Therapeutic proteins and peptides have potential to elicit immune responses resulting in anti-drug antibodies that can pose problems for both patient safety and product efficacy. During drug development immunogenicity is usually examined by risk-based approach along with specific strategies for developing “fit-for-purpose” bioanalytical approaches. Enzyme-linked immunosorbent assays and electrochemiluminescence immunoassays are the most widely used platform for ADA detection due to their high sensitivity and throughput. During the past decade, LC/MS has emerged as a promising technology for quantitation of biotherapeutics and protein biomarkers in biological matrices, mainly owing to its high specificity, selectivity, multiplexing, and wide dynamic range. In fully taking these advantages, we describe here an immunocapture-LC/MS methodology for simultaneous isotyping and semiquantitation of ADA in human plasma. Briefly, ADA and/or drug-ADA complex is captured by biotinylated drug or anti-drug Ab, immobilized on streptavidin magnetic beads, and separated from human plasma by a magnet. ADA is then released from the beads and subjected to trypsin digestion followed by LC/MS detection of specific universal peptides for each ADA isotype. The LC/MS data are analyzed using cut-point and calibration curve. The proof-of-concept of this methodology is demonstrated by detecting preexisting ADA in human plasma.
Therapeutic proteins and peptides have potential to elicit immune responses [
Characterization and analysis of ADA are a vital element of immunogenicity assessment. Enzyme-linked immunosorbent assays (ELISA) and electrochemiluminescence (ECL) immunoassays [
Recommendations for ADA assay development, method validation, and testing strategies have been published by the Ligand-Binding Assay Bioanalytical Focus Group (LBABFG) of American Association of Pharmaceutical Scientists (AAPS) [
Since the late 1990s, liquid chromatography coupled to mass spectrometry (LC/MS) has been a dominant tool for sensitive, accurate, and rapid analysis of small-molecule drugs, metabolites, and biomarkers [
Recently, Furlong et al. developed a universal peptide method to quantitate human antibody Fc region-containing therapeutic protein candidates in nonclinical species [
LC/MS has also been reported to assess ADA in the presence of excess protein therapeutic in support of clinical programs addressing the safety and tolerability of human growth hormone analogues [
In another application, LC/MS was used to evaluate neutralizing Ab (NAb) assay by simultaneously quantitating residual mAb-drug, endogenous IgG, and NAb-positive control in BEAD eluates [
We describe here an immunocapture-LC/MS-based approach for simultaneous ADA isotyping and semiquantitation in human plasma. Biotinylated drug or anti-drug Ab is used to capture ADAs or drug-ADA complexes in plasma, respectively. The resulting ADA-drug or ADA-drug-Ab complexes are then immobilized on streptavidin magnetic beads and separated from matrix by a magnet. After washing, ADA is released from the beads and subjected to trypsin digestion followed by LC/MS detection of specific universal peptides for each ADA isotype.
Protein Z (containing no human Fc) was a proprietary experimental biotherapeutic of Boehringer Ingelheim Pharmaceuticals (Ridgefield, CT) and produced in-house. The mouse anti-Protein Z monoclonal Ab (mAb) was supplied in-house. Human IgG1, IgG2, IgG3, IgG4, and IgM as well as bovine serum albumin (BSA) were purchased from Sigma Aldrich (St. Louis, MO), human IgE was from MP Biomedicals (Solon, OH), human IgA1 was from Abcam (Cambridge, MA), and human IgA2 was from EMD Millipore (Billerica, MA). Internal standard peptides with stable labeled C-terminal [13C6,15N4]Arg or [13C6,15N2]Lys were synthesized at GenScript (Piscataway, NJ). Streptavidin magnetic beads (1
Biotinylation of Protein Z and the mouse anti-Protein Z mAb was performed using an EZ-Link Sulfo-NHS-LC Biotinylation kit. A 1 mg/mL solution of the drug or mAb was combined with a 10-fold molar excess of biotin and allowed to react at room temperature for 60 minutes. Excess biotin was removed using desalting columns provided in the kit. A HABA assay was used to determine the amount of biotin incorporation in the sample after desalting. Typical biotin incorporation was approximately 2 biotins per drug and 4–7 biotins per mouse mAb. The biotinylated drug and biotinylated mouse mAb solutions were diluted to 0.1 mg/mL in water and stored at −80°C prior to use.
