WO2009022001A1

WO2009022001A1 - Improvement of drug tolerance in immunogenicity testing

Info

Publication number: WO2009022001A1
Application number: PCT/EP2008/060724
Authority: WO
Inventors: Annette Zaar; Denise SICKERT; Kerstin Kroeger; Christophe ZICKLER; Edwige CHOKOTE
Original assignee: Novartis Ag
Priority date: 2007-08-16
Filing date: 2008-08-14
Publication date: 2009-02-19

Abstract

The invention a new method based on surface plasmon resonance that uses an acid dissociation step to allow for anti-drug antibody (ADA) detection and overcomes drug interference in the presence of drug in serum. Removal of drug excess after acid treatment is not required. The method is generally applicable for anti-drug antibody detection to therapeutic proteins.

Description

IMPROVEMENT OF DRUG TOLERANCE IN IMMUNOGENICITY TESTING

FIELD OF THE INVENTION

[01] This invention relates generally to the analytical testing of antibodies in vitro, and more particularly to aspects of immunogenicity testing in presence of drug by label free optical detection methods.

BACKGROUND OF THE INVENTION

[02] As more therapeutic proteins (drugs) become available on the market, the incidence of unwanted anti-drug antibodies (ADAs) responses in treated patients is rising. Hermeling S et ah, Pharmaceutical Research 21(6): 897-903 (2004). Such anti-drug antibodies can neutralize the biological activity of the drug, alter pharmacokinetics and in rare cases have been responsible for life threatening conditions. Casadevall N et ah, The New England Journal of Medicine 346:1584-1586 (2002). In pre-clinical studies, ADAs can seriously impact interpretation of pharmacokinetic, pharmacodynamic and toxicology data. Therefore, it is important to monitor and evaluate antibody responses during clinical and pre-clinical studies. Patton A et ah, Journal of Immunological Methods 504:189-195 (2005); Wadhwa M et al, Dev. Biol. (Basel) 122:155-70 (2005).

[03] Surface plasmon resonance, an optical phenomenon that enables detection of unlabeled interactants in real time, is frequently used for the detection of antibodies directed against therapeutic proteins (drug). A well-known line of machines for performing surface plasmon resonance are the Biacore machines (e.g., BIACORE 2000 and BIACORE TlOO), which are commercially available from BIAcore AB (Uppsala, Sweden). Thus, it is possible to detect low-affinity binding events and anti-drug antibodies (ADAs) regardless of their isotype and species of origin.

[04] However, detection of ADAs can be difficult when the therapeutic protein of interest is present in the sample to be assayed. Biological drugs present in patient's blood can complex with ADAs and hence reduce the amount of antibody detectable by immunogenicity assays. This is a particular concern with therapeutic monoclonal antibodies (mAbs), which have typically longer half- lives than other proteins. For detection of ADAs in presence of high drug concentrations, assay choice is limited to methods like Enzyme-Linked Immunosorbent Assay (ELISA), that are capable of incorporating acid dissociation procedures to separate drug- ADA immune complexes, before analysis. Patton A et al, Journal of Immunological Methods 504:189-195 (2005).

[05] It has been shown that dissociation procedures used in ELISA methods can reduce drug interference and tolerate the presence of- 100-fold molar excess of drug. Lofgren JA et al, The Journal of immunology 178:1461-1412 (2007); Patton A et al, Journal of Immunological Methods 504:189-195 (2005); Lofgren JA et al, Journal of Immunological Methods 308: 101-108 (2006). However, it has not been shown to date that acid dissociation procedures can be used with label free optical detection methods. As a consequence, steps to ensure adequate tolerance to drug excess remain a prerequisite to enable sensitive detection of ADAs by surface plasmon resonance.

SUMMARY OF THE INVENTION

[06] The invention a new surface plasmon resonance method that uses an acid dissociation step to allow for anti-drug antibody (ADA) detection and to overcome drug interference in the presence of drug in serum. Removal of drug excess after acid treatment is not required. [07] The method has been investigated with ADAs derived from rabbits and humans. In one embodiment, the therapeutic monoclonal chimeric antibody has anti-cancer activity. In a particular embodiment, the therapeutic antibody is LBY 135.

[08] This method is generally applicable for the detection of anti-drug antibodies to therapeutic proteins.

BRIEF DESCRIPTION OF THE DRAWINGS

[09] FIG. 1 is a graph showing the determination of a negative cut-off (NCO). Twenty- five individual lots of human blank sera were used to establish the NCO for each sensor surface.

The NCO was calculated as mean of the human blank serum values + 2SD.

[10] FIG. 2 is a typical calibration curve with position of quality control samples (QCs) and specific negative cut-off (NCO).

[11] FIG. 3 is a curve showing a drug interference of strong immune response. The impact of LBY135 on signal detection was evaluated by an inhibition curve. One inhibition curve was generated for each chip surface. One inhibitory LBY135 concentration was defined for each level of immune response. [12] FIG. 4 is a curve showing a drug interference of mid immune response. The impact of LBY 135 on signal detection was evaluated by an inhibition curve. One inhibition curve was generated for each chip surface. One inhibitory LBY 135 concentration was defined for each level of immune response.

