PROCESSES FOR OBTAINING A HIGHLY CONCENTRATED ANTIBODY SOLUTION

Abstract

The present invention relates to processes for obtaining a highly concentrated antibody solution. In particular, to processes for obtaining a highly concentrated therapeutic antibody solution that may be used for highly concentrated therapeutic antibody formulations, e.g. suitable for subcutaneous administration.

Claims

1. A process for obtaining a highly concentrated antibody solution comprising the steps of subjecting a clarified cell harvest to an affinity chromatography step, and subjecting the obtained eluate to at least two ion exchange chromatography steps and at least three UF/DF steps.

2. The process of claim 1, wherein said highly concentrated antibody solution has an antibody concentration equal to or greater than about 120 g/L.

3. The process of claims 1 and 2, wherein said at least three UF/DF steps are performed with a tangential flow filtration cassette and comprise a first UF/DF performed after the first of said at least two ion exchange chromatography, a second UF/DF performed after the second of said at least two ion exchange chromatography, and a third UF/DF performed after the second UF/DF, and wherein said highly concentrated antibody solution has an antibody concentration equal to or greater than about 150 g/L.

4. The process of claim 3, wherein said third UF/DF comprises the steps of (a) equilibration of the cassette by an equilibration buffer; (b) loading of the cassette with an antibody solution with antibody concentration comprised between about 50 g/L and about 90 g/L; (c) first ultrafiltration to concentrate the antibody to a concentration comprised between about 80 g/L and about 120 g/L; (d) diafiltration using a diafiltration buffer; (e) second ultrafiltration to concentrate the antibody to a concentration comprised between about 200 g/L and about 300 g/L; (f) flushing of the cassette with a flushing buffer; (g) obtaining a highly concentrated antibody solution with antibody concentration comprised between about 150 g/L and about 200 g/L.

5. The process of claim 4, wherein the antibody solution loaded onto the third UFDF cassette has an antibody concentration of about 70 g/L and/or the first ultrafiltration concentrates the antibody to a concentration of about 100 g/L and/or the second ultrafiltration concentrates the antibody to a concentration of about 260 mg/mL and/or the obtained highly concentrated antibody solution has an antibody concentration of about 170 g/L.

6. The process of any one of the preceding claims, wherein the third UF/DF is performed using an equilibration buffer comprising histidine-HCl at a concentration of about 5 mM and having pH about 6, a diafiltration buffer comprising histidine-HCl at a concentration of about 25 mM and arginine-HCl at a concentration of about 150 mM and having pH of about 6 and flushing buffer comprising histidine-HCl at a concentration of about 25 mM and arginine-HCl at a concentration of about 150 mM and having pH of about 6.

7. The process of any one of the preceding claims wherein said affinity chromatography is protein A affinity chromatography.

8. The process of any one of the proceeding claims, wherein said two steps of ion exchange chromatography steps comprise a first step of cation exchange chromatography and a second step of anion exchange chromatography.

9. A stable pharmaceutical formulation obtained by adding excipients to said highly concentrated antibody solution obtained by the process of any one of claims 1 to 8.

10. The stable pharmaceutical formulation of claim 9, comprising an a antibody or fragment thereof present within said pharmaceutical formulation at a concentration of about 150 g/L, histidine-HCl buffer present within said pharmaceutical formulation at a concentration of about 25 mM, arginine-HCl present within said pharmaceutical formulation at a concentration of about 150 mM and Polysorbate 80 present within said pharmaceutical formulation at a concentration of about 0.036% (w/v).

11. A process of production of a bulk drug substance or a drug product comprising the steps of: (a) Protein A chromatography of a clarified cell harvest comprising an antibody; (b) Viral inactivation of the resulting protein A eluate; (c) Neutralization of the protein A eluate to pH 5.2, followed by 0.2 μm filtration; (d) Cation exchange chromatography of the neutralized protein A eluate, followed by 0.2 μm filtration; (e) First UF/DF of the cation exchange chromatography eluate, followed by 0.2 μm filtration; (f) Anion exchange chromatography in flow through mode performed by membrane adsorption, followed by 0.2 am filtration; (g) Viral nanofiltration; (h) Second UF/DF of the nanofiltrated solution, followed by 0.2 um filtration; (i) Third UF/DF of the antibody solution obtained by the second UF/DF according to the processes of claims 1 to 6, followed by 0.2 um filtration; (j) Obtaining a stable pharmaceutical formulation by adding excipients to the highly concentrated antibody solution obtained by the third UF/DF, followed by 0.2 um filtration.

12. The process of claim 11, wherein said third UF/DF is performed according to claim 6, and said stable pharmaceutical formulation is the formulation of claim 9 or 10.

Description

[0155] FIG. 1: UFDF 3 process performance (permeate flux/TMP vs. concentration)

EXAMPLES

Example 1: Concentration of an Antibody Solution

Downstream Process (DSP)

[0156] To develop a process for obtaining a highly concentrated antibody solution, first a cell culture harvest obtained from 12 to 14 days culture of CHO cells expressing the antibody Ab1 was clarified and subjected to the following downstream processing steps.

[0157] First the clarified cell harvest was subjected to a protein A chromatography following standard procedure, next the obtained protein A chromatography eluate was subjected to a viral inactivation step performed by holding the solution at pH 3.5±0.1, for 60 minutes. The viral inactivation was then neutralized to pH 5.2, and followed by a depth and a 0.2 μm filtration. Next, cation exchange chromatography (CEX) was carried out in a bind-eluate where the antibody was eluted with a 25 mM Citrate, 100 mM NaCl pH 5.2 elution buffer, and was followed by a 0.2 μm filtration. In order to concentrate antibody in the CEX eluate to 20 g/L a first ultrafiltration/diafiltration (UFDF-1) step was carried out using a tangential flow filtration cassette with a nominal molecular weight limit (NMWL) of 30 kDa, a diafiltration buffer comprising 50 mM Histidine at pH 6.5 was used for this step. UFDF-1 was followed by a 0.2 μm filtration. Next anion exchange chromatography was carried out by membrane absorption (MA) in flow-through mode using 50 mM Histidine pH 6.5 buffer, and was followed first by a 0.2 μm filtration, next by a virus nanofiltration carried out using 50 mM Histidine pH 6.5. A second UFDF (UFDF-2) was then used to concentrate the obtained antibody solution to 70 g/L. A 5 mM L-Histidine buffer at pH 6.0 was used for the UFDF-2 step.

