Anal. Chem. 2005, 77, 5529-5533
Investigation of the Immunogenicity of a Protein Drug Using Equilibrium Dialysis and Liquid Chromatography Tandem Mass Spectrometry Detection Qin C. Ji,*,† Ramona Rodila,† Sherry J. Morgan,‡ Rod A. Humerickhouse,§ and Tawakol A. El-Shourbagy†
Department of Drug Analysis, Preclinical Safety, and Global Oncology Development, Abbott Laboratories, Abbott Park, Illinois 66064
Biotherapeutics such as protein and peptide drugs have attracted significant attention in the medical community and pharmaceutical industry in recent years. Immunogenicity is one of the major concerns in the development and application of biotherapeutics. Although great efforts have been put forth in reducing immunogenicity, monitoring the free (“active”) drug concentration and the antibody formation is critical for preclinical and clinical studies. Currently, it is still a challenging task to have a standardized test method monitoring immunogenicity when biotherapeutic compounds such as proteins and peptides are administrated. Combined with liquid chromatography/ tandem mass spectrometry detection, the equilibrium dialysis technique that is conventionally used for measuring the free and bound concentration of small organic molecules was extended to the application of measuring the free and bound concentrations of a protein drug with a relative molecular mass over 10 000 from plasma samples containing antibody. This novel approach could also be used for accurately measuring the antibody concentration when a reference standard of the antibody is available. With recent advances in the field of pharmaceutical research, an increasing number of peptides and proteins have been identified as potential therapeutic compounds for drug development.1,2 Different from traditional small organic compound drug candidates, the biological drug candidates such as proteins and peptides often induce an immune response and antibody formation. The binding of the drug to an antibody may reduce the available free drug concentration in the systemic circulation that is related to possible toxicity effects in preclinical studies and the associated coverage over expected exposures in clinical studies as well as resulting in a potential decrease in clinical efficacy. In addition, * Corresponding author: (phone) (847) 937 5786; (fax) (847) 938 7789; (email)
[email protected]. † Department of Drug Analysis. ‡ Preclinical Safety. § Global Oncology Development. (1) Crommelin, D. J. A.; Storm, G.; Verrijk R.; Leede L. D.; Jiskoot, W.; Hennik, W. E. Int. J. Pharm. 2003, 266, 3-16. (2) Rosenberg, A. S.; Woroboc, A. Biopharm. Int. 2004, 17, 22-26. 10.1021/ac050243f CCC: $30.25 Published on Web 08/09/2005
© 2005 American Chemical Society
in some extreme cases, induced antibodies could lead to serious adverse effects in clinical studies. Although great efforts have been put forward to alter immunogenicity, analytical methods to monitor the concentration of free drugs and the related antibodies are important for understanding the immunogenic risks and effects of these protein and peptide drugs.3 Traditionally, binding assays such as directive binding and competitive ELISA are often used to monitor the protein/peptide drug and anti-drug antibody concentrations. Depending on the antigen or antibody reagent used in the assay, the measurement of the free drug concentration and antibody concentration varies significantly from the true meaning of the intended measurement (free drug concentration versus “active” drug concentration, “neutralizing” antibody versus “total” antibody). Nevertheless, these results are often the best available information for understanding the immunogenicity of the drug candidate. For the drug development of the small organic candidates, although the total drug concentration in plasma is generally measured, it is the free (unbound) drug concentration that equilibrates with the concentration at the site(s) of action. Drug tends to have nonspecific binding with plasma proteins even when there is no antibody generated against small organic drugs. The study of the drug-protein binding is a required component in the traditional drug development process. Ultracentrifugation, ultrafiltration, and equilibrium dialysis are some of the most common techniques for the measurement of the free and bound drug concentrations in plasma or serum.4-9 Because equilibrium dialysis has fewer disturbances to the target system, it is often used as the “reference method” for the determination of free and bound drug concentrations in a drug-protein binding profile.4 However, to our knowledge, there is no publication that reports (3) Chamberlain, P.; Mire-Sluis, A. R. In Immunogenicity of Therapeutic Biological Products; Brown, F., Mire-Sluis, A. R., Eds.; Developments in Biologicals 112; Karger: Basel, 2003; pp 3-11. (4) Oravcova, J.; Bohs, B.; Lindner, W. J. Chromatogr., B 1996, 677, 1-28. (5) Collins, J. M.; Klecker, R. W., Jr. J. Clin. Pharmacol. 2002, 42, 971-975. (6) Lin, Z. J.; Desai-Krieger, D.; Shum, L. J. Chromatogr., B 2004, 801, 265272. (7) Fung, E. N.; Chen, Y. H.; Lau, Y. Y. J. Chromatogr., B 2003, 795, 187-194. (8) Nakai, D.; Kumamoto, K.; Sakikawa, C.; Kasaka, T.; Tokui, T. J. Pharm. Sci. 2004, 93, 847-854. (9) Tsai, T. H. J. Chromatogr., B 2003, 797, 161-173.
Analytical Chemistry, Vol. 77, No. 17, September 1, 2005 5529
the use of equilibrium dialysis and LC-MS/MS for the determination of free concentration of protein drugs over a relative molecular mass of 10 000. Here we report an innovative approach using equilibrium dialysis and LC-MS/MS detection for the measurement of free and bound drug concentrations of a protein drug candidate rK5 in plasma in the presence of anti-drug antibody. rK5 is a small protein with a relative molecular mass of 10 464. It is a specific and potent angiogenesis inhibitor that may be effective in the treatment of human brain glioma and other human tumors.10,11 The efforts described in this report were undertaken with the aim of obtaining an estimate of the free drug concentration in preclinical toxicity study samples in order to provide a conservative estimation of the safety margin and set dosages for a clinical study. rK5 in plasma takes three main forms: (1) free rK5, (2) plasma protein-bound rK5, and (3) antibody-bound rK5. The free rK5 reaches equilibrium on both sides of the dialysis membrane, and the antibody-bound rK5 will remain on the sample side of the membrane. For the preclinical study, we believe that the amount of the antibody-free rK5 concentration (free rK5 plus plasma protein-bound rK5) can be used as the safety coverage for the starting dose of a clinical study. Although this antibody-free rK5 concentration is not achieved strictly by this approach, the dialysis will provide an even lower concentration (free rK5 plus a fraction of the protein-bound rK5 lower than 100K relative molecular mass), which can be used as a guide for the starting dose of the clinical study. In addition, this dialysis method will provide an alternative way of measuring anti-K5 antibody concentrations in plasma samples, based on the binding activity of the hyperimmunized monkey antibody. Furthermore, the innovative approach using equilibrium dialysis and LC-MS/MS for free and bound protein drug concentration and antibody concentration measurement may provide an alternative for the study of immunogenicity for other protein drug candidates. EXPERIMENTAL SECTION Chemicals and Reagents. Guanidine hydrochloride was purchased from Sigma-Aldrich (St. Louis, MO). Sodium chloride (0.9%) water solution was supplied by Abbott Laboratories (North Chicago, IL). High-purity methanol, acetonitrile, and water, all Omnisolv grade, were purchased from EMD, formerly EM Science (Gibbstown, NJ), along with HPLC grade hexanes. Glacial acetic acid was purchased from Aldrich (St. Louis, MO). Trifluoroacetic acid was purchased from EM Science. The stock solutions of rK5 and internal standard were produced at Abbott Laboratories (Abbott Park, IL). Normal Cyno plasma with potassium EDTA as anticoagulant and normal Cyno serum were purchased from Lampire Biological Laboratories (Pipersville, PA). Instrumentation. Microequilibrium dialyzers, two-chamber systems with a 100-µL chamber volume, together with the 100 000 MWCO cellulose acetate membranes were purchased from Harvard Apparatus (Holliston, MA). A reciprocating shaker from (10) Lu, H.; Dhanabal, M.; Volk, R.; Waterman, M. J. F.; Ramchandran, R.; Knebelmann, B.; Segal, M.; Sukhatme, V. P. Biochem. Biophys. Res. Commun. 1999, 258, 668-673. (11) Davidson, D. J.; Haskell, C.; Majest, S.; Kherzai, A.; Egan, D. A.; Walter, K. A.; Schneider, A.; Gubbins, E. F.; Solomon, L.; Chen, Z.; Lesniewski, R.; Henkin, J. Cancer Res 2005, 65, 4663-4672.
5530 Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
Figure 1. Schematic of microequilibrium dialysis of rK5.
Eberbach Corp. (Ann Arbor, MI) was used for dialyzer incubation. Plasma solutions were transferred using a single-channel positive displacement handheld pipet from Gilson (Middleton, WI). Multichannel handheld electronic pipets were from BioHit (Helsinki, Finland) for the manual addition of reagent for solid-phase extraction (SPE). The SPE plates were from Waters Corp (Milford, MA). A Hamilton (Reno, NV) MicroLab AT 2 Plus automated liquid handler was used for the addition and mixing of the guanidine hydrochloride solution and the internal standard. A Beckman-Coulter (Fullerton, CA) square well plate collar and vacuum manifold base were used in the SPE process. The Shimadzu (Kyoto, Japan) HPLC system included a Shimadzu LC10 AD HPLC pump, a Shimadzu SIL-10A XL autosampler, and a Shimadzu SCL-10A system controller. A Hot Pocket column heater from Keystone Scientific (Bellefonte, PA) was also used. The valves used to control LC flow between the mass spectrometer inlet and waste line were from Valco Instruments (Houston, TX). An 1100 series HPLC pump and degasser system from Agilent Technologies (Palo Alto, CA) was used to deliver backwash solvent for the precolumn regeneration. A 2.1 × 150 mm Symmetry 300 C18, 5-µm column from Waters was used as the analytical column, along with a precolumn consisting of a 3.9 × 20 mm Symmetry 300 C18, 5-µm cartridge, also from Waters, and an inline filter with an A-110 × 2 µm titanium frit from Upchurch Scientific Inc. (Oak Harbor, WA). An API-3000 mass spectrometer and the computer control system were from PE Sciex (Toronto, ON, Canada). MassChrom version 1.1.1 (or Analyst version 1.3.2) was used as the data acquisition software. Equilibrium Dialysis. The schematic describing the equilibrium dialysis experiment is shown in Figure 1. A prewet membrane separates two chambers of identical volume. At the beginning, the sample that needs to be dialyzed is added into the sample side (sample chamber) and the product side (assay chamber) is filled with dialysis solution. The dialyzer is put into a horizontal shaker and kept at the desired temperature, while the drug of interest equilibrates between the two chambers. The membrane allows only the free drug to pass through while the bound drug, being too large, is retained on the sample side of the membrane. Multiple samples can be dialyzed at same time (each using one dialyzer) using multiple dialyzers. The concentrations of the analyte in both sample side and product side are analyzed using SPE and LC-MS/MS detection as described in the following section. Once the equilibrium is reached, the free drug concentration can be calculated as twice the concentration found in the product side of the dialyzer, while the bound drug
concentration can be calculated as the difference between the concentration found in the sample side of the dialyzer and the concentration found in the product side. Percent dialyzed (another parameter for the description of the amount of analyte dialyzed) is calculated as the percentage of drug amount found in the product side divided by the total amount of drug of both sides. The percent equilibration at a specific dialysis time is calculated as the percent dialyzed at the specific dialysis time divided by the percent dialyzed at equilibrium. Concentration Determination of rK5 Samples from the Dialysis Cell. The samples taken from the sample and product sides were analyzed using SPE and LC-MS/MS analysis, with or without a denaturing step in sample preparation as described in detail in our other publications.12,13 Briefly, a gradient HPLC method was utilized for separation with mobile phases A and B. Mobile phase A consisted of 0.1% acetic acid and 0.02% trifluoroacetic acid in water, and mobile phase B consisted of 0.1% acetic acid and 0.02% trifluoroacetic acid in 80/20 (v/v) acetonitrile/water. The analytical column was maintained at a temperature of 40 °C, and the injection volume was 40 µL. While rK5 gives multiply charged ions at m/z 1308.7(8+), 1495.5(7+), 1744.6(6+), and 2093.2 (5+), the SRM channel of m/z 1495.5(7+) > 1462.9(7+) was used for the LC-MS/MS detection of rK5. 15N-Labeled rk5 was used as internal standard. While 15N-labeled rk5 gives multiply charged ions at m/z 1324.0 (8+), 1514.0(7+), 1766.0(6+), and 2118.0(+), the SRM channel of m/z 1514.0(7+) > 1481.0(7+) was used for the LC-MS/MS detection for internal standard. The peak area ratio of rK5 against internal standard was used for the calculation of the rK5 concentration. When samples from the dialysis of rK5 in saline sample against saline product were analyzed, direct injection of the sample without SPE was used. For the analysis of total drug concentration in plasma samples with both free and bound drug, guanidine hydrochloride was used to denature the complex of rK5 and antibody so the binding between the antibody and rK5 was broken apart during the sample process period. After a 100-µL plasma sample was dialyzed against 100 µL pf saline, 50 µL of solution were taken from each side of the dialyzer for analysis. To compensate for the matrix differences between the analyte solutions before subjected to SPE, 50 µL of saline solution was added to each plasma sample, standard, and QC samples for the quantitation, while 50 µL of blank plasma was added to each saline sample in the 96-well plate. RESULTS AND DISCUSSION Equilibrium Dialysis of rK5 in Saline or Plasma without Antibody. The performance of the dialysis was first evaluated using rK5 diluted in saline solution in the sample side. The product side was filled with blank saline, and then the dialyzers were incubated in a shaker at ∼10 °C. As shown in Figure 2, rK5 moved from the sample side to the product side rapidly at the beginning and gradually slowed thereafter. A complete equilibration with (12) Ji, Q. C.; Rodila, R.; Gage, E. M.; El-Shourbagy, T. A. Anal. Chem. 2003, 75, 7008-7014. (13) (a) Ji, Q. C.; Rodila R.; El-Shourbagy, T. A. Proceedings of the 53rd ASMS Conference on Mass Spectrometry. San Antonio, TX, 2005. (b) Ji, Q. C.; Rodila R.; El-Shourbagy, T. A. A Sample Preparation Process for LC-MS/ MS Analysis of Total Protein Drug Concentration in Monkey Plasma Samples with Antibody. Manuscript in preparation.
Figure 2. Equilibration curve of an rK5 saline solution dialyzed against saline at 10 °C.
Figure 3. Equilibration curve of rK5 plasma samples dialyzed against saline at room temperature.
half and half (50:50) rK5 in both sides was reached when it was tested after 6 days of dialysis. While ∼90% equilibration can be achieved within 48 h, ∼96% equilibration can be reached within 72 h. An extensive experiment of the dialysis of rK5 in plasma was performed while rK5 plasma sample was in the sample side of the dialysis cell and saline was added to the product side of the cell. Saline was selected over other buffer solutions because saline causes minimal pH change of the matrix in the plasma sample side and no additional background material is added to the plasma sample side except salt that is already part of the matrix. Plasma was also tested as the dialysis matrix in the product side; however, the rate of the dialysis was greatly reduced. The result of the dialysis is shown in Figure 3. A total of 91 data points were generated. The rate of dialysis is rapid in the first 48 h and gradually slows down as the dialysis of rk5 reaches equilibration. Αn equilibrium of ∼75% is achieved after 48 h of dialysis and greater than 95% equilibration is achieved in 72 h. At equilibrium, only ∼37% of the rK5 moved to the product side because the rK5 nonspecifically bound to the plasma protein is not available for equilibrium. The percentage of the nonspecific plasma protein binding was not changed from ∼96 to 336 h of dialysis. Similar experiments were performed at 10 and 37 °C. While no significant improvement of the dialysis rate was observed at 37 °C, the dialysis rate was much slower at 10 °C. In addition, the percentage of the nonspecific rK5 plasma binding at equilibration stays approximately same. Dialysis was performed when blank plasma was added to the product side of the dialysis cell and rK5 saline solution was added into the sample side. The concentration of rK5 was higher at the side of the matrix of the plasma and the equilibration with ∼37% rK5 in the saline side was achieved (results not shown). This further proves that the binding of rK5 Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
5531
Figure 4. Regression of found rK5 concentrations from dialyzed and undialyzed calibration standards.
Figure 5. Dialysis reproducibility evaluation of rK5 plasma samples at ∼72 (n ) 4) and ∼96 (n ) 4) h.
with plasma background proteins changes the distribution of rK5 on both sides of the dialyzer. The equilibrium dialysis was further evaluated by changing the rK5 plasma concentration at the sample side; saline was used as the dialysis solution in the product side. The dialysis incubation time at room temperature with shaking was ∼47 h. As shown in Figure 4, the rK5 concentrations in both sample side and product side increased linearly with the increase in the concentration of rK5 in the initial plasma samples. The total rK5 concentration from both sides of the dialyzer was compared with the total rK5 concentration without going through the dialysis process. The excellent agreement between the two curves demonstrates the complete recovery of rK5 of the dialysis processes and no loss of the rK5 by binding to the dialysis cell or membrane. The reproducibility of the dialysis of rK5 in plasma was evaluated. As shown in Figure 5, multiple samples with rK5 in plasma were dialyzed against saline solution for ∼72 and ∼96 h at room temperature. The reproducibility of the percent of rk5 dialyzed into the product side was evaluated using multiple samples with a same sample concentration. The standard deviation of percent rK5 distribution in both sample side and product side was 1.1 at ∼72 h and 2.5 at ∼96 h. Equilibrium Dialysis of rK5 in Plasma with Antibody. rK5 samples with antibody were made by adding a known amount of rK5 in blank plasma and spiking with hyperimmunized monkey serum. The samples were added in the sample side of the dialyzers, and saline was added to the product sides. After the completion of the dialysis, the samples from both sides of the dialyzer were initially analyzed using the SPE LC-MS/MS method without the “denaturing” step.12 Because of binding of the antibody to rK5 as well as to the internal standard, the internal standard response was significantly lower than the internal 5532 Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
Figure 6. Measured rK5 free and bound concentration profile after 4 days of dialysis for rK5 plasma sample containing different concentrations of anti-rK5 antibody.
standard used in the calibration curve. In addition, the recovery of the total rK5 concentration from both sides of the dialyzer was much lower than when the same rK5 sample without spiked antibody was analyzed. The experiment was repeated, and samples from both sides of the dialyzer were analyzed with a denaturing step.13 The denaturing enabled the dissociation of the binding between the rK5 and anti-rK5 antibody and therefore allowed for the measurement of total rK5 from samples in the presence of anti-rK5 antibody. As shown in the Figure 6, when increasing amounts of the hyperimmunized monkey antibody were added to plasma samples containing the same amount of rK5, the free rK5 concentration was decreasing while bound rK5 concentration was increasing continuously until no free rK5 could be detected. Since serum antibody was added to some of the plasma samples, blank serum was added to all other samples to compensate for the difference of the background matrixes in the sample side of the dialysis chamber. The total bound rK5 concentration measured here likely resulted from the rK5 bound to anti-rK5 antibody and to nonspecific plasma background protein. Therefore, the free concentration here is less than the actual “active” drug concentration. We believe that this free drug concentration obtained from preclinical study samples will provide a most conservative estimation of the safe concentration coverage for the initial dose of the clinical studies. Furthermore, as shown in Figure 6, the distribution of rK5 in the sample and product sides is directly related to the amount of the antibody in the plasma. These response curves could be used for the quantitative analysis of antibody concentration in an unknown plasma sample, assuming the binding activity is equivalent to the antibody standard used to generate the calibration curve. By optimization of the amount of rK5 in the plasma used for creating a response curve of the antibody amount versus either bound or free rK5 concentration, the amount of the antibody in the sample can be accurately determined. Compared to traditional binding assays, this could provide more information for understanding the immunogenicity of the protein drug compound. Even when no antibody reference standard is available, a significant increase of the bound drug may indicate the generation of an antibody. The amount and strength of the antibody may change due to sample collection and the variation between study subjects. However, the experiment by measuring the free and bound rK5 could be used for the standardization of this antibody-antigen binding. Stability and Reproducibility. The following experiment was performed for the evaluation of the binding stability between rK5
for the above experiment. The CVs of bound rK5 concentration are 6.4 and 4.6% for stability and control samples, respectively, the CVs of free rK5 concentration are 35.0 and 15.1% for stability and control samples, and the CVs of total rK5 concentration are 5.6 and 5.0% for stability and control samples.
Figure 7. Evaluation of the reproducibility of 4 days of dialysis and 5 days binding stability of rK5 with anti-rk5 antibody.
with antibody in room temperature. One set of the rK5 and antibody mixture samples (stability samples) were stored in the sample side of a dialyzer and kept at room temperature. Five days later, new sets of rK5 and antibody mixture were prepared as control samples. Both stability and control samples were set for dialysis at the same time. The dialysis time was ∼4 days. The free and bound rK5 concentrations were obtained for both stability and control samples. As shown in Figure 7, there is no significant difference of free and bound rK5 concentrations between control and stability sample. This indicates that the binding between rK5 and antibody is stable. The unique stability property of rK5 and the binding between rK5 and antibody allows for the determination of the free and bound rK5 concentrations after being dialyzed for at least 5 days. However, dialysis for 3 days should still give acceptable results as 95% of the equilibration is achieved as shown in Figure 6. Development of new types of dialyzers may allow for faster dialysis for these applications. The reproducibility of the dialysis was further evaluated with four replicates of both stability and control samples being used
CONCLUSIONS While it is still quite a challenge to have a standardized test method for monitoring immunogenicity when biotherapeutic compounds such as proteins and peptides are administered, here we report a complimentary approach to the conventional binding assay. Combined with liquid chromatography tandem mass spectrometry detection, an equilibrium dialysis technique, which is conventionally used for measuring free and bound concentrations for small organic molecules, was extended for the application of measuring the free and bound concentrations of a protein drug with a relative molecular mass over 10 000 in plasma samples with antibody. The work also demonstrated that this novel approach could be used to measure the antibody concentration accurately when a reference standard of the antibody is available. With further development of the dialysis membrane and related techniques for reducing the dialysis time, this approach could find a wide application in monitoring the immunogenicity for preclinical and clinical studies. ACKNOWLEDGMENT The authors thank Drs. Donald J Davidson, Jack Henkin, Robert Carr, and Laurie Iciek for the helpful discussions and Joseph C. Kim for performing part of the experiments. Received for review February 8, 2005. Accepted July 6, 2005. AC050243F
Analytical Chemistry, Vol. 77, No. 17, September 1, 2005
5533