[13] FIG. 5 is a curve showing a drug interference of weak immune response. The impact of LBY135 on signal detection was evaluated by an inhibition curve. One inhibition curve was generated for each chip surface. One inhibitory LBY135 concentration was defined for each level of immune response.

[14] FIG. 6 shows a rabbit anti-LBY135-antibody concentration curve in pooled human serum, analyzed on the Biacore. The ADAs were spiked in different concentrations into neat pooled human serum: 0, 5, 7.5, 10, 25 and 100 μg/ml. Shown are the untreated, acidified and the untreated minus non specific binding (NSB) concentration curves. The decrease of signal between acid treated (pH 2.5 at 37 ⁰C for 30 min) and untreated blank sample suggests the reduction of non-specific binding by denaturation of non-specific serum components. In addition, the results evidence that ADAs remain biologically active after the acidification step.

[15] FIG. 7 shows the recovery of ADAs from anti-LBY135-LBY135 immune complexes in human serum on Biacore by acid dissociation. In untreated serum samples spiked with immune complexes (open circles), no ADAs were detected on Biacore. When treated with acid (pH 2.5, 37 ⁰C, 30 min) followed by dilution in Biacore running buffer to reach a pH of around 6, the percentage of recovery from each sample was 97 - 100.0 % (black circles) as comparing to the corresponding control sample spiked with equivalent amount of ADAs without LBY135 (triangles). The NSB was subtracted to obtain the 100% signal of the reference.

[16] FIG . 8 is a bar graph showing the recovery of ADAs from anti-LB Y 135 -LBY 135 immune complexes in rabbit serum on Biacore by acid dissociation. After acidification (pH 2.5, 37 ⁰C, 30 min) of the drug spiked rabbit sera, the level of ADAs binding on the Biacore chip surface were recovered by comparing to the acid treated rabbit sera without drug. About 97-99 % of ADAs were recovered in the binding assay. [17] FIG. 9 is a bar graph showing the recovery of ADAs from clinical samples. Serum samples were obtained from one patient, after dosing with 0.3 mg/kg LBY135 on day 1, 41, 61, 83, 102 and 124. Serum: untreated patient samples (measured drug levels below 18.6 μg/ml); serum + HCl: patient samples treated with acid pH3; IC: patient samples spiked with 1 mg/ml LBY135; IC + HCl: acid treated IC samples; serum minus NSB: patient samples minus non-specific binding. The non specific binding is the difference between non-treated pre-dose and acid treated pre-dose sample. 88 - 99 % of patient ADAs generated against the drug were recovered after acid treatment using "serum minus NSB" as a reference. [18] FIG. 10 is a bar graph showing the results of an investigation of drug interference in clinical samples containing different drug levels. To assess the impact of drug to the recovery of ADAs, different concentrations of LBY135 (0.2, 1, 2 mg/ml) were spiked in an immunogenicity positive patient sample (chapter 3.3, after dosing with 0.3 mg/kg LBY135 on day 124). All three drug levels inhibited binding of ADAs to the drug on the solid phase, whereas acid treatment allowed dissociating of the complex between ADAs and the soluble drug and finally favored binding of ADAs to the drug of the solid phase. The acid dissociation step resulted in a 92 - 99 % recovery of ADAs.

DETAILED DESCRIPTION OF THE INVENTION

[19] LBY 135 is a chimeric monoclonal antibody directed against human DR5. It is currently being developed for the treatment of cancer as single agent or in combination therapy.

EXAMPLE 1

QUANTIFICATION OF ANTI-LBY135 ANTIBODIES IN HUMAN SERUM BY A

BIACORE BINDING ASSAY

[20] A screening method on a BIAcore 3000 instrument to quantify anti-LBY135 antibodies in human serum was developed and validated. Protein G is used to immobilize

LBY135 on the sensor chip, making sure that the potentially immunogenic CDR regions are exposed. Samples are injected and binding of anti-LBY135 to LBY135 is measured in real time. Since LBY 135 is bound to protein G reversibly, it is removed from the sensor surface after every cycle by regeneration and injected newly before the next sample is applied. This procedure makes sure that bound LBY 135 is not denatured by acid treatment during chip regeneration.

[21] The assay was validated based on the criteria presented in Mire-Sluis AR et ah, Journal of Immunological Methods 289:1-16 (2004).

[22] The following parameters were validated according to specific acceptance criteria: (a) sensor surface activation (200-300 RU), (b) protein G coupled to the sensor surface (400-800 RU), (c) reproducibility of drug injection (< 10 RU difference between 2 consecutive cycles for LBY135), (d) sensor surface stability over time (< 15 % loss of activity over 100 cycles), (e) ratio of positive control signal to negative cut-off (NCO) (> 5), (f) intra- and inter-surface accuracy and precision (accuracy values between 85-115% (80-120% at the LLOQ and ULOQ), and (g) precision-values < 15% (< 20% at the LLOQ and ULOQ). [23] Accuracy values for calibration samples (Cs) within the working range: deviation < 15% from nominal concentration for at least ²A of minimally 6 non-zero calibration samples. All acceptance criteria were met. As shown below, the following parameters were validated according to specific acceptance criteria with the following results: sensor surface activation (205.5 - 214.3 RU), protein G coupling to the sensor surface (485.1 - 519.9 RU), reproducibility of drug injection (< 4 RU difference between 2 consecutive cycles), sensor surface stability over time (< 2% loss of activity over 93 cycles), ratio of positive control signal to negative cut-off (> 9), intra- and inter-surface accuracy 95.2 - 114.0% and precision 0.5 - 7.8%.