[0158] All the described steps were performed according to the provider recommendation. As we targeted a higher concentration of the antibody upon the purification process, we developed a further concentration step.

[0159] In particular we developed a third UFDF (UFDF-3) wherein starting from an antibody solution with an antibody concentration between about 60 to 80 g/L, it is possible to reach a higher target concentration such as 170 g/L. The developed UFDF is described in the next sections.

UFDF-3 for High Concentration of and Antibody Solution—Conditions Selection

[0160] A UFDF-3 step has been developed to concentrate the product and reach the target concentration of 170 g/L after flushing, starting from an antibody solution with an antibody concentration between about 65 and about 75 g/L. The concentration was performed by Tangential Flow Filtration (TFF).

UFDF Load and Product Storage Conditions

[0161] Freeze/thaw and hold time studies were performed on UFDF-3 load and product as described in Table 1:

TABLE-US-00001 TABLE 1 Hold time study conditions for Abl UFDF-3 product Timepoints Storage T 0 1 week 2 weeks  +5 ± 3° C. x x x +22.5 ± 2.5° C. x x

[0162] Thus UFDF 3 load and product were firstly frozen at −20° C., then thaw at room temperature and to finish re-frozen. UFDF-3 loads and products were analyzed by SE-HPLC, CGE (reduced and non-reduced), iCE and CEX-HPLC to assess the product quality. pH, conductivity, concentration and UFDF 3 loads and products osmolality were also measured.

Results and Discussion

[0163] The selected cassette for the UFDF-3 step is a tangential flow filtration (TFF) cassette with nominal molecular weight limit (NMWL) of 30 KDa, named Cassette 1. The diafiltration was done using 25 mM Histidine, 150 mM Arginine pH 6.0 (DF buffer) during 7 DVs. The first concentration was made up to 100 g/L prior to be diafiltered into the aforementioned pre-formulation buffer. Then, the diafiltered product was concentrated again until the feed pressure reaches 3 bars (˜ 220 g/L). Even if higher pressures (until 5 bars) did not show an impact on product quality, the 3 bars maximum pressure corresponds to the limit of the TFF system (tubing limitation) according to recommendation.

[0164] This choice leads to certain constraints which were taken into account during development. They are described in the Table 2.

TABLE-US-00002 TABLE 2 Equipment characteristics and associated operating parameters to consider during the UFDF-3 development System characteristics taken into account Operating parameters to consider Holder capacity Cassettes surface Feed pump maximum capacity CFR selection Retentate tank maximum capacity Diafiltration concentration start Retentate tank minimum capacity Minimum load volume Hold-up volume Flush and minimum target pre-flushed concentration Pressure Process pressures (especially feed pressure)

Cross Flow Rate (CFR)

[0165] CFR is calculated using feed and permeate flow rates (CFR=Feed flow rate−Permeate flow rate). For this reason, a pump calibration was performed prior to this run (not shown) in order to ensure an accurate measurement of CFR. CFRs (from 100 LMH to 360 LMH), FIG. 1 shows the UFDF-3 process performances (permeate flux and TMP evolution throughout the UFDF-3 step) obtained with different CFRs (from 100 LMH to 360 LMH). As demonstrated, increasing the CFR tends to improve the permeate flux. Indeed the starting permeate flux obtained with 100 LMH CFR is approximately 13 LMH compared to that obtained with 360 LMH CFR which is approximately 25 LMH. Nevertheless, no significant permeate flux improvement from 240 LMH to 360 LMH CFRs is observed (˜ 20 LMH and 25 LMH for respectively 240 LMH and 360 LMH CFRs).

[0166] Regarding the pressure, a pressure drop is observed at the end of concentration. For runs performed with a CFR comprised between 240 and 360 LMH CFRs, this phenomenon occurs around 210 g/L whereas for lower CFR (.sup.˜100 LMH), it appears later at about 250 g/L. However, even if there is an earlier pressure drop for CFR>100 LMH, working with higher CFR (>240 LMH) allows twice higher permeate flux, at the beginning and during diafiltration, thus allowing much lower process duration (gain of more than 2 hours of process time). Moreover, the reached pre-flushed concentration of 210 g/L is sufficient for a flush of 3 HUV. Consequently, a CFR range of ≥240 LMH-≤360 LMH is the best compromise to reach highest flux without having too much pressure.

Maximum Feed Pump

[0167] The flow rate selection must take into account the maximum feed pump speed capacity of the selected TFF skid. Indeed, as the CFR is expressed in L/h per square meter, it is dependent on the cassette surface. Consequently, depending of the cassette surface used for a run, there is a risk to be above the maximum limit of the feed pump speed (i.e. 1000 L/h). The maximum feed pump (L/h) is the multiplication of the maximum feed flow rate (LMH) and the maximum surface area (m.sup.2):

