METHOD FOR TRANSFERRING A BATCH PRODUCTION PROCESS TO A CONTINUOUS PRODUCTION PROCESS

Abstract

Described herein is a method for transferring of batch production process for a monoclonal antibody to a continuous production process for the same monoclonal antibody.

Claims

1. A method for transferring of batch production process for a monoclonal antibody to a continuous production process for the same monoclonal antibody comprising the steps a) providing a particle-free fluid (product stream) from a heterogeneous cell culture-fluid mixture containing the monoclonal antibody, in the form of a product stream, b) at least one continuous Protein A chromatography, characterized in that the aseptic processing is ensured in the continuous mode via sanitization of the Protein A resin with a caustic substance, c) at least one anion exchange chromatography (AEX) in flow through mode, characterized in that the flow of the product stream in the batch production process is 1-20 membrane volumes per minute and the flow of the product stream in the continuous production process for the monoclonal antibody is 0.1-0.99 membrane volumes per minute OR d) at least one anion exchange chromatography characterized in that the batch production process for the monoclonal antibody comprises a membrane absorber for AEX and that in the continuous production process for the same monoclonal antibody said AEX is carried out in a pulsatile manner.

2. The method according to claim 1, wherein at least one filtration providing a filtrate is carried out during the production process.

3. The method according to claim 1, wherein the at least one continuous Protein A chromatography is further characterized in that the flow in continuous mode is 0-8, preferably 0.2-3.8 times less than in batch mode and/or wherein the cycles per column in continuous mode is 10-20 times higher than in batch mode.

4. The method according to claim 1, further comprising at least one cation exchange (CEX) chromatography step in parallel batch mode, wherein the number of chromatography columns used in the CEX step to carry out the parallel batch mode is chosen in such a manner that the time required for regeneration and elution of a given number of columns is smaller than the load time of a given column, thereby ensuring that always at least one column can be used for loading and thus achieving a continuous loading product stream.

5. The method according to claim 1, wherein the at least one continuous Protein A chromatography is further characterized in that no peak cutting is performed during elution.

6. The method according to claim 1, wherein the peak cutting conditions developed for the CEX step in batch mode are also applied in the continuous mode.

7. The method according to claim 1, wherein the continuous process is carried out in a closed system using disposable equipment.

8. A method for transferring a batch chromatography process for a protein of interest to a continuous chromatography process for the same protein of interest, comprising at least one continuous bind and elute type chromatography in the continuous process is monitored using asymmetry factor analysis.

Description

BRIEF DESCRIPTION OF THE FIGURES

[0097] FIG. 1 shows the process overview for the exemplary continuous DSP including sampling spots and techniques

[0098] FIGS. 2a and 2b show for one example process which was transferred from batch to continuous mode the differences in batch and continuous DSP and possible impact on yield, product quality and impurities

EXAMPLE

1. General

[0099] In general after establishing the method of transferring of the batch production process for a monoclonal antibody to a continuous production process for the same monoclonal antibody both process modes i.e. batch and continuous were compared experimentally by purifying the protein of interest, here a monoclonal antibody, from the same harvest material.

[0100] For the continuous process thus a combination of a fed-batch USP—as employed for the batch process—and a continuous down-stream process (DSP) were used. This variant of the continuous process mode is known as a so-called hybrid process. However, a method for a reliable and efficient transfer of a batch down-stream production process of a given monoclonal antibody to a continuous down-stream process production process for the monoclonal antibody did not exist so far.

[0101] Samples were drawn at different time points and positions in the process for batch and continuous mode. Product quality attributes and process performance attributes were determined. The resulting polished material was processed to drug substance and further evaluated regarding storage stability and degradation behavior. Minor differences between batch and continuous samples were expected as different processing conditions were unavoidable due to the different nature of batch and continuous processing. All tests revealed no significant differences in the intermediates and comparability in the drug substance between the samples of both process modes. The stability study of the final product also showed no differences in the stability profile during storage and forced degradation.

[0102] Therefore, the work described herein does not only provide a method for a reliable and efficient transfer of a batch production process of a given monoclonal antibody to a continuous production process for the same monoclonal antibody but also demonstrated that the process mode does not influence product quality which is an essential prerequisite for successful implementation of the technology at every stage of the product life cycle

2. Materials

[0103] Harvest was generated in a 200 L Sartorius fed-batch reactor (Sartorius Stedim Biotech, Gottingen, Germany) using a Chinese hamster ovary (CHO) cell culture. The fermenter was harvested using a Pall Stax depth filter (Pall GmbH, Dreieich, Germany). The 0.2 μm filtered product had an IgG titer of 2.8 g/L.

