A Novel and Cost Effective Method of Removing ... - ACS Publications

Aug 1, 2014 - Bioanalytical Sciences, Research & Development, Bristol-Myers Squibb, Co.; Route ... all these approaches are either not cost-effective ...
0 downloads 0 Views 795KB Size
Download PDF
Article pubs.acs.org/ac

A Novel and Cost Effective Method of Removing Excess Albumin from Plasma/Serum Samples and Its Impacts on LC-MS/MS Bioanalysis of Therapeutic Proteins Guowen Liu,* Yue Zhao, Aida Angeles, Lora L. Hamuro, Mark E. Arnold, and Jim X. Shen Bioanalytical Sciences, Research & Development, Bristol-Myers Squibb, Co.; Route 206 and Province Line Road, Princeton, New Jersey 08543, United States S Supporting Information *

ABSTRACT: We have developed an innovative method to remove albumin from plasma/serum samples for the LC-MS/MS quantitation of therapeutic proteins. Different combinations of organic solvents and acids were screened for their ability to remove albumin from plasma and serum samples. Removal efficiency was monitored by two signature peptides (QTALVELVK and LVNEVTEFAK) from albumin. Isopropanol with 1.0% trichloroacetic acid was found to be the most effective combination to remove albumin while retaining the protein of interest. Our approach was compared with a commercial albumin depletion kit on both efficiency of albumin removal and recovery of target proteins. We have demonstrated that our approach can remove 95% of the total albumin in human plasma samples while retaining close to 100% for two of three therapeutic proteins tested, with the third one at 60−80%. The commercial kit removed 98% of albumin but suffered at least 50% recovery loss for all therapeutic proteins when compared to our approach. Using BMS-C as a probe compound, the incorporation of the albumin removal approach has improved both assay sensitivity and ruggedness, compared to the whole plasma protein digestion approach alone. An LC-MS/MS method was developed and validated based on this new approach for the analysis of BMS-C in monkey serum. This assay was successfully applied to a toxicological study. When the albumin removal method was used in another clinical LC-MS/MS method, the sensitivity improved 10-fold to 50 ng/mL LLOQ comparing to a typical pellet digestion method.

P

cost efficiently remove these highly abundant background proteins from a biological sample for protein bioanalysis using LC-MS/MS. As it is generally understood, serum albumin is the most abundant protein in plasma and serum samples. For example, in human plasma, serum albumin constitutes about 50% of all the plasma proteins.8 Albumin removal is a logical target to decrease the total protein load in human plasma/serum. A number of different strategies have proved efficient in selectively extract therapeutic proteins from these albumin reach plasma/serum samples. These approaches includes techniques, such as albumin depletion using plates or columns,7 immunoaffinity purification using antibodies from generic protein A/G,9 or specific antidrug antibody, etc.5 However, all these approaches are either not cost-effective or not operational friendly to high throughput sample analysis, which is a default requirement for modern bioanalysis. A simpler and more cost-effective way to reduce the complexity of plasma samples is desired. Cohn et al.8 invented a process (referred as the Cohn’s process) to purify albumin from serum

rotein bioanalysis using LC-MS/MS have proliferated in the recent years because of improvement in separation and detection approaches.1,2 A number of sample preparation approaches, including pellet digestion,3 direct digestion of the whole plasma/serum sample after reduction and alkylation,4 immunoaffinity purification before digestion,5,6 and albumin depletion approach7) have been reported by scientists across the biopharmaceutical industry. Both pellet digestion and direct digestion are popular since the digestion of all plasma proteins followed by measuring one or more surrogate peptides from the target protein for the quantitation is simple and fast. These approaches work well in cases where assay sensitivity is not paramount and assay ruggedness is not of the critical importance. However, scientists using these techniques have recognized a number of drawbacks. Most of these are related to the over abundance of the endogenous proteins, which have several unwanted effects on LC-MS/MS assays. These include but are not limited to (1) complicated the digestion step, (2) interference with the MS signal of a surrogate peptide from a target protein caused by peptides from the endogenous proteins, and (3) limiting sample loading capability on the LC-MS/MS system due to saturation of the mass spectrometric signal from high abundance of peptide in reconstitution solution following digestion. Therefore, it is imperative to © 2014 American Chemical Society