[00001]$\mathrm{Maximum}\mathrm{feed}\mathrm{pump}\left(L\text{/}h\right)=\mathrm{Maximum}\mathrm{feed}\mathrm{flow}\mathrm{rate}\left(\mathrm{LMH}\right)*\mathrm{Maximum}\mathrm{surface}\left(m^2\right)$$\mathrm{And}\mathrm{consequently}\text{:}$$\mathrm{Maximum}\mathrm{feed}\mathrm{flow}\mathrm{rate}\left(\mathrm{LMH}\right)=\frac{\mathrm{Maximum}\mathrm{feed}\mathrm{pump}\left(L\text{/}h\right)}{\mathrm{Maximum}\mathrm{surface}\left(m^2\right)}$$\mathrm{Considering}a\mathrm{maximum}\mathrm{holder}\mathrm{capacity}\mathrm{of}2.85m^2,\mathrm{the}\mathrm{maximum}\mathrm{feed}\mathrm{flow}\mathrm{rate}\mathrm{calculation}\mathrm{is}\text{:}$$\mathrm{Maximum}\mathrm{feed}\mathrm{flow}\mathrm{rate}\left(\mathrm{LMH}\right)=\frac{\mathrm{Maximum}\mathrm{feed}\mathrm{pump}\left(L\text{/}h\right)}{\mathrm{Maximum}\mathrm{surface}\left(m^2\right)}=\frac{1000L\text{/}h}{2.85m^2}=351\mathrm{LMH}$

[0168] Knowing that, at this feed flow rate (FFR), the permeate flow rate (PFR) is about 25 LMH, it means that the CFR would be:

CFR (LMH)=FFR (LMH)−PFR (LMH)=351 LMH−25 LMH=326 LMH

[0169] Consequently, a maximum CFR limit of 326 LMH should be established.

[0170] Moreover, to avoid working at the maximum of the pump which is not desirable, a set point at 290 LMH was defined. Based on small scale runs data, the permeate flow rate obtained at the beginning of the concentration with a CFR of 290 LMH was 25 LMH. Considering a maximum total surface area of 2.85 m.sup.2, the associated feed flow rate for the defined CFR set point would be:

FFR (LMH)=CFR (LMH)+PFR (LMH)=290 LMH+25 LMH=315 LMH

[0171] Which corresponds to a feed pump speed of:

[00002]$\mathrm{Feed}\mathrm{pump}\mathrm{speed}\left(L\text{/}h\right)=\mathrm{FF}R\left(LMH\right)\times \max \mathrm{cassettes}\mathrm{surface}\left(m^2\right)=315\mathrm{LMH}\times 2.85m^2=898L\text{/}h$

[0172] As the cassettes surface of 2.85 m.sup.2 is the maximum stood for the manufacturing TFF system, the feed pump would speed is still suitable whatever the surface area used. Furthermore the maximum pump speed will never be reached. Thus the recommendation for the CFR is ≥240 LMH−≤325 LMH with a set point at 290 LMH.

Volumetric Loading Factor (VLF)

[0173] The tested VLFs are described in the Table 3:

TABLE-US-00003 TABLE 3 Ab1 UFDF-3 VLF evaluation Parameters Sample A Sample B Sample C Sample D Sample E CFR [LMH] 360 240 290 290 290 Volumetric loading factor [L/m.sup.2] 15 20 25 30 25 UFDF-3 process duration [hours] 5.2 9.4 9.8 11.8 9.3

[0174] Table 3 shows that higher the VLF is, longer is the process duration: 5 hours at 15 L/m.sup.2, 9 hours at 20 L/m.sup.2, and 11.8 hours at 30 L/m.sup.2. According to SBL mass balance for a 5000 L batch, the highest expected UFDF-3 load volume would be 75 L. If the maximum VLF is set at 25 L/m.sup.2, a cassette surface of 3 m.sup.2 would be required which is not suitable for the maximum capacity of the holder (2×1.14 m2+1×0.57 m.sup.2). Therefore, the VLF of 30 L/m.sup.2 was chosen as the upper limit to fit with potential highest volumes. Moreover, no product quality impact was noticed whatever the VLF. Thus the recommendation for the VLF is 30 L/m.sup.2 with a set point at 25 L/m.sup.2.

Diafiltration

[0175] The scope of this experiment was to define the most appropriate product concentration to start the diafiltration. The choice was made on a compromise between sufficient volume reduction (to fit with the retentate tank volume capacity) while keeping the lowest DF duration (i.e. high permeate flux).

[0176] Previous experiments allowed to select a start of the DF for a product concentration of 100 g/L. Targeting this concentration allowed to put the product into the diafiltration buffer (25 mM L-Histidine, 150 mM L-Arginine-HCl pH 6.0) before reaching too high pressure caused by the increasing viscosity. Indeed, the diafiltration buffer allows a reduction of viscosity which is high when the product is in the equilibration buffer (5 mM L-Histidine pH 6.0).

[0177] According to SBL mass balance, for a 5000 L batch, the UFDF-3 load volume and concentration upper limits are 75 L at 75 g/L. Targeting 100 g/L for the diafiltration would lead to a DF product volume of 56.3 L which is higher than the maximum volume recommended by the supplier (55 L for the selected retentate tank of 50 L). Consequently, target DF concentration was optimized to mitigate this volume constraint. The results are presented in Table 4.

TABLE-US-00004 TABLE 4 Ab1 UFDF-3 DF concentration evaluation. Parameters Sample C Sample E CFR [LMH] 290 290 Volumetric loading factor [L/m.sup.2] 25 25 Diafiltration concentration [g/L] 110 100 UFDF-3 process duration [hours] 9.8 9.3

[0178] The only difference observed is the process duration which is slightly longer for a DF performed at 110 g/L compared to one performed at 100 g/L (9.8 hours than 9.3 hours respectively). However, this difference can be considered as negligible and no impact on product quality was observed. Thus, for high starting volumes, the recommendation is to start the diafiltration at a product concentration of 110 g/L. If this is not the case, targeting 100 g/L is reducing the whole process duration. The recommendation for the DF concentration is 110 g/L with a set point at 100 g/L. The actual value to target should be calculated while taking into account the initial volume and the retentate tank capacity.

Worst Case Scenario 1: Initial Volume Upper Limit

[0179] Following the recommendation mentioned in the previous section, calculations were made to indicate which concentration to choose for different initial volumes (Table 5).