[0104] GE MabSelect Sure was used as Protein A (ProtA) capture resin (GE Healthcare Life Sciences, Little Chalfont, UK). GE Capto SP impres was used in the intermediate bind & elute (B/E) chromatography process. Sartorius Sartobind Q membrane adsorber capsules were installed as final polishing step.

[0105] Citric buffers were employed for all chromatography steps. In case of continuous processing, buffers were 0.2 um filtered into 200 L Sartorius Flexboy bags. Intermediates for batch processing were 0.2 um filtered using Sartorius Sartopore 2 filters before storage. Sartorius Sartoguard NF filters were used as intermediate filters in continuous production to remove precipitates.

[0106] Pall BioSMB PD systems (Pall Biotech, Dreieich, Germany) were utilized for continuous chromatography, whereas Akta process systems (GE Healthcare Life Sciences, Little Chalfont, UK) were used for batch chromatography.

2.2. Analytics

[0107] Important product attributes were determined offline from samples drawn at different positions and time points within the batch and the continuous process. The product concentration was either determined by POROS-A high performance liquid chromatography (HPLC) or, after the ProtA capture step, through the determination of the absorbance at 280 nm wavelength (A280). The concentration of monomer, high molecular weight (HMW) and low molecular weight (LMW) isoforms was determined by size exclusion chromatography (SEC) HPLC. As the resolution for LMWs were not sufficient with this method, only the results for monomer concentration and HMWs were further used. In order to evaluate the degree of fragmentation of the product into light and heavy chain (LC and HC) during the process, capillary gel electrophoresis under non-reducing conditions was performed (CGE-nr). Moreover, CGE under reducing conditions was conducted to determine the degree of fragmentation into smaller debris (CGE-r). The charge variant distribution was investigated by capillary isoelectric focusing (cIEF). The impact of the process mode on the protein oxidation was determined by idES-HPLC. The activity of the mAb was investigated through a binding enzyme-linked immunosorbent assay (ELISA). The n-glycan profile was investigated with the help of HILIC uHPLC using Glykoprep®-plus Rapid N-Glycan sample preparation with the InstantAB kit (GPPNG-LB).

[0108] Additionally, process-related impurities such as host cell proteins (HCPs), deoxyribonucleic acid (DNA) and leached Protein A were determined. DNA was detected by polymerase chain reaction (PCR), whereas HCPs and leached ProtA were investigated through different ELISA methods.

[0109] Finally, the bioburden of the process in batch and continuous mode was determined. Therefore, the total aerobic microbial count (TAMC) and the total yeast/mold count (TYMC) were measured. Additionally, the endotoxin level was identified according to the harmonized testing method of Ph.Eur., USP, JP and TGA. The test kit N588 Pyrogent-5000 from Lonza was used for quantification

2.3. Plant Setup

[0110] The steps that were transferred from batch to continuous were capture (ProtA), low pH viral inactivation, intermediate chromatography by cation exchange (CEX) and polishing chromatography by anion exchange (AEX). FIG. 1 shows the process overview for the continuous DSP including sampling spots and techniques. Prior to the first chromatography step, the harvest was pre-filtered by a 0.2 μm filter. The ProtA chromatography was executed in B/E mode as in the batch process, but by using a BioSMB chromatography with 5 columns a continuous feed stream and a continuous eluate stream is generated. After the ProtA step, pH was adjusted for the following low pH viral inactivation (VI). In order to generate the necessary hold time for the step, a coiled flow inverter (CFI) was introduced into the system. Afterwards, pH and conductivity were adjusted for the following flow through (FT) UO and intermediate CEX. Another filtration step took place prior to the intermediate chromatography step. The CEX was executed in parallel batch mode to enable peak cutting to gain the required product purity in this step. The CEX eluate was again adjusted regarding pH and conductivity and 0.2 μm filtered prior to the polishing AEX in FT mode. The material after the polishing AEX chromatography of both process modes was further processed through VF and ultrafiltration/diafiltration (UF/DF), all in batch mode.

[0111] In the continuous DSP, pH and conductivity were inline-monitored and feedback-controlled if needed. UV was only inline monitored. All modules were highly automated and synchronized using a PCS7 distributed control system. A steady-state fluid connection existed between all modules and the entire process was a closed system. Sampling for bioburden and instantaneous grab samples were executed in either manual or automated mode; sampling for integral samples was achieved in automated mode. Grab samples were taken in a very short time and are only representative for the current status, whereas integral samples were drawn over a long period of time and were thus representative for this prolonged time.

[0112] In the batch DSP, pH and conductivity adjustments were performed manually. No fluid connection existed between the different UOs. Samples were taken manually from the pooled process intermediate between the different UOs. As the purification process took several days, process intermediates were stored at 2-8° C. between UOs, whereas no such process hold times exist in the continuous process.

