Bacterial endotoxin test for the determination of endotoxins

Abstract

Herein is reported a method for determining bacterial endotoxin at low concentrations in a sample of an antibody (that has been produced using bacterial cells) comprising the following steps in the following order: i) adding magnesium ions to the sample, ii) diluting the sample, iii) dialyzing the sample having a pH-value of 5.7-8.0 against an endotoxin-flee aqueous solution, and iv) determining bacterial endotoxin in the sample using a bacterial endotoxin test, particularly the Limulus amoebocyte lysate assay.

Claims

1. A method for the reduction of endotoxin-masking and/or for overcoming Low Endotoxin Recovery (LER effect) in a limulus amoebocyte lysate (LAL) assay of a sample comprising an antibody, wherein the method comprises the following steps in the following order: (a) adding magnesium ions to the sample, (b) diluting the sample, and (c) dialyzing the sample having a pH-value of 5.7-8.0 against an endotoxin-free aqueous solution, wherein: the antibody is a monoclonal therapeutic antibody, said endotoxin masking and/or said LER effect is caused by endotoxin-binding proteins present in said sample and/or formulation ingredients or buffer components, and the formulation ingredients or buffer components comprise an amphiphilic compound combined with citrate buffer or phosphate buffer.

2. The method of claim 1, wherein said sample is a formulation sample.

3. The method of claim 1, wherein said magnesium ions in step (a) are added in form of MgCl.sub.2.

4. The method of claim 1, wherein in step (a) magnesium ions are added to a final concentration of about 10 to 100 mM.

5. The method of claim 1, wherein the antibody is formulated with polysorbate 80 and with a citrate buffer.

6. The method of claim 5, wherein the antibody is formulated with about 25 mM sodium citrate buffer and about 700 mg/l polysorbate 80 and has a pH value of about 6.5.

7. The method of claim 1, wherein said method comprises additionally the production of a low endotoxin recovery (LER) positive control by spiking a known amount of endotoxin into an aliquot of the sample comprising the antibody.

8. The method of claim 7, wherein said LER positive control exhibits a LER effect if steps (a) to (c) of the method of claim 1 has not been performed on the positive control.

9. The method of claim 8, wherein said production of said low endotoxin recovery (LER) positive control comprises shaking of the endotoxin spiked aliquot.

10. The method of claim 9, wherein said production of said low endotoxin recovery (LER) positive control is by spiking a known amount of endotoxin into an aliquot of the sample and shaking the endotoxin spiked aliquot of the sample for 60 min to 2 hours.

11. The method of claim 7, wherein said production of said low endotoxin recovery (LER) positive control comprises spiking said aliquot of the sample comprising the antibody with Controlled Standard Endotoxin (CSE).

12. The method of claim 11, wherein said CSE spiked to said aliquot is in a defined concentration.

13. The method of claim 1, wherein the antibody is the anti-CD20 antibody rituximab.

14. The method of claim 1, wherein in step (b) the pH-value of the sample is adjusted by diluting the sample with 10-50 mM Tris/HCl buffer pH 6.0-9.0.

15. The method of claim 1, wherein in step (b) the sample is diluted at a ratio of 1:10.

16. The method of claim 1, wherein during dialysis in step (c) the sample has a pH-value of 6.0-8.0.

17. The method of claim 1, wherein in step (c) the dialysis takes about 24 hours at room temperature.

18. The method of claim 1, wherein for the dialysis in step (c) a membrane with a molecular-weight cut-off of 10 kDa is used.

19. The method of claim 1, wherein for the dialysis in step (c) a cellulose acetate membrane is used.

20. The method of claim 1, further comprising changing water twice in dialysis step (c).

21. The method according to claim 1, wherein said amphiphilic compound is a non-ionic detergent.

22. The method according to claim 21, wherein said non-ionic detergent is polysorbate.

23. The method according to claim 22, wherein said polysorbate is polysorbate 80.

24. The method according to claim 4, wherein said magnesium ions are added to a final concentration of about 25 to 75 mM.

25. The method according to claim 9, wherein said shaking is for about 45 min to about 2 hours.

26. The method according to claim 12, wherein said defined concentration is about 0.5 or about 5 EU/ml.

27. The method according to claim 14, wherein said buffer is at pH 6.0-8.0.

Description

DESCRIPTION OF THE FIGURES

(1) FIG. 1 Time dependency of dialysis of NeoRecormon® containing phosphate and polysorbate 20 by using a MWCO of 12-16 kDa. Shown is the weight of the inner dialyzate obtained after the indicated dialysis time at room temperature, lyophilization and weighting. Data on top of gray bars show the average amount of 2 measurements (%).

(2) FIG. 2 Dialysis of NeoRecormon® containing phosphate and polysorbate 20 by using a membrane MWCO of 12-16 kDa treated with or without (w/o) bovine serum albumin (BSA) prior to dialysis. Shown is the content of phosphate (P) in the inner dialyzate obtained after the time indicated. Left bars show the amount of P when the membrane was treated with 0.2% BSA before dialysis. Right bars correspond to the amount of P without BSA-treatment. The photometric test of the phosphate recovered from the inner dialyzate was performed according to Strominger (1959, J. Biol. Chem. 234: 3263-3267).

(3) FIGS. 3A-3B: [Rituximab 115] and [Rituximab 117] Recovery rates (%) obtained by performing the protocol for overcoming the LER effect as described in Example 2.1 by using 3A Rituximab and 3B Rituximab placebo as sample. In the Figures “fast spin” and “slow spin” means the frequency of the stirrer (i.e. “fast spin” means that the frequency of the stirrer is high). This exemplary protocol is also useful for routine quality control of other samples, preferably for specimen containing sodium citrate buffer and polysorbate 80 as detergent.

(4) FIGS. 4A-4B: Schematic representation of a modified protocol for overcoming the LER effect and recovery rates obtained by performing said protocol. 4A Schematic representation of a protocol according to the invention for overcoming the LER effect (e.g. in rituximab and rituximab placebo). The detailed protocol is described in Example 2.2. This exemplary protocol is also useful for routine quality control of other samples, preferably for specimen containing sodium citrate buffer and polysorbate 80 as detergent. 4B [rituximab 046] Recovery rate (%) of rituximab and rituximab placebo obtained by the LER assay after performing the protocol according to FIG. 4A. For further details, see protocol described in Example 2.2.

(5) FIG. 5: [Rituximab 059] Recovery rates (%) obtained by performing the protocol as described in Reference Example 2 by using Rituximab as sample.

(6) FIG. 6: Recovery rates (%) obtained by performing the protocol as described in Reference Example 3 by using Rituximab as sample. Recovery rates obtained by performing the protocol as described in Reference Example 2.2 [rituximab 061].

(7) FIGS. 7A-7B: Recovery rates (%) obtained by performing the protocol as described in Reference Example 4 by using Rituximab as sample. 7A Recover) rates obtained by performing the protocol as described in Reference Example 3.1 [rituximab 062]. 7B Recover rates obtained by performing the protocol as described in Reference Example 3.2 [rituximab 063].

(8) FIGS. 8A-8B: Recovery rates (%) obtained by performing the protocol as described in Reference Example 5 by using Rituximab as sample. 8A Recover)/rates obtained by performing the protocol as described in Reference Example 4.1 [rituximab 064]. 8B Recover rates obtained by performing the protocol as described in Reference Example 4.2 [rituximab 065].

(9) FIG. 9: Recovery rates (%) obtained by performing the protocol as described in Reference Example 6 by using rituximab and rituximab placebo as sample. [rituximab 072].

(10) FIGS. 10A-10D: Recovery rates (%) obtained by performing the protocol as described in Reference Example 7 by using rituximab and rituximab placebo as sample. 10A [rituximab 079] no incubation; 10B [rituximab 080] 4 h incubation; 10C [rituximab 081] 1 day incubation; 10D [rituximab 082] 3 days incubation.

(11) FIGS. 11A-11C: Recovery rates (%) obtained by performing the LAL assay as described in Reference Example 8 by using Rituximab as sample. 11A [rituximab 002] LAL assay with different dilutions; 11B [rituximab 004] comparison of Lonza and ACC CSE spiking; 11C [rituximab 005] LAL assay with different dilutions and pH adjustment.

(12) FIG. 12: Recovery rates (%) obtained by performing the LAL assay as described in Reference Example 8 by using Rituximab as sample. Dialysis and dilution alone does not overcome the LER effect [rituximab 011].

(13) FIG. 13: Time dependency of the LER effect. The Figure shows recovery rates (%) obtained by performing spiking and the LAL assay as described in Reference Example 1 by using Rituximab as sample [rituximab 027]. The shaking time after spiking (i.e. 2 sec to 60 min) is indicated.

(14) FIGS. 14A-14B: Importance of the buffer system on the LER effect. Recovery rates (%) obtained by performing the LAL assay as described in Reference Example 10 are shown. 14A LAL assay when Rituximab or sodium citrate are used as sample and diluted at a ratio of 1:2, 1:5, 1:10, or 1:20 [rituximab 006]; 14B LAL assay when sodium citrate; polysorbate 80 or sodium citrate and polysorbate 80 are used as sample and diluted at a ratio of 1:2, 1:5 or 1:10 [rituximab 029].

(15) FIGS. 15A-15D: Effect of MgCl.sub.2 on the LER effect. Recovery rates (%) obtained by performing the LAL assay as described in Reference Example 13 are shown. 15A Addition of MgCl.sub.2 to a concentration of 10 mM [rituximab 030]; 15B Addition of MgCl.sub.2 to a concentration of 50 mM [rituximab 031]; 15C Addition of MgCl.sub.2 to a concentration of 25 mM [rituximab 032]; 15D Addition of MgCl.sub.2 to a concentration of 75 mM [rituximab 033].

(16) FIG. 16: Effect of mechanical treatments on the LER effect. Recovery rates (%) obtained by performing the LAL assay as described in Reference Example 14 are shown [rituximab 034]. In the Figure, “shaken” means shaken for 60 min.

(17) The following Examples are provided to aid the understanding of the present invention, the true scope of which is set forth in the appended claims. It is understood that modifications can be made in the procedures set forth without departing from the spirit of the invention.