TABLE-US-00005 TABLE 5 DF concentration recommendation for different UFDF-3 load volumes Initial volumes 70 L 75 L 80 L 85.5 L Cassette surface 2 × 1.14 m.sup.2 + 1 × 0.57 m.sup.2 Initial concentration 75 g/L VLF 25 L/m.sup.2 26 L/m.sup.2 28 L/m.sup.2 30 L/m.sup.2 DF volumes @ 100 g/L 51.3 L 55.1 L 58.8 L 62.9 L @ 110 g/L 46.5 L 49.9 L 53.3 L 57.1 L

[0180] According to the scale up calculation provided above and in order to fit with the maximum retentate tank capacity of 55 L, it appears that the diafiltration step must be performed at 100 g/L up to 70 L initial volume. Above 70 L, the diafiltration has to be performed at 110 g/L.

Worst Case Scenario 2: Initial Volume Lower Limit

[0181] In the case of lowest initial volume, the manufacturing constraint is the minimum working volume of the system. Indeed, the risk is to have a product pre-flushed volume below the minimum working volume which may lead to increase in foam and shear stress.

[0182] Therefore, an initial volume lower limit must be established to avoid having a pre-flushed volume below the minimum recommended recirculation volume. Table 6 shows estimated pre-flushed volumes for different potential initial volumes and assuming that the pre-flushed concentration is 220 g/L.

TABLE-US-00006 TABLE 6 UFDF-3 volumes estimation related to low initial volumes and concentration Initial volume 40 L 45 L 50 L 60 L Cassette surface 1 × 1.14 m2 + 1 × 0.57 m2 2 × 1.14 m2 Initial concentration  65 g/L VLF      23 L/m2 29 L/m2 22 L/m2 26 L/m2 Pre-flushed concentration 220 g/L Pre-flushed volume  10.9 L 12.4 L  .sup.   13.7 L  .sup.   16.7 L  .sup.  

[0183] According to the scale up calculation provided above, it appears that processing low initial volumes can be risky in term of the minimum retentate tank capacity and pre-flushed volumes achieved. This point need to be considered and hence determine before running step at scale. The recommendation here would be not to processed less than 40 L. Initial volumes able to be processed during this UFDF-3 step as well as the DF concentration are: [0184] ≥40 L-<70 L with a diafiltration performed at 100 g/L (maximum tank capacity) [0185] ≥70 L-≤80 L with a diafiltration performed at 110 g/L (minimum tank capacity).

Product Quality

[0186] No impact on product quality has been observed during UFDF-3 development.

[0187] Table 7 shows two examples of product quality results obtained at the selected optimized conditions:

TABLE-US-00007 TABLE 7 UFDF-3 analytical results Run Sample C Sample D Sample F Scale Small Pilot LOAD PH 6.6 6.6 6.5 Conductivity [mS/cm] 0.5 0.5 0.5 Concentration [g/L] 71.3 67.3 SE-HPLC Monomer [%] 98.4 98.4 97.7 Non-reduced IgG [%] 92.1 93.4 92.5 CGE Assigned peaks [%] 6.6 6.5 7.0 Reduced CGE LC [%] 34.2 34.2 34.2 HC [%] 65.2 65.2 65.0 Other peaks [%] 0.7 0.6 0.9 cIEF Acidic [%] 29.6 30.3 27.0 Main [%] 47.8 47.1 48.6 Basic [%] 22.6 22.6 24.4 CEX-HPLC Acidic [%] 17.7 17.6 19.7 Main [%] 48.3 48.1 46.1 Basic [%] 34.0 34.3 34.2 Density [g/cm.sup.3] 1.0185 1.0186 1.0183 Viscosity [mPa/s] 3.405 2.292 2.708 Osmolality [mosm/kg] 10 12 11 Quantity L/m.sup.2 membrane 25 30 20 Quantity mAb/m.sup.2 membrane 1783 2139 1356 Tested parameters CFR [LMH] 290 290 290 VLF [L/m.sup.2] 25 30 20 DF concentration [g/L] 110 100 100 PRODUCT PH 6.0 6.1 6.1 Conductivity [mS/cm] 8.2 8.0 7.8 Concentration [g/L] 172.5 174.5 177.9 Flush [HUV] 3.5 2.8 3.2 SE-HPLC Monomer [%] 98.5 98.6 98.3 Non-reduced IgG [%] 92.3 92.8 91.9 CGE Assigned peaks [%] 6.5 6.6 7.6 Reduced CGE LC [%] 34.2 34.1 34.0 HC [%] 65.2 65.3 64.8 Other peaks [%] 0.6 0.7 1.2 cIEF Acidic [%] 29.4 29.4 26.9 Main [%] 47.8 48.0 49.1 Basic [%] 22.8 22.5 24.1 CEX-HPLC Acidic [%] 18.0 17.8 19.4 Main [%] 47.9 48.1 46.6 Basic [%] 34.2 34.1 34.0 Density [g/cm.sup.3] 1.0583 1.0623 1.0626 Viscosity [mPa/s] 10.61 13.45 15.43 Osmolality [mosm/kg] 317 339 325 Yield [%] 98 93 97

[0188] High viscosity observed for Sample C UFDF-3 load is unexpected. However, high deviation level (23%) occurred during the analysis. The yield obtained for Sample D (93%) is slightly lower than expected but it is easily explained by the lack of flush (2.8 HUV), lower than the specification set (≥3 HUV). Indeed the flush is an important point to consider, especially for such high concentrations, in order to recover as much product as possible. No significant differences were observed with respect to SE-HPLC (loads and products ≥98%) and CGE (loads and products ≥92% with less than 1% differences in all conditions). Regarding charged variant profiles by clEF, differences of about 1.5% were observed between small and pilot runs. No significant differences between respective load and product were detected, indicating that UFDF-3 does not charged variant profiles.