Streptavidin magnetic beads were prepared freshly for each assay batch. The magnetic beads (10 mg/mL) were transferred to a polypropylene tube and placed on a magnetic stand to remove supernatant and collect the beads. The beads were then washed with 10x volume of Tris buffered saline containing 0.1% Tween-20 (TBS-T) and resuspended in 2x volume of TBS-T to yield a final working bead concentration of 5 mg/mL.
An aliquot of 95
An aliquot of 144
Commercial stock solutions of the different immunoglobulins (Igs) ranged from 1 to 4.18 mg/mL. A series of 50–10,000 ng/mL spiking calibration standards were prepared in 0.1% BSA using the stock solutions and stored at −80°C prior to use. Pooled human blank plasma was processed using either immunocapture procedure. Matrix calibration standards were prepared by adding 10
To the immunocapture eluent, matrix calibration standards, or neat Ig PBS solution, 5
Eksigent Ekspert MicroLC 200 coupled with AB Sciex 6500 triple quadrupole mass spectrometer (AB Sciex, Framingham, MA) was used. Chromatographic separation was performed using ACQUITY UPLC Peptide BEH C18 column (1 mm × 50 mm, 1.7
The mass spectrometer was operated in in positive electrospray ionization mode. Key instrument parameters were as follows: +5000 V electrospray voltage, 65 nebulizer gas units, 30 axillary gas units, 375°C ion source temperature, 10 collision gas units, and unit resolution on both Q1 and Q3. For identifying unique peptides for each ADA isotype with IDA, up to 4 multiple-reaction-monitoring (MRM) pairs were used for screening followed by enhanced product ion scan. For ADA isotyping and semiquantitation, 13 MRM pairs of universal peptides were used along their respective internal standards.
Despite its many advantages and potential, the use of LC/MS for protein quantitation is not as straightforward as for small-molecules and oftentimes demands comprehensive method development. Plasma and serum are very complex matrices that contain several hundreds of thousands of proteins and protein isoforms in a wide concentration range [
Antibodies are secreted by plasma cells and come in different isotypes with genetic variations or differences in the constant regions of the heavy and light chains. In humans, there are five heavy chain isotypes (
ADA isotyping and semiquantitation were based on the surrogate peptides of ADAs instead of whole molecules, mainly in consideration of sensitivity [
Among the Ig isotypes, only IgG1–IgG3 and IgA2 have different allotypes [
Additional criteria that are commonly used in selecting a surrogate peptide include the following [
To identify candidates for unique peptides, 15
After suitable surrogate peptide candidates were identified, immunocapture was performed, and sensitivity and matrix interference were assessed for each MRM pair. MRMs with the best sensitivity, specificity, and selectivity were selected. Two different immunocapture approaches were used for immunopurification. One was to capture ADAs with biotinylated Protein Z, while the other was to capture drug-ADA complexes with a biotinylated mouse mAb. Each has its own advantages and shortfalls and is discussed below in detail. The workflow of the immunocapture-LC/MS assay is depictured in Figure
Immunocapture-LC/MS workflow chart.
Using drug as capture reagent is straightforward, similar to traditional drug-bridging ECL assays [
The primary goal of the immunocapture was to isolate ADA from the plasma matrix. The efficiency of the immunopurification should be assessed using human ADA positive controls. Because there were no human ADA standards available, commercial human Igs were used as surrogates. Although different in their variable regions, the surrogate human Igs have the same constant regions as human ADAs and thus produce the unique peptides of human ADA isotype/subclass upon trypsin digestion which can then be detected and quantitated by LC/MS. Unlike human ADA, however, the surrogate Igs would not bind to biotinylated Protein Z to form (biotinylated)drug-Ig complexes. Therefore, instead of directly spiked into blank human plasma, the surrogate Igs were spiked into the magnetic beads eluent of blank human plasma after the immunocapture steps. The spiked and unspiked magnetic bead eluent were then digested and assayed by LC/MS. Only the MRM pairs that had the best sensitivity with minimal matrix interferences were chosen for immunocapture-LC/MS assay. Table
List of unique peptides and MRMs for ADA isotopes/subclasses used in the immunocapture-LC/MS assay.