EXAMPLE 1

Technical Equipment and Reagents

(18) 1. Technical Equipment

(19) 1.1 Microplate Reader System (Herein Also Designated as “Reader”) Infinite® 200 PRO, Multimode Microplate Reader; Tecan, Switzerland/Tecan Deutschland GmbH, Germany, P/N: 30050303. Magellan V. 7.1 Software Costar™ Cell Culture Plates, 96 Wells, Fisher Scientific, P/N: 07-200-89.

(20) 1.2 Shaker System and Glass Vials Multi Reax; Heidolph, Germany, P/N: 545-10000-00. 1.5 ml Screw Neck Glass Vials (N8); Macherey-Nagel GmbH & Co. KG, Germany, P/N: 702004 (Qty. of 100). N 8 PP screw cap, black, closed top; Macherey-Nagel GmbH & Co. KG, Germany, P/N: 70250 (Qty. of 100). 4 ml Screw Neck Glass Vials (N13); Macherey-Nagel GmbH & Co. KG, Germany, P/N: 702962 (Qty. of 100). N 13 PP screw cap, black, closed top; Macherey-Nagel GmbH & Co. KG, Germany, P/N: 702051 (Qty. of 100).

(21) 1.3 Dialysis Equipment SpinDIALYZER™, chamber volume 1000 μl; Harvard Apparatus, U.S.A., P/N 740314 (Qty. of 1) and 740306 (Qty. of 5), local distributor: Hugo Sachs Elektronik Harvard Apparatus, GmbH, Germany, P/N SP1 74-0306 (Qty. of 5). Remark: Use of Lot No: 032613. The dialyzer is a simple single-sided device for dialysis of biological samples. A broad range of dialyzer sizes are available to accommodate sample volumes ranging from 20 μl to 5 ml. The catalogue Nb. for 1 ml (as used herein) is 74-0314. The MWCO of the membrane ranges from 100 to 300,000 Da. The entire unit is constructed of PTFE, a virtually unreactive material. Fast SpinDIALYZER, chamber volume 1000 μl, Harvard Apparatus, U.S.A., P/N 740510 (Qty. of 1) or 740504 (Qty. of 5), Remark: Two-sided membrane system, top plus bottom membrane. The dialyzer is a reusable sample chamber made of PTFE for high sample recovery and has been redesigned to provide larger membrane surface areas for an even faster dialysis rate. The Ultra-Fast Dialyzers are of 50 μl to 1500 μl volume and have been used here with 1000 μl. The catalogue Nb. for 1 ml (as used herein) is 74-0412. Cellulose acetate membranes, 500 Da MWCO, Harvard Apparatus, U.S.A., P/N: SP1 7425-CA500, local distributor: Hugo Sachs Elektronik Harvard Apparatus GmbH, Germany, P/N: SP1 7425-CA500. Cellulose acetate membranes, 10 kDa MWCO, Harvard Apparatus, U.S.A., P/N: SP1 7425-CA10K, local distributor: Hugo Sachs Elektronik Harvard Apparatus GmbH, Germany, P/N: SP1 7425-CA10K. Remark: Tested in addition to the ‘standard’ 500 Da MWCO membranes in LER investigations on Rituximab as well as in the LER experiments on NeoRecormon®. Cellulose acetate membranes, 25 kDa MWCO, Harvard Apparatus, U.S.A., P/N: SP1 7425-CA25K, local distributor: Hugo Sachs Elektronik Harvard Apparatus GmbH, Germany, P/N: SP1 7425-CA25K. Remark: Tested in addition to the ‘standard’ 500 Da MWCO membranes in LER investigations on Rituximab as well as in the LER experiments on NeoRecormon®. Aqua B. Braun, sterile pyrogen-free water, 1 l, B. Braun Melsungen A G, Germany, P/N: 14090586. Crystallizing Dishes, 900 ml, OMNILAB, Germany, P/N: 5144008. (Remark: Use for rinsing of dialysis membranes) DURAN® Beakers, tall form, 2000 ml, OMNILAB, Germany, P/N: 5013163. DURAN® Beakers, tall form, 250 ml, OMNILAB, Germany, P/N: 5013136.

(22) 1.4 Routine Laboratory Equipments Autoclaving System (Remark: Use for sterilisation of dialyses chambers) epT.I.P.S.® LoRetention-Reloads, PCR clean, 0.5-10 μl, Eppendorf, Germany, P/N: 0030072.057 epT.I.P.S.® LoRetention-Reloads, PCR clean, 2-200 μl, Eppendorf, Germany, P/N: 0030072.022 epT.I.P.S.® LoRetention-Reloads, PCR clean, 50-1000 μl, Eppendorf, Germany, P/N: 0030072.030 Stripettes®, Individual, 5 ml, Paper/Plastic Wrap, Fisher Scientific, P/N: 10420201.

(23) 2. Reagents

(24) 2.1 Kinetic Chromogenic LAL Assays and LAL-Associated Reagents Kinetic-QCL™ Kit; Lonza, Switzerland, P/N: 50-650U or 50-650H (i.e. “Lonza kit”). CHROMO-LAL von Associates of Cape Cod (AAC) Inc., USA, P/N: C0031-5 (i.e. “ACC kit”). Endotoxin E. coli O55:B5 for K-QCL; Lonza, Switzerland, P/N: E50-643. Endotoxin E. coli O55:B5, 2.5 mg/vial; Lonza, Switzerland, P/N: N185. LAL Reagent Water—100 ml; Lonza, Switzerland, P/N: W50-100. MgCl.sub.2, 10 mM solution for use with LAL, 30 ml vial; Lonza, Switzerland, P/N: S50-641. Magnesium chloride hexahydrate for analysis EMSURE® ACS, ISO, Reag. Ph. Eur., 250 g; Merck, Germany, P/N: 1.05833.0250. Tris buffer, 50 mM solution for use with LAL, 30 ml vial; Lonza, Switzerland, P/N: S50-642.

(25) 2.2 Protein Reagents Albumin bovine Fraction V, very low endotoxin, fatty acid free, 25 g; Serva, Germany, P/N: 47299.04. Albumin, human serum, fraction V, high purity; 1 g; Merck, Germany, P/N: 126658-1GM.

(26) 3. Tested Pharmaceuticals

(27) For the herein described Examples, Rituximab (which comprises) and NeoRecormon® (which comprises epoetin-beta) were used. In addition, the respective placebos of Rituximab and NeoRecormon® were also applied in the herein described methods.

(28) The placebo of the respective sample is identical to the sample except for the absence of the active therapeutic ingredient, i.e. rituximab placebo does not contain rituximab but all other component of the formulation.

EXAMPLE 2

Methods of the Invention for Overcoming the LER Effect

EXAMPLE 2.1

Protocol for Overcoming the LER Effect

(29) In this Example rituximab and rituximab placebo were used as sample. However, as discussed below, the herein described protocol is useful for overcoming the LER in all typical formulations of pharmaceutical antibodies.

(30) Materials Used for this Example Membranes: 10 kDa cellulose acetate (CA) membranes from Harvard Apparatus, U.S.A., P/N: SP1 7425-CA10K Dialyzer: FastSpinDIALYZER, chamber volume 1000 μl, Harvard Apparatus, U.S.A., P/N 740510 (Qty. of 1) or 740504 (Qty. of 5) Sample vials: 1.5 ml Screw Neck Glass Vials (N8); Macherey-Nagel GmbH & Co. KG, Germany, P/N: 702004 N 8 PP screw cap, black, closed top; Macherey-Nagel GmbH & Co. KG, Germany, P/N: 70250 Crystallizing dishes: 900 ml, Duran, VWR Germany, P/N: 216-1817 MgCl.sub.2—stock solution: 1M MgCl.sub.2 dissolved in water (Magnesium chloride hexahydrate for analysis EMSURE® ACS, ISO, Reag. Ph. Eur., 250 g; Merck, Germany, P/N: 1.05833.0250) Tris-Buffer, 50 mM solution for use with LAL (i.e. endotoxin-free), 30 ml vial; Lonza, Switzerland, P/N: S50-642 Samples: rituximab placebo and LAL water

(31) Step by Step Protocol:

(32) Step 1: Preparation of the samples 1× rituximab placebo 900 μl+100 μl LAL water 1× rituximab placebo 900 μl+100 μl CSE conc. 50 EU/ml=final conc. 5.0 EU/ml 1×LAL water 1000 μl 1×LAL water 900 μl+100 μl CSE conc. 50 EU/ml=final conc. 5.0 EU/ml Shake the samples 60 min at RT (room temperature) [i.e. in a Heidolph Multi Reax shaker, high speed (2,037 rpm)],

(33) Step 2: Washing of dialysis membrane Use ten 10 kDa cellulose acetate (CA) membranes and put them into the crystallizing dish with 300 ml Aqua Braun (i.e. distilled water of the manufacturer B. Braun, Melsungen) Shake them for 1 h (Shaker SG 20. IDL GmbH, Germany or equivalent, 50 to 300 rpm, preferably 100 rpm) Transfer the membranes into an new crystallizing dish with fresh Aqua Braun (also 300 ml) Shake them for 1 h (Shaker SG 20. IDL GmbH, Germany or equivalent, 50 to 300 rpm, preferably 100 rpm)

(34) Step 3: Addition of MgCl.sub.2 to a final MgCl.sub.2 concentration of about 50 mM MgCl.sub.2 Add 50 μl of the 1M MgCl.sub.2 stock solution to the samples of step 1 Shake them 1 min. [i.e. in a Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature] Incubate the samples for 60 min at room temperature Shake them 1 min. [i.e. in a Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(35) Step 4: Dilution Take one of the samples of step 3 and dilute it 1:10 with buffer, pH˜7 (i.e. 50 mM Tris/HCl buffer pH˜7) 895 μl 50 mM Tris-buffer+105 μl sample Perform it twice for a repeat determination (i.e. determination in duplicates): 2× rituximab placebo 1:10 with Tris-buffer 2× rituximab placebo 5.0 EU/ml 1:10 with Tris-buffer 2×LAL water 1:10 with Tris-buffer 2×LAL water 5.0 EU/ml 1:10 with Tris-buffer