[24] Since no positive anti-LBY135 positive human samples were available, an anti-human IgG depleted rabbit anti-LBY135-Fab serum was used as a positive control. Since IgG in serum is very stable, stability was not assessed. [25] The laboratory equipment used in developing the assay is listed in TABLE 2.

[26] For further information regarding the use of the Biacore equipment, see BIAcore®

3000 Instrument Handbook (Edition October 2003); BIAcore® 3000 GxP Handbook (Edition

April 2003); BIAcore® 3000 GxP Getting Started (Edition April 2003); and BIAevaluation,

Software Handbook (Edition April 2003).

[27] All chemicals and reagents whose reference is given in TABLE 3 were of analytical grade.

[28] The ultra pure distilled water used in the assay development was from Gibco, 10977- 015. The EDC-solution was reconstituted according to the manufacturer in filtered, deionized water to 11.5 mg/mL and stored in aliquots at -20⁰C. The NHS-solution was reconstituted according to the manufacturer in filtered, deionized water to 75.0 mg/mL and stored in aliquots at -20⁰C. The ethanolamine solution, IM, pH 8.5, was stored in aliquots at 4°C. The non specific binding reducer solution (NSB-Reducer) was from BIAcore AB, BR- 1006-91, containing 10 mg/mL of Carboxymethyl Dextran Sodium Salt in 0.15 M NaCl containing 0.02% NaN₃. The 10x PBS buffer was from Roche, 1666 789. The 10x Running buffer was 10x PBS-EDTA 10 mM, 0.5% Tween 20. To make 1 L of 10x Running buffer, 5 mL of Tween 20 and 3.7224 g of EDTA were dissolved in 10x PBS buffer. The Ix Running buffer was PBS-EDTA 1 mM, 0.05% Tween 20. To make the Ix Running buffer, the 10x Running buffer was diluted 1 :10 with distillated water and filtrate the buffer on 0.2 μm filter. Sodium acetate, 10 mM, pH 4.0 was from BIAcore, BR-1003-49. Hydrochloric acid (HCl), 100 mM, was made by diluting 4M HCL in double distilled water.

[29] The LBY135 stock solution was 43.8 mg/mL, stored at < -70⁰C in aliquots in polypropylene tubes. The working solution (LBY-WS) was 1 mg/ml in running buffer, pH 7.4. The LBY-WS was prepared freshly on day of use.

[30] The Protein G stock solution had a concentration of 2 mg/mL in ultra pure water, stored in aliquots at -20⁰C in polypropylene tubes. The working solution (protein G-WS): concentration 100 μg/mL in 10 mM Sodium Acetate, pH 4.0, prepared freshly on day of use.

[31] Use of new, uncoated sensor chip. A sensor chip CM5 was docked according to the manual's instructions (BIAcore 3000, Instrument Handbook) and "Prime" was run (Tools, Working Tools) with running buffer. The temperature of the BIAcore 3000 was set to 25 ⁰C. [32] Use of coated sensor chip. If a sensor chip was already coated with protein G before, the chip was stored in running buffer at 4 ⁰C. Before use, it was rinsed carefully with double distilled water. Blocking and priming were performed as described above for the use of new, uncoated sensor chip.

[33] Covalent coupling of Protein G to sensor chip. Protein G was covalently bound to the dextran matrix of the sensor chip by amino coupling. Protein G was coupled to sensor surface (FC) 2, 3 and 4 but not to sensor surface 1. [34] Procedure of a reference sensor surface preparation (FCl). To generate a reference surface for background subtraction, the sensor surface 1 (FCl) was only activated with EDC/NHS and deactivated with ethanolamine (EtOH-NH). Formation of air bubbles while preparing the samples should be avoided. Vials were sealed with a cap. All steps were performed manually:

1. Activation of carboxyl residues on the sensor chip:

• 100 μl EDC solution and 100 μl NHS solution were mixed in a BIAcore tube and vortexed for a few seconds.

• 35 μl at 5 μl /min was injected to reach a signal of 200-300 RU

2. Deactivation of non-coupled carboxyl residues:

• 35 μl ethanolamine solution was injected at 5 μl /min.

[35] Procedure of an active sensor surface preparation (FC2, 3 and 4). All steps were performed manually:

1. Activation of carboxyl residues on the sensor chip:

• 35 μl at 5 μl /min was injected to reach a signal of 200-300 RU

2. Coupling of protein G on FC-2, FC-3 and FC-4:

• Protein G (6 μl of 100 μg/ml) in sodium acetate buffer, pH 4.0, was injected at 5 μl/min to reach a signal of 400 - 800 RU.

3. Deactivation of non-coupled carboxyl residues:

• 35 μl ethanolamine solution was injected at 5 μl /min.

[36] Step 1-3 were completed for each active sensor surface before proceeding to the next one. Parallel surface preparation is not recommended.