Isotype/subclass | Unique peptide sequence | MRM pairs |
---|---|---|
|
||
IgG1 | GPSVFPLAPSSK | 593.83→699.40 |
IgG2 | GLPAPIEK | 412.75→654.38 |
IgG3 | WYVDGVEVHNAK | 708.85→698.48 |
IgG4 | GLPSSIEK | 415.73→660.36 |
IgE | AEWEQK | 395.69→590.29 |
IgM | GQPLSPEK | 428.23→670.38 |
IgA1, IgA2 | YLTWASR | 448.73→620.32 |
|
||
|
||
IgG1, IgG3 | ALPAPIEK | 419.76→654.38 |
IgG1, IgG3, IgG4 | VVSVLTVLHQDWLNGK | 603.34→1110.57 |
IgE | LEVTR | 309.18→375.24 |
IgM | VSVFVPPR | 450.77→615.36 |
IgA1, IgA2 | VAAEDWK | 409.71→648.30 |
IgA1 | TFTC |
688.31→765.38 |
The final MRM pairs were grouped into two types: quantitation and confirmation. The quantitation MRMs were the most sensitive and used for isotyping and quantitation. The confirmation MRMs were less sensitive and results from the confirmation peptides are expected to be similar to those from quantitation peptides. In case there is a large discrepancy between the quantitation and confirmation peptide results, investigation on the assay may be needed. The AA sequence length was 6–12 for the quantitation peptides and 5–16 for the confirmation peptide. The quantitation peptides for IgG1, IgG2, IgG3, and IgG4 were GPSVFPLAPSSK, GLPAPIEK, WYVDGVEVHNAK, and GLPSSIEK, respectively. In addition, peptides ALPAPIEK and VVSVLTVLHQDWLNGK were also sensitive and found in IgG1/IgG3 and IgG1/IgG3/IgG4, respectively. These 2 peptides were not unique to any one single Ig isotype/subclass and were included as confirmation peptides. Likewise, quantitation peptides AEWEQK and GQPLSPEK and confirmation peptides LEVTR and VSVFVPPR were identified for IgE and IgM, respectively.
Both IgA1 and IgA2 shared the same peptide, YLTWASR, which was much more sensitive than any other unique peptides and thus used for quantitating the total of IgA1 and IgA2. In addition, peptide TFTC[CAM]TAAYPESK was unique to IgA1. However, TFTC[CAM]TAAYPESK was much less sensitive (2500 ng/mL limit of detection human plasma after immunocapture) and thus not very useful for low level ADA detection. For both IgA1 and IgA2, VAAEDWK was also used as a confirmation peptide.
After the final MRMs were selected, LC was optimized and total runtime was shortened from 20 min to 8 min. All the selected peptide came between 0.8 and 4 min in the 8 min run. Representative LC/MS chromatograms of IgG1, IgE, and IgM from the pooled ADA-free blank human plasma and low limit of quantitation standards (see discussions later) are shown in Figure
LC/MS chromatograms of unique peptides of IgG1 (top), IgM (middle), and IgE (bottom) from blank human plasma (left) and LLOQ samples (right) after immunocapture when using drug as ADA capture reagent.
The LC/MS assay was tested with blank human plasma with and without preexisting ADA (PEA) for Protein Z. The blank plasma was obtained from 20 in-house healthy donors and had been screened for PEA with a drug-bridging ECL assay. Based on the ECL assay, plasma from 9 donors was PEA negative whereas it was PEA positive from the remaining 11 donors.
The 20 plasma samples were processed using the immunocapture procedures with biotinylated Protein Z as the capture reagent. Along with the 20 plasma samples, a set of calibration standards were prepared in beads eluent from the pooled PEA negative blank plasma. After digestion, the plasma samples and the calibration standards were assayed using the LC/MS ADA method. The results were evaluated using either a cut-point or a calibration curve.
The LC/MS responses (analyte/IS peak area ratio) of each ADA isotype/subclass from the 9 PEA negative and the 11 PEA positive human plasma samples are listed in Tables
LC/MS peak area ratio response and cut-points of ADA isotopes/subclasses in PEA negative human plasma with drug as ADA capture reagent.
