and reproducible UPLC-MS/MS method for quantification of Bu in human plasma samples is described. The present tandem mass spectrometric method is appropriate for routine analysis of Bu in plasma samples of patients being prepared for HSCT therapy. The described method is routinely employed in our TDM lab for quantification of Bu in human plasma samples of both hematologic malignant and non-malignant patients. Utilization of TDM to Bu therapy minimizes toxicity, maximizes efficacy and improves transplantation outcome. In addition, the pharmacokinetic profile of Bu in hematologic malignant and non-malignant patients have been investigated. To identify the factors that influence the variation in the pharmacokinetic profile of Bu in hematologic malignancies and non-malignancies, population pharmacokinetic studies are warranted.