[37] Sensor chip acceptance criteria. The suitability of a sensor chip was tested at four levels: (a) the activation of sensor surface; (b) the amount of protein G coupled to a sensor chip; (c) stability of sensor chip surface over time; and (d) the sensor chip suitability test. [38] For the surface activation, an activation level between 200 and 300 RU was accepted. If this criterion cannot be fulfilled a new chip must be prepared with freshly diluted reagents. [39] For the protein G coupling, an immobilization level of protein G to FC 2, 3 and 4 between 400 and 800 RU was accepted. If this criterion cannot be fulfilled, a new chip is prepared with freshly diluted reagents.

[40] For the sensor chip stability test, the chip surface was stabilized before use as follows: 15 μl of 1000 μg/ml LBYl 35-WS was injected at 5 μl/min onto the reference and the protein G surfaces. The surfaces were regenerated twice with 15 μl of 100 mM HCL, at 50 μl/min. The specific signal for an LBY135 injection after ten consecutive cycles was accepted when between 1000 and 2000 RU. The specific signal is the relative signal obtained on the LBY 135 surface minus the relative signal obtained on the reference surface (report points: baseline, 15 sec before injection; LBY135 value, 4 min +/- 5 sec after injection). A loss of signal of 10 RU between the last 3 injections was accepted. If this criterion cannot be fulfilled LBY 135 must be injected as many times as required to reach a stable signal. If the criterion cannot be fulfilled, a new chip is prepared with freshly diluted reagents.

[41] For the sensor chip suitability test, after chip stabilization the signal of positive and negative control sera were assessed in duplicate. NSB-Reducer (final concentration: 1 mg/mL), samples and running buffer were mixed 2:1 :17 and injected at 5 μl/min on the reference and protein G surfaces. One sample cycle consists of LBY 135 injection, sample injection and surface regeneration. The specific signal obtained for the human blank serum negative control (HS53 was used for this validation) was accepted when < 200 RU (report points: baseline, 15 sec before injection; human blank serum value, 4 min +/- 5 sec after injection). In a specific test, we re-measured the negative control several times during the validation run. The change of signal over 91 cycles was calculated. [42] The specific signal obtained with the rabbit positive control was accepted when > 1000. If these criteria cannot be fulfilled a new chip must be prepared with freshly diluted reagents. In a specific test, we re-measured the positive control several times during the validation run. The change of signal over 91 cycles was calculated. [43] Preparation of human blank sera. NSB-Reducer (final concentration: 1 mg/mL), human blank serum and running buffer were mixed 2:1 :17 prior to starting the assay run. [44] Preparation of calibrators (Cs) and quality control samples (QCs). Cs and QCs were freshly prepared on the day of analysis by successive dilution of the rabbit positive control in a human serum pool. The following concentration ranges were covered: (a) 6 Cs, concentration range 1.60 to 4.61 AU and 1 anchor point at 1.3 AU; (b) 5 QCs, concentration range 1.60 to 4.01 AU.

[45] Positive control and preparation of calibrators (Cs). Since no human anti-LBY135 samples were available, anti-human IgG depleted anti-LBY135-Fab rabbit serum was used as a positive control. To obtain the positive control, a rabbit was hyper-immunized with the Fab- fragment of LBY 135 and adjuvant. The resulting anti-serum was depleted of anti-human IgG on a human IgG coupled affinity column. The resulting anti-serum is supposed to recognize the variable mouse region (including the complementary determining region (CDR)) of LBY135. This positive control was used as the anchor point for the calibration curve. It was further serially diluted in 1 :2 steps in a human serum pool (HS58 and HS66) to give the 6 Cs. Cs were given as loglO of their final dilution and are expressed in arbitrary units (AU). NSB- Reducer (final concentration : 1 mg/mL), Cs and running buffer were mixed 2:1 :17 before injection on the sensor surface resulting in a dilution for the anchor point of 1 :20 (1.30 AU) , the highest C 1 :2 x 1 :20 (1.60 AU), the third C 1 :8 x 1 :20 (2.20 AU), the forth C 1 :32 x 1 :20 ( 2.81 AU), the fifth C 1 :128 x 1 :20 (3.41 AU), the sixth C 1 :512 x 1 :20 (4.01 AU) and the lowest C 1 :2048 x 1 :20 (4.61 AU).

[46] Preparation of quality control samples (QCs). NSB-Reducer (final concentration: 1 mg/mL), QCs and running buffer were mixed 2:1 :17 before injection on the sensor surface. The dilution for the highest QC was 1 :2 x 1 :20 (1.60 AU), the second QC 1 : 16 x 1 :20 (2.51 AU), the third QC 1 :64 x 1 :20 (3.11 AU), the forth QC 1 :256 x 1 :20 (3.71 AU) and the lowest QC 1 :512 x 1 :20 (4.01 AU).

[47] Assay procedure. After surface activation, protein G immobilization, surface inactivation and chip stabilization, the chip was ready for suitability check and sample measurement.

[48] The program of TABLE 5 was developed and used for suitability check and validation runs. The program is intended to be used unaltered for in-study sample measurement. Negative (HS53) and positive control were run every 10-20 cycles.