Plasma lot # | IgG1 | IgG2 | IgG3 | IG4 | IgE | IgM | IgA1 + IgA2 |
---|---|---|---|---|---|---|---|
1 | 0.0277 | 0.0099 | — | 0.0107 | — | 0.1130 | 0.0041 |
2 | 0.0107 | 0.0008 | — | 0.0019 | — | — | 0.0059 |
3 | 0.0263 | — | — | 0.0012 | 0.0150 | 0.0084 | 0.0078 |
4 | 0.0093 | 0.0023 | — | 0.0053 | — | 0.0072 | 0.0057 |
5 | 0.0249 | 0.0244 | 0.0210 | 0.0099 | 0.0110 | 0.0474 | 0.0136 |
6 | 0.0060 | 0.0044 | 0.0110 | 0.0035 | — | 0.0326 | — |
7 | 0.0129 | — | — | 0.0037 | 0.0160 | 0.0159 | 0.0062 |
8 | 0.0637 | 0.0205 | 0.0570 | 0.0353 | 0.0140 | 0.0066 | 0.0174 |
9 | 0.0127 | 0.0008 | — | 0.0111 | — | 0.0035 | 0.0075 |
|
|||||||
Mean | 0.0216 | 0.0070 | 0.0099 | 0.0092 | 0.0061 | 0.0261 | 0.0086 |
SD | 0.0177 | 0.0093 | 0.0192 | 0.0105 | 0.0074 | 0.0361 | 0.0043 |
|
|||||||
Cut-point (95%) | 0.0507 | 0.0224 | 0.0416 | 0.0265 | 0.0183 | 0.0854 | 0.0156 |
—: no LC/MS response was detected.
LC/MS peak area ratio responses and calculated ADA isotope/subclass levels in PEA positive human plasma with drug as ADA capture reagent (numbers in bold italic are above respective cut-points).
Lot # | IgG1 | IgG1 conc. ( |
IgG2 | IgG3 | IG4 | IgE | IgM | IgA1 + IgA2 |
---|---|---|---|---|---|---|---|---|
1 |
|
— |
|
|
0.0096 | — | 0.0499 |
|
2 |
|
— |
|
|
0.0070 | — | 0.0421 |
|
3 |
|
0.660 | 0.0061 | 0.0263 | 0.0088 | — | 0.0111 |
|
4 |
|
— | 0.0047 | — | 0.0020 | 0.0113 | 0.0068 |
|
5 | 0.0239 | — | 0.0044 | 0.0119 | 0.0020 | — | 0.0007 |
|
6 |
|
— | 0.0141 | 0.0205 | 0.0192 | 0.0054 | 0.0316 |
|
7 | 0.0188 | — | 0.0041 | — | — | — | 0.0127 |
|
8 | 0.0150 | — | 0.0109 | 0.0315 | 0.0017 | 0.0119 | 0.0027 |
|
9 |
|
— | 0.0009 | — | 0.0009 | 0.0150 | 0.0099 |
|
10 |
|
— | 0.0076 | 0.0192 | — | 0.0155 | 0.0314 |
|
11 |
|
0.680 | 0.0041 | 0.0303 | 0.0031 | — |
|
|
|
||||||||
Cut point (95%) | 0.0507 | — | 0.0224 | 0.0416 | 0.0265 | 0.0183 | 0.0854 | 0.0160 |
—: below the limit of quantitation for IgG1 concentration or no peak was detected for other isotypes/subclasses.