(36) Step 5: Dialysis Shake all diluted samples for 1 min [i.e. in a Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature], Transfer into the FastSpinDIALYZER Put one dialyzer per beaker (i.e. DURAN® baker, tall form, 2000 ml, OMNILAB, Germany, P/N: 5013163) on a magnetic stirrer plate. Adjust the frequency of the stirrer to be high (i.e. “fast spin”). A high frequency of the stirrer means 50-300 rpm, preferably 200-300 rpm. The stirrer is a heat-sterilized (4 hours at 250° C.) magnetic stirrer having a length of about 40 mm and a diameter of about 14 mm. Fill the beaker with 200 ml Aqua Braun Dialyze 24 h and exchange the Aqua Braun after 2 h and 4 h at room temperature (21±2° C.) After dialysis transfer the sample into new 1.5 ml screw vials

(37) Step 6: Shaking Shake the samples for 20 min. [i.e. in a Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(38) Step 7: Preparation of the LER positive control (i.e. the positive LER control) and of further water controls Prepare the LER positive control 1 h before the dialysis ends 1. rituximab placebo 900 μl+100 μl LAL water 2. rituximab placebo 900 μl+100 μCSE conc. 50 EU/ml=final conc. 5.0 EU/ml 3. LAL water 1000 μl 4. LAL water 900 μl+100 μl CSE conc. 50 EU/ml=final conc. 5.0 EU/ml Shake 1 h [i.e. in a Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature] Dilute samples 1:10 (sample:LAL water) with LAL water Shake [i.e. in a Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature] for 1 min

(39) Step 8: LAL assay Prepare the standard and start the LAL assay according to the instructions of the manufacturer (Kinetic-QCL™ assay; Lonza)

(40) Results and Discussion

(41) As can be seen in FIGS. 3A (i.e. [rituximab 117]) and 3B (i.e. [rituximab 115]) the above described method is able to overcome the LER effect. In addition, by using this method, the LER effect can be overcome in rituximab as well as in rituximab placebo. This indicates that the above described protocol in not dependent on a formulation comprising a particular monoclonal antibody but can be used to obviate the LER effect in every formulation comprising polysorbate 80 and a chelating buffer (such as sodium citrate). This formulation is typical for antibodies, in particular monoclonal antibodies. Thus, the above described method is expected to be useful to overcome the LER effect in every antibody formulation.

(42) It has been found that Mg.sup.2+ is the divalent cation of choice to restore LAL reactivity in formulations containing chelating buffers (such as sodium citrate) and showing the LER effect.

(43) In order to remove the chelating buffer (e.g. the Sodium citrate buffer), a second step (after addition of Mg.sup.2+) is to perform dialysis. The spinDIALYZER™ of Harvard is the preferred equipment for the dialysis.

(44) The detergent (e.g. polysorbate 80) represents the second reason for the LER effect. In general, the presence of detergents (such as polysorbate 80) in a biological sample leads to micelle formation in case the critical micellar concentration (CMC) of the detergent (usually in the μM range) is reached. Micelles may inhibit the LPS-mediated activation of factor C, a serine protease representing the first enzyme in the LAL-cascade reaction (Nakamura (1988a) J. Biochem. 103: 370-374). In monoclonal antibody preparations, the undiluted sample is usually above the CMC in order to obtain a functional solubilisation of the antibody. In the products which were investigated here, the CMC of the detergents indeed exceeded their CMC (polysorbate 80: 700 mg/l (50 fold excess)) leading to the assumption that polysorbate 80 is present in form of micelles. In the above described protocol the concentration of the detergent is reduced by dilution so that the concentration of the detergent is near/drops below the CMC value (polysorbate 80: 14 mg/l or 10.6 μM). Dilution of the detergent to near-CMC concentrations may eliminate the micellar compartmentalization, and therefore, render the CSE molecules spiked accessible for the LAL enzymes.

(45) Accordingly, the problem of the LER effect, (e.g. in the event sodium citrate and polysorbate 80 are used for the formulation of a pharmaceutical product) can now be considered as being solved. In conclusion, herewith provided is a safe, robust and reproducible testing method for pharmaceutical products.

(46) In summary, in rituximab and rituximab placebo the above described protocol surprisingly overcomes the LER effect. By contrast, the same protocol could not reveal satisfactory results for NeoRecormon® (which does not comprise an antibody but epoetin-beta) indicating that the herein provided methods are particularly useful for antibody formulations, preferably for formulations with monoclonal antibodies, citrate buffer and polysorbate 80.

EXAMPLE 2.2

Modified Protocol (1) for Overcoming the LER Effect

(47) In this Example a modified protocol has been used which nevertheless overcomes the LER effect. The most important changes compared to Example 2.1 are as follows: 1. In Example 2.2 the Spin Dialyzer has been used. In contrast, in Example 2.1 the FastSpinDIALYZER is used which has more efficient dialysis chambers and increases the efficiency of the dialysis (the membranes are on both sides of the cylinder). 2. In Example 2.2 the MWCO of the dialysis membrane is 500 Da. In contrast, in Example 2.1 the MWCO of the dialysis membrane is 10 kDa. 3. In Example 2.2 the dilution is 1:10 with endotoxin-free water. In contrast, in Example 2.1 the dilution 1:10 with Tris-buffer pH˜7 (i.e. Tris/HCL buffer pH˜7). By diluting the sample with endotoxin-free water at a ratio of 1:10 the pH value of the sample is adjusted to about pH 6.0. 4. In Example 2.2 the dialysis time is 4 h. In contrast, in Example 2.1 the dialysis time is 24 h. In this Example rituximab and rituximab placebo were used as sample. However, for the same reasons as discussed with respect to Example 2.1, this protocol is useful for overcoming the LER in all typical antibody formulations.

(48) In particular, the protocol used in Example 2.2 is detailed as follows.

(49) Protocol Overview Step 1: “Setting up the LER effect” (see also below “LER positive control”): Rituximab and rituximab placebo samples were spiked with 5 EU/ml or 0.5 EU/ml (CSE; Lonza, E. coli O055:B5) and the mixture was shaken for 60 min at room temperature at maximum speed [Shaker: Heidolph Multi Reax, high speed (2,037 rpm)] to obtain a “positive LER-effect” sample. Step 2: Adding MgCl.sub.2: Before dialysis, add 2 M MgCl.sub.2 stock solution so that the final conc. is about 50 mM MgCl.sub.2; 1 min shaking as in step 1. Step 3: 1:10 Dilution [one sample without dilution (undiluted) as reference]; shaking for 1 min as in step 1. Step 4: Dialysis for 4 h using a 500 Da membrane (30 min pre-incubated with 0.2% BSA; optionally but not mandatory), exchange of water after 2 h once. Transferring of the solution from the dialysis chamber into a glass vial and shaking as in step 1 for 20 min at RT (room temperature, i.e. 21±2° C.). Step 5: kinetic LAL-assay measurement.

(50) Detailed Protocol

(51) Step 1: Preparation of the samples Preparation antibody solution (rituximab) for 50 mM MgCl.sub.2: Fill 1 tube with 877.5 μl rituximab+97.5 μl CSE (stock solution of 50 (5) EU CSE/ml.fwdarw.5 (0.5) EU/ml final concentration). Unspiked control: 877.5 μl rituximab placebo+97.5 μl water Unspiked water control: 975 μl water (for blank subtraction) used vials: clear flat bottom small opening 1.5 ml Macherey & Nagel, Ref. Nr. 70213 1 h shaking on Heidolph Multi Reax, high speed (2,037 rpm) at room temperature.

(52) Step 2: Addition of MgCl.sub.2 to a final concentration of 50 mM MgCl.sub.2 Stock solution 1M MgCl.sub.2.6H.sub.2O: add 50 μl of a 1M MgCl.sub.2-stock solution to the spiked sample as well as to the unspiked sample (blank).

(53) Step 3: Dilution Sample rituximab with 50 mM MgCl.sub.2: prepare a 1:10 dilution by adding 900 μl endotoxin-free water (i.e. LAL water)+100 μl sample Water control is treated the same with endotoxin-free water (i.e. LAL water) instead of rituximab: dilute 1:10

(54) Step 4: Dialysis 1 min shaking before dialysis [i.e. in a Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]. Put samples into the 1 ml dialyzer chambers (Harvard Spin Dialyzer) to which a membrane with MWCO 500 Da (optionally, 30 min pre-incubated with 0.2% BSA) is fixed. Dialyze 4 h against 1 l Aqua Braun (i.e. sterile, pyrogen free water; as supplied by B. Braun, Melsungen) at 24° C.; change water after 2 h. Changing water has been tempered also to 24° C. The Spin Dialyzers are distributed (depending on the number of dialysis chambers) over multiple 21 beakers filled with 1 l Aqua Braun under stirring (magnetic Teflon stirrer). There are at most 5 Dialyzer in one 2 l beaker.

(55) Step 5: Preparation of the LER positive control

(56) Also in this Example a LER positive control is used in the LAL assay. This LER positive control can be prepared at any time, provided that it is ready if the LAL assay starts. Advantageously, the LER positive control is prepared 1 h before the end of 4 h dialysis, so that all samples are ready for testing at the same time. For preparing the LER positive control the following protocol is used: rituximab 900 μl+100 μl CSE.fwdarw.final conc. CSE: 5.0 EU/ml. Shaking in a Heidolph Multi Reax, high speed (2,037 rpm) at RT for 1 h. Only under these conditions the max. LER effect (<1% recovery rate) will be obtained. In parallel prepare the following blanks: Water with 5.0 EU/ml CSE Water with 5.0 EU/ml CSE; diluted 1:10 (0.5 EU/ml).

(57) Step 6: LAL assay Start test after all samples are prepared. From all samples two aliquots of 100 μl are used for repeat determination (2×, i.e. determination in duplicates) in a plate which is incubated 10 min at 37° C. in the Tecan Reader. Add 100 μLAL Reagent (Kinetic-QCL™ Assay; Lonza) to each sample in a well-defined sequence (according to the read-out of the machine).

(58) Results and Discussion:

(59) The protocol described in Example 2.1 resulted in best reproducible recovery rates (also with respect to the water controls). However, the protocol described in Example 2.2 resulted in a good CSE recovery-rate ranging from 50 to 95% for both CSE concentrations spiked (see FIG. 4B [rituximab 046]). Therefore, it can be concluded that the protocol used in Example 2.2 represents a functional equivalent to the protocol described in Example 2.1.