Stability Studies

[0189] During the first assessment hold time and freeze and thaw studies have been performed using UFDF-3 load and product from Sample G. For each sample (load and product), all tested conditions were analyzed within the same sequence in order to minimize the variability. Table 8 and Table 9 represent hold time study results of respectively UFDF-3 load and product at different time points, namely 1 week (w) and 2 weeks, and storage temperatures.

TABLE-US-00008 TABLE 8 UFDF-3 load hold time study UFDF-3 LOAD Sample G +22.5 ± 2.5° C. +5 ± 3° C. Conditions T 0 1 w 2 w 1 w 2 w SE-HPLC Aggregate [%] 2.3 2.2 2.1 2.2 2.0 Monomer + Tailing 97.6 97.7 97.8 97.7 97.9 [%] Fragment [%] 0.1 0.1 0.1 0.1 0.1 Non-reduced LC [%] 1.8 1.7 1.6 1.7 1.7 CGE HC [%] 0.1 0.1 0.1 0.1 0.2 75 kD [%] 0.0 0.1 0.1 0.0 0.0 100 kD [%] 0.6 0.5 0.7 0.6 0.7 125 kD [%] 4.7 4.8 4.9 4.6 5.0 IgG′ [%] 2.4 2.7 3.3 2.8 2.7 IgG [%] 90.3 90.1 89.3 90.2 89.9 Total IgG [%] 92.7 92.8 92.6 93.0 92.6 Reduced CGE LC [%] 38.2 38.2 37.8 38.3 38.7 HC aglycosyl [%] 0.5 0.5 0.6 0.0 0.0 HC [%] 61.3 61.3 61.6 61.7 61.3 cIEF Acidic [%] 29.5 29.9 30.5 29.4 29.4 Main [%] 49.7 49.0 48.8 49.2 49.2 Basic [%] 20.8 21.1 20.7 21.4 21.4 CEX-HPLC Acidic [%] 14.7 15.0 15.3 14.8 14.8 Main [%] 56.5 56.2 56.4 56.6 56.4 Basic [%] 28.8 28.8 28.3 28.7 28.8

TABLE-US-00009 TABLE 9 UFDF-3 product hold time study UFDF-3 PRODUCT Sample G +22.5 ± 2.5° C. +5 ± 3° C. Conditions T 0 1 w 2 w 1 w 2 w SE-HPLC Aggregate [%] 1.6 1.5 1.5 1.5 1.5 Monomer + 98.4 98.5 98.4 98.4 98.4 Tailing [%] Fragment [%] 0.1 0.1 0.1 0.1 0.1 Non-reduced LC [%] 1.8 1.8 1.7 1.8 1.8 CGE HC [%] 0.1 0.1 0.1 0.0 0.1 75 kD [%] 0.0 0.1 0.0 0.1 0.1 100 kD [%] 0.7 0.7 0.7 0.8 0.7 125 kD [%] 4.9 5.3 5.1 5.4 5.4 IgG′ [%] 2.9 2.9 3.1 3.4 3.0 IgG [%] 89.6 89.3 89.2 88.6 89.0 Total IgG [%] 92.5 92.2 92.3 92.0 92.0 Reduced CGE LC [%] 39.0 38.6 38.8 37.4 37.4 HC aglycosyl [%] 0.0 0.4 0.3 0.4 0.7 HC [%] 61.0 61.0 60.9 62.1 61.8 cIEF Acidic [%] 29.5 28.8 29.1 29.0 29.3 Main [%] 49.3 49.7 49.7 49.7 49.5 Basic [%] 21.2 21.6 21.2 21.1 21.2 CEX-HPLC Acidic [%] 14.4 14.3 14.5 14.5 14.5 Main [%] 56.8 56.5 56.4 56.4 56.5 Basic [%] 28.8 29.2 29.1 29.1 29.1

[0190] Whatever the sample (load or product), no significant differences were detected regarding charge variant profiles neither by clEF nor by CEX-HPLC. SE-HPLC are all within method variability (≤0.3%) as well, observed variations are thus not significant. Slight variations observed with respect to CGE (reduced and non-reduced) are inherent to the method and thus results are considered as comparable. To conclude, UFDF-3 load and product are both stable up to 2 weeks at room temperature (+22.5±2.5° C.) as well as at +5±3° C.

Freeze and Thaw Study

[0191] As said before, all tested conditions were analysed within the same sequence in order to minimize the variability. Table 10 represents freeze and thaw study results of the UFDF-3 product.

TABLE-US-00010 TABLE 10 UFDF-3 product freeze and thaw study UFDF-3 PRODUCT Sample G −20° C. Conditions 1 F/T 2 F/T 3 F/T SE-HPLC Aggregate [%] 1.6 1.6 1.6 Monomer + Tailing [%] 98.4 98.3 98.3 Fragment [%] 0.1 0.1 0.1 Non-reduced CGE LC [%] 1.5 1.8 1.7 HC [%] 0.1 0.1 0.1 75 kD [%] 0.0 0.0 0.0 100 kD [%] 0.6 0.5 1.0 125 kD [%] 5.2 5.1 5.0 IgG' [%] 2.5 3.2 3.0 IgG [%] 90.0 89.3 89.3 Total IgG [%] 92.5 92.5 92.3 Reduced CGE LC [%] 35.8 35.6 36.2 HC aglycosyl [%] 0.7 0.6 0.7 HC [%] 62.5 62.7 62.1 cIEF Acidic [%] 29.3 29.7 29.4 Main [%] 49.7 49.1 49.5 Basic [%] 21.1 21.1 21.1 CEX-HPLC Acidic [%] 14.3 14.4 14.1 Main [%] 56.9 56.8 56.9 Basic [%] 28.8 28.8 28.9

[0192] There are no significant differences regarding charge variant profiles neither by clEF nor by CEX-HPLC. The product purity by SE-HPLC is highly similar whatever the F/T number. Fragments content with respect to CGE (reduced and non-reduced) is also comparable. To conclude, the freeze and thaw at −20° C. (up to 3 times) of the UFDF-3 product sample has no significant impact.