Likewise, cut-points were calculated for all other ADA isotypes/subclasses (Table
ADA levels in the 11 PEA positive samples were semiquantitatively determined using a calibration curve. A series of matrix calibration standards ranged from 0.05 to 10
During the immunocapture process, the (biotinylated)drug-ADA complexes were separated from the plasma by a magnet. However, as endogenous plasma Igs were at much higher levels [
The higher limit of quantitation (HLOQ) of the calibration curve range depended on the beads and capture reagent capacity. However, as the calibration standards were prepared after immunocapture, the beads and capture reagent capacity could not be readily assessed. Based on our experiences with immunocapture using similar experimental settings, the HLOQ was arbitrarily set at 10
The calibration linear range was defined from LLOQ to HLOQ. The curve linear regression correlation coefficients (
Calibration curve ranges and regression coefficients (
Isotype/subclass | Unique peptide | Calibration curve parameters | |||
---|---|---|---|---|---|
Drug capture | mAb capture | ||||
Range ( |
|
Range ( |
| ||
IgG1 | GPSVFPLAPSSK | 0.5–10 | 0.9919 | 0.5–10 | 0.9940 |
IgG2 | GLPAPIEK | 0.25–10 | 0.9939 | 0.1–10 | 0.9964 |
IgG3 | WYVDGVEVHNAK | 0.25–10 | 0.9858 | 0.25–10 | 0.9952 |
IgG4 | GLPSSIEK | 0.1–10 | 0.9919 | 0.25–10 | 0.9909 |
IgE | AEWEQK | 0.1–10 | 0.9947 | 0.1–10 | 0.9976 |
IgM | GQPLSPEK | 0.25–10 | 0.9988 | 0.25–10 | 0.9909 |
IgA1 + IgA2 | YLTWASR | 0.1–10 | 0.9974 | 0.1–10 | 0.9966 |
Calibration curves of IgG1 in human plasma eluent after immunocapture when using either drug (a) or mouse mAb as ADA capture reagent (b).
Using the calibration curve, only 2 out of the 11 PEA positive samples had ADA levels (for IgG1 only) above the LLOQ 0.5
The biggest caveat of the semiquantitation approach was that the calibration standards did not go through the immunocapture process whereas the study samples did. The calibration thus did not take immunocapture recovery into account. The recovery could be estimated using well characterized polyclonal human ADA positive controls, which were not available for Protein Z. Based on our experiences with immunocapture in similar experimental setting and those reported in literature [
Different from ECL assays, the ADA levels measured by the immunocapture-LC/MS represent absolute amounts. This allows one to compare ADA isotype levels between samples, studies, and different biotherapeutics. Database of such information could be gradually built and provide valuable insight to better understand immunogenicity and immunology of biotherapeutics.
As with traditional ECL drug-bridging assays, the immunocapture-LC/MS method could be hampered by drug tolerance issues when drug is present [
It should be noted that if drug contains the human Ig Fc region, it may also bind to the beads via nonspecific binding just like endogenous proteins and could contribute to interference in the LC/MS assay. On the other hand, biotinylated drug that binds to streptavidin beads will not be eluted out under the elution conditions due to very strong biotin-streptavidin interaction [
It was evident that the results from the immunocapture-LC/MS assay confirmed PEA positive results in most of these samples and were in good agreement with the drug-bridging ECL assay.
The second immunocapture approach was using a mouse anti-Protein Z mAb to capture ADA in human plasma. In this approach, all ADAs had to be first completely converted to drug-ADA complexes by adding excessive Protein Z to the samples [
The merit using anti-drug Ab as the capture reagent lies on that drug no longer interferes with the assay. This offers a huge advantage when drug levels in the study samples are high enough such that drug tolerance becomes a concern in other types of assays. The most important element of this approach is that the capture Ab should not compete with ADA for the drug; that is, the two should not share the same binding domain on the drug. To confirm this for the mouse mAb, the immunocapture recovery of Protein Z from ADA positive samples was assessed. Protein Z was spiked at 5 ng/mL to the pooled PEA negative and the 11 positive human plasma samples and its concentration was determined using an immunocapture-LC/MS PK assay. The PK assay was developed in our lab to support clinical studies. In the PK assay, Protein Z was captured using the mouse mAb, and the resulting drug-mAb complex was then immobilized on magnetic beads, separated from plasma, eluted out from beads, digested, and analyzed by LC/MS. A unique peptide from Protein Z was monitored by LC/MS and used to quantitate Protein Z. The immunocapture recovery was determined by comparing Protein Z concentrations in the PEA positive human plasma with the pooled PEA negative plasma. No difference in Protein Z concentration was observed between the PEA positive and PEA negative samples (data not shown), and Protein Z recovery was more than 82% with averaging 97%. It was evident that the mouse mAb was indeed able to capture Protein Z regardless of whether it is in ADA-Protein Z complexes or free form. However, one has to be cautious as human ADAs come in many different forms and some may bind to the same domain on the drug as the capture Ab. Therefore, it is recommended to run this test using ADA positive samples from the study.