REFERENCE EXAMPLE 1

Time Dependency of the LER Effect

(60) In the prior art it is assumed that the LER effect appears immediately after spiking of the sample with a defined amount of CSE (C. Platco, 2014, “Low lipopolysaccharide recovery versus low endotoxin recovery in common biological product matrices”. American Pharmaceutical Review, Sep. 1, 2014, pp. 1-6). Therefore, first the samples were shaked after LPS spiking for a rather short time of about 2-10 min at room temperature. However, this kind of spiking turned out to be inefficient and some experiments indicated that the masking effect of the material spiked has not yet reached its maximum during this short time interval (<10 min). It was found that the mechanism of spiking is one of the fundamental processes in analyzing the LER effect in a correct way (see, e.g., FIG. 13 [rituximab 027]). According to these data, the LER effect is a kinetic phenomenon, which requires time to mask the CSE molecules e.g. by penetrating into the micelles of the formulation mixture. Thus, shaking for 2-10 min prior to the next step for analyzing the LER effect, as it represents the routine practice, are inappropriate and cannot be considered to be representative for the LER effect, because the conditions for its formation has not yet been reached. Therefore, an internal standard to test the “positive LER effect” (defined to be present in case the recovery rate of 0% by the LAL test has been reached) was included in the experiments. By performing a kinetic study on the LER effect in rituximab, the positive LER effect was demonstrated to need ≥60 min incubation time.

(61) In particular, it was analyzed how long shaking has to be carried out [max. frequency (i.e. vortexing) in a Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature in (21° C.±2° C.) a 1.5 ml clear glass, crimp neck, flat bottom vessel], in order to achieve the maximum LER effect. Therefore, rituximab samples were spiked with CSE in a vial, so as to obtain 0.5 and 5.0 EU/ml (vials by Macherey-Nagel, 1.5 ml). After spiking, the samples were shaked for 60 min, 30 min, 10 min, 5 min, or 2 sec, respectively. Afterwards, 1:10 dilutions were prepared by mixing 900 μl endotoxin-free water (i.e. LAL water) with 100 μl sample. After dilution, the samples were again shaked for 1 min. Subsequently, the samples were tested in the LAL assay in duplicates. In particular, 100 μl of each sample was applied onto a plate and incubated in the reader for 10 min at 37° C. Then, 100 μl chromogen was added to each sample and the measurement was carried out. In this experiment, all solutions had room temperature. As can be seen in FIG. 13 [rituximab 027], the LER effect is lower (i.e. the recovery rate is higher) if the sample is diluted at a ratio of 1:10 as compared to the corresponding undiluted sample. In addition, after 2 sec shaking the recovery values for the diluted samples with 5.0 EU/ml endotoxin were still at approximately 50%. However, increasing shaking (i.e. vortexing) time results in a constant decrease of the recovery rate (with exception of the 30-min value), see FIG. 13 [rituximab 027]. In contrast, the undiluted samples show the maximum LER effect already after 2 sec. However, also the diluted samples showed a significant LER effect in the samples which have been shaked (i.e. vortexed) for 60 min.

(62) From this result it was concluded that spiking needs time to mask the LPS molecules into the detergent micelles. The “positive LER effect” is complete when about 100% masking or <0.5% recovery rates of CSE are obtained. This process requires a minimum of 1 h during shaking at room temperature [e.g., shaker: Heidolph Multi Reax, high speed (2,037 rpm) for 1 h at room temperature in a 1.5 to 5 ml clear glass, crimp neck, flat bottom] or alternatively storage at 4° C. for a longer time period >24 h. The resulting “positive LER control” is shown in all graphical plots as one bar in the graphical presentations at the right side of the diagram.

REFERENCE EXAMPLE 2

Influence of Human Serum Albumin (has) and Different MgCl.SUB.2 .Concentrations on the Recovery Rate

(63) In order to determine the effect of HSA and different MgCl.sub.2 concentrations on the recovery rate of endotoxin spiked rituximab samples, the following experiment has been performed. In addition, in this experiment the influence of dialysis on the recovery rate has been analyzed. More specifically, rituximab spiked samples were shaken for 60 min in order to obtain the “positive LER effect”. Prior to the dialysis, 10-75 mM MgCl.sub.2 were added, subsequently, a dilution was performed. No BSA-blocked membrane was used. After the dialysis, 0.01 μg/ml HSA is either added or not added. Subsequently, shaking for 20 min is performed. In addition, some samples were not dialyzed at all. In particular, the different samples which have been tested in the LAL assay are shown in FIG. 5 (i.e. [rituximab 059]). In this experiment, the LER effect could be overcome in some samples without dialysis. However, further experiments demonstrated that without dialysis the LER effect cannot reproducibly been overcome. Or, in other words, without dialysis, the LER effect is sometimes overcome and sometimes not. Thus, dialyzing the samples results in a more robust method for overcoming the LER effect.

(64) The samples have been prepared in a 1.5 ml screw neck vial by Macherey-Nagel.

(65) Step 1: Preparation of the samples Preparation of spiked rituximab for 10 mM MgCl.sub.2: 897 μl rituximab+99.8 μl CSE so that 5.0 EU/ml are obtained Preparation of spiked rituximab for 50 mM MgCl.sub.2: 889 μl rituximab+98.8 μl CSE so that 5.0 EU/ml are obtained Preparation of spiked rituximab for 75 mM MgCl.sub.2: 883 μl rituximab+98.1 μl CSE so that 5.0 EU/ml are obtained Preparation of spiked water for 10 mM MgCl.sub.2: 897 μl water+99.8 μl CSE so that 5.0 EU/ml are obtained Preparation of spiked water for 50 mM MgCl.sub.2: 889 μl water+98.8 μl CSE so that 5.0 EU/ml are obtained Preparation of spiked water for 75 mM MgCl.sub.2: 883 μl water+98.1 μl CSE so that 5.0 EU/ml are obtained Shake for 60 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(66) Step 2: Addition of MgCl.sub.2 A 4 M MgCl.sub.2 stock solution (i.e. 511.437 mg MgCl.sub.2.6H2O in 0.629 ml water) was used. For 10 mM MgCl.sub.2 2.5 μl of the 4 M solution are added to the spiked sample. For 50 mM MgCl.sub.2 12.5 μl of the 4 M solution are added to the spiked sample. For 75 mM MgCl.sub.2 19 μl of the 4 M solution are added to the spiked sample. Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(67) Step 3: Dilution Dilutions at a ratio of 1:10 were prepared as follows: Preparation rituximab 1:10: always 900 μl LAL water+100 μl sample The water was not diluted 1:10 since there are not enough dialyzers available. Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(68) Step 4: Dialysis The samples were put into a 1 mL dialyzer. A 500 Da membrane. However, the membrane was washed in LAL water. Dialysis was performed against 1 L Aqua Braun at 24° C. for 4 h, and after 2 h the water was changed. The new water also had a temperature of 24° C. The dialyzers were located in three 2 L beakers and rotated since there was a long stirrer (i.e. stir bar) in each beaker. There are always 4 dialyzers in each beaker.

(69) Step 5: Addition of HSA after dialysis After the dialysis, the samples were portioned. For the preparation of HSA-samples 396 μl of each sample were added to a separate vial. For the preparation of samples without HSA 400 μl were added to a separate vial. To obtain a HSA concentration of 0.01 μg/ml, 4 μl of a 1 μg/ml solution were added to the 396 μl samples The HSA stock solution was newly prepared. Shake for 20 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(70) Step 6: Preparation of the LER positive control The LER positive control is prepared 1 h before the end of the 4 h dialysis so that it is ready at the same time as the other samples. rituximab 900 μl+100 μl CSE of different CSE stock solutions so that 5.0 EU/ml are obtained shaken at room temperature for 1 h [Heidolph Multi Reax shaker, high speed (2,037 rpm)]

(71) Step 7: LAL assay 100 μl of each samples were applied onto a plate in double determination Incubation in the reader at 37° C. for 10 min. 100 μl chromogen were applied to each sample. Starting the measurement in the reader.

(72) Results and Discussion:

(73) The results are shown in FIG. 5 [rituximab 059]. This experiment demonstrates that HSA treatment reduces the recovery rate and is therefore less useful in context of the herein provided methods. In addition, the results show that BSA treatment of the dialysis membrane is not necessary to obtain satisfactory recovery rates. In addition, this experiment also demonstrates that 50 mM MgCl.sub.2 is the optimum value for recovery, 10 and 75 mM MgCl.sub.2 result in lower recovery. However, also with 75 mM MgCl.sub.2 a satisfactory recovery rate was obtained. Moreover, this experiment shows that addition of MgCl.sub.2 leads to a recovery within a satisfactory range (70-100%) even without dialysis. However, as mentioned above, without dialysis the LER effect cannot reproducibly been overcome. Thus, dialyzing the samples results in a more robust method for overcoming the LER effect. It is indicated that in the experiment [rituximab 059] the water control values were high (part of the values >220%). The LER positive control is satisfactory; i.e. 0% recovery.

REFERENCE EXAMPLE 3

4 Hours Incubation Time After Addition of MgCl.SUB.2

(74) In this experiment rituximab samples were shaken for 60 min in order to achieve the “positive LER effect”. After addition of MgCl.sub.2 the undiluted samples were incubated for 4 h at room temperature). After this incubation, the samples were shaked for 2 min. The different samples which have been tested in the LAL assay are shown in FIG. 6 [rituximab 061].

(75) The samples have been prepared in a 1.5 ml screw neck vial by Macherey-Nagel.