Case Study

Case Study 1

[0193] Inputs: Volume to process=80 L and Concentration=75 g/L

[0194] In this particular case, main concerns are the retentate tank capacity, the cassettes surface and hence the related pump speed. The first step is to define the surface required for this UFDF-3 step by targeting 25 L/m.sup.2 (VLF set point):

[00003]$\frac{\mathrm{Volume}\mathrm{to}\mathrm{process}}{\mathrm{Target}\mathrm{VLF}}=\frac{80L}{25L\text{/}{m}^2}=3.2m^2$

[0195] The next calculation will be to assess if by using this surface, the VLF will still fit with the specifications:

[00004]$\frac{\mathrm{Volume}\mathrm{to}\mathrm{process}}{\mathrm{Cassettes}\mathrm{surface}}=\frac{80L}{2.85m^2}=28L\text{/}{m}^2$

[0196] This VLF remains within specifications (≤30 L/m2) even it is higher than the set point of 25 L/m.sup.2. The second step is to ensure if the system pump speed is suitable with this cassettes surface taking into account the maximum feed flow rate of 315 LMH (i.e. 290 LMH CFR). Even if the feed flow rate to be applied (898 L/h) is closed to the maximum pump speed limit (≤1000 L/h), it remains suitable for this UFDF-3 step. The third step is to estimate the volume at the start of the diafiltration. Indeed, this volume should be lower than 50 L to fit with the maximum retentate tank capacity (≤55 L). Knowing that the diafiltration concentration set point is 100 g/L, the DF volume is:

[00005]$\frac{C_{initiale}\times V_{initial}}{C_{DF}}=\frac{75g\text{/}L\times 80L}{100g\text{/}L}=60L$

[0197] As explained earlier, if this volume does not fit with the tank capacity (higher than 55 L), the target DF concentration should be 110 g/L instead of 100 g/L set point:

[00006]$\frac{C_{initiale}\times V_{initial}}{C_{DF}}=\frac{75g\text{/}L\times 80L}{110g\text{/}L}=54.5L$

[0198] Moreover, this calculated volume corresponds to the total retentate volume without including the system hold-up volume (HUV) which includes the flow path void volume (FVV) and the cassette/device void volume (DVV)

[0199] The DF volume actually contained into the retentate tank is thus:

[00007]$V_{DF}-\mathrm{System}\mathrm{HUV}=V_{DF}-\left(\mathrm{FVV}+\left(2\times {\mathrm{DVV}}_{1.14m^2}+1\times {\mathrm{DVV}}_{0.57m^2}\right)\right)=54.5L-\left(0.650L+\left(2\times 0.227L+1\times 0.118L\right)\right)=54.5L-1.222L=53.3L$

[0200] This volume fit with the retentate tank maximum capacity and can be processed (55 L).

[0201] The final step is to estimate the preflushed volume into the tank assuming a related concentration of 220 g/L to ensure that it is higher than the minimum retentate tank capacity:

[00008]$\frac{C_{initiale}\times V_{initial}}{C_{\mathrm{preflushed}}}=\frac{75g\text{/}L\times 80L}{220g\text{/}L}=27.3L$

[0202] There is no issue regarding the minimum retentate tank capacity. It is not a concern with such volumes.

Case Study 2

[0203] Inputs: Volume to process=45 L, Concentration=65 g/L

[0204] In this case study, there is only one main concern which is the retentate tank capacity and more precisely its minimum recirculation volume. The same exercise as the case study 1 above is performed to assess the suitability of these inputs by considering same parameters.

[0205] The first step is to define the surface required for this UFDF-3 step by targeting 25 L/m2 (VLF set point):

[00009]$\frac{\mathrm{Volume}\mathrm{to}\mathrm{process}}{\mathrm{Target}\mathrm{VLF}}=\frac{45L}{25L\text{/}{m}^2}=1.8m^2$

[0206] Based on two cassettes sizes suitable with the manufacturing scale (0.57 m2 and 1.14 m2) and in order to achieve this required surface, 2×1.14 m2 is used (total surface of 2.28 m2). The next calculation will be to assess if by using this surface, the VLF will still fit with the specifications:

[00010]$\frac{\mathrm{Volume}\mathrm{to}\mathrm{process}}{\mathrm{Cassettes}\mathrm{surface}}=\frac{45L}{2.28L\text{/}{m}^2}=20m^2$

[0207] This VLF remains within specifications (≤30 L/m2) even it is lower than the set point of 25 L/m2. In this case, the feed flow rate to applied targeting a set point of 315 LMH (i.e. 290 LMH CFR) is not on the critical path. Indeed the maximum pump capacity (≤1000 L/h) is not reached using this surface (2.28 m2). It is nevertheless calculated for information only:

Feed flow rate×Cassettes surface=315 LMH×2.28 m.sup.2=718 L/h

[0208] Within the same way, the diafiltration step can be performed at 100 g/L without any issue. Anyway, the volume content into the retentate tank at this stage does not reach the minimum tank capacity. However, the critical point to consider here is the pre-flushed volume remaining into the retentate tank at the end of concentration. The final step is to estimate this volume assuming a pre-flushed concentration of 220 g/L. The goal is to ensure that it is higher than the minimum retentate tank capacity:

[00011]$\frac{C_{initiale}\times V_{initial}}{C_{\mathrm{preflushed}}}=\frac{65g\text{/}L\times 45L}{220g\text{/}L}=13.3L$

[0209] This calculated volume corresponds to the total retentate volume without including the HUV thus the pre-flushed volume actually contained into the retentate tank is:

[00012]$V_{\mathrm{pre}-\mathrm{flushed}}-\mathrm{System}\mathrm{HUV}=V_{\mathrm{pre}-\mathrm{flushed}}-\left(\mathrm{FVV}+\left(2\times {\mathrm{DVV}}_{1.14m^2}\right)\right)=13.3L-\left(0.650L+\left(2\times 0.227L\right)\right)=13.3L-1.104L=12.2L$

[0210] Although it is really closed to the minimum, this pre-flushed volume fit with the retentate tank minimum capacity and can be processed.