Received: May 16, 2014 Accepted: August 1, 2014 Published: August 1, 2014 8336

dx.doi.org/10.1021/ac501837t | Anal. Chem. 2014, 86, 8336−8343

Analytical Chemistry

Article

spectrometer (Applied Biosystems/MDS SCIEX, Concord, Ontario, Canada) coupled with a Shimadzu (Columbia, MD, USA) Nexera UHPLC system. The Shimadzu UHPLC system consists of two LC-30AD pumps, two DGU-20A5 degassers, one SIL-30ACMP autosampler and one CTO-30AS column heater. A Waters (Milford, MA, USA) Acquity UPLC CSH C18 column (130 Å, 1.7 μm, 2.1 × 50 mm) was used for chromatographic separation. Eppendorf (Hauppauge, NY, USA) Thermomixer R and MTP Microblock was used for the digestion. A JANUS Mini automated liquid handler from PerkinElmer (Waltham, MA, USA) was used for all liquid transfers. Methods and Experiments. LC-MS/MS Methods. Mobile phase A (0.1% formic acid (FA) in water) and mobile phase B (0.1% FA in methanol) were used for all LC-MS/MS analysis. A Waters CSH C18 UPLC column was used for the chromatographic separation with a flow rate of 0.8 mL/min. The mass spectrometer was operated in positive ion electrospray mode with selective reaction monitoring (SRM). Either doubly or triply charged parent ions of the surrogate/signature peptides and their corresponding internal standards were selected for Q1. The probe temperature was set at 650 °C. The turbo ionspray voltage was set at 4000 V and GS1 and GS2 were both set at 65 psi. HPLC conditions for the two surrogate peptides QTALVELVK (QTAL) and LVNEVTEFAK (LVNE) for albumin monitoring are as followings: the gradient started at 10% mobile phase B and continued to 0.5 min; then changed to 50% B from 0.5 to 3.95 min; changed to 100% at 4.0 min and maintained at 100% B until 5.0 min; then changed back to 10% at 5.01 min until to the end of the run at 6.0 min. For the recovery test, signature tryptic peptides EEQY (for BMS proprietary peptide sequences, we will use the first four amino acids to represent each peptide, for all other signature peptides, the complete sequence were listed for their first appearance and later were quoted by their first four amino acids) for BMS-A, GQGT for BMS-B, and ITYG for BMS-C were monitored. For the method validation and sample analysis of drug candidate C in monkey serum, two surrogate peptides, ITYG and VVSVLTVLHQDWLNGK (VVSV) were monitored. And for the assay validation in human serum of drug candidate C, only the signature peptide ITYG was used as peptide VVSV can also be found in all human IgG1, IgG3, and IgG4. The same LC conditions used for albumin monitoring were used for peptide GQGT and ITYG monitoring to evaluate the recoveries of BMS-B and BMS-C. However, peptide EEQY is an extremely polar peptide and a different gradient was used for its quantitation. Specifically, from 0 to 0.5 min, the gradient was set as 100% mobile phase A; from 0.5 to 2.5 min, changed from 100% A to 45% A; at 2.6 min to 0% A and maintained at 0% A until 3.5 min; then changed to 100% A at 3.6 min through the end of the run at 4.0 min. The HPLC conditions were further optimized for peptide ITYG and VVSV for the method validation and sample analysis of BMS-C in monkey and human serum. The gradient started at 15% B through 0.50 min, then mobile phase B was changed from 15% to 48% from 0.50 to 6.0 min, then at 6.01 min, the gradient was changed to 100% B and maintained until 7.00 min, then changed back to 15% at 7.01 min and stopped at 8.00 min. The other peptide-specific LC-MS/MS parameters can be found in the Supporting Information (see Table 1S for details). Selective Albumin Removal. Different combinations of organic solvent and acids were screened for their abilities to remove albumin from plasma and serum samples. Specifically,