(76) Step 1: Preparation of the samples Preparation of spiked rituximab/water for 10 mM MgCl.sub.2: 897 μl rituximab/water+99.8 μl CSE so that 5.0 EU/ml are obtained Preparation of spiked rituximab/water for 50 mM MgCl.sub.2: 889 μl rituximab/water+98.8 μl CSE so that 5.0 EU/ml are obtained Preparation of spiked rituximab/water for 75 mM MgCl.sub.2: 883 μl rituximab/water+98.1 μl CSE so that 5.0 EU/ml are obtained Preparation of spiked rituximab/water for 100 mM MgCl.sub.2: 877 μl rituximab/water+97.5 μl CSE so that 5.0 EU/ml are obtained Preparation of spiked rituximab/water for 150 mM MgCl.sub.2: 866 μl rituximab/water+96.3 μl CSE so that 5.0 EU/ml are obtained Shake for 60 mM [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(77) Step 2: Addition of MgCl.sub.2 A 4 M MgCl.sub.2 stock solution (i.e. 534.661 mg MgCl.sub.2.6H.sub.2O in 0.657 ml water) was used. For 10 mM MgCl.sub.2 2.5 μl of the 4M solution are added to the spiked sample. For 50 mM MgCl.sub.2 12.5 μl of the 4M solution are added to the spiked sample. For 75 mM MgCl.sub.2 19 μl of the 4M solution are added to the spiked sample. For 100 mM MgCl.sub.2 25 μl of the 4M solution are added to the spiked sample. For 150 mM MgCl.sub.2 37 μl of the 4M solution are added to the spiked sample. Shake for 1 min [high speed (2,037 rpm) at room temperature]

(78) Step 3: Dilution Dilutions at a ratio of 1:10 were prepared: 900 μl LAL water+100 μl sample (i.e. rituximab sample or water sample) Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(79) Step 4: Preparation of the LER positive control rituximab 900 μl+100 μl CSE so that 5.0 EU/ml are obtained Another LER positive control was prepared by mixing 900 μl rituximab+100 μl CSE so that 5.0 EU/ml are obtained; and subsequently diluting the sample 1:10 with endotoxin-free water

(80) Step 5: Shaking All samples as well as the LER positive controls were shaken at room temperature for 1 h [Heidolph Multi Reax shaker, high speed (2,037 rpm)]

(81) Step 6: LAL assay 100 μl of each sample were applied onto a plate in double determination Incubation in the reader at 37° C. for 10 min. 100 μl chromogen was applied to each sample. Starting the measurement in the reader.

(82) Results and Discussion:

(83) The result of this experiment is shown in FIG. 6 (i.e. [rituximab 061]). This Figure demonstrates again that 50 mM MgCl.sub.2 is the reproducible optimum value for CSE recovery; 10, 75 and 150 mM show slightly inferior results. When an incubation time after addition of MgCl.sub.2 was performed, the undiluted rituximab samples did not lead to a CSE recovery at all, see FIG. 6 (i.e. [rituximab 061]). However, the 1:10 dilutions resulted in approximately 50-60% recovery (in particular when 10, 50, 75 or 100 mM MgCl.sub.2 was added). The water control values as well as the LER positive control were satisfactory. In this experiment no dialysis was performed. However, several experiments showed that dialysis is necessary for reproducibly overcoming the LER effect.

REFERENCE EXAMPLE 4

Comparison of 2 and 4 Hours Incubation Time After Addition of MgCl.SUB.2

(84) The rituximab samples were shaken for 60 min in order to achieve the “positive LER effect”. After addition of MgCl.sub.2 the undiluted samples were incubated for 2 or 4 h, then 1:10 diluted and measured in the LAL assay. The different samples which have been tested in the LAL assay are shown in FIGS. 7A and 7B (i.e. [rituximab 062] and [rituximab 063], respectively).

(85) The samples have been prepared in a 1.5 ml screw neck vial by Macherey-Nagel.

(86) Step 1: Preparation of the samples Preparation of rituximab/water for 10 mM MgCl.sub.2: 897 μl rituximab/water+99.8 μl of different CSE stock solutions so that 0.5 and 5.0 EU/ml are obtained. Preparation of rituximab/water for 50 mM MgCl.sub.2: 889 μl rituximab/water+98.8 μl of different CSE stock solutions so that 0.5 and 5.0 EU/ml are obtained. Preparation of rituximab/water for 75 mM MgCl.sub.2: 883 μl rituximab/water+98.1 μl of different CSE stock solutions so that 0.5 and 5.0 EU/ml are obtained.

(87) Step 2: Preparation of two LER positive controls rituximab 900 μl+100 μl CSE so that 0.5 and 5.0 EU/ml are obtained. Dilution of one of the LER positive controls at a ratio of 1:10 with endotoxin-free water.

(88) Step 3: Shaking All samples as well as the LER positive control are shaken for 1 h: [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(89) Step 4: Addition of MgCl.sub.2 A 4 M MgCl.sub.2 stock solution was used. For 10 mM MgCl.sub.2 2.5 μl of the 4M solution are added to the spiked sample For 50 mM MgCl.sub.2 12.5 μl of the 4M solution are added to spiked sample For 75 mM MgCl.sub.2 19 μl of the 4M solution are added to spiked sample Shake for 1 mM [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(90) Step 5: Incubation time The (undiluted) samples (as well as the LER positive controls) are portioned. One half of each sample (about 500 μl) was incubated for 2 h and the other half was incubated for 4 h, respectively.

(91) Step 6: Dilution Shake for 2 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature] Dilutions at a ratio of 1:10 were prepared: 900 μl LAL water+100 μl sample (i.e. rituximab sample or water sample). Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(92) Step 7: LAL assay 100 μl of each samples were applied onto a plate in double determination Incubation in the reader at 37° C. for 10 min. 100 μl chromogen were applied to each sample. Starting the measurement in the reader

(93) Results and Discussion:

(94) The results are shown in FIGS. 7A and 7B (i.e. [rituximab 062] and [rituximab 063], respectively). Here the recovery rates of 0.5 and 5.0 EU/ml endotoxin was measured. When 10-75 mM MgCl.sub.2 was added to the samples, the recovery was the same in the samples which were incubated for 2 h (FIG. 7A, i.e. [rituximab 062]) and in the samples which were incubated for 4 h (FIG. 7B, i.e. [rituximab 063]). In both experiments the recovery rates are very similar. In addition, in the samples which were spiked with 5.0 EU/ml endotoxin satisfactory recovery rates (80-90%) were obtained, even without dialysis. In the samples which were spiked with 0.5 EU/ml endotoxin the recovery rates were approximately 35-45%. Importantly, without dilution (at a ratio of 1:10), complete LER is observed, i. e. 0% recovery, also in the presence of 10-75 mM MgCl.sub.2. The water controls as well as the LER positive controls were satisfactory. In these experiments no dialysis has been performed. However, further experiments demonstrated that without dialysis, the LER effect is sometimes overcome and sometimes not. Thus, dialyzing the samples results in a more robust method for overcoming the LER effect.

REFERENCE EXAMPLE 5

Comparison of 2 Hours Incubation Time After Addition of Different Amounts of MgCl.SUB.2 .with No Incubation Time After Addition of Different Amounts of MgCl.SUB.2

(95) The rituximab samples were shaken for 60 min in order to achieve the “positive LER effect”. After addition of MgCl.sub.2 the undiluted samples were either not incubated or incubated for 2 h. Then 1:10 diluted and measured in the LAL assay. The different samples which have been tested in the LAL assay are shown in FIGS. 8A and 8B (i.e. [rituximab 064] and [rituximab 065], respectively).

REFERENCE EXAMPLE 5.1

No Incubation Time After Addition of MgCl.SUB.2

(96) In this experiment, no incubation was performed after addition of MgCl.sub.2 to the samples.

(97) The samples have been prepared in a 1.5 ml screw neck vial by Macherey-Nagel.

(98) Step 1: Preparation of the samples Preparation of rituximab/water for 10 mM MgCl.sub.2: 897 μl rituximab/water+99.8 μl of different CSE stock solutions so that 0.5 and 5.0 EU/ml are obtained Preparation of rituximab/water for 25 mM MgCl.sub.2: 895 μl rituximab/water+99.4 μl of different CSE stock solutions so that 0.5 and 5.0 EU/ml are obtained. Preparation of rituximab/water for 50 mM MgCl.sub.2: 889 μl rituximab/water+98.8 μl of different CSE stock solutions so that 0.5 and 5.0 EU/ml are obtained.

(99) Step 2: Preparation of two LER positive controls rituximab 900 μl+100 μl CSE so that 0.5 and 5.0 EU/ml are obtained. Dilution of one of the LER positive controls at a ratio of 1:10 with endotoxin-free water

(100) Step 3: Shaking All samples as well as the LER positive control were shaken (i.e. vortexed) for 60 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(101) Step 4: Addition of MgCl.sub.2 A 4 M MgCl.sub.2 stock solution was used. For 10 mM MgCl.sub.2 2.5 μl of the 4M solution are added to the spiked sample For 25 mM MgCl.sub.2 6.25 μl of the 4M solution are added to spiked sample For 50 mM MgCl.sub.2 12.5 μl of the 4M solution are added to spiked sample Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(102) Step 5: Dilution Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature] Dilutions at a ratio of 1:10 were prepared. Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(103) Step 6: LAL assay 100 μl of each samples were applied onto a plate in double determination Incubation in the reader at 37° C. for 10 min. 100 μl chromogen were applied to each sample. Starting the measurement in the reader.

(104) Results and Discussion: The results are shown in FIG. 8A (i.e. [rituximab 064]). For the discussion of the results see Reference Example 4.2.

REFERENCE EXAMPLE 5.2

Incubation of 2 h After Addition of MgCl.SUB.2

(105) In this experiment, the samples were incubated for 2 h after addition of MgCl.sub.2. Steps 1 to 4 were performed as described above under Reference Example 4.1. However, after addition of MgCl.sub.2 the undiluted samples were incubated for 2 h at room temperature (21° C.)). After the incubation, the following steps 5 and 6 were performed. The different samples which have been tested in the LAL assay are shown in FIG. 8B (i.e. [rituximab 065]).

(106) Step 5: Dilution Shake for 2 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature] Dilutions at a ratio of 1:10 were prepared: 900 μl LAL water+100 μl sample (i.e. rituximab sample or water sample) Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(107) Step 6: Preparation of two LER positive controls rituximab 900 μl+100 μl CSE so that 0.5 and 5.0 EU/ml are obtained. Dilution of one of the LER positive controls at a ratio of 1:10 with endotoxin-free water

(108) Step 7: Shaking All samples as well as the LER positive control were shaken for 1 h [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(109) Step 8: LAL assay 100 μl of each samples were applied onto a plate in double determination Incubation in the reader at 37° C. for 10 min. 100 μl chromogen were applied to each sample. Starting the measurement in the reader.