[49] Assay validation. As described above, the assay to quantify anti-LBY135 in human serum was validated based on Mire-Sluis AR et ah, Journal of Immunological Methods 289:1-16 (2004). QCs were used to establish the working range of the assay. Negative cut- off (NCO), A- factor and intra- and inter-assay variability were assessed on 3 independently prepared sensor surfaces on the same sensor chip.

[50] Negative cut-off (NCO) determination. The NCO is the border line between immunogenicity positive and negative. Values above the NCO are considered as positive and below as negative. Twenty- five individual lots of human blank sera were used to establish the NCO for each sensor surface (see FIG. 1). All samples were run in duplicate. The calculated NCO value was set at the mean of all blank values + 2SD, which represents the 95^th percentile of a normal distribution. The NCO for FC2 was 104.4 RU, for FC3 was 114.4 RU, and for FC4 was 189.4 RU.

[51] Negative cut-off normalization factor (A). A factor (A- factor) is required to normalize the NCO amongst chip surfaces. The difference between the NCO value and one selected human blank serum (HS53) was calculated for each chip surface (Al, A2, A3) (see FIG. 1). The mean A-factor of the three chip surfaces (A1+A2+A3) / 3 = A) was calculated and was used to normalization the NCO between surfaces. The A-factor determined was 18.8 RU. The NCO for a specific chip is calculated as: HS53 (RU) + A-factor (RU) = NCO (RU). [52] Calibration curve. Three independent calibration curves (C-curve a-c) were prepared and run on all surfaces in parallel (FCl, FC2, FC3 and FC4) using 6 calibrators and 1 anchor point in the range from 1.30 to 4.61 AU. The quality of the individual C-curves was assessed by the accuracies and the precisions of the back-calculated concentrations of the calibration standards.

[53] Accuracy, precision, lower limit of quantification (LLOQ). Three independent QC sets (QC sets a-c) were prepared and run on all surfaces in parallel (FCl, FC2, FC3 and FC4) using 5 different levels in the range from 1.60 to 4.01 AU. Accuracy and precision of anti- LBY 135 detection were assessed by calculating the intra-surface accuracy and precision and the inter-surface accuracy and precision of the QCs with respect to the Cs.

[54] Performance results obtained during method validation. Parameters were calculated according to the following definitions:

[55] Interference of the drug. In this EXAMPLE, LBY 135 in the sample interferes with the antibody detection. The drug captures the antibody in solution and thereby inhibits partially or totally binding of the antibody to its target molecule used for detection in the assay. [56] To investigate the impact of LBY 135 on anti-LBY135 detection, three levels of immune response (strong, mid, weak) were spiked with increasing amounts of LBY 135 (from 0.46 μg/ml to 3000 μg/ml). The spiked samples were incubated over the week-end at 4°C to allow complex formation between LBY135 and the anti-LBY135 antibodies. The samples were analyzed by the usual procedure on the next day. The impact of LBY 135 on signal detection was evaluated by an inhibition curve. The determination of the NCO value for each surface was calculated as the mean value of HS53 measured before and after each level of immune response + the normalization factor ( A = 18.8 RU). Inhibition curves were fit using the software Origin 7.5.

[57] To mimic samples with strong, mid or weak physiological immune response, the anti- human IgG depleted rabbit anti-LBY135-Fab serum (positive control) was used. Titers take the 1 :20 dilution in NSB-Reducer and buffer before sample injection into consideration. [58] -For strong immune response (positive control diluted 1 : 16 in human serum (1.90 AU)) the LBY 135 concentration to inhibit the positive signal to the NCO was determined at 111 μg/ml. (See FIG. 3). For mid immune response (positive control diluted 1 : 16 in human serum (2.51 AU)) the inhibitory LBY 135 concentration to inhibit the positive signal to the NCO was determined at 37 μg/ml. (See FIG. 4). For weak immune response (positive control diluted 1 :32 in human serum (2.81 AU)) the inhibitory LBY 135 concentration to inhibit the positive signal to the NCO was determined at 12 μg/ml. (See FIG. 5). [59] Inhibitory drug concentrations were defined as the drug concentration required to reach full inhibition of the signal (inhibition plateau). This plateau was almost overlapping with the NCO.

[60] Thus, strong immunogenicity (1.90 AU) can be detected up to 111 μg/ml LBY135, mid immunogenicity (2.51 AU) up to 37 μg/ml LBY 135 and weak immunogenicity (2.81 AU) up to 12 μg/ml LBY 135 in the sample.

[61] Conclusion. We have developed a quantitative screening method for anti-LBY135 antibodies in human serum. The principle of the method is the interaction of anti-LBY135 with LBY 135 based on plasmon resonance spectroscopy (BIAcore 3000 instrument). Many samples can be run on one sensor surface. LBY 135 is immobilized in a reversibly manner to the sensor surface via protein G, which is covalently bound to the chip. Protein G coupling makes sure that the CDR regions of LBY135 are accessible. The fact that LBY135 is newly injected after every cycle, and not exposed to gradual denaturation by acid regeneration, has lead to a very precise and stable assay. EXAMPLE 2

IMPROVEMENT OF DRUG TOLERANCE BY ACID TREATMENT ON BIACORE

[62] Described herein is an improved surface plasmon resonance assay, which can be applied to analyze drug-complexed ADA. A chimeric therapeutic monoclonal antibody

(LBY135) with tumoricidal activity in humans was selected as the drug. The assay uses an acid treatment step but does not remove excess antigen.