Similar to the first approach using drug as capture reagent, the 11 PEA positive and 9 PEA negative blank human plasma samples were used to evaluate immunocapture using the mouse mAb as the capture reagent. Besides using a different capture reagent, the only difference between the two approaches was that in the second approach there was an additional step to convert ADA to ADA-drug complexes. To ensure a complete conversion, 6
Tables
LC/MS peak area ratio response and cut-points of ADA isotopes/subclasses in PEA negative human plasma with mouse mAb as ADA capture reagent.
Lot # | IgG1 | IgG2 | IgG3 | IG4 | IgE | IgM | IgA1 + IgA2 |
---|---|---|---|---|---|---|---|
1 | 0.0495 | 0.0107 | — | 0.1110 | — | 0.0422 | 0.0224 |
2 | 0.0340 | 0.0054 | — | 0.0357 | 0.0152 | 0.0702 | 0.0223 |
3 | 0.0400 | 0.0035 | — | 0.0548 | — | 0.0277 | 0.0538 |
4 | 0.0243 | 0.0084 | — | 0.0259 | — | 0.0358 | 0.0264 |
5 | 0.0265 | 0.0108 | — | 0.0677 | — | 0.0593 | 0.0237 |
6 | 0.0210 | 0.0511 | — | 0.0868 | — | 0.0341 | 0.0688 |
7 | 0.0815 | 0.0384 | — | 0.0009 | — | 0.0622 | 0.0087 |
8 | 0.0231 | 0.0294 | — | 0.0866 | — | 0.0029 | 0.0274 |
9 | 0.0253 | 0.0027 | — | 0.0719 | — | 0.0015 | 0.0094 |
|
|||||||
Mean | 0.0361 | 0.0178 | — | 0.0601 | 0.0017 | 0.0373 | 0.0292 |
SD | 0.0194 | 0.0175 | — | 0.0345 | 0.0051 | 0.0245 | 0.0197 |
|
|||||||
Cut-point (95%) | 0.0680 | 0.0466 | 0.0000 | 0.1168 | 0.0100 | 0.0775 | 0.0617 |
—: no LC/MS response was detected.
LC/MS peak area ratio responses and calculated ADA isotope/subclass levels in PEA positive human plasma with mouse Ab as ADA capture reagent (numbers in bold italic are above respective cut-points). Plasma samples were spiked with addition of excessive drug (+drug) or without (−drug) addition of excessive drug.
Lot # | IgG1 | IgG2 | IgG3 | IgG4 | IgE | IgM | IgA | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
−drug | +drug | −drug | +drug | −drug | +drug | −drug | +drug | −drug | +drug | −drug | +drug | −drug | +drug | |
1 | 0.0369 |
|
0.0077 | 0.0106 | — | — | 0.0668 | 0.0823 | — | — | 0.0159 | 0.0094 | 0.0194 | 0.0411 |
2 | 0.0145 | 0.0563 | 0.0072 | 0.0066 | — | — | 0.0584 | 0.0458 | — | — | 0.0316 | 0.0340 | 0.0033 | 0.0113 |
3 | 0.0354 |
|
0.0074 | — | — | — | 0.0634 | 0.0594 | — | — | 0.0268 | 0.0098 | 0.0405 | 0.0483 |
4 | 0.0212 |
|
0.0240 | 0.0314 | — | — | 0.0484 | 0.0369 | — | — | 0.0045 | 0.0190 |
|
|
5 | 0.0350 | 0.0381 | — | 0.0059 | — | — | 0.0182 | 0.0150 | — | — | 0.0053 | 0.0118 | 0.0325 |
|
6 |
|
|
0.0122 | 0.0237 | — | — | 0.0325 | 0.0544 | — | — | 0.0575 | 0.0310 |
|
|
7 | 0.0231 | 0.0343 |
|
|
— | — | 0.0685 | 0.0768 | — | — | 0.0234 | 0.0321 | 0.0339 | 0.0476 |
8 | 0.0463 | 0.0474 | 0.0086 | 0.0045 | — |
|
0.0572 | 0.0528 | — | — | 0.0089 | 0.0394 | 0.0142 | 0.0329 |
9 | 0.0396 |
|
0.0011 | 0.0026 | — | — | 0.0661 | 0.0404 | — | — | 0.0094 | 0.0125 | 0.0437 |
|
10 | 0.0188 |
|
0.0020 | 0.0039 | — | — | 0.0151 | 0.0225 | — | — | 0.0275 | 0.0329 | 0.0029 | 0.0208 |
11 |
|
|
0.0094 | 0.0075 | — | — | 0.0294 | 0.0305 | — | — |
|
|
|
|
|
||||||||||||||
Cut-point (95%) | 0.0680 | 0.0466 | 0.0000 | 0.1168 | 0.0100 | 0.0775 | 0.0617 |
—: no LC/MS response was detected.