(110) Results and Discussion:

(111) The results of Reference Example 4.1 are shown in FIG. 8A (i.e. [rituximab 064]); the results of Reference Example 4.2 are shown in FIG. 8B (i.e. [rituximab 065]). In these two experiments, dilution and LAL measurement was either carried out immediately after addition of MgCl.sub.2 (FIG. 8A, [rituximab 064]) or after leaving the sample to rest for 2 h after addition of MgCl.sub.2 (FIG. 8B, [rituximab 065]). All recovery values were very similar and in the samples which were spiked with 5.0 EU/ml CSE, satisfactory (60-80%) recovery rates have been obtained even without dialysis. Interestingly, after an incubation time of 2 h, the 25 mM MgCl.sub.2 concentration resulted in 100% recovery. Thus, an incubation time after addition of MgCl.sub.2 seems to be a valuable measure to decrease the LER effect. However, the recovery values for the samples which were spiked with 0.5 EU/ml CSE were low with approximately 20-35%. This indicates that beside addition of Mg.sup.2+ and dilution, dialysis represents a necessary step for reliably overcoming the LER effect. In these experiments the water control values were satisfactory. The undiluted LER positive control was also satisfactory, i.e. 0%.

REFERENCE EXAMPLE 6

Comparison of Different Dilutions With Rituximab and Rituximab Placebo Samples

(112) After spiking, rituximab and rituximab placebo samples were shaken for 60 min in order to achieve the “positive LER effect”. After addition of MgCl.sub.2 the undiluted samples were shaked for 1 h and diluted at a ratio of 1:2, 1:5, 1:10 or 1:20 Afterwards the LAL assay was performed. The different samples which have been tested in the LAL assay are shown in FIG. 9 (i.e. [rituximab 072]).

(113) In particular, the following experiment has been performed:

(114) Step 1: Preparation of the samples Preparation of rituximab/rituximab placebo/water for 25 mM MgCl.sub.2: 895 μl rituximab/rituximab placebo/water+99.4 μl of different CSE stock solutions so that 0.5 and 5.0 EU/ml are obtained.

(115) Step 2: Preparation of three LER positive controls rituximab placebo 450 μl+50 μl CSE so that 0.5 und 5.0 EU/ml are obtained (first LER positive control). rituximab 450 μl+50 μl CSE so that 0.5 und 5.0 EU/ml are obtained (second LER positive control). rituximab 450 μl+50 μl CSE so that 0.5 und 5.0 EU/ml are obtained. Afterwards, this sample was diluted at a ratio of 1:10 (third LER positive control).

(116) Step 3: Shaking All samples as well as the LER positive controls were shaken (i.e. vortexed) for 60 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(117) Step 4: Addition of MgCl.sub.2 A 4 M MgCl.sub.2 stock solution was used. For 25 mM MgCl.sub.2 6.25 μl of the 4M solution are added to spiked sample Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(118) Step 5: Dilution The samples were diluted as follows: Dilution at a ratio of 1:5: 400 μl water+100 μl sample Dilution at a ratio of 1:10: 450 μl water+50 μl sample Dilution at a ratio of 1:20: 475 μl water+25 μl sample One of the three LER positive controls was diluted at a ratio of 1:10. The water without MgCl.sub.2 was not diluted. Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(119) Step 6: LAL assay 100 μl of each samples were applied onto a plate in double determination Incubation in the reader at 37° C. for 10 min. 100 μl chromogen was applied to each sample. Starting the measurement in the reader.

(120) Results and Discussion:

(121) The results are shown in FIG. 9 (i.e. [rituximab 072]). Importantly, the results for rituximab and rituximab placebo show no significant differences. This indicates that the LER effect in rituximab is mainly based on the buffer system (i.e. citrate buffer with polysorbate 80) and that the antibody (i.e. rituximab) does not have a significant impact on the LER effect. However, in this experiment the recovery rates are unsatisfactory for both rituximab and rituximab placebo. In the experiment described above, the water control values were satisfactory. The undiluted LER positive controls were satisfactory, too, with recovery rates of 0%.

REFERENCE EXAMPLE 7

Influence of Incubation Time Before Addition of MgCl.SUB.2 .on the Recovery Rate in Rituximab and Rituximab Placebo Samples

(122) In the following experiment it was tested whether incubation times before addition of MgCl.sub.2 have an influence on the recovery rate of rituximab and rituximab placebo samples. In particular, rituximab and rituximab placebo samples were shaken for 60 min in order to achieve the “positive LER effect”. Then the samples were incubated at 4° C. for 0 h to 3 days. Afterwards, MgCl.sub.2 was added to a concentration of 50 mM and the samples were diluted. Then, dialysis was performed with a dialysis membrane which was not BSA-blocked. The different samples which have been tested in the LAL assay are shown in FIGS. 10A-10D (i.e. [rituximab 079], and [rituximab 082]).

(123) The samples have been prepared in a 1.5 ml screw neck vial by Macherey-Nagel.

(124) Step 1: Preparation of the samples Preparation rituximab/rituximab placebo/water for 50 mM MgCl.sub.2: 5,346 μl rituximab/rituximab placebo/water+596 μl of different CSE stock solutions so that 0.5 or 5.0 EU/ml were obtained. Vortex for 60 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature] For the further procedure 1 ml of each sample was transferred into new vials

(125) Step 2: Incubation time The samples were put into the refrigerator at 4° C. for 0 h, 4 h, 1 day, 3 days, or 7 days. After the incubation time of 1 day, 3 days or 7 days the samples were shaked for 2 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]. After 0 h and 4 h incubation time the samples were not shaked.

(126) Step 3: Addition of MgCl.sub.2 A 5 M MgCl.sub.2 stock solution (i.e. 0.9055 g MgCl.sub.2 in 0.891 ml water) was used For 50 mM MgCl.sub.2 10 μl of the 5 M solution were added to the spiked samples Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(127) Step 4: Dilution Dilutions at a ratio of 1:10 were prepared: 900 μl LAL water+100 μl sample (i.e. rituximab sample, rituximab placebo sample or water sample) The water without MgCl.sub.2 was not diluted. Shake for 1 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(128) Step 5: Dialysis The samples were added into a 1 ml dialyzer. Dialysis was performed against 1 l Aqua Braun at 24° C. for 4 h, and after 2 h the water was changed. The new water also had a temperature of 24° C. A 500 Da membrane was used which has not been incubated with BSA before. The dialyzers were located in three 2 l beakers and rotated since there was a long stirrer (i.e. stir bar) in each beaker. After dialysis the samples have been transferred into new 1.5 ml vials.

(129) Step 6: Shaking Shake for 20 min [Heidolph Multi Reax shaker, high speed (2,037 rpm) at room temperature]

(130) Step 7: Preparation of the LER positive control The LER control was prepared 1 h before the end of the 4 h dialysis so that it is ready at the same time as the other samples. rituximab or rituximab placebo 900 μl+100 μl CSE to obtain 5.0 EU/mL CSE Shake at room temperature for 1 h [Heidolph Multi Reax shaker, high speed (2,037 rpm)]

(131) Step 8: LAL assay 100 μl of each samples were applied onto a plate in double determination Incubation in the reader at 37° C. for 10 min. 100 μl chromogen were applied to each sample. Starting the measurement in the reader.

(132) Results and Discussion:

(133) The results are shown in FIG. 10A (i.e. [rituximab 079], no incubation), FIG. 10B (i.e. [rituximab 080], 4 h incubation), FIG. 10C (i.e. [rituximab 081], 1 day incubation), and FIG. 10D (i.e. [rituximab 082], 3 days incubation). The results for the 7 days incubation are not shown. The results again demonstrate that by using the herein described protocols good recovery rates can be obtained for rituximab as well as for rituximab placebo samples. In addition, this experiment demonstrates that an incubation time before addition of MgCl.sub.2 does not improve the recovery rates. In particular, while the incubation time of 0-4 h led to recovery rates which are within the desired range (50-200%), the incubation times of 1 day and 3 days led to lower recovery rates (approximately 20-30%). If no (i.e. 0 h) incubation was performed, very good recovery rates were obtained for rituximab (70-80%). Also for rituximab placebo which was spiked with 5.0 EU/ml CSE, the recovery rate was satisfactory. The recovery rate for rituximab placebo which was spiked with 0.5 EU/ml CSE was negative since the blank was very high. This may indicate that this blank sample was contaminated with endotoxin. In this experiment also recovery rates of the water control (80-120%) as well as of the LER positive control (˜0%) were satisfactory.

REFERENCE EXAMPLE 8

The Protocol of the State of the Art (“LAL Assay”) and Modifications Thereof Cannot Overcome the LER Effect in Rituximab or Rituximab Placebo

(134) In this Example it was determined whether the commonly known LAL assay is able to detect endotoxins in rituximab and rituximab placebo preparations. Therefore, the following materials have been used: [rituximab 002]: Lonza CSE+Lonza reagent (i.e. Lonza kit) [rituximab 004]: ACC CSE or Lonza CSE, respectively+ACC reagent (i.e. ACC kit) [rituximab 005]: Lonza CSE+ACC reagent (i.e. ACC kit)

(135) The LAL assays have been precisely been performed as described by the manufacturer.

(136) As can be seen from FIGS. 11A-11C (i.e. [ rituximab 002], [rituximab 004] and [rituximab 005], respectively), the standard LAL assay did not lead to satisfactory recovery rates, even if several different dilutions are tested. In particular, in one experiment (i.e. [rituximab 002]), rituximab was pipetted into a 96-well plate (i.e. a microtiter plate) and spiked with Lonza CSE to a final concentration of 0.5 EU/ml or 5.0 EU/ml. Subsequently, dilutions with water as shown in FIG. 11A (i.e. [rituximab 002]) were carried out in the 96-well plate. Then a LAL assay was performed. However, as can be seen in FIG. 11A (i.e. [rituximab 002]), 50% recovery was not reached.

(137) In a similar experiment (i.e. [rituximab 004]) rituximab was pipetted into the wells of a microtiter plate and spiked with Lonza CSE and ACC CSE to a final concentration of 0.5 EU/ml or 5.0 EU/ml. Subsequently, dilutions with water as shown in FIG. 11B (i.e. [rituximab 004]) were carried out in the 96-well plate. Afterwards, measurement was performed. However, as can be seen in FIG. 11B, with ACC CSE a recovery which is more than 200% was obtained and also the Lonza CSE spiking did not result to satisfactory recovery rates.