and plasma using combination of different percentage of ethanol at different temperature and pHs. It has also been reported10,11 that the trichloroacetic acid (TCA)-generated precipitates of certain plasma proteins dissolve in alcohols whereas those of other proteins do not. In addition, Rajalingam et al.12reported U-shaped TCA-induced protein precipitation curves for different proteins when mixing with aqueous solution with increasing TCA concentrations. Chen et al.13 developed a modified protein precipitation procedure for efficient removal of albumin from serum using combination of TCA and acetone. All these work points to the direction that a combination of TCA with organic solvent may be able to remove albumin from plasma/serum samples. On the basis of this work, we developed a simple solvent extraction approach to efficiently remove albumin from plasma/serum samples while retaining the target proteins. In this work, we applied the same principle of albumin purification reported decades ago8 to remove albumin from serum/plasma samples. We developed a semiquantitative LCMS/MS method14 to measure the efficiency of the albumin removal. Our ultimate goal was to utilize a cost-effective and simple protocol to remove albumin but retaining therapeutic proteins of interest to facilitate LC-MS/MS quantitation of these drugs. Following the development of the protocol, we applied our clean up protocol to three types of proprietary drug candidates [one monoclonal antibody (mAb) of molecular weight (MW) of 145.3 kDa (BMS-A), one domain antibody (dAb) with a MW of 78.0 kDa (BMS-B) and a fusion-protein with a MW of 80.0 kDa (BMS-C)] to test the efficiency and ruggedness of our protocol. In addition, simplicity, cost and throughput were modeled as important parameters of the clean up procedure. We also evaluated our procedure against a commercial albumin removal kit to compare its effectiveness in removing albumin while retaining proteins of interest. The benefits resulting from removing albumin to LC-MS/MS assays, such as assay sensitivity and ruggedness, were also evaluated. We then applied this new approach to the method development, validation, and sample analysis for BMS-C in monkey serum. Using albumin removal overcame a sensitivity limitation as the desired LLOQ had not been feasible using a traditional digestion followed by LC-MS/MS approach.

■

EXPERIMENTAL DETAILS Chemicals, Reagents, Materials, and Apparatus. Isopropanol (IPA) and HPLC grade acetonitrile and methanol were purchased from J.T. Baker (Phillipsburg, NJ, USA). Formic acid (SupraPur grade) was purchased from EMD Chemicals (Gibbstown, NJ, USA). Bovine trypsin and trichloroacetic acid (TCA) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Deionized water was generated using a NANOpure Diamond ultra pure water system from Barnstead International (Dubuque, IA, USA). Monkey and human plasma or serum were purchased from Bioreclamation Inc. (Hicksville, NY, USA). Protein drug candidates, BMS-A, BMS-B, and BMSC, and the stable isotope-labeled surrogate peptide internal standards (SIL-ITYG, SIL-GQGT, and SIL-VVSV, for each surrogate peptide, were obtained internally at Bristol-Myers Squibb Co. In each stable labeled surrogate peptide internal standard, one amino acid was labeled with the stable isotopes 13 C and 15N. The Pierce albumin depletion kit was purchased from Thermo Scientific (Rockford, IL, USA). Equipment and Apparatus. All sample analyses were performed on a AB SCIEX Triple Quad 5500 mass 8337

dx.doi.org/10.1021/ac501837t | Anal. Chem. 2014, 86, 8336−8343

Analytical Chemistry

Article

Sensitivity and Assay Ruggedness Evaluation. BMS-C was used to evaluate the benefits of albumin removal for quantitative LC-MS/MS bioanalysis. Samples containing 0.05, 0.5, and 5 μg/mL BMS-C were prepared in human serum. Aliquot sizes of 20, 50, and 100 μL of serum samples were processed following the pellet digestion or IPA-TCA protocols. The resultant samples were analyzed on a LC-MS/MS system with an injection volume of 40 μL. Five replicates of each sample were tested. The final results were computed for absolute sensitivity. Peak shape and retention time of the surrogate peptide were evaluated for chromatographic performance.