[0211] To conclude, the Table 11 is a summary of process parameters to be applied for performing this UFDF-3 case study:

TABLE-US-00011 TABLE 11 summary of UFDF-3 case studies Parameters Operating ranges Case study 1 Case Study 2 Inputs Volume ≤80 L 80 L 40 L Concentration ≥65 g/L-≤75 g/L 75 g/L 65 g/L Process Cassette Surface ≤2 × 1.14 m2 + 2.85 m.sup.2 1.71 m.sup.2 parameter 1 × 0.57 m.sup.2 (2 × 1.14 m.sup.2 + (1 × 1.14 m.sup.2 + to applied 1 × 0.57 m.sup.2 1 × 0.57 m.sup.2) VLF 25 L/m2 set point 28 L/m.sup.2 23 L/m.sup.2 ≤30 L/m2 CFR 290 LMH set point 290 LMH (≥240 LMH-≤360 LMH) Feed flow rate 315 LMH set point 315 LMH (≥260 LMH-≤390 LMH) Pump speed ≤1000 L/h 898 L/h 539 L/h DF concentration 100 g/L set point 110 g/L 100 L/h (≤110 g/L)

UFDF-3 for High Concentration of and Antibody Solution—5000 L Processing

Materials, Equipment and Methods

[0212] Buffers were prepared at room temperature (≥+19 and ≤+25° C.) and pH was adjusted to the target at room temperature, the list of used buffer for the concentration of Ab1 with the developed UFDF-3 starting from the an antibody solution with an Ab concentration between about 65 and about 75 g/L are listed in Table 12.

TABLE-US-00012 TABLE 12 List of buffers used for the UFDF-3 and formulation process Buffer description Step 5 mM L-Histidine pH 6.0 Equilibration 25 mM L-Histidine, 150 mM Diafiltration, L-Arginine-HCl, pH 6.0 flush 25 mM L-histidine, 150 mM Excipient addition L-arginine-HCl, 2.5% v/v Polysorbate 80, pH 6.0

TABLE-US-00013 TABLE 13 Characteristics of equilibration buffer Parameter Set Point Acceptable Range Unit pH (20-25° C.) 6.0 ≥5.9-≤6.1 NA Conductivity (24-26° C.) 0.30 ≥0.15-≤0.45 mS/cm.sup.  Bioburden — ≤3 (Alert Limit) CFU/10 mL ≤50 (Action Limit) Endotoxin — ≤1 EU/mL .sup. 

TABLE-US-00014 TABLE 14 Characteristics of diafiltration buffer Parameter Set Point Acceptable Range Unit pH (20-25° C.) 6.0 ≥5.9-≤6.1 NA Conductivity (24-26° C.) 12.2 ≥12.0-≤12.4 mS/cm.sup.  Osmolality.sup.1 293 NA mOsm/kg .sup.    Bioburden — ≤3 (Alert Limit)  .sup. CFU/10 mL ≤50 (Action Limit) Endotoxin — ≤1 EU/mL

TABLE-US-00015 TABLE 15 Characteristics of formulation buffer Parameter Set Point Acceptable Range Unit pH (20-25° C.) 6.0 ≥5.9-≤6.1 NA Conductivity (24-26° C.) 11.5 ≥11.1-≤11.9 mS/cm.sup.  Osmolality 313 NA mOsm/kg .sup.    Bioburden — ≤3 (Alert Limit)  .sup. CFU/10 mL ≤50 (Action Limit) Endotoxin — ≤1 EU/mL

[0213] The prepared buffers were tested for bioburden and then filtered through 0.22 μm filter and tested for Endotoxins post filtration.

[0214] The TFF cassettes with nominal molecular weight limit (NMWL) of 30 KDa were used for the UFDF-3 at 5000 L scale antibody production (see paragraph “Downstream process (DSP)” for steps details). The surface required can be recalculated according to the material quantity at the beginning of the step. Single-use or reusable cassettes can be used.

[0215] In our case the starting material for the developed UFDF-3 step was the antibody solution obtained by a after the UFDF2 of the Downstream process (DSP) described before, which had an antibody concentration between about 65 and about 75 g/L.

UFDF-3 Step Description

[0216] The objective of this step was to reach a high concentration of the antibody Ab1 (170 g/L) and to buffer exchange the product into the diafiltration buffer made of 25 mM L-Histidine, 150 mM L-Arginine-HCl, pH6.0) required before excipient addition.

[0217] The cassettes need to be installed in the holder following the supplier recommendations (torque value). The number of cassettes to be used depends on the volume of material to be used. This needs to be calculated using the mass balance equations:

[00013]$•\mathrm{Total}\mathrm{Product}\mathrm{from}\mathrm{UFDF}2\left(g\right)=\mathrm{UFDF}2\mathrm{Pool}A280\mathrm{Concentration}\left(g\text{/}L\right)\times \mathrm{UFDF}\text{-}2\mathrm{Pool}\mathrm{Volume}\left(L\right)$$•\mathrm{Total}\mathrm{Membrane}\mathrm{surface}\mathrm{required}\left(m^2\right)=\frac{\mathrm{Total}\mathrm{Product}\mathrm{from}\mathrm{VFUFDF}\text{-}2\left(g\right)}{\mathrm{Maximum}\mathrm{Loading}\mathrm{factor}\left(g\text{/}{m}^2\right)}$$•\mathrm{Number}\mathrm{of}0.57m^2\mathrm{cassettes}\mathrm{required}\left(\mathrm{rounded}\mathrm{up}\mathrm{value}\right)=\frac{\mathrm{Total}\mathrm{Membrane}\mathrm{surface}\mathrm{required}\left(m^2\right)}{\mathrm{Membrane}\mathrm{surface}\left(m^2\right)=0.57}$

