Using Mass Spectrometry to Quantify Rituximab ... - ACS Publications

Subscriber access provided by UNIV OF NEBRASKA - LINCOLN

Article

Using Mass Spectrometry to Quantify Rituximab and Perform Individualized Immunoglobulin Phenotyping in ANCA-associated Vasculitis John R. Mills, Divi Cornec, Surendra Dasari, Paula M. Ladwig, Amber M. Hummel, Melissa Cheu, David L Murray, Maria Alice Willrich, Melissa R. Snyder, Gary S. Hoffman, Cees G.M. Kallenberg, Carol A. Langford, Peter A. Merkel, Paul A. Monach, Philip Seo, Robert F. Spiera, E. William St. Clair, John H. Stone, Ulrich Specks, and David R. Barnidge Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b00544 • Publication Date (Web): 26 May 2016 Downloaded from http://pubs.acs.org on May 26, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

msID_ac-2016-0544w_Manuscript File Using Mass Spectrometry to Quantify Rituximab and Perform Individualized Immunoglobulin Phenotyping in ANCA-associated Vasculitis

John R. Mills1, Divi Cornec2,3, Surendra Dasari4, Paula M. Ladwig1, Amber M. Hummel2, Melissa Cheu5, David L. Murray1, Maria A. Willrich1, Melissa R. Snyder1, Gary S. Hoffman6, Cees G.M. Kallenberg7, Carol A. Langford6, Peter A. Merkel8, Paul A. Monach9, Philip Seo10, Robert F. Spiera11, E. William St. Clair12, John H. Stone13, Ulrich Specks2, *David R. Barnidge1 1

Department of Laboratory Medicine and Pathology, Mayo Clinic Rochester, MN, 55905 Division of Pulmonary and Critical Care Medicine, Mayo Clinic Rochester, MN, 55905 3 Rheumatology Department, Brest University Hospital, 29609 Brest, Cedex, France 4 Department of Health Sciences Research, Mayo Clinic Rochester, MN, 55905 5 Genentech Inc., South San Francisco, CA, 94080 6 Cleveland Clinic Foundation, Cleveland, OH, 44195 7 University of Groningen, 9712 CP Groningen, the Netherlands, 8 University of Pennsylvania, Philadelphia, PA, 19104 9 Boston University Medical Center, Boston, MA, 02115 10 Johns Hopkins University, Baltimore, MD, 21218 11 Hospital for Special Surgery, New York, NY, 10021 12 Duke University, Durham, NC, 27710 13 Massachusetts General Hospital, Boston, MA, 02114 2

*Corresponding Author: Dr. David R. Barnidge Mayo Clinic 200 1st St SW Rochester, MN 55905 Phone: (507) 266-4777 Email: [email protected]

Running title: monoclonal therapeutic antibody, mAb, quantification, serum, mass spectrometry, liquid chromatography, ELISA, rituximab, miRAMM 1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT Therapeutic monoclonal immunoglobulins (mAbs) are used to treat patients with a wide range of disorders including autoimmune diseases. As pharmaceutical companies bring more fully humanized therapeutic mAb drugs to the healthcare market analytical platforms that perform therapeutic drug monitoring (TDM) without relying on mAb specific reagents will be needed. In this study we demonstrate that liquidchromatography-mass spectrometry (LC-MS) can be used to perform TDM of mAbs in the same manner as smaller non-biologic drugs. The assay uses commercially available reagents combined with heavy and light chain disulfide bond reduction followed by light chain analysis by microflow-LC-ESI-Q-TOF MS. Quantification is performed using the peak areas from multiply charged mAb light chain ions using an in-house developed software package developed for TDM of mAbs. The data presented here demonstrate the ability of an LC-MS assay to quantify a therapeutic mAb in a large cohort of patients in a clinical trial. The ability to quantify any mAb in serum via the reduced light chain without the need for reagents specific for each mAb demonstrates the unique capabilities of LC-MS. This fact, coupled with the ability to phenotype a patient’s polyclonal repertoire in the same analysis further shows the potential of this approach to mAb analysis.