[63] Materials: Normal human sera were obtained from healthy and untreated individuals.

For the development of the acid dissociation procedure, immune complexes needed to be prepared.

[64] During assay development, an antibody positive human sample was not available for use as a positive control. Therefore, a monoclonal antibody raised in rabbit was used as a surrogate ADA (anti LBY135) for indicating assay performance.

[65] Pooled human serum samples spiked with ADAs of known concentration were used for assessing the ADA binding level on the Biacore chip in RU (Resonance Unit) at a range of concentrations to ensure confidence in the results obtained after acid treatment of the immune complexes.

[66] In addition, clinical samples obtained from patients treated with LBY 135 were used for the assessment of immunogenicity in the presence of drug excess. All used clinical samples (see TABLE 11) had been previously tested for immunogenicity in a clinical study.

[67] Preparation of immune complexes: Immune complexes were prepared by spiking ADAs and drug into neat pooled human serum. The final concentration of the ADAs in the different sample amounted 0, 5, 7.5, 10, 25, 100 μg/ml. Drug at a concentration of 1 mg/ml was added into the samples containing the ADAs and gently mixed on a vortex. An incubation time of 1 h at 37°C in a thermomixer at 300 rpm allowed the drug - ADA binding (the immune complex formation). The immune complex-spiked serum was treated with IM hydrochloric acid with a final pH of 2.5 and 3.0 in the sample containing rabbit ADA and human ADA, respectively. After incubation at 37°C for 30 min in a thermomixer at 300 rpm the samples were neutralized by adding 1 x PBS, NSB reducer and Biacore running buffer at a final pH of around 6.0. The samples were then measured on a Biacore machine. [68] Biacore assay: The ADAs were assessed using real-time Biomolecular Interaction Analysis (BIA) on Biacore based on surface plasmon resonance as detection method. Antibody immunoassay measurements were performed using a BIACORE 2000 and BIACORE TlOO.

[69] In the Biacore assay, the sensor chip consists of a carbomethyl dextran layer covalently attached to a gold film coated on to a glass slide. Fagerstam LG et ah, Journal of Chromatography 597(l-2):397-410 (1992); Malmqvist M, Nature 361 (6408): 186- 187 (1993). Protein G was covalently bound to the dextran matrix of the sensor chip via amino coupling and was used to bind the therapeutic monoclonal antibodies (LBY 135, 1 mg/ml) to the sensor chip, making sure that the potentially immunogenic CDR regions are exposed. Protein G was immobilized to sensor surface (FC) 2, 3 and 4 but not to sensor surface 1. To generate a reference surface for background subtraction, the sensor surface 1 (FCl) was only activated with EDC/NHS and deactivated with EtOH-NH. Samples are injected and binding of ADAs (anti-LBY135) to drug (LBY135) was measured in real time. Since LBY135 is bound to protein G reversibly, it was removed from the sensor surface after every cycle and injected newly before the next sample is applied. The surface was regenerated using of 100 mM HCl. [70] Thus, this Biacore assay was successfully used for the detection of ADAs (anti- LBY135) directed against a monoclonal chimeric therapeutic antibody (LBY135) in presence of high drug (LBY135) concentration.

[71] Impact of acid treatment on human serum and rabbit ADAs. The impact of acid treatment on human serum and polyclonal rabbit anti-LBY135-antibodies (ADAs) was investigated. In a first step, the affinity purified ADAs were spiked into neat pooled human serum to generate a concentration curve ranging from 5 to 100 μg/ml. Samples were measured on a Biacore chip coupled with LBY135. In a second step, blank pooled and AD As-spiked samples were treated with hydrochloric acid adjusting a final pH of 2.5 before injection onto the Biacore chip. All samples were diluted with Biacore running buffer containing NSB reducer to reach a final pH of around 6. Acid treated samples lead to a signal reduction (2.5 - 57 %) as compared to untreated samples (FIG. 6). This reduction was also observed in blank serum, which suggests that acidification did not have a major impact on ADAs themselves, but rather denaturated unspecific matrix proteins. In consequence, the signal reduction obtained with the blank serum was defined as non specific binding (NSB) to the sensor surface and was subtracted from signals obtained with untreated ADAs spiked samples, in order to obtain a reference signal acid treated ADAs containing samples. After NSB subtraction from the samples, we recovered around 97 -107 % of spiked ADAs, which indicates that ADAs remained biologically active after the acidification step. [72] Recovery of ADAs from anti-LBY135-LBY135 immune complexes on Biacore by acid dissociation. Immune complexes were prepared by spiking LBY 135 (1 mg/ml) together with the surrogate rabbit ADAs in different concentrations into neat pooled human serum. The inhibition of anti-drug antibody binding on the Biacore surface evidenced that the immune complex formation was successful (FIG. 7). After forming of immune complexes, the serum was acidified and further treated as described above. 97 - 100.0 % of the spiked ADAs were recovered by this procedure, using the ADA spiked sample without LBY135 and minus NSB as a reference. The successful recovery at such high drug levels might be facilitated by the fact that acid-dissociated ADAs preferably re-bind to the immobilized drug surface, than to the soluble drug.

