Sep-Pak classic C-18 cartridges (Waters Corporation, Milford, MA, USA) was used for analyte extraction from plasma samples. Chromatography was performed on an LCMS/MS system consisting of Water Alliance 2695 separation module (Waters Associates, Inc Milford, MA, USA) for solvent delivery and sample introduction, and a Micromass Quattro micro API bench-top triple quadruple mass spectrometer (Micromass, Manchester, UK) interfaced with a Z-spray ESI source as detector. Analysis was performed using reversed phase Atlantis dC₁₈ column (2.1 × 100 mm, 3 µm) protected by guard column Symmetry C₁₈ (3.9 × 20 mm, 5 µm) at room temperature. Mass Lynx software (Ver 4.0) working under Microsoft Window XP professional environment was used to control the instrument, data acquisition, peak integration, peak smoothing, and signal-to-noise ratio measurement.

Colistin sulfate and clarithromycin pure standards were purchased from Sigma-Aldrich, MO, USA. Formic acid, methanol, and acetonitrile (HPLC grade) were purchased from Fisher Scientific, NJ. USA. Water HPLC grade was prepared by reverse osmosis and further purified by passing through a Milli-Q, Synergy System (Millipore, Bedford, MA, USA). Drug-free human plasma was obtained from the blood bank of King Faisal Specialist Hospital & Research Centre (KFSHRC) Riyadh, Saudi Arabia.

The mobile phase consisted of 0.2% formic acid in acetonitrile and water (50:50, v:v). It was degassed before use and delivered at a flow rate of 0.2 ml/minute at ambient temperature with a run time of 4 minutes. The electrospray ionization source was operated in the positive-ion mode at a capillary voltage of 3.5 kV and a cone voltage of 35 V. Nitrogen was used as nebulizing and desolvation gas at a flow rate of 50 and 600 L/hr, respectively. Argon was used as the collision gas at a pressure of 1.28 × 10⁻³ mbar. The optimum collision energy for colistin and the IS was 25 eV. The ion source and the desolvation temperatures were maintained at 120˚C and 350˚C, respectively. Colistin A, colistin B, and the IS were detected in the positive ion using multiple reactions monitoring mode at the following transitions of mass to charge (m/z): 585.6 → 101.4, 578.7 → 101.3, and 748.6 → 158.4, respectively.

Stock solution of colistin (A and B) and IS (1.0 mg/ml) were prepared in methanol. They were further diluted with plasma (colistin A and B) and water (IS) to produce working solutions of 10 µg/ml for colistin (A & B) and 0.1 µg/ml for IS. Ten calibration standards in the range of 0.05 - 10.0 μg/ml and four quality control (QC) samples (0.05, 0.15, 5.0, and 9.0 μg/ml) were prepared in human plasma. Calibration standards and QC samples were vortexed for one minute and 1.0 ml aliquots were transferred into Teflon-lined, screw-capped, borosilicate glass (13 × 100 mm) culture tubes and stored at −20˚C until used.

Aliquots of 1 ml blank plasma, calibration curve, or QC samples were allowed to equilibrate to room temperature. To each tube, 150 µl of the IS 0.1 µg/ml solution was added and vortexed for 20 seconds. Before loading samples, Sep-Pak C-18 cartridges were conditioned with 1 ml methanol followed by 2 ml water. Then the sample (1 ml) was loaded followed by 1 ml water. Samples were eluted with 1 ml methanol containing 0.1% formic acid. The eluates were then evaporated to dryness under a gentle nitrogen stream in a heating block at 45˚C, the residues were reconstituted in 100 µl of 0.1% formic acid in water, transferred into auto-sampler vial, and 10 µl were injected into the LC-MS/MS system.

A total of 40 aliquots of four QC samples (0.05, 0.15, and 9.0 μg/ml) were used for stability studies. Five aliquots of each QC sample were extracted and immediately analyzed (baseline), five aliquots were allowed to stand on the bench-top for 24 hours at room temperature before being processed and analyzed (counter stability, 24 hours at room temperature), five aliquots were stored at −20˚C for eight weeks before being processed and analyzed (long term freezer storage stability), and five aliquots were processed and stored at room temperature for 24 hours or 48 hours at −20˚C before analysis (autosampler stability). Finally, fifteen aliquots of each QC sample were stored at −20˚C for 24 hours, they were then left to thaw completely at room temperature unassisted. Five aliquots of each QC sample were extracted and analyzed and the rest returned to −20˚C for another 24 hours. The cycle was repeated three times (freeze-thaw stability).

The method was validated according to standard procedures described in the US Food and Drug Administration bioanalytical method validation guidance [18]. The validation parameter included specificity, linearity, accuracy, precision, recovery, and stability.