2


Page 2 of 30

Page 3 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


INTRODUCTION During the last 20 years, therapeutic monoclonal antibodies (mAbs) have emerged as one of the main new classes of drugs. Therapeutic mAbs have radically changed the prognosis of patients with a wide range of disorders allowing them to live longer and healthier lives. As a class of drugs, mAbs have the exceptional ability to target specific antigens resulting in improved efficacy and fewer side effects compared to other drugs. Accordingly, pharmaceutical companies are continually bringing therapeutic mAbs to the healthcare market. However, the clinical response to mAbs varies from patient to patient creating a need for validated analytical methods for therapeutic drug monitoring (TDM) of mAbs to guide patients’ care towards individualized therapy. The study of mAb pharmacokinetics is typically the first step in defining an individual’s response to a drug. However, the relationships between serum concentration and therapeutic efficacy, as well as unwanted side effects of the drugs, are ill defined. Analytical methods that can provide the greatest level of information from a single analysis will create the most efficient path to understanding the sources of variability in dose–response relationships between patients. To date, immunoassays (i.e. ELISA) have been the primary analytical tool for quantifying therapeutic mAbs in biologic fluids. These assays are fast, inexpensive to run, and are routinely used in clinical laboratories. However, they rely on reagents designed specifically for each therapeutic mAb, such as recombinant target antigen or anti-idiotypic capture antibodies. The time needed to develop these reagents may range from months to years. In most cases therapeutic mAb specific immunoassays are not commercially available since the assays are developed by pharmaceutical companies in-house as 3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

part of the regulatory approval process. Consequently, laboratories interested in TDM of mAbs must develop their own immunoassay for each therapeutic mAb that comes to market. Our group has focused on developing mass spectrometry (MS) based analytical methods that can quantify any therapeutic mAb without the need for mAb specific reagents in a format that can be used in a clinical laboratory setting. Our group’s first mass spectrometry (MS) based method used the clonotypic approach to quantify mAbs in human sera for therapeutic drug monitoring of infliximab in clinical samples 1. This approach is based on quantifying a tryptic peptide from the variable region of the heavy chain and/or light chain from the mAb (i.e. clonotypic peptide) using a multiplexed LC system connected to a triple quadrupole mass spectrometer. This MS platform is commonly used in clinical laboratories for TDM of smaller drugs but has recently been adapted for quantifying endogenous proteins using protein specific tryptic peptides (i.e. proteotypic peptides) 2-8. However, when applied to therapeutic mAbs, the clonotypic peptide approach is limited to quantifying only chimeric mAbs that contain non-human CDR regions that will produce quantifiable peptides when cleaved with an enzyme. Pharmaceutical companies are bringing fully humanized mAbs to market eliminating the clonotypic peptide approach to mAb TDM 9. Our group concluded that a methodology originally developed in our laboratory for identifying a monoclonal immunoglobulin in patients with a monoclonal gammopathy would be ideally suited to quantify therapeutic mAbs. The method, referred to by its acronym miRAMM – monoclonal immunoglobulin Rapid Accurate Mass Measurement, uses LC-MS to identify and quantify monoclonal immunoglobulins in human serum and urine 10,11. The methodology is different than 4


Page 4 of 30

Page 5 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


most LC-MS based TDM analyses that typically monitor a select group of exogenous drug related compounds. In contrast, miRAMM monitors not only the therapeutic mAb of interest, but also the endogenous polyclonal light chains present in the sample, all in the same analysis 12,13. The technique is based on measuring molecular mass, a first principle of monoclonal immunoglobulins due to the fact that each monoclonal immunoglobulin has a fixed amino acid sequence and therefore has an exact molecular mass that can be monitored using a mass spectrometer. In addition, therapeutic mAbs are homodimers and therefore have equimolar concentrations of heavy and light chains allowing for the concentration of one chain to represent the concentration of the whole molecule. The light chain portion is used for quantification since heavy chain glycosylation creates heterogeneity in molecular mass complicating quantification. Light chains are detected in serum using a commercially available immunoglobulin purification step combined with heavy and light chain disulfide bond reduction followed by light chain analysis by microflow-LC-ESI-Q-TOF MS. Assay specificity is obtained by combining the known retention time with the known mass/charge values of multiply charged ions specific to the therapeutic mAb. The concentration of the mAb is determined using the peak areas from the specific multiply charged light chain ions found using a unique software package developed in-house for therapeutic drug monitoring of mAbs. In this study, we used miRAMM to quantify serum levels of rituximab (Rituxan®), a therapeutic mAb that induces depletion of B cells by binding to CD20 14. One goal of the study was to evaluate the analytical performance of miRAMM using serial serum samples from a large therapeutic trial (total samples = 775) 15,16. The results 5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