[73] The same procedure was applied to two different rabbit sera containing anti-human IgG depleted LBY135-specific antibodies. The rabbit sera were also spiked with excess of drug (1 mg/ml). After formation of anti-LBY135-LBY135 complexes, the samples were treated with hydrochloric acid by the above mentioned procedure. (FIG. 8). About 97- 99 % of ADAs were recovered in the binding assay.

[74] Recovery of ADAs from clinical samples. Using rabbit surrogate ADAs spiked in pooled human serum, we were able to recover nearly 100% of ADAs from fully drug inhibited samples (10 - 200 times excess of LBY 135 over ADAs). To investigate the variability between rabbit positive control ADAs and human derived ADAs with respect to acid treatment conditions, clinical samples with previously detected immunogenicity were used. One patient, with low PK values, who had been treated 6 times over 18 weeks with LBY 135 at 0.3 mg/kg, was chosen. One mg/ml LBY 135 was spiked into the serum samples (day 1, day 41, day 61, day 83, day 102, day 124) to form immune complexes. Without acid treatment, the spiked LBY 135 completely inhibited ADA detection. Applying the above described treatment conditions as used in experiments with the rabbit surrogate ADAs, we recovered only around 25 % of human ADAs. However, by changing the acid treatment conditions to pH 3 instead of pH 2.5, we obtained a recovery of about 95 - 100 % in the clinical samples.

[75] Due to affinity maturation of ADAs after repeated drug applications, we concluded that the affinity of ADAs was increasing over time. We found no significant difference in the recovery of ADAs for the different sampling time points of the patient after acid treatment. This suggests that the immune complex dissociation at low pH is probably not affected by the antibody affinity. At the same time, it seems that human derived antibodies seem to be more sensitive to acid denaturation than rabbit derived antibodies.

[76] Investigation of the influence of different drug concentration on ADA recovery. The impact of drug concentration on ADAs recovery from drug-AD As complexes was evaluated, by spiking increasing amounts of drug (0.2, 1 and 2 mg/ml) into a clinical serum sample with known immunogenicity. The serum after the last LBY 135 treatment was chosen, because we assumed that in the course of immune response maturation the affinity of the anti-drug antibodies, developed after the last drug application, is the highest. It is established that Biacore assays are less sensitive for detection of high affinity antibodies compared to low affinity antibodies in presence of drug. Lofgren JA et ah, The Journal of immunology 178:7467-7472 (2007). None of the three different drug concentrations were tolerated in the assay and hence the binding of ADAs on the drug coupled on the solid phase was inhibited. The acid dissociation step resulted in a recovery of 92 % ADAs for 0.2 mg/ml LBYl 35, 96 % for 1 mg/ml LBY135 and 99 % for 2 mg/ml LBY135 (FIG. 10).

[77] Specificity. To confirm the specificity of the rabbit positive to LBY 135, a non-specific monoclonal antibody (ACZ885, a human anti-human ILl-beta antibody) was coupled to a chip sensor surface. The positive control was run over this surface and did not bind. This confirmed the specificity of the rabbit control to LBY 135. This also provided information of the real specificity of the ADAs by binding to the drug and not some other moieties present in the serum.

[78] Analysis. Immunogenicity of therapeutic antibodies is an important topic, since it can affect safety, pharmacokinetics (PK) and efficacy. Thus, it has been suggested that during development of biotherapeutics, immunogenicity should always be monitored and potential clinical consequences evaluated. Wadhwa M et al, Dev. Biol. (Basel) 122:155-70 (2005). [79] A hurdle for immunogenicity assays has been the detection of ADAs in the presence of high drug concentrations. Under such conditions, ADAs are complexed by the drug and therefore not detectable by the ADA screening assay anymore.

[80] To asses the applicability of the acid dissociation procedure to a Biacore assay, we first evaluated the acid stability of blank human serum and drug specific rabbit ADAs, which served as a positive control. Our data (see above) suggests that acidification did not have a major impact on ADAs themselves, but rather denatured unspecifϊc serum proteins, thereby reducing non-specific binding of serum samples to the Biacore chip.

[81] We then investigated the potential of sample acidification to recover ADAs from drug- ADAs complexes using the rabbit positive control as well as immunogenicity positive patients' samples with negligible LBY135 concentrations. The drug (LBY135) was spiked at a final concentration of 1 mg/ml into human serum samples spiked with different concentrations of rabbit ADAs (5 - 100 μg/ml) corresponded to a 10-200 - fold excess of LBY35. LBY135 was also spiked into patient samples. Since the ADA concentration of patient samples was unknown, the LBY 135 excess could not be calculated. ADA-drug complex formation was performed in serum at 37°C to mimic physiological conditions. Wadhwa M et al, Dev. Biol. (Basel) 122:155-70 (2005). LBY135 inhibited the detection of ADAs in untreated samples completely. After acid treatment 97 -100 % of rabbit ADAs and 88 -99 % of patient ADAs were recovered, including all patient samples and rabbit ADAs samples tested. This high yield recovery was irrespective of ADAs concentration or LBY 135 excess.