[0113] In the batch process, all chromatography steps were sanitized with sodium hydroxide. In the continuous process, the majority of the parts were gamma irradiated with at least 25 kGy, including the chromatography columns. Where no gamma irradiation was applicable, the parts were treated with ethylenoxide (EO) or autoclaved before introduction into the system.

3. Experimental Procedure

[0114] In order to preclude the impact of storage on the starting material, the harvest stored for 5 days at 2-8° C. was split into two parts for the two process modes immediately before processing. The harvest part for the batch purification was equilibrated to room temperature and processed on the ProtA column within approximately 2 h. The batch production until the AEX step was carried out in 7 days comprising intermediate storage at 2-8° C. Batch chromatography processes followed standard procedures for installation and processing. This included the sanitization prior to each step with sodium hydroxide.

[0115] During the continuous DSP, the first step was the inline calibration and adjustment of the pH and conductivity sensors inline in the conditioning modules using citrate buffers at the operating point. The entire DSP was then primed with process buffers and controlled with respect to flow, pH and conductivity. Once steady state conditions were reached, the harvest was connected to the first filtration module of the DSP. During the whole continuous production time (40 h), the harvest was supplied out of a chilled vessel. The feed material was equilibrated to room temperature during the initial 0.2 μm filtration before entering the ProtA UO.

[0116] The shut-down phase was initialized by replacing the harvest bag with ProtA equilibration buffer. ProtA chromatography operation continued until all 5 columns were eluted. After stopping the ProtA module, a low pH buffer flush at the inlet of the VI module continued chasing product. CEX and AEX chromatography continued until the inlet protein concentrations were below 0.1 g/L. The process was fully automated including start-up, shut-down and handling of events. The established parameter control strategy ensured that only product within normal operating ranges (NORs) was further processed.

[0117] Process parameters were kept identical between batch and continuous production wherever possible. This includes chromatographic buffers, resins and membrane types. Furthermore, column load conditions with respect to conductivity and pH were the same.

4. Process Control

[0118] As similar sensors were used for batch and continuous processing, the measurement accuracy in both process modes was comparable. For the conductivity sensors, an accuracy of ±5% was assumed, whereas the accuracy for pH meters in general is ±0.1 pH units. During the continuous process, sensors were adjusted every 72 h before and after processing, enabling the identification of the sensor drift. For pH sensors, the maximum drift was 0.076. The conductivity sensors drifted by 0.02 mS/cm at maximum

[0119] Consequently, the process control accuracy shown in this study guarantees the same process conditions for the entire DSP material based on the collected measurement data

5. Quality Attributes

[0120] In order to ensure side-by-side comparability between batch and continuous downstream processing, the product quality attributes must be similar through all stages of the process and over the entire production time taking expected differences into account. Therefore, several parameters such as product concentration, product-related impurities and process-related impurities were analyzed for the samples drawn from the batch and the continuous process.

[0121] In order to evaluate the comparability between both process modes, the final drug products were extensively tested and compared to each other. Moreover, a stability study was performed with the final drug product to evaluate the comparability of storage and degradation behavior. Finally, the microbial load of both process modes was tested.

[0122] Additionally, the microbial load of both process modes was tested. Finally, a stability study was performed with the final product to evaluate the comparability of storage and degradation behavior.

6. Process Intermediates

[0123] Sampling took place before and after every chromatography step in batch as well as in continuous mode.

[0124] Comparing batch and continuous results, differences in the mAb concentration were apparent. The highest differences were observed at the AEX FT samples. This process behavior was expected and the explanation lies in the combination of CEX and AEX attributes. Lower protein concentrations in the CEX eluate are observed because of a larger elution volume of column no. 1. Lower concentration in the AEX pool was attributed to a longer chase at the end of an AEX cycle, whereas in batch the chase was fractionated according to an A280 inline signal.

[0125] Regarding the monomer ratio, minor differences between the different process steps were observed. This can be explained by the low resolution of this method for monomer and LMWs. Regarding the HMWs there are no significant differences between the batch and continuous processes within the assay variability.

[0126] The results of CGE under reducing and non-reducing conditions showed that IgG purity remains constant throughout the purification process and is comparable with the batch data. The differences are within the assay deviation.

[0127] Additionally to the results for product-related impurities, the charge heterogeneity was evaluated with the help of cIEF. No differences between the different samples could be observed over the entire process of the two process modes. The charge variant distribution was stable.