[0218] The membrane surface available are 0.57 m.sup.2 and 1.14 m.sup.2. Regarding the 5000 L scale production, the recovery and expected volume UFDF-3 load to be processed can be from 40 L to 80 L. Prior to any UFDF-3 run, a sanitization needs to be performed if the system flow path and/or cassettes are not single-use or non-gamma irradiated. After the sanitization, the cassettes was rinsed using water for injection (WFI) and then equilibrated using the 5 mM L-Histidine pH 6.0 buffer (≥10 L/m.sup.2) until buffer pH and conductivity of retentate and permeate lines are within specifications of the equilibration buffer (Table 13). After the loading of Ab1 in the retentate container, the first ultrafiltration (UF1) operation can directly start. The product needs to be concentrated from about 70 g/L to approximatively 100 g/L.

[0219] The step of first concentration is followed by the diafiltration (DF), the product needs to be buffer exchanged to the diafiltration buffer 25 mM L-Histidine, 150 mM L-Arginine-HCl pH 6.0 using at minimum 6 DVs. The pH and conductivity of permeate need to be checked at this point and need to be in the DF buffer specifications (Table 14) before to proceed to next step. After the ultrafiltration 2 (UF2) operation is performed, the product must reach the exact concentration calculated before to start UFDF-3 based on the quantity of product in UFDF-3 load and the volume required for the flush. The pre-flushed product needs to be sufficiently concentrated to allow an effective flush of the lines and the cassettes and hence to reach the concentration of 170 g/L at the end.

[0220] As described in paragraph “UFDF-3 for high concentration of and antibody solution—condition selection”, a concentration of 260 g/L was reached with a feed pressure <3 bars without any impact on product quality. Due to the high viscosity of the product, a gradient with 2 levels of viscosity was observed which induce an important feed pressure increase. During UF2 of the UFDF-3 stage the pressure and viscosity increased with the increase of the product concentration. Therefore the feed pump flow rate was decreased to maintain the pressure approximately at 0.8 bars. When the minimum pump flow is reached, the retentate valve is opened to maintain the retentate pressure >0.0 bar and allow the TMP to increase to approximately 1.5 bars and a feed pressure up to 3 bars, to be able to concentrate the product to the target. At the end of ultrafiltration 2, only the feed pressure will be controlled to avoid to exceed the maximum.

[0221] Due to the high concentration of the product before flush (approximatively 220 g/L), the 0.2 μm filtration was performed just after the pool (product after UF2 and flush), in order to maintain a low bioburden level on UFDF-3 product (at 170 g/L), followed immediately by the excipient addition and concentration adjustment.

[0222] The flushing volume for this stage was 3 HUV. The flush volume must be re-calculated after ultrafiltration 2 based on the final concentration reached, to be optimal to reach the defined UFDF-3 concentration target after flush while maintaining an acceptable product recovery (Step yield specification 85%).

CONCLUSION

[0223] In summary the developed UFDF3 steps allowed to concentrate an antibody solution with a starting antibody concentration of about 70 g/L up to about 170 g/L. The steps of the developed UFDF3 are summarized below:

Sanitation:

[0224] Pre-Use water for injection (WFI) flush [0225] Sanitation with 0.5 NaOH-30 min recirculation [0226] Pre-use WFI rinsing 0.1 mS/cm [0227] Feed pressure: ≤3 bars [0228] TMP target 0.8 bars (≥0.6 bars-≤1. bars)

Equilibration:

[0229] Equilibration buffer: 5 mM L-Histidine pH 6.0 [0230] Feed pressure: ≤3 bars [0231] TMP target 0.8 bars (≥0.6 bars-≤1. bars)

Loading:

[0232] Volumetric loading factor: target 25 L/m2 (≤30 L/m2) [0233] Concentration: ≥65 g/L-≤75 g/L

First Ultrafiltration:

[0234] Cross flow rate: target 290 LMH (≥100 LMH-≤325 LMH) [0235] Feed flow rate: target 315 LMH (≥110 LMH-≤350 LMH) [0236] Feed pressure: ≤3 bars [0237] TMP: target 0.8 bars (≥0.6 bars-≤1.0 bars) [0238] Concentration: target 100 g/L (≥90 g/L-≤110 g/L)

Diafiltration:

[0239] Cross flow rate: target 290 LMH (≥100 LMH-≤325 LMH) [0240] Feed flow rate: target 315 LMH (≥110 LMH-≤350 LMH) [0241] Feed pressure: ≤3 bars [0242] TMP: target 0.8 bars (≥0.6 bars-≤1.0 bars) [0243] Diafiltration buffer: 25 mM L-Histidine, 150 mM L-Arginine-HCl pH 6.0, performed at ≥6 DVs
Second ultrafiltration: [0244] Cross flow rate: target 290 LMH (≥7 LMH≤325 LMH) [0245] Feed flow rate: target 315 LMH (≥7 LMH-≤350 LMH) [0246] Feed pressure: 3 bars [0247] TMP: ≤1.5 bars [0248] Concentration: ≤260 g/L

Flushing:

[0249] Flushing buffer: 25 mM L-Histidine, 150 mM L-Arginine-HCl pH 6.0-≥3 HUV [0250] Concentration reached: ≥162 g/L-≤179 g/L

[0251] The UFDF-3 product at 170 g/L was then formulated with Polysorbate 80 and diluted to 150 g/L. In particular the final antibody formulation contained 150 g/L of Ab1, 25 mM of L-histidine-HCl, 150 mM of L-arginine-HCl and Polysorbate 80 present at a concentration of about 0.036% (w/v).