demonstrated that rituximab concentrations found by miRAMM were in good agreement with those found by ELISA (R2=0.853/m=0.70) and that the accuracy and precision of the miRAMM based assay was within acceptable limits. Moreover, the endogenous polyclonal kappa and lambda light chain phenotypes were examined over the course of treatment in patients and examples are presented that demonstrate the unique adaptive capabilities of miRAMM as a quantitative and qualitative clinical diagnostic tool.

EXPERIMENTAL SECTION Safety Considerations All patient samples and flammable reagents should be handled with the appropriate personal protective equipment. Patient samples This study utilized serum samples from the 99 rituximab-treated patients included in the rituximab versus cyclophosphamide for the treatment of ANCA-associate vasculitis in the RAVE trial 15,16. All patients gave their informed consent to participate in the trial and for the future use of biospecimens collected during the conduct of the trial and the study was conducted in accordance with the Declaration of Helsinki. Serum was stored at 80°C until use. Overall, 775 serum samples were analyzed, 98 of which were sampled before treatment with rituximab and therefore serve as true negatives. Reagents Ammonium bicarbonate, dithiothreitol (DTT), and formic acid (FA) were purchased from Sigma-Aldrich (St. Louis, MO). Water and acetonitrile were purchased from Honeywell Burdick and Jackson (Muskegon, MI). Melon™ Gel and CELLect pooled human serum 6


Page 6 of 30

Page 7 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


were purchased from Thermo Fisher Scientific (Waltham, MA). Therapeutic mAbs (rituximab and vedolizumab) were obtained from the Mayo Clinic pharmacy. Vedolizumab is a humanized immunoglobulin IgG1-kappa monoclonal antibody and was used as an internal standard (IS) to correct for imprecision in the volume of melon gel dispensed for each sample and was used at a fixed concentration 50 µg/mL 17.

ELISA for rituximab quantification An enzyme-linked immunosorbent assay (ELISA) validated in-house at Genentech was used to measure rituximab concentrations in serum. Briefly, patient samples, quality controls, and standards were diluted 100-fold prior to incubation for 1 hour on plates pre-coated with polyclonal goat anti-rituximab antibodies (Genentech, Inc., South San Francisco, CA, USA), followed by washing. Bound samples were detected by incubation with goat anti-mouse IgG F(ab')2 conjugated to horseradish peroxidase (Jackson ImmunoResearch, Inc., West Grove, PA, USA). Following a further wash to remove any unbound conjugate, a substrate solution (tetramethylbenzidine/hydrogen peroxide) was added to the wells, resulting in a color development in proportion to the amount of rituximab in the samples. The reaction was stopped, and absorbance was measured photometrically. Patient samples that did not fit within the measureable range of the assay (5 ng/mL to 125 ng/mL) were rediluted accordingly. The lower limit of quantification (LLOQ) for the assay was 0.5 µg/mL. Sample preparation for mass spectrometry Serum was enriched for IgG immunoglobulins using Melon™ Gel (Thermo Fisher Scientific, Waltham, MA). Briefly, a volume of 20 µL of serum was mixed with 20 µL of 7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