[82] Acid treatment conditions for serum spiked with rabbit ADAs and immunogenicity positive patient samples differed slightly. For rabbit ADAs, a pH of pH 2.5 was required, whereas for patient samples pH 3.0 was used to obtain the maximum ADAs recovery. The discrepancy in pH may be explained by the different sensitivity of antibodies to acid between species. Since the affinity of ADA-drug complexes could be another source of different behavior to acid treatment, several samples of one patient were analyzed after repeated drug application (TABLE 11). Multiple drug applications are assumed to increase ADA affinity, due to the maturation of the immune response. Agur Z et al., Proceedings of the Biological Science, 245(1313):\47-\50 (1991). Addition of 200 μg/ml LBY135 to each sample demonstrated that after each drug application the affinity of ADAs increased. Although the ADA signal increased over time, the spiked LBY 135 did inhibit the ADAs signal in samples after the first drug application by 46 %, whereas ADAs occurring after the last drug application were fully inhibited. This can be explained by the fact that higher affinity ADAs, at a later stage of treatment, bind better to the drug than low affinity ADAs at beginning of treatment (data not shown). However, after acid treatment of all LBY135 spiked patient samples, nearly 100 % of the ADAs were recovered for all patient samples, independent of ADAs affinity (FIG. 9). [83] We found that using acid treatment in combination with a Biacore ADA assay, nearly 100 % of ADAs could be recovered in human serum spiked with rabbit ADAs and in immunogenicity positive patient samples, after complete inhibition of the ADA signal with excess drug. In addition, acid treatment reduced non-specific binding of serum samples to the Biacore chip and is likely to also reduce undesired specific binding by denaturing soluble target in samples in case of antibody drugs. Based on the obtained data, acid treatment of preclinical and clinical samples followed by Biacore analysis should allow for a more reliable analysis of immunogenicity at high drug concentrations.

[84] Thus, surface plasmon resonance and comparable flow system methods are now applicable to detect ADAs directed against therapeutic proteins in presence of excess drug. Using such approaches, immunogenicity sampling time points should not be limited to low drug levels anymore, but should only be guided by clinical considerations.

EQUIVALENTS

[85] The details of one or more embodiments of the invention are set forth in the accompanying description above. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention, the preferred methods and materials are now described. Other features, objects, and advantages of the invention will be apparent from the description and the claims. In the specification and the appended claims, the singular forms include plural referents unless the context clearly dictates otherwise. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. All references cited herein are incorporated herein by reference in their entirety and for all purposes to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference in its entirety for all purposes.

[86] The present invention is not to be limited in terms of the particular embodiments described in this application, which are intended as single illustrations of individual aspects of the invention. Many modifications and variations of this invention can be made without departing from its spirit and scope, as will be apparent to those skilled in the art. Functionally equivalent methods and apparatuses within the scope of the invention, in addition to those enumerated herein, will be apparent to those skilled in the art from the foregoing descriptions. Such modifications and variations are intended to fall within the scope of the appended claims. The present invention is to be limited only by the terms of the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims

CLAIMSWe claim:

1. A method based on label free optical detection for detecting anti-drug antibodies present in a sample containing residual drug, comprising the steps of:

(a) obtaining a sensor for a label free optical detector, wherein the sensor has been modified to permit the binding of a drug, wherein the drug is a therapeutic agent;

(b) applying the therapeutic protein reversibly or non-reversibly to the modified sensor, such that the drug is bound by the modified sensor, resulting in a drug- sensor complex;

(c) detecting the presence and amount of the drug-sensor complex;

(d) applying an acidic solution to the patient sample suspected of containing drug and anti-drug antibody (ADA);

(e) neutralizing the acidified patient sample suspected of containing drug and ADA;

(f) applying the neutralized patient sample suspected of containing drug and ADA to the modified sensor, such that the ADA are bound by the modified sensor, resulting in an ADA-drug-sensor complex;

(g) detecting the presence and amount of the ADA-drug-sensor complex; (h) determining the difference between the amount of the ADA-drug-sensor complex and the prior drug-sensor complex, wherein the difference between both is a measurement of the presence of anti-drug antibodies in the patient sample; (i) evaluating the specificity and identity of the bound material under consideration of a reference sensor and reference samples that account for non- specific binding and biological variability between samples and individuals.

2. The method of claim 1, wherein the label free optical detection method is surface plasmon resonance

3. The method of claim 1, wherein the acid solution has a pH of pH 3.0 or lower.

4. The method of claim 1, wherein the acid solution has a pH of pH 2.5 or lower.

5. The method of claim 1, wherein the neutralized patient samples has a pH of pH 6 or higher.

6 The method of claim 1 , wherein the drug is a therapeutic protein

7. The method of claim 1, wherein the drug is a therapeutic antibody.

8. The method of claim 7, wherein the drug is a monoclonal therapeutic antibody.

9. The method of claim 7, wherein the drug is an anti-cancer therapeutic antibody.

10. The method of claim 7, wherein the drug is LBY135.

11. The method of claim 7, wherein the drug is ACZ885.

12. The method of claim 1 , wherein the modified sensor comprises protein G covalently bound to the sensor surface.

13. The method of claim 12, wherein the drug is LBY135 and the drug-sensor complex comprises an LBY135-protein G complex, wherein the protein G is covalently bound to the sensor surface.