0.1%, 0.5%, 1%, 2%, and 5% formic acid (FA) or trichloroacetic acid (TCA) in methanol, ethanol, isopropanol, and acetone were tested with human plasma at a ratio of 1:10 (plasma to organic solvent) to evaluate the efficiency of albumin removal and target protein recovery. The previously reported pellet digestion approach2,3 was used as a reference of 0% albumin removal. Specifically, two signature peptides (QTAL and LVNE)14 following tryptic digestion were monitored and their LC-MS/MS responses were used to estimate the relative amount of albumin presented in each samples. The reason that peptides QTAL and LVNE were chosen as the surrogate peptides for albumin was because they are unique peptides in serum albumin found in many species, including human and cynomolgus monkey. Both peptides gave good sensitivity on the mass spectormeter and excellent peak shapes on a regular C18 column. For convenience, we define our final approach as IPA-TCA throughout this paper and the reference approach was defined as “pellet digestion”. One signature peptide of each target protein was monitored for their relative recovery compared to the standard pellet digestion approach,2,3 which was used as 100% recovery of each protein. Sample Digestion. Previously reported pellet digestion3 was used as a reference. Briefly, 50 μL of plasma/serum samples were mixed with 150 μL of methanol to generate protein pellets, which were resuspended with 100 μL of 100 mM ammonium bicarbonate buffer and digested with 200 μg trypsin in 30 min at 60 °C. For detailed procedure, please refer to previously published paper.2,3 Our new approach combines the albumin removal with the pellet digestion described previously. Automation was incorporated to improve throughput. Here, one volume of plasma/ serum samples (e.g., 20 μL) were mixed with ten volumes (e.g., 200 μL) of IPA with 1.0% (by weight) TCA solution. The mixture was vortexed vigorously for 2 min and then centrifuged at 1500g at 5 °C for 5 min. The supernatant was removed using a liquid handler and the protein pellets were washed with 200 μL methanol. The wash was accomplished by resuspending the pellets in methanol, centrifuging at 2000 rpm for 2 min and then disposing of the supernatant with a liquid handler. The final protein pellets were resuspended in 200 μL of 100 mM ammonium bicarbonate aqueous buffer. Digestion was accomplished by adding 500 μg of trypsin (25 μL at 20 μg/ μL) to each sample and gently mixing the solution before incubating at 60 °C for 30 min with the shaker set at 750 rpm. The digestion was stopped by adding 50 μL of IS working solution (prepared in 10% FA in water) to the mixture. A portion, 10−40 μL, of the final solution was injected for LCMS/MS analysis. Comparison between the IPA-TCA Albumin Removal with the Pierce Albumin Depletion Kit. The IPA-TCA approach was compared with the Pierce Albumin Depletion Kit on both efficiency of albumin removal and recovery of the target therapeutic protein. Three therapeutic proteins were spiked simultaneously to a concentration of 50 μg/mL in human plasma. Twenty microliters of the plasma samples were tested in five replicates using either Pierce Kit or IPA-TCA approach to remove albumin. The provided, standard protocol for the Pierce Albumin Depletion Kit was followed. The detailed procedure described in the Sample Digestion section was used for IPA-TCA albumin removal. After albumin removal, the depleted plasma samples were then digested using a similar procedure as that described in the Sample Digestion section.

■

RESULTS Challenges Related to Whole Plasma Digestion. Two of the common challenges confronting scientists during quantitative determination of the digestion of the plasma/ serum samples are (1) limitations in the amount (e.g., microliters of plasma/serum samples) of each plasma/serum sample that can be processed while maintaining consistency and (2) the lack of corresponding sensitivity improvement expected from injecting larger amounts of the extracts into a LC-MS/MS system. Both of these issues limit the ultimate sensitivity obtainable using a LC-MS/MS method.2 It is not difficult to understand that digestion efficiency may plateau and not improve linearly with increased amount of sample. While more enzymes could be added to maintain the same protein to enzyme ratio, trypsin has a tendency to undergo autolysis (selfdigestion). More concentrated enzymes will also lead to faster autolysis and result in more interference from the trypsin itself. Lower concentration and longer digestion time is an alternative but may not work under many conditions since some type of trypsin (e.g., from different vendors) loses activities quickly. In addition to this digestion limitation, another significant issue hampering sensitivity is the lack of a linear relationship between instrument peak response and increasing injection volume for extracts obtained from direct pellet digestion (see the Supporting Information Figure 1s). In this experiment, 50 μL plasma samples containing the three target proteins was pellet digested. The final extracts were analyzed using injection volumes of 1, 2, 5, 10, 20, 35, and 50 μL. As shown in Supporting Information Figure 1s, the absolute instrument responses for all three signature peptides reached a plateau at an injection volume of 10 μL and increased injection volumes beyond 10 μL did not result in further improvement in mass spectrometer response. This observation is similar to another previously reported experiment where increased chromatography background and absence of sensitivity gain was observed at larger injection volumes for a different protein.2 As demonstrated in the previous publication,2 this phenomenon is believed to be caused by matrix ion suppression in the mass spectrometer where any signal gain from the increased analyte load was offset by an increase in ion suppression. Further, due to large amount of digested peptides present in the extracts, high resolution chromatography capacity became saturated and resulting in reduced ability to resolve the peptide of interest from background analytes. We postulated that providing a better sample cleanup before digestion will decrease the sample mixture complexity; thereby improve specificity and sensitivity of the final assay. We further believe that digestion efficiency would improve by removing competing, unwanted proteins in the sample matrix. To this end, we focused our efforts in developing a simple and effective 8338