(138) In further experiments, the effect of pH adjustment on the LAL assay was analyzed. In particular, in one experiment [rituximab 005] rituximab was pipetted into a microtiter plate and Lonza CSE spiking was performed in the plate. Subsequently, the dilutions with water or the pH adjustment as indicated in FIG. 11C [rituximab 005] were performed. However, neither the dilution nor the pH adjustment resulted in a recovery of 50% (see FIG. 11C [ rituximab 005]).

(139) Also dialysis alone does not result in a satisfactory recovery rate. More specifically, in a further experiment, rituximab was spiked with CSE to result in a final concentration of 0.5 and 5.0 EU/ml (i.e. 900 μl rituximab solution was mixed with 100 μl CSE). Subsequently, the samples were dialysed in a 1 ml Spin Dialyser (in 1 ml Teflon chambers) for 4 hours at 4° C. with one change of water after 2 h. The dialysis membrane had a MWCO of 100 Da. Then, dilutions as shown in FIG. 12 (i.e. [rituximab 011]) were performed in the plate. Subsequently, endotoxin recovery was measured by using the LAL assay. However, good recovery rates could only be obtained for the water controls. In the case of rituximab the maximum recovery was <5% (see FIG. 12, [rituximab 011]). Thus, only dialysis and dilution does not overcome the LER effect.

REFERENCE EXAMPLE 9

Hold Time Studies

(140) To identify and monitor the LER effect, endotoxin contents have been monitored over time in an endotoxin hold time study. Therefore, an undiluted sample of various buffers has been spiked with endotoxin and stored over time (up to 28 days). Acceptable endotoxin values recovered in the PPC after spiking with the appropriate sample mixture are defined to be in the range of 50-200% of the theoretical spike value (100%). The LER effect is indicated by a significant loss of endotoxins over time. In particular, an adverse trend of endotoxin values <50% of the theoretical spike value are indicative for the LER effect.

(141) Several formulation buffer components were studied in an endotoxin hold time study (for results see the following table).

(142) TABLE-US-00001 TABLE 1 Hold time studies endotoxin recovery endotoxin [EU/ml] at time spike start excipient [EU/ml] (T.sub.0) day 7 day 14 day 21 day 28 α,α-trehalose 5 4.85 4.65 3.48 4.32 4.51 NaH.sub.2PO.sub.4 5 5.27 4.64 3.3  3.4  3.37 Na.sub.2HPO.sub.4 5 5.99 5.72 5.07 5.63 5.06 Polysorbate 20 5 4.04 4.03 4.23 3.94 4.16 Polysorbate 20 + 5 0.13 0.18 n.d. n.d. n.d. Na.sub.2HPO.sub.4 Polysorbate 20 + 5 0.37 0.77 n.d. n.d. n.d. NaH.sub.2PO.sub.4 Na.sub.2HPO.sub.4 + 5 5.12 4.64 n.d. n.d. n.d. NaH.sub.2PO.sub.4 Polysorbate 20 + 5 0.91 <0.1 n.d. n.d. n.d. Na.sub.2HPO.sub.4 + NaH.sub.2PO.sub.4 sodium citrate- 5 5.2 4.7 4   3.72 4.28 dihydrate Polysorbate 80 5 3.25 3.15 3.27 3.05 3.17 NaCl 5 5.76 5.72 5.45 4.74 6.49 Na citrate + 5 1.33 0.16 n.d. n.d. n.d. polysorbate 80 + NaCl Na citrate + 5 0.8 <0.1 n.d. n.d. n.d. polysorbate 80 polysorbate 80 + 5 2.94 2.45 n.d. n.d. n.d. NaCl Urea 5 5.14 6.21 5.66 5.28 5.29 L-Leu 5 5.72 5.61 5.87 5.01 5.88 L-Ile 5 5.70 5.82 6.24 5.17 6.17 L-Thr 5 5.54 5.65 5.66 4.76 5.88 L-Glu 5 5.28 5.03 5.50 4.43 4.24 L-Phe 5 5.49 5.50 5.96 4.98 6.34 Gly 5 4.64 4.57 4.75 4.27 5.00 n.d. = not determined

(143) As can be seen from the above table, the buffers comprising polysorbate 20 and Na.sub.2HPO.sub.4; polysorbate 20 and NaH.sub.2PO.sub.4; Polysorbate 20, Na.sub.2HPO.sub.4.sup.+ and NaH.sub.2PO.sub.4; Na citrate, polysorbate 80 and NaCl; Na citrate and polysorbate 80; as well as polysorbate 80 and NaCl exhibit a LER effect.

REFERENCE EXAMPLE 10

Influence of Buffer and Detergent on the LER Effect

(144) In several experiments the effect of citrate and/or polysorbate 80 on the LER effect was analyzed. In particular, in one experiment rituximab and 25 mM sodium citrate buffer were used as samples. Before spiking, the pH was adjusted to pH 7. Subsequently, CSE spiking was performed in the plate, and the samples were diluted with water. As can be seen from FIG. 14A (i.e. [rituximab 006]), a satisfactory recovery rate could be obtained for sodium citrate by using a dilution of 1:10. However, in the case of rituximab a recovery of 50% could not be reached.

(145) In another experiment 25 mM sodium citrate buffer, polysorbate 80 and a combination of both were used as samples. In particular, the concentrations as present in Rituximab were used (i.e. polysorbate 80: 0.7 mg/ml; sodium citrate: 9 mg/ml). These buffer systems were spiked with 0.5 and 5.0 EU/ml of Lonza CSE or with Cape cod CSE (except of sodium citrate, which was spiked with Lonza only, as ACC spiking of sodium citrate buffer was already performed in experiment described above and shown in FIG. 14A (i.e. [rituximab 006]). After spiking, a 1:2 or 1:5 dilution with water was performed in the plate. In the case of polysorbate 80, several samples led to a satisfactory recovery rate between 50% and 200%. In contrast, in the case of sodium citrate buffer, only the 1:10 dilution led to a recovery rate between 50% and 200%. This indicates that the citrate buffer has a more significant impact on the LER effect as compared to polysorbate 80. Moreover, the LER effect could not be overcome in this experiment if a combination of sodium citrate and polysorbate 80 was used as a sample. This experiment indicates that in monoclonal antibody formulations the LER effect is caused by the buffer formulation (i.e. by the combination of sodium citrate buffer and polysorbate 80.

(146) These results have been verified by another experiment wherein several different dilutions were tested. In particular, samples comprising either 25 mM sodium citrate buffer (pH 6.5), 700 mg/L polysorbate 80 or both (i.e. the formulation of rituximab) were prepared. These preparations as well as water controls were spiked with Lonza CSE to a final concentration of 0.5 or 5.0 EU/ml. All samples were shaken for 1 hour at room temperature in the vortex machine shaker: Heidolph Multi Reax, high speed (2,037 rpm) in a 1.5 clear glass, crimp neck, flat bottom vessel Subsequently, the dilutions as indicated in FIG. 14B (i.e. [rituximab 029]) were performed with endotoxin-free water in 1.5 ml vials and are shaken (as before) for 1 min. After shaking, the LAL assay was performed. In particular, 100 μl of each of the samples were added to a 96-well plate and incubated in the reader for 10 min at 37° C. Then, 100 μl chromogen was added to each of the samples and the measurement was carried out. As can be seen in FIG. 14B (i.e. [rituximab 029]) the LER effect of the buffer (i.e. the sodium citrate buffer) is stronger as compared to the LER effect of the detergent (i.e. polysorbate 80). While polysorbate 80 shows a relatively constant recovery rate (˜40-90%, see FIG. 14B, [rituximab 029], columns 2-5 from the right), in citrate-buffer the LER effect is dependent on the dilution. Most importantly, a strong and reproducible LER effect is expressed if polysorbate 80 is combined with citrate buffer (as it is the case in monoclonal antibody formulations), see FIG. 14B (i.e. [rituximab 029]). In these samples, also with high dilution, the recovery rate is only ˜5-10%. Accordingly, this experiment demonstrates how a positive LER effect can be obtained. This result is pioneering in the field of endotoxin determination, as it allows for testing of several means and methods for their ability to overcome the LER effect.

(147) When analyzing the effect of buffer and detergent separately, the effect of the buffer on the LER effect was more pronounced in both NeoRecormon® and Rituximab (see, e.g. FIG. 14B, i.e. [rituximab 029]). These data surprisingly indicate that the removal of the buffer is more critical than removal of the detergent. Taking into account that the concentrations of buffers in both formulations is comparable (rituximab: 25 mM sodium citrate and NeoRecormon: 27.8 mM for sodium phosphate), the reason for this effect has to be found in the structure of the buffer and/or its physico-chemical properties. Sodium citrate is a well-known chelating anion, whereas in phosphate this effect is less pronounced. Therefore, these observations may also explain why the addition of Mg.sup.2+ is important for overcoming the LER effect, since Mg.sup.2+ is complexed by the chelating buffer reducing its concentration in the LAL test.

REFERENCE EXAMPLE 11

Standard Physical and Biochemical Methods Do Not Recover Endotoxins Masked by the LER Effect

(148) In order to overcome the LER effect (in the buffers identified as having the LER effect in Reference Example 9), different physical and biochemical methods were tested:

(149) Freezing of endotoxin spiked samples at −30° C. This study is based on the initial finding that LER is more pronounced at room temperature as compared to 2-8° C. Result: freezing of endotoxin spiked samples does not overcome LER.

(150) Heating of endotoxin spiked samples for 30 minutes at 70° C. This study was conducted because heating has shown to overcome endotoxin masking effects for some products (Dawson, 2005, LAL update. 22:1-6). Result: Heat treatment of endotoxin spiked samples does not overcome LER.

(151) Dilution of endotoxin spiked samples to maximum valid dilution (MVD). This study was conducted since sample dilution is the standard method to overcome LAL inhibition. Result: As can be seen from FIGS. 11A-11C (i.e. [rituximab 002], [rituximab 004], [rituximab 005]), dilution alone does not overcome the LER effect.