LC/MS chromatograms of unique peptides of IgG1 (top), IgM (middle), and IgE (bottom) from blank human plasma (left) and LLOQ samples (right) after immunocapture when using mouse mAb for ADA capture.
Calculated cut-points for all other ADA isotypes/subclasses are listed in Table
The mechanism of ADA capture using the mAb was more complicated than using drug. The ADA must be bound to the drug first, and the resulting drug-ADA complexes had to be bound to the mouse mAb and survive the immunocapture procedure in order to be detected by LC/MS. Endogenous components such as Igs that cross-reacted with the mAb could also interfere with the assay and give false positive results. Although this potential interference was already accounted for in the cut-point determination, it was further assessed for the PEA positive plasma samples without the addition of excessive Protein Z. The “−drug” plasma samples were spiked with the mAb and then processed with the immunocapture procedure followed by LC/MS analysis. The results are provided in Table
Calibration curves were established for each ADA isotype in the same way as in the first approach. The calibration linear ranges and curve regression correlation coefficients are provided in Table
Using the calibration curve, ADA (IgG1) level in lot 3 and lot 11 plasma was determined to be 0.570 and 1.25
Besides IgG1, ADA levels for other ADA isotypes/classes were all BLQ in these 11 PEA positive samples. This was consistent with the first approach.
It should be noted that the anti-drug Ab capture approach may not be used if the biotherapeutic proteins contain constant human Fc regions. Unlike using drug as capture reagent, anti-drug Ab captures both free drug and drug-ADA complexes and during the ADA elution step drug is also eluted out from magnetic beads and thus interferes with LC/MS detection. For instance, Humira (adalimumab), a TNF inhibiting anti-inflammatory drug and the first fully human monoclonal antibody drug approved by the FDA, is an IgG1 made by phage display technology with amino acid sequences only from the human germline, making it indistinguishable in structure and function from natural human IgG1 [
We demonstrated for the first time that immunocapture-LC/MS can be used for simultaneous ADA isotyping and semiquantitation in human plasma. Either biotinylated drug or biotinylated anti-drug Ab could be used as the immunocapture reagent, each with its own merits and shortfalls. Biotinylated drug can readily capture ADA but drug interference could be an issue if drug levels in the samples are high. On the other hand, immunocapture using an anti-drug Ab eliminates drug interference, providing that the Ab is able to capture drug-ADA complex in addition to free drug. With this method, unique peptides from each ADA isotype/subclass were identified and monitored by LC/MS. ADA isotyping was performed by the detection of isotype-unique peptides. Absolute ADA amount was determined semiquantitatively using surrogate calibration standards. Similar to traditional drug-bridging ELISA assay, cut-points at 95% were established. The assay was used for screening, isotyping, and semiquantitating preexisting ADAs in human plasma. It could be also used as a confirmatory assay. Endogenous Ig interferences need to be reduced in order to improve the assay sensitivity and specificity, and human positive ADA controls will be needed for more accurate ADA quantitation.
Owing to LC/MS’s advantages such as high specificity, selectivity and reproducibility, wide dynamic range, and multiplexing capability, it is expected that, with further improvements, immunocapture-LC/MS will become an invaluable tool in immunogenicity assessment. It can be easily implemented in bioanalytical lab settings for routine ADA isotyping and semiquantitation. As ADA levels measured by immunocapture-LC/MS represent absolute amounts, one can compare ADA isotype levels between samples, studies, and different biotherapeutics, providing that consistency in positive controls is achieved to determine recovery. Database of such information could be gradually built and provide valuable insight to better understand immunogenicity and immunology of biotherapeutics.
The authors declare that there is no conflict of interests regarding the publication of this paper.
The authors would like to thanks Drs. J. Duggan, S. Norris, and C. Kane for fruitful discussions and valuable input. Special thanks go to Ms. C. Quatrano for providing human ADA plasma and ECL data.