dx.doi.org/10.1021/ac501837t | Anal. Chem. 2014, 86, 8336−8343

Analytical Chemistry

Article

approach to remove albumin, the most abundant endogenous protein, from plasma/serum samples before the digestion step. LC-MS/MS Method Development for Monitoring Serum Albumin. To assess the impact of albumin removal has on quantitative protein analysis, we required a simple method to monitor albumin in the extracts. To this end, we developed a semiquantitative LC-MS/MS method using two surrogate peptides from the serum albumin. Since there are no synthetic reference standards available for both peptides, we confirmed their identities via their tandem mass spectrum (see Supporting Information Figure 2s). The determination is considered semiquantitative since no standard curves were used for albumin quantitation. The MS responses from the signature peptides in each sample were used to estimate the amount of albumin present. Since the data from both peptides (QTAL, LVNE) agree from the method development experiments, we only used MS responses from the peptide QTAL for all of the subsequent comparisons. For comparison purpose, we have assumed the digestion efficiency of albumin and MS ionization of the surrogate peptides in all samples are the same. It should be noted there is a possibility that in the cleaner sample following albumin removal, the MS signal of surrogate peptides may be stronger due to less ion suppression and higher digestion efficiency. This is not a concern for us since our assumption will only underestimate the albumin removal efficiency. Specific Albumin Removal Using Acidified Organic Solvents. To obtain the best extraction condition to efficiently remove albumin from plasma/serum samples, we screened combinations of four organic solvents (acetone, methanol, ethanol and IPA) and two different organic acids (FA and TCA) at different concentrations. As shown in Figure 1, both the type of organic solvent and acid have significant impact on albumin removal efficiency and differences were observed across organic solvent types when combined with either TCA or FA. In general, TCA was much better than FA on albumin removal at each condition tested. Overall, all four organic

solvents with 1% or 2% TCA could effectively remove albumin from human plasma. Specifically, as shown in Figure 1a, the amount of albumin detected from pellet digestion with protein precipitation using pure methanol (labeled as MeOH_Ref in Figure 1a) was defined as reference or 100% albumin remaining. Each data point is an average of three measurements and %CV is shown as an error bar at each data point. The most efficient removal was obtained at a concentration of 1% or 2% TCA in all of the solvents. It should also be noted that as the TCA concentration further increased to 5%, the amount of albumin remaining increased. This clearly indicated that there is a sweet spot for TCA concentration. Specifically, a combination of IPA with 1% TCA removed about 95% of total albumin from plasma. For FA, as shown in Figure 1b, its ability to remove albumin was not as efficient as those obtained by TCA. The best removal condition was observed with ethanol with 5% FA, for which about 80% of albumin was removed. Based on these results, TCA was chosen for target protein recovery evaluation. Target Protein Recovery with Albumin Removal. Our goal was to maximize albumin removal, while retaining the analyte of interests. Since it was demonstrated that the combination of TCA with different organic solvents was much better than FA at albumin removal, we focused on using TCA for our recovery experiments. Three therapeutic proteins, BMS-A (mAb), BMS-B, (dAb) and BMS-C (fusion protein), were monitored for recovery in combinations of different the organic solvents and different TCA concentrations. Again, as shown in Figure 2, the recovery of all three target proteins using a methanol protein precipitation method was defined as the reference or 100% (labeled as MeOH_Ref in each Figure 2). The peak area ratio of each surrogate peptide to its SIL-IS (SP/IS) under each testing condition was compared with the peak area ratio in the reference sample to estimate the relative recovery. In general, excellent recoveries were observed across all conditions tested for all three proteins. Specifically, as show in Figure 2a, IPA demonstrated good recoveries for both BMSA and BMS-B with more than 100% for BMS-A and close to 100% for BMS-B. The observed more than 100% relative recoveries for BMS-A can be explained by increased digestion efficiency after albumin removal. As shown in Figures 1a and 2a, the relative recovery increase correlated well with the albumin removal. For BMS-C, a reasonable recovery (60−80%) was observed for IPA with 1% or 2% TCA. Recovery around 60% for BMS-C was demonstrated later on to provide sufficient ruggedness to the bioanalytical method. In general, 1% or 2% TCA gave comparable results with regard to albumin removal and target proteins recovery. Although 2% TCA gave a slightly better recovery than 1% TCA for BMS-C, the variation was also bigger. On the basis of the results for both albumin removal and target analyte recovery, 1% TCA was chosen to achieve homogeneity of variance. Comparison between the IPA-TCA Method with a Commercial Kit. A Pierce albumin depletion kit was used to compare with the new IPA-TCA method on both albumin removal and target analyte recovery. For the Pierce kit, the kitprovided protocol was followed for sample preparation. As shown in Supporting Information Figure 3s-a, the IPA-TCA method was as effective as the Pierce kit on albumin removal where close to 98% of the total albumin were removed compared to the reference method. Each data point shown was an average of three measurements and %CV is shown as an error bar at each data point. As shown in Supporting Information Figure 3s-b, IPA-TCA demonstrated better