(152) Use of Endo Trap Columns for endotoxin spiked samples. These columns serve to remove endotoxins from solutions via affinity chromatography. A test was carried out with an aqueous endotoxin solution. Result: Endotoxins could not be recovered from the column.

REFERENCE EXAMPLE 12

Removal of Detergents by Dialysis

(153) Several dialysis chambers and membranes (including different sizes of the molecular weight cut-off, MWCO) available on the market have been tested as detailed below.

(154) Suitable membranes for dialysis chambers are commercially available are, e.g., cellulose acetate (MWCO 100 to 300,000 Da), regenerated cellulose (MWCO 1,000 to 50,000 Da), or cellulose ester (MWCO of 100 to 500 Da). Herein cellulose acetate and cellulose ester are preferred, cellulose acetate is most preferred.

(155) The test samples (i.e. rituximab) were diluted prior to the dialysis, this way approaching the CMC and creating increasing levels of monomers of the detergent which were expected to diffuse through the dialysis membrane. Investigations on the recovery rate of the CSE spiked revealed that in case of rituximab the regenerated cellulose was not so efficient as compared to the cellulose acetate. In a series of experiments with rituximab it was identified that the MWCO is preferably ˜10 kDa. This size is preferred because this size is thought to i) speed up the dialysis process and ii) allow also higher oligomeric aggregates (but not micelles) of the detergent to pass through the membrane, in case the hydrophobic character of the cellulose acetate (acetyl esters on the glucose polymers) will not inhibit such kind of transportation process.

(156) The experiment to determine the optimum for dialysis has been performed to mimic the situation in NeoRecormon®. As outlined earlier, buffer and detergent were those compounds in the sample formulation which mostly influenced the LER effect. In order to mimic the formulation of NeoRecormon® a defined amount of phosphate buffer in a total volume of 0.5 ml (2.7 mg) in the presence of 0.1 mg/ml polysorbate 20 was prepared and subjected to dialysis (in a spin dialyzer). In FIG. 1 is shown a very simple example of such experiment using a dialysis membrane of cellulose acetate with a MWCO of 12-16 kDa. Here the material remaining in the inner dialysate after a given time is shown, this way demonstrating the efficiency of the dialysis. In particular, the weight of material remaining in the inner dialysis chamber over a period of 72 h (3 d) was analyzed. The result is rather surprising as it shows that complete and effective dialysis of NeoRecormon® is only achieved after a longer dialysis period at room temperature (>24-48 h). Based on this experiment the preferred dialysis time is 20 h to overnight (e.g. 24 h). In addition, this result indicates that Harvard Fast Dialyzer is preferred over the Harvard Spin Dialyzer, the former having the double area of dialysis membrane, and thus leads to a quicker dialysis.

(157) In FIG. 2, there is shown the efficiency of dialysis in case phosphate is placed into the inner dialysate compartment of the dialysis. In this experiment, the dialysis membrane (MWCO 12-16 kDa) was washed with 0.2% BSA (30 min) prior to its use, in order to avoid unspecific absorbance of the CSE spiked to the sample. However, in the herein provided inventive methods a dilution (e.g. a dilution at a ratio of 1:10) of the samples reduces the concentration of the detergent.

REFERENCE EXAMPLE 13

Influence of MgCl.SUB.2 .on the LER Effect

(158) It has been found that the LER effect could be reduced by addition of MgCl.sub.2 to the sample (see, e.g., FIG. 15B (i.e. [rituximab 031]). In particular, best results were observed when the concentration of Mg.sup.2+ was twice the concentration of the sodium citrate (i.e. 50 mM Mg.sup.2+).

(159) In particular, samples comprising either 25 mM sodium citrate buffer, pH 6.5 (i.e. sodium citrate buffer, pH 6.5), 0.7 mg/ml polysorbate 80, or both (with pH 6.5, i.e. the formulation of rituximab) were prepared. These preparations as well as water controls were spiked with Lonza CSE to a final concentration of 0.5 or 5.0 EU/ml. All samples were shaken for 1 hour at room temperature [shaker: Heidolph Multi Reax, high speed (2,037 rpm) in a 1.5 clear glass, crimp neck, flat bottom vessel]. Then, MgCl.sub.2 to reach a concentration of 10 mM, 25 mM, 50 mM or 75 mM was added to the samples. Subsequently, the dilutions as indicated in FIGS. 15A, 15B, 15C, and 15D (i.e. [rituximab 030-rituximab 033]) were performed with endotoxin-free water in a 1.5 ml vial and shaken (i.e. vortexed as before) for 1 min. After shaking, the LAL assay was performed. In particular, 100 μl of each of the samples was added in a 96-well plate and incubated in the reader for 10 min at 37° C. Afterwards, 100 μl chromogen was added to each of the samples and the measurement was carried out. As can be seen in FIG. 15A (i.e. [rituximab 030]), in all diluted samples, MgCl.sub.2 (10 mM) can neutralize the complexing effect of citrate. Magnesium ions reduce the LER effect in samples which comprise polysorbate 80 as well as citrate buffer. In this case, a recovery of approximately 50% with 5.0 EU/ml and 25% with 0.5 EU/ml was achieved. The control values of water range around the theoretically expected value (i.e. 70-130%). Moreover, a comparison of FIGS. 15A, 15B, 15C, and 15D (i.e. [rituximab 030], [rituximab 031], [rituximab 032] and [rituximab 033]) show that a MgCl.sub.2 concentration which is twice the concentration of the citrate buffer (i.e. 50 mM MgCl.sub.2) leads to best recovery rates. Although 25 mM and 75 mM MgCl.sub.2 are not the optimal concentration of MgCl.sub.2, these concentrations nevertheless overcome the LER of the citrate buffer (recovery: 75-190%). In a similar experiment wherein rituximab was used as a sample, it was demonstrated that only the addition of MgCl.sub.2 to a concentration of 10 mM, 50 mM, or 75 mM MgCl.sub.2 and a subsequent dilution at a ratio of 1:10 (without dialysis) was able to lead to a satisfactory recovery of endotoxin which was spiked to a final concentration of 5.0 EU/ml (see FIG. 5, i.e. [rituximab 059]).

REFERENCE EXAMPLE 14

Effects of Mechanical Treatments on the LER Effect

(160) It was tested whether mechanical treatments (such as shaking and ultrasonification) are useful for dispersing the micelles, and thus for reducing the LER effect.

(161) In particular, endotoxin-free water (i.e. LAL water) and rituximab were spiked with Lonza CSE to achieve a final concentration of 0.5 and 5.0 EU/ml. Then, the samples were either sonicated for 1 hour or shaked for 1 hour [i.e. vortexed in the Heidolph Multi Reax shaker at high speed (2,037 rpm) at room temperature in a 1.5 ml clear glass, crimp neck, flat bottom vessels]. Then 1:10 (sample:water) dilutions were prepared with endotoxin-free water. Subsequently, the diluted samples were dialyzed by using a 12-16 kD membrane (which, before dialysis, had been incubated in 0.2% BSA for 30 min). The dialysis took place in two 2 l beakers for 4 hours. The external dialysate was 1 l Aqua Braun and the water was changed after 2 hours of dialysis. After dialysis MgCl.sub.2 was added to some of the samples (as indicated in FIG. 16, i.e. [rituximab 034]) so as to result in a final concentration of MgCl.sub.2 of 50 mM. After addition of MgCl.sub.2, all samples (also the samples without MgCl.sub.2) were shaken for 20 min (e.g. vortexed as before). Subsequently, the LAL assay was performed. In particular, 100 μl of each of the samples was added to a 96-well plate and incubated in the reader for 10 min at 37° C. Then, 100 μl chromogen was added to each of the samples and the measurement was carried out. As can be seen in FIG. 16 (i.e. [rituximab 034], the recovery rates are neither improved by shaking nor by ultrasound. Therefore, it can be concluded that mechanically dispersing the LER causing micelles by shaking or ultrasonification is compared thereto inefficient. However, addition MgCl.sub.2 (50 mM) reduced the LER effect as it resulted in improved recovery values of up to 5-20%. In addition, this experiment also demonstrates that the order of the different performed steps is important for overcoming the LER effect. In particular, in the experiment described above (and shown in FIG. 16, i.e. [rituximab 034]), the order of the steps was (a) dilution, (b) dialysis, and (c) addition of MgCl.sub.2. This order did not result in satisfactory recovery rates (see FIG. 16, i.e. [rituximab 034]). However, as demonstrated in FIGS. 3A-3B and 4A-4B (i.e. [rituximab 046], [rituximab 115] and [rituximab 117]), the order (a) addition of MgCl.sub.2, (b) dilution, and (c) dialysis results in recovery rates which fulfill the requirements of the FDA (i.e. 50%-200%).

(162) The present invention refers to the following nucleotide and amino acid sequences:

(163) TABLE-US-00002 SEQ ID NO: 1: Rituximab heavy chain, amino acid sequence QVQLQQPGAELVKPGASVKMSCKASGYTFTSYNMHWVKQTPGRGLEWIGA IYPGNGDTSYNQKFKGKATLTADKSSSTAYMQLSSLTSEDSAVYYCARST YYGGDWYFNVWGAGTTVTVSAASTKGPSVFPLAPSSKSTSGGTAALGCLV KDYFPEPVTVSWNSGALTSGVHTFPAVLQSSGLYSLSSVVTVPSSSLGTQ TYICNVNHKPSNTKVDKKVEPKSCDKTHTCPPCPAPELLGGPSVFLFPPK PKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGVEVHNAKTKPREEQY NSTYRVVSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREP QVYTLPPSRDELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPP VLDSDGSFFLYSKLTVDKSRWQQGNVFSCSVMHEALHNHYTQKSLSLSPG K SEQ ID NO: 2: Rituximab light chain, amino acid sequence QIVLSQSPAILSASPGEKVTMTCRASSSVSYIHWFQQKPGSSPKPWIYAT SNLASGVPVRFSGSGSGTSYSLTISRVEAEDAATYYCQQWTSNPPTFGGG TKLEIKRTVAAPSVFIFPPSDEQLKSGTASVVCLLNNFYPREAKVQWKVD NALQSGNSQESVTEQDSKDSTYSLSSTLTLSKADYEKHKVYACEVTHQGL SSPVTKSFNRGEC