[0128] Concerning the process-related impurities it was found that the initial ProtA concentration was higher for the continuous process than for the batch process. This could be caused by the different sanitization methods used. The batch column was sanitized once with 0.1 M NaOH. The columns used for the continuous process were gamma-irradiated once and sanitized with 0.1 M NaOH in every cycle. Additionally, more cycles were used in the continuous DSP and the collected elution volume was higher than in the batch DSP. In both process modes the leached ProtA concentration decreased rapidly due to the FT UO before the CEX. Afterwards, the ProtA concentration was stable close to the limit of quantification for the used assay.

[0129] Also the host cell DNA concentration over the course of the process was measured (data not shown). Comparing the four continuous and the batch samples to each other, the curves were almost identical. The major depletion of DNA took place during the ProtA chromatography step. The remaining DNA was then removed by the FT UO afterwards. Consequently, no DNA was detected in the CEX Load samples and beyond.

[0130] Measurement of the HCP concentration over the course of the process showed that the main reduction of HCPs took place during the ProtA chromatography, followed by the FT UO and CEX. The analysis results of ProtA Load and Eluate were almost identical. Beginning with the CEX Load samples, the results of the four continuous and the batch sample differed. As the specific load of mAb and the flow rate on the FT UO was higher in batch mode than in continuous mode the HCP breakthrough was higher as well. Nevertheless, the resulting final measurements of the AEX FT samples showed similar low HCP concentrations.

[0131] To conclude, all tested product parameters and process-related impurities showed only minor differences between batch and continuous processing and between continuous processing at different time points. The occurred alterations were explainable by differences in sanitization methods or process parameters. This proved that batch and continuous downstream processing of the mAb results in a comparable product over the entire course of the process.

7. Final Drug Substance

[0132] In order to evaluate the comparability of the final drug substance, the final product was extensively tested.

[0133] The post-AEX process intermediates of both process modes were further processed to bulk drug substance via viral filtration, ultrafiltration and diafiltration. The resulting material was further analyzed for product quality attributes such as charge variants, pI range, n-Glycan profile, activity, molecular weight, idES-HPLC and IgG purity. Additionally, product related impurities were analyzed as well. It was clearly demonstrated that most results are identical or only show minor differences. The same quality of results was achieved for all other tested attributes as well (data not shown). All samples were within the given specifications. Tests for host cell DNA, endotoxins and microbial load were negative.

[0134] To conclude, the results of the final drug substance for batch and continuous processing show full comparability between both process modes. Consequently, the study design and execution can be regarded successful.

8. Stability Studies Final Drug Substance

[0135] Stability tests were conducted at room temperature (18-26° C.) and 40° C. for 4 weeks each. The final product was stored in sterile low density polyethylene (LDPE) bags during the study. Several quality attributes were tested during the study, including protein concentration (OD A280), monomer and HMW concentration via SEC-HPLC, % intact IgG through CGE-nr, charge variant distribution (% main peak, % total basic and acidic isoforms) with the help of cIEF, the impact on protein oxidation (idES-HPLC) and finally activity through binding-ELISA. Exemplary, the results for the charge variants distribution did not show any significant impact on the charge heterogeneity. Moreover, no difference between batch and continuously produced mAb solution were observed. Also at elevated temperature (40° C.) no significant difference between batch and continuous samples was observed. To conclude, the stability study showed that also the degradation profile of the final product was comparable between batch and continuously produced mAb considering charge heterogeneity.

[0136] Regarding the other tested quality attributes including protein concentration, monomer concentration, HMWs, protein oxidation and activity, no significant difference between both samples was observed at room temperature as well as at 40° C. Consequently, the comparability of batch and continuously produced final drug substance was also shown in a stability study including a wide range of relevant quality attributes.

9. Microbial Control

[0137] In order to evaluate the bioburden of the processes, the TAMC and TYMC were determined from certain samples (see FIG. 1 for bioburden sampling positions). Additionally, the endotoxin level was measured. In the continuous process the samples were derived from the retentate side of the 0.2 μm filters, as possible microbial contaminations would concentrate at these positions. For all tested samples, the results for TAMC, TYMC and endotoxin were below the limits of quantification. This leads to values <1 cfu/100 mL for TAMC and TYMC and <1.0 EE/mL for endotoxin. Consequently the batch intermediates as well as the entire continuous process can be considered as bioburden-free during the current runs.

10. Conclusion

[0138] Overall the above results demonstrate that by employing the method of transfer of batch production process for a monoclonal antibody to a continuous production process for the same monoclonal antibody described herein leads to comparable results regarding product quality attributes. Thus, said method represents a major step in the implementation process towards continuous production of mAbs, as already existing batch processes can be transferred into their continuous mode without affecting the comparability of final drug substance characteristics. Therefore, choosing the process mode individually and flexible independent from the product life cycle, from clinical trials material to large-scale production, should be possible in the future.