Figure 1. Albumin removal effect under different combinations of organic solvents and acids (A). Combinations of organic solvents with different concentrations of trichloroacetic acid (TCA). (B) Combinations of organic solvents with different concentrations of formic acid (FA). Albumin detected in pellet digestion (MeOH_Ref) was used as the reference for no albumin removed. Each data point is an average of three measurements. 8339

dx.doi.org/10.1021/ac501837t | Anal. Chem. 2014, 86, 8336−8343

Analytical Chemistry

Article

Figure 2. Relative recoveries of BMS-A (A), BMS-B (B), and BMS-C (C) under different combinations of organic solvents with different concentrations of TCA. The instrument responses for each drug candidate using pellet digestion (MeOH_Ref) were used as the reference for 100%. Each data point is an average of three measurements.

injected at different injection volumes for LC-MS/MS analysis. Their absolute peak areas were compared for sensitivity. As shown in Figure 3, peak area for samples using “pellet digestion” reached a plateau quickly as the injection volume increased, but this is not true for the IPA-TCA method. At an injection volume of 50 μL, the absolute signal for IPA-TCA was twice that of the pellet digestion method. This was because of a less complicated digestion mixture obtained after albumin removal. In addition, because of the fewer total proteins present in the samples before digestion, IPA-TCA would be expected to allow a larger aliquot size of serum sample to be processed. This concept is demonstrated as shown in Figure 3, where the absolute sensitivity was further increased by increasing the sample aliquot size from 50 to 100 μL when using IPA-TCA method. We observed 4−5-fold sensitivity improvement when IPA-TCA approach was used in conjunction with a larger sample volume. Also as expected, by decreasing the total amount of peptides in the final extracts, the LC method ruggedness was significantly improved. When injecting samples

recoveries for all three drug candidates, while less than 50% of BMS-A and BMS-B and less than 5% of BMS-C were recovered when using the Pierce kit compared to the IPA-TCA method. It should be noted that the standard protocol for the Pierce kit was used without method development and it is possible further refinements to the protocol could improve recoveries for these test compounds. The IPA-TCA method was deemed to be more cost-effective, as the cost solvent usage is negligible compared to the Pierce kit. Furthermore, IPA-TCA has the virtue of being automation compatible, therefore, suitable for high throughput bioanalysis. Assessing Albumin Removal’s Impact on Quantitative LC-MS/MS Protein Bioanalysis. To assess the impact of albumin removal on quantitative LC-MS/MS analysis, we used BMS-C as a probe compound to evaluate both sensitivity and assay ruggedness. Pellet digestion with methanol was used as the reference. Specifically, 50 μL of human serum spiked with 5 μg/mL BMS-C were processed using either the IPA-TCA method or the pellet digestion method. The final samples were 8340

dx.doi.org/10.1021/ac501837t | Anal. Chem. 2014, 86, 8336−8343

Analytical Chemistry

Article

precision, sensitivity, selectivity, ruggedness and analyte stability were validated according to current regulatory guidance (FDA15 and EMA16). Assay validation details are provided in the Supporting Information (Table 2s). The assay was used to measure samples from a toxicological study. The LC-MS/MS method itself performed well, where incurred sample reanalysis testing indicated that the reanalysis results of 38 out of 45 samples tested were within 10% from the mean of the initial and the reanalysis results. LC-MS/MS data were also found to be comparable with those generate using a validated immunoassay assay. Representative PK profiles of two animals based on LC-MS/MS and immunoassay data were shown (Figure 4). The PK profiles matched well to each other at the early time points. Specifically, after excluding the three late time points for animal A, the differences between LC-MS/MS results and the mean of LC-MS/MS and immunoassay results for 21 out 26 samples were within ±15.0% of the mean. For the three late time points for animal A, the LC-MS/MS measured concentrations were significantly higher. This could be explained by the severe immunogenicity developed in animal A interfering the immunoassay measurements. As shown in Figure 4, the titer number (an immunogenicity indicator) for animal A was 1 000 000, comparing to 1000 for animal B, suggesting more severe the immunogenicity. Furthermore, we have evaluated the feasibility of using an LC-MS/MS method to support clinical studies, where an LLOQ of