vedolizumab internal standard dissolved in 50 mM ammonium bicarbonate. The sample and IS were added to 160 µL of gel and allowed to shake on an orbital mixer for 5 minutes. After mixing, the gel was allowed to settle for 5 minutes and 20 µL of the supernatant containing the immunoglobulins was removed and heavy and light chain disulfide bonds were reduced by adding 10 µL of 200 mM DTT along with 20 µL of 50 mM ammonium bicarbonate buffer. The solution was vortexed for 5 seconds and then incubated at 55ºC for 15 minutes before injection. Samples were placed into 96 deepwell PCR plates kept at 9ºC in an autosampler. Standard curves were obtained for each plate by diluting known quantities of rituximab and vedolizumab IS in pooled normal serum with a concentration range of; 10, 50, 100, and 250 µg/ml. Two levels of quality controls (QC), intermediate (50 µg/mL) and high (200 µg/mL), were included in each plate and were run before and after patient samples. Liquid chromatography An Eksigent Ekspert 200 microLC (Foster City, CA) was used to separate immunoglobulin light chains prior to MS detection. Mobile phases were aqueous mobile phase A (water+0.1% FA), and organic mobile phase B (80% acetonitrile+10% 2propanol+0.1% FA). A 2 µL injection was made onto a 1.0 x 75 mm Poroshell 300SBC3 column with 5 µm particle size flowing at 25 µL/minute kept at 60°C in a column heater. A 25 minute gradient was started at 80%A / 20%B, held for 1 minute, ramped to 75%A / 25%B over 1 min, then ramped to 65%A / 35%B over 10 min, then ramped to 50%A / 50%B over 4 min, then ramped to 5%A / 95%B over 2 min held for 5 min, then ramped to 80%A / 20% over 1 minute, then equilibrating at 80%A / 20% for 1 minute.

8


Page 8 of 30

Page 9 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ESI-Q-TOF MS Spectra were collected on an SCIEX TripleTOF 5600 quadrupole time-of-flight (Q-TOF) mass spectrometer (SCIEX, Concord, Canada) in electrospray ionization (ESI) positive mode with a Turbo V dual ion source with an automated calibrant delivery system (CDS). Source conditions were: ionspray voltage: 5500, temperature: 500°C, gas settings: curtain 45, gas 1: 35, gas 2: 30, and a collision energy setting of 10. TOF MS scans were acquired from m/z 600-2500 with an acquisition time of 100 milliseconds in high sensitivity mode. Under these instrument conditions multiply charged light chain ions from rituximab and vedolizumab had a resolution of approximately 10,000 while the average molecular mass calculated for rituximab and vedolizumab using light chain ion peak centroids with mass measurement accuracy < 10 ppm. The instrument was calibrated every 5 injections through the CDS using calibration solution supplied by the manufacturer. Data analysis Analyst TF v1.6 was used for instrument control. Data were viewed using Analyst TF v1.6 and PeakView v1.2.0.3. The m/z ranges for each peak representing the +11 through +20 charge states for rituximab and vedolizumab were found using material obtained from the pharmacy. In-house developed software (miRAMMWARE) was used to determine the peak area for each of the +11 through +20 charges states obtained from summed mass spectra. The summed mass spectra were acquired by an automated summing operation using the LC retention time window for the rituximab light chains (5.0±0.1) and vedolizumab light chains (6.0±0.1 minutes). Each charge state peak present in summed mass spectra was modeled using a cubic smoothing spline 9



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

over a quadratic background. The area under the fitted curve for each charge state from each kappa light chain was recorded independently in a mass spectrum and as a value in an Excel spreadsheet. This process was repeated for all raw files. A rituximab/vedolizumab peak area ratio was computed for the kappa light chain from each mAb using the summed peak area values for charge states +11 to +20. This ratio was transformed into rituximab concentrations according to the standard curves built with the data of the wells containing known concentrations of rituximab in each plate.

RESULTS and DISCUSSION Standards in normal pooled serum The total ion chromatogram (TIC) and selected ion chromatograms (SIC) for a 50 µg/mL standard of rituximab spiked into normal pooled serum analyzed by miRAMM can be seen in Figure S1. The retention time window highlighted in gray in the TIC in the figure represents the retention time window where mass spectra were summed to be used by the miRAMMWARE program to determine peak areas for quantitative analysis. Representative mass spectra from this process are shown in Figure 1 where +11 through +14 charge states from the light chains of rituximab (top) and vedolizumab (bottom) are labeled and clearly visible above the polyclonal kappa and lambda normally distributed molecular mass peaks as previously described 12. Figure 2 shows the output from miRAMMWARE displaying the modeling of charge state peak shapes to determine peak areas along with the corresponding peak area values for the +11 through +20 charge states that were summed and then used for quantification. The performance of miRAMMWARE was compared to manual peak integration using the 10


Page 10 of 30

Page 11 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


peak integration script in Analyst TF. Two sets of rituximab standards run on two different days were used to compare the two integration methods. Peak areas determined using the two methods were in excellent agreement as observed by linear regression analysis (R2 = 0.9999 and slope = 0.9832). MS-based assay characteristics and validation Intraday imprecision was evaluated by preparing 10 individual medium level (50 µg/mL) QC samples analyzed in succession. For each sample miRAMMWARE calculated the peak area for charge states +11 through +20 (10 per sample) and summed them together. The resulting %CV for these 10 independent summed peak area calculations was 5.0% for rituximab and 7.6% for vedolizumab. Interday imprecision was calculated using the rituximab concentrations determined for 10 medium (50 µg/mL) and 10 high level (200 µg/mL) QC samples taken over 6 weeks. Each of the 10 plates analyzed contained standards, QC samples, and patient samples. The analysis/preparation dates for each plate were – Plate 1 (12/12/2014); Plates 2, 3, and 4 (12/15/2014); Plates 5 and 6 (12/22/2014); Plate 7 (1/7/2015); Plate 8 and 9 (1/12/2015) and Plate 10 (1/20/2015). QC sample concentrations were determined using the standard calibration curves aliquoted at the start of the study and then prepared for each plate over the course of 6 weeks. The interday %CV values were 16.4% for the medium QC and 15.6% for the high QC. Linearity of the standard calibration was assessed for each plate analyzed over the 6 week period. Linear regression analysis of these standards gave R² values over a range of 0.987-0.999 with a mean of 0.994. The known concentrations for each standard were compared to the values found using the linear regression equation derived from the standard curve for each plate and these values 11



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

are shown in Table S1. Included in the table are the average concentrations observed for the standards over the 6 week period along with their respective %CV values. The %CV values range from very good (0.8 for the 250 µg/mL standard) to acceptable (19 for the 10 µg/mL standard). The large spread in the imprecision of the standards demonstrates that the miRAMMWARE algorithm had difficulty determining the peak areas for rituximab at low concentrations due to the low signal-to-noise (S/N) ratio at that concentration. Figure 3 illustrates this point by showing an unsmoothed mass spectrum labeled A for the Blank along with a 10 µg/mL standard labeled B. In both A and B the +11 charge states of the polyclonal kappa and lambda light chain molecular mass distributions are clearly visible and are labeled κ and λ. In section B the monoclonal kappa light chain from rituximab is labeled with an arrow and is visible above the polyclonal background. Below each mass spectrum are the miRAMMWARE peak area determinations for the +11 charge state. The background observed in 10 Blank samples, one from each of the 10 plates was used to determine the LLOQ. The LLOQ was set as the mean concentration determined for these 10 Blanks plus 1 standard deviation (S.D.). The mean for the 10 Blanks was found to be 6.9 µg/mL with an S.D. of 2.8 µg/mL resulting in an LLOQ of: 6.9 + (2.8) = 9.7 µg/mL. The specificity of miRAMMWARE was assessed using the 98 patient baseline serum samples obtained before the first rituximab infusion. A total of 91 of these samples were negative according to the concentrations obtained using the miRAMMWARE peak area calculations. However, 7 of these samples had peak areas as determined by the miRAMMWARE software that resulted in rituximab concentrations above the LLOQ. The raw data from these samples were manually inspected and it was found that the 12


Page 12 of 30

Page 13 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


miRAMMWARE software had incorrectly assigned peak areas for several of the charge states in these samples resulting in a false positive result. Future versions of the miRAMMWARE software will incorporate modifications to the peak selection algorithm to eliminate false positive results. Comparison of ELISA and MS results The concentrations of rituximab found using miRAMM and ELISA were compared using a least squares linear regression plot and a Bland-Altman difference plot and the results are shown in Figure S2. The RAVE trial patient samples used in the figure (n=411) had rituximab concentrations greater than the LLOQ for miRAMM. The results using the two methods had a correlation coefficient of R²=0.8308 with miRAMM showing higher concentrations than ELISA as illustrated by a slope of 0.72. The differences in the rituximab concentrations generated by miRAMM vs. ELISA over the time course of the study were also evaluated using a paired t-test assuming equal variances and normally distributed data. The results were compiled for the sample dates Week 2, Week 4, Month 2, Month 4, and Month 6 after infusion and are displayed in Table S2. The pvalues were calculated for each methodology using the mean rituximab concentrations from patient serum ± the standard error of the mean at each time point. The results in Table S2 show that the means are statistically different between the two methodologies, except for Month 6, when determined by miRAMM compared to ELISA. Samples with large discrepancies between the rituximab concentrations observed by miRAMM vs. ELISA were reexamined to determine if any bias was apparent in the miRAMMWARE software integration. A total of 14 outliers were checked where the rituximab concentration observed by miRAMM was >100 µg/mL higher than 13



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

by ELISA. In all outlier cases the peak area integration calculated by miRAMMWARE had correctly integrated the peak area for rituximab and vedolizumab as confirmed by visual inspection of the raw data. Visual inspection of raw data files and miRAMMWARE peak integrations were done on additional outliers and the integrations were found to be correct. Examples of these outliers include a patient sample from Month 2 that had a rituximab concentration of 111 µg/mL by miRAMM vs. 1 µg/mL by ELISA and a patient from Month 6 with a rituximab concentration 100 µg/mL. While the cause for this is not known, we speculate that the difference may be due to the influence of other immunoglobulins in the sample on the ELISA assay 19. Since the miRAMM sample preparation involves the use of heat and reducing agent to generate light chains it is able to measure total rituximab (bound and free) and it not affected by ADA’s or immune-complexes. Also, not all patients with rituximab concentration >100 µg/mL have discordant results between miRAMM and ELISA supporting the theory that the discrepancy is likely biological in nature. Further examination of this data set will be needed to determine if there is a correlation between higher rituximab concentrations by miRAMM and noticeable changes in the endogenous light chain phenotype. Top-down and bottom-up MS may also be used in the future to study the amino acid sequences of the variable regions of endogenous light chains produced after administration of rituximab in order to document germline sequences used and if possible outline modifications to the CDR regions 13,20. Another difference between the methodologies is the data acquisition time. While the preanalytical portion of the miRAMM methodology (Melon Gel + reduction with DTT 17



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

= 20 minutes) is efficient and can be automated, the LC-MS method run time is 25 minutes, substantially longer than the analysis time for the ELISA assay. Adjustments to the LC gradient are being investigated to determine if the LC run can be shortened while maintaining assay integrity combined with the evaluation of multiplexed LC to improve sample throughput 21. We feel that the trade-off in acquisition time is justified considering the data can be mined not only to quantify any therapeutic mAb, but also to obtain quantitative data on a patient’s secreted monoclonal and polyclonal repertoire – all in the same assay and without the development time needed for mAb specific reagents. CONCLUSIONS The data presented in this study illustrate the unique attributes of mass spectrometry as a tool to quantify a therapeutic mAb along with endogenous monoclonal immunoglobulins secreted into circulation. The patient serum samples and patient information from the RAVE trial for ANCA-associated vasculitis represent an exceptionally well characterized data set that enabled our group to show that the miRAMM methodology can be expanded beyond our original application to monoclonal gammopathies 10. The excellent performance of an in-house designed miRAMMWARE algorithm resulted in the absolute quantification of rituximab using peak areas from mass spectra acquired using microflow LC-ESI-Q-TOF MS. Future studies will focus on the quantification of endogenous monoclonal immunoglobulins observed in this unique patient cohort with the goal of generating clinically relevant prognostic information.

18


Page 18 of 30

Page 19 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Acknowledments: -The rituximab versus cyclophosphamide for ANCA-associated vasculitis (RAVE) trial, from which this data was obtained, was supported by a grant from the National Institute of Allergy and Infectious Diseases to the Immune Tolerance Network (N01-AI-15416; protocol no. ITN021AI). Genentech, Inc. and Biogen IDEC, Inc. provided the study medications and partial funding for the trial. Genentech, Inc. performed the rituximab concentration measurements by ELISA for the trial. -This project also received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 668036 (RELENT). -Divi Cornec received a fellowship grant from the French Society of Rheumatology. *The authors declare no competing financial interest. Supporting Information Available: Table S1 – Standard concentrations found using calibration curves Table S2 – Sample concentrations determined using miRAMM and ELISA Figure S1 – Chromatograms of rituximab spiked into serum analyzed by miRAMM Figure S2 – Plots comparing rituximab concentrations found using ELISA and miRAMM This material is available free of charge via the Internet at http://pubs.acs.org.

19



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

REFERENCES (1) Willrich, M. A.; Murray, D. L.; Barnidge, D. R.; Ladwig, P. M.; Snyder, M. R. Int Immunopharmacol 2015, 28, 513-520. (2) Barnidge, D. R.; Goodmanson, M. K.; Klee, G. G.; Muddiman, D. C. J Proteome Res 2004, 3, 644-652. (3) Ladwig, P. M.; Barnidge, D. R.; Snyder, M. R.; Katzmann, J. A.; Murray, D. L. Clin Chem 2014, 60, 1080-1088. (4) Chen, Y. H.; Snyder, M. R.; Zhu, Y.; Tostrud, L. J.; Benson, L. M.; Katzmann, J. A.; Bergen, H. R. Clin Chem 2011, 57, 1161-1168. (5) Bondar, O. P.; Barnidge, D. R.; Klee, E. W.; Davis, B. J.; Klee, G. G. Clin Chem 2007, 53, 673-678. (6) Kumar, V.; Barnidge, D. R.; Chen, L. S.; Twentyman, J. M.; Cradic, K. W.; Grebe, S. K.; Singh, R. J. Clin Chem 2010, 56, 306-313. (7) Hoofnagle, A. N.; Becker, J. O.; Oda, M. N.; Cavigiolio, G.; Mayer, P.; Vaisar, T. Clin Chem 2012, 58, 777-781. (8) Barr, J. R.; Maggio, V. L.; Patterson, D. G., Jr.; Cooper, G. R.; Henderson, L. O.; Turner, W. E.; Smith, S. J.; Hannon, W. H.; Needham, L. L.; Sampson, E. J. Clin Chem 1996, 42, 1676-1682. (9) Nelson, A. L.; Dhimolea, E.; Reichert, J. M. Nat Rev Drug Discov 2010, 9, 767-774. (10) Barnidge, D. R.; Dasari, S.; Botz, C. M.; Murray, D. H.; Snyder, M. R.; Katzmann, J. A.; Dispenzieri, A.; Murray, D. L. J Proteome Res 2014, 13, 1419-1427.

20


Page 20 of 30

Page 21 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(11) Botz, C. M.; Barnidge, D. R.; Murray, D. L.; Katzmann, J. A. Br J Haematol 2014, 167, 437-438. (12) Barnidge, D. R.; Dasari, S.; Ramirez-Alvarado, M.; Fontan, A.; Willrich, M. A.; Tschumper, R. C.; Jelinek, D. F.; Snyder, M. R.; Dispenzieri, A.; Katzmann, J. A.; Murray, D. L. J Proteome Res 2014, 13, 5198-5205. (13) Barnidge, D. R.; Lundstrom, S. L.; Zhang, B.; Dasari, S.; Murray, D. L.; Zubarev, R. A. J Proteome Res 2015, 14, 5283-5290. (14) Liu, A. Y.; Robinson, R. R.; Hellstrom, K. E.; Murray, E. D., Jr.; Chang, C. P.; Hellstrom, I. Proc Natl Acad Sci U S A 1987, 84, 3439-3443. (15) Specks, U.; Merkel, P. A.; Seo, P.; Spiera, R.; Langford, C. A.; Hoffman, G. S.; Kallenberg, C. G.; St Clair, E. W.; Fessler, B. J.; Ding, L.; Viviano, L.; Tchao, N. K.; Phippard, D. J.; Asare, A. L.; Lim, N.; Ikle, D.; Jepson, B.; Brunetta, P.; Allen, N. B.; Fervenza, F. C.; Geetha, D.; Keogh, K.; Kissin, E. Y.; Monach, P. A.; Peikert, T.; Stegeman, C.; Ytterberg, S. R.; Mueller, M.; Sejismundo, L. P.; Mieras, K.; Stone, J. H. N Engl J Med 2013, 369, 417-427. (16) Stone, J. H.; Merkel, P. A.; Spiera, R.; Seo, P.; Langford, C. A.; Hoffman, G. S.; Kallenberg, C. G.; St Clair, E. W.; Turkiewicz, A.; Tchao, N. K.; Webber, L.; Ding, L.; Sejismundo, L. P.; Mieras, K.; Weitzenkamp, D.; Ikle, D.; Seyfert-Margolis, V.; Mueller, M.; Brunetta, P.; Allen, N. B.; Fervenza, F. C.; Geetha, D.; Keogh, K. A.; Kissin, E. Y.; Monach, P. A.; Peikert, T.; Stegeman, C.; Ytterberg, S. R.; Specks, U. N Engl J Med 2010, 363, 221-232. (17) Sandborn, W. J.; Feagan, B. G.; Rutgeerts, P.; Hanauer, S.; Colombel, J. F.; Sands, B. E.; Lukas, M.; Fedorak, R. N.; Lee, S.; Bressler, B.; Fox, I.; Rosario, M.; 21



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Sankoh, S.; Xu, J.; Stephens, K.; Milch, C.; Parikh, A. N Engl J Med 2013, 369, 711721. (18) Mills, J. R.; Barnidge, D. R.; Murray, D. L. Methods 2015, 81, 56-65. (19) Kelley, M.; Ahene, A. B.; Gorovits, B.; Kamerud, J.; King, L. E.; McIntosh, T.; Yang, J. Aaps J 2013, 15, 646-658. (20) Bergen, H. R., 3rd; Dasari, S.; Dispenzieri, A.; Mills, J. R.; Ramirez-Alvarado, M.; Tschumper, R. C.; Jelinek, D. F.; Barnidge, D. R.; Murray, D. L. Clin Chem 2016, 62, 243-251. (21) Yang, L.; Mann, T. D.; Little, D.; Wu, N.; Clement, R. P.; Rudewicz, P. J. Anal Chem 2001, 73, 1740-1747.

22


Page 22 of 30

Page 23 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure Legends Figure 1. Mass spectra generated by summing spectra over the specific ±0.5 minute SIC retention time window for rituximab (top) and vedolizumab (bottom). The +11, +12, +13, and +14 charge states from the light chains of rituximab (top) and vedolizumab (bottom) are labeled. The normally distributed polyclonal kappa and lambda light chain background are clearly visible as broad Gaussian shaped peaks particularly at the +11 charge state. Figure 2. miRAMMWARE output file displaying the modeling of each of the charge states from the 50 µg/mL rituximab (RTX) standard and the vedolizumab (VDL) internal standard IS along with the corresponding peak areas used for quantification. The charge state for each peak is labeled on the top while the mass/charge range covered for each peak is labeled on the bottom. The red line shows the equation used to fit the peak while the blue line shows the baseline. Figure 3. miRAMMWARE peak area determination for a Blank (left) and a 10 µg/mL standard (right) on the bottom. The mass spectrum labeled A is from the Blank and the mass spectrum labeled B is from the 10 µg/mL rituximab standard. The polyclonal kappa and lambda light chain molecular mass distributions are clearly visible and are labeled κ and λ. In section B the monoclonal kappa light chain from rituximab is labeled with an arrow and is visible above the polyclonal background. Below each mass spectrum is the smoothed data used by the miRAMMWARE algorithm to generate the peak areas used in creating the standard curves.

23



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 4. Overlaid mass spectra from the 50 µg/mL standard (black trace) and a serum sample (red trace) taken 1-year after a patient’s first dose of rituximab. The inset labeled RTX is a close up of the m/z range for rituximab showing the patient’s endogenous monoclonal light chain with a similar mass to rituximab. The inset labeled IS shows the overlaid vedolizumab IS. Figure5. Mass spectra showing the +11 charge state from the 5-6 minute LC retention time window over the course of a year from a single patient. The base peak intensity is listed on the left side of each spectrum while the sampling date is located on the right. Rituximab is labeled as RTX and vedolizumab is labeled as IS. A large lambda light chain observed in this patient one month after peak rituximab concentration is also labeled and the peak from this clone is observed from 4/16/2008 until 2/2/2009.

24


Page 24 of 30

Page 25 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


25



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

26


Page 26 of 30

Page 27 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


27



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

28


Page 28 of 30

Page 29 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


29



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For TOC Only

30


Page 30 of 30

Using Mass Spectrometry to Quantify Rituximab ... - ACS Publications

Recommend Documents