The product and precursor ions of colistin A, colistin B, and clarithromycin (IS) were determined by infusing a standard mixture (1.0 µg/ml in methanol) in the mass spectrometer using a syringe pump at a flow rate of 20 µl/ml. Mass spectrometry acquisition was performed in multiple reaction monitoring (MRM) with positive ion transition mode set at (precursor [M-2H]²⁺ to product ion) mass to charge (m/z): 585.6 → 101.4, 578.7 → 101.3, 748.6 → 158.4, for colistin A, colistin B, and the IS, respectively. The optimum peak height response were obtained by applying capillary voltage of 3.5 KV, cone voltage of 35V, and collision energy of 25 eV. The optimal chromatographic condition consisted of 0.2% formic acid in acetonitrile and water (50:50, v:v), delivered at a flow rate of 0.2 ml/minute. Analysis was completed within 4 minutes run time. The retention time of colistin A, colistin B and the IS were around 1.4, 1.4, and 2.8 minutes, respectively. Figure 1 represents total ion chromatogram (TIC) of colistin A, colistin B and IS extracted from plasma.

We screened six blank plasma samples and seven commonly used drugs including aspirin, acetaminophen, ranitidine, nicotinic acid, caffeine, diclofenac and omeprazole for interferences with colistin A and colistin B, or IS. No drug or endogenous component co-eluted with colistin A, colistin B or the IS. Figure 2(A) depicts a representative chromatogram of drug free human plasma (blank) used in preparation of standards and quality control samples. Figure 2(B) depicts blank plasma spiked with the IS (0.015 µg).

Linearity of the assay was evaluated by analyzing a series of standard mixtures containing colistin A and colistin B in human plasma at nine concentrations over the range of 0.05 - 10.0 μg/ml. Corresponding peak height ratios to the IS and concentrations were subjected to regression analysis. Mean equations obtained were y = 0.0979x − 0.0029, r² = 0.9901 (n = 10) and y = 0.0461x − 0.0012, r² = 0.9921, (n = 10) for colistin A and colistin B, respectively. The suitability of the calibration curves was confirmed by back calculating the concentration of colistin A and colistin B in human plasma from calibration curves (Table 1). All calculated concentrations were well within the acceptable limits. Figure 3 represents

*Coefficient of variation (CV) = standard deviation (SD) divided by mean measured concentration × 100. **Accuracy = measured level divided by nominal level × 100.

mean calibration curve of colistin A and colistin B. Limit of detection of colistin A and colistin B in plasma were 0.030 µg/ml and 0.035 µg/ml, respectively, whereas, limits of quantification were 0.05 µg/ml for both compounds.

Precision and bias were determined for four QC concentrations (0.05, 0.15, 5.0, and 9.0 μg/ml). Intra-day precision (n = 10) was ≤11.4% for colistin A and ≤11.2% for colistin B. Inter-day precision (n = 20, over three consecutive days) was ≤11.5% for colistin A, and ≤9.9% for colistin B. Inter-day bias was in the range of −9.1% to 10.5% for colistin A and −8.6% to 10.2% for colistin B. Inter-day bias was −3.0% to 6.0% for colistin A, and −4.7% to 3.0% for colistin B. The results are summarized in Table 2. Figure 4 depicts the LC-MS/MS chromatogram of low QC (0.15 µg/ml), and high QC (9 µg/ml) both spiked with IS (0.015 µg).

Extraction recovery of colistin A and colistin B at four concentrations (0.05, 0.15, 5.0 and 9.0 µg/ml) and the IS at one concentration (0.1 µg/ml) were determined by comparing peak heights of spiked-before-extraction samples and spiked-after-extraction samples (5 sets). Mean measured extraction recovery of colistin A, and colistin B was 97%, and 94%, respectively. Recovery of IS was 97%.

Matrix effect was quantitatively evaluated by comparing peak heights of colistin

SD, standard deviation. CV, standard deviation divided by mean measured concentration × 100. Bias, measured level − nominal level divided by nominal level × 100.

A and colistin B (0.05, 0.15, 5, and 9 µg/ml) and the IS (0.1 µg/ml) prepared in human plasma to those prepared in water. Mean ion suppression effect was −4.9% for colistin A, −4.5% for colistin B and −1.0% for the IS.

Colistin (A and B) and the IS stability in processed and unprocessed plasma samples at 3 concentrations (0.05, 0.15 and 9.0 μg/ml) was investigated. Colistin was stable in processed samples for at least 24 hours at room temperature (≥98% for colistin A and ≥92% for colistin B) and 48 hours at −20˚C (≥105% for colistin A and ≥91% for colistin B). Colistin in unprocessed plasma samples was stable for at least 24 hours at room temperature (≥87% for colistin A and ≥92% for colistin B), at least eight weeks at −20˚C (≥90% for colistin A ≥94% for colistin B), and after three freeze-and thaw cycles (≥100% colistin A and colistin B. Table 3 summaries the results of stability studies.

Stability (%) = mean measured concentration (n = 5) at the indicated time divided by mean measured concentration (n = 5) at baseline × 100. Spiked plasma samples were processed and analyzed immediately (baseline, data not shown), after 24 hours at room temperature (24 hrs RT), after freezing at −20˚C for 8 weeks (8 wks, −20˚C), or processed and then analyzed after storing for 24 hours at room temperature (24 hrs, RT) or 48 hours at −20˚C (48 hrs, −20˚C).