Systems and methods for monitoring charge isomers of monoclonal antibodies using Raman spectroscopy

Abstract

A system and method for determining charge isomers in monoclonal antibodies using Raman spectroscopy is disclosed.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This patent application claims the benefit of Indian Patent Application No. 202211047354 filed on August 19, 2022, which is incorporated by reference into this patent application. Background Art

[0002] Charge variants are considered one of the critical quality attributes (CQAs) of monoclonal antibodies (mAbs) because they can significantly affect the safety and efficacy of the final drug product and must therefore be strictly controlled. Charge variants can be broadly divided into acidic and basic forms. Acidic variants are produced by modifications including deamidation, high mannose content, saccharification, fragmentation, disulfide structural heterogeneity, sialylation, etc., because these modifications lower the isoelectric point of the mAb. Similarly, basic variants are produced by C-terminal lysine truncation, amidation, glycosylation, incomplete cyclization of N-terminal glutamine or glutamic acid, methionine oxidation, succinate formation, incomplete removal of the leader sequence, or polymer formation, because these increase the isoelectric point. These charge variants may affect protein stability and alter the function and immunogenicity of the final product and may lead to different protein interactions. Most commercial mAb processes require the use of cation exchange chromatography (CEX) in the downstream process to further remove charge variants in order to achieve the final quality target product profile (QTPP). At-line HPLC has been reported as an option for monitoring and controlling charge variants.

[0003] In the manufacture of monoclonal antibodies, a need exists for improved determination of charge variants.

[0004] The present invention provides improvements to at least some of the shortcomings of the prior art.These and other advantages of the present invention will be apparent from the description as set forth below. Summary of the Invention

[0005] Aspects of the present invention provide a method for determining charge variants in a monoclonal antibody, the method comprising: (a) continuously passing a fluid containing the monoclonal antibody through a chromatography column, including sampling the fluid containing the monoclonal antibody before passing the fluid containing the monoclonal antibody through the chromatography column, and binding the monoclonal antibody in the fluid containing the monoclonal antibody to the chromatography column; (b) eluting the bound monoclonal antibody from the chromatography column to provide an eluted monoclonal antibody-containing fluid, and sampling the eluted monoclonal antibody-containing fluid; (c) analyzing a sample of the fluid containing the monoclonal antibody before passing the fluid containing the monoclonal antibody through the chromatography column and a sample of the eluted monoclonal antibody-containing fluid by means of Raman spectroscopy; and (d) quantifying the charge variant distribution present in the sampled monoclonal antibody-containing fluid before passing the fluid containing the monoclonal antibody through the chromatography column. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] Figure 1 is a graph showing analytical cation exchange chromatography (CEX) of a monoclonal antibody prior to charge separation according to aspects of the present invention, illustrating the presence of different charge isomers (acidic, major, and basic), with the acidic eluting first, followed by the major, and then the basic.

[0007] Figures 2A-2B is a graph showing the elution profile of a preparative scale CEX monoclonal antibody according to another aspect of the present invention, showing the acidic-major species ( Figure 2A ) and mainly alkaline substances ( Figure 2B ), but limited coelution between acidic-basic species.

[0008] Figures 3A-3B is a diagram showing the process of pre-processing ( Figure 3A ) and after preprocessing ( Figure 3B ) of a monoclonal antibody mixture, where the sample has high acidic, high primary, and high basic content.

[0009] Figures 4A-4D Figures illustrating observed and predicted scatter plots for the proposed CNN architecture: (A) acidic charge variants; (B) major charge variants; (C) basic charge variants; and (D) total protein.

[0010] Figures 5A-5D Graphs illustrating Taylor plots of predicted charge variant concentrations during the validation phase: (A) acidic charge variants; (B) major charge variants; (C) basic charge variants; and (D) total protein (all charge variants).

[0011] Figures 6A-6DGraphs illustrating the comparison of actual and predicted charge variant concentrations using different machine learning models: (A) acidic charge variants; (B) major charge variants; (C) basic charge variants; and (D) total protein.

[0012] Figures 7A-7B Figures illustrating the schematic design of integrating Raman spectroscopy in a continuous process: (A) after protein A elution; (B) after CEX elution.

[0013] Figure 8 Graph illustrating comparison of Raman quantification and HPLC determination of charge variant concentrations in different preparative CEX elution fractions. DETAILED DESCRIPTION

[0014] According to aspects of the present invention, a method for determining charge variants in a monoclonal antibody is provided, the method comprising: (a) continuously passing a fluid containing the monoclonal antibody through a chromatography column, comprising sampling the fluid containing the monoclonal antibody before passing the fluid containing the monoclonal antibody through the chromatography column, and binding the monoclonal antibodies in the fluid containing the monoclonal antibody to the chromatography column; (b) eluting the bound monoclonal antibodies from the chromatography column to provide an eluted monoclonal antibody-containing fluid, and sampling the eluted monoclonal antibody-containing fluid; (c) analyzing a sample of the fluid containing the monoclonal antibody before passing the fluid containing the monoclonal antibody through the chromatography column and a sample of the eluted monoclonal antibody-containing fluid by means of Raman spectroscopy; and (d) quantifying the charge variant distribution present in the sampled monoclonal antibody-containing fluid before passing the fluid containing the monoclonal antibody through the chromatography column.

[0015] In aspects, the method comprises adjusting the elution rate of the monoclonal antibody from the chromatography column based on the charge variant distribution quantified in (d).

[0016] In some aspects of the method, (d) comprises quantifying the charge variant distribution present in the sampled eluted monoclonal antibody-containing fluid.

[0017] Aspects of the method can be utilized with in-line and at-line sampling of monoclonal antibodies.

[0018] Advantageously, Raman spectroscopy can be used for the real-time quantification of charge isomers in refined chromatography, and can be used for real-time process analytical technology (PAT) tools to control the pooling decision-making of key quality attributes (CQAs) of monoclonal antibodies (mAbs) in the downstream chain of continuous production. In contrast, the analog conventional device for charge isomer determination is high performance liquid chromatography (HPLC). However, HPLC methods usually require an analysis time of 5-30 minutes, and when manufacturers are trying to achieve real-time control of CQAs in continuous processing, this can become a challenge.

[0019] Unlike infrared spectroscopy, Raman spectroscopy exhibits minimal interference from water absorption bands, which often obscure the spectral signals of component molecules in aqueous solutions—a common problem when deploying spectroscopic methods in biopharmaceutical manufacturing, compared to solid-state pharmaceutical processes. Raman spectroscopy can be used in both low- and high-concentration aqueous solutions and is able to capture unique vibrations from different bonds to provide molecular fingerprints of different molecules in aqueous mixtures. It is a rapid, non-invasive, and non-destructive technique, making it suitable as a PAT tool.

[0020] Pre-processing of Raman spectra may include baseline subtraction, cosmic ray removal, smoothing and normalization.

[0021] According to aspects of the present invention, various artificial intelligence (AI)-based models were made to determine acidic, primary, alkaline and total protein concentrations, wherein the correlation coefficient was greater than 0.9 in each case, such as 0.94, 0.99, 0.96 and 0.99, respectively. Raman spectroscopy can provide a fast and accurate determination of charge isomer percentages, thereby enabling it to be used for real-time decision making regarding the pooling strategy of mAb products and thereby achieving a consistent charge isomer profile for the resulting product. According to aspects of the present invention, Raman spectroscopy can be used for charge isomer detection in liquid and lyophilized forms.

[0022] In some aspects, lasers with wavelengths of 500-800 nm can be used for spectral acquisition, where the spectral range includes but is not limited to 100-2000 cm -1 If desired, Raman spectroscopy with exposure times of 10 seconds or longer and powers of 30 mW or higher can be used for rapid and accurate charge variant determination of monoclonal antibodies.

[0023] Each of the components of the present invention will now be described in more detail below, wherein like components have like reference numerals.

[0024] 1. Raman spectroscopy 1.1 Selection of spectral acquisition settings

[0025] An inVia™ confocal Raman microscope with a 785 nm laser providing 300 mW of power was used. To collect Raman spectra. Wire software was used to control the instrument and capture spectra. Samples were collected in 96-well plates and the images were taken using a 50x long-range objective (Leica ), 200-second exposure time and 300mW power, and acquired spectra in the range of 300-1800cm-1. These parameters were adjusted experimentally over the range of 10-300 second exposure time and 50-1000mW power to determine the optimal conditions for capturing the spectral features of the mAb samples. Before exporting the spectra for model development, baseline subtraction and height- and width-based cosmic ray removal were performed using built-in features in the Wire software. 1.2 Data Preprocessing

[0026] Data preprocessing can be useful in creating and training machine learning models. To ensure robustness, raw data is converted into a format that can be understood by the machine learning model. In this paper, three different preprocessing techniques are utilized: data smoothing, scaling, and data augmentation. The recorded raw spectra have a certain degree of noise that we wish to remove before further analysis. The Savitzky-Golay filter is a well-known method for smoothing signals. This method relies on solving a linear least-squares problem on a subset of data points, attempting to locally fit a low-degree polynomial with a specified segment width and polynomial order. The following parameter set yielded the best results for this application: polynomial order = 1 and frame length = 19. The dynamic range of the spectral data is quite high, and it is desirable to reduce it for numerical stability. Spectra preprocessed using a standard pipeline are used for further processing. In the context of deep learning, synthetic training datasets are generated from the original dataset. This data augmentation often involves introducing random noise, rotations, scaling, and other transformations to the training dataset while retaining meaningful information. This helps the model gain insight into details and improves its ability to handle noisy inputs. To improve the generation of new training samples, Gaussian noise is added to existing samples without significantly changing the structure of the spectra. According to aspects of the present invention, performance data augmentation is applied only on the training set, as applying it to the entire dataset before the cross-validation split can cause information leakage into the test set. 1.3 Development of Artificial Intelligence-Machine Learning (AI-ML) Framework

[0027] According to aspects of the present invention, first consider the calibration library designed for quantifying the charge isomers in mAb samples. Designed a collection of samples for the model calibration library. Target is to design a sample collection that independently covers the variation of acidic, main and alkaline substances across the expected chromatographic elution space. The scope of the model is 0-20g / L mAb, where the charge isomer concentration is from 0 to 100%. As discussed further below, a total of 270 library standards prepared using 10% spectra randomly selected for internal validation were fitted to the calibration model.

[0028] To determine the charge variant content of the mAb, the protein was loaded onto an analytical scale CEX column. Figure 1 The presence of different charge variants in the mAb load prior to any preparative-scale purification is depicted. As determined by analytical-scale CEX-HPLC, the charge variant profile of the mAb load prior to any preparative-scale purification contained 29.8% acidic species, 50.6% major species, and 19.6% basic species. Semi-preparative-scale CEX was then performed to prepare samples for AI-ML model calibration. Preparative-scale CEX was run using Eshmuno CPX resin, and the collected samples were used for AI-ML model validation. Figure 2A-2B A preparative scale CEX chromatogram during elution is shown. Significant co-elution was observed between acidic-major and major-basic species, but limited co-elution between acidic-basic species.

[0029] After semi-preparative scale purification of mAb, all acidic, major, and basic fractions were pooled accordingly to obtain nearly pure charge variant species. The pooled acidic content was found to be 97.4% pure, the major species was 81.9% pure, and the basic isomer was 94.8% pure. These pooled charge variants were mixed in different ratios to generate different charge variant contents at different concentrations.

[0030] Four machine learning frameworks are included to estimate charge variants: support vector regression (SVR), random forest (RF), deep neural network (DNN), and convolutional neural network (CNN).

[0031] In the Raman spectrum of 300-1800 cm -1 In the region of , four calibration models were fitted for acidic species concentration, major species concentration, basic isomer concentration, and total mAb concentration. Figure 3A-Figure 3B Shown are spectra before and after pre-treatment of samples with high acidic, high essential and high alkaline contents.

[0032] Four machine learning frameworks are compared and adopted, such as R 2, RMSE, MAE, MSE and Taylor diagram to evaluate the model prediction performance. Figures 4A-4D Scatter plots are shown for SVR, RF, DNN, and CNN for different charge variants and total protein. Based on statistical values, the CNN architecture using data augmentation provides better performance than the SVR, RF, and DNN models. Pearson's correlation coefficients of 0.9692, 0.9956, 0.9801, and 0.994 were obtained for acidic, major, basic, and total proteins, respectively, and for the charge variant CNN architecture. Similarly, Pearson's correlation coefficients were evaluated for SVR (0.7190 for acidic, 0.9872 for predominant, 0.7642 for basic, and 0.9711 for total protein), RF (0.7953 for acidic, 0.9901 for predominant, 0.8983 for basic, and 0.9836 for total protein), and DNN (0.9121 for acidic, 0.9918 for predominant, 0.9559 for basic, and 0.9901 for total protein).

[0033] The results of the four different models used to estimate the concentrations of acidic, major, basic charge variants, and total protein are reported in Table 1. 2 , RMSE, MAE and MSE values. As indicated by the statistical values shown in Table 1, it is noted that the CNN-based model has higher prediction accuracy compared to the SVR-, RF- and DNN-based models. In terms of the CNN architecture, the highest R 2 values (0.9401, 0.9908, 0.9604 and 0.9881) and the minimum RMSE values (0.1846, 0.1627, 0.1029 and 0.2483). However, in the DNN architecture, the RMSE value increased by 9% and the corresponding R 2 The value is reduced by about 11%. From Table 1, it is obvious that the CNN architecture surpasses all other current process-level techniques (such as SVR, RF and DNN-based models for acidic, primary, basic and total proteins). Table 1. Statistical results of the SVR, RF, DNN, and CNN models for predicting acidic, major, and basic charge variants.

[0034] As can be seen from the lower levels of the scatter plot and the fact that the estimated data are Figures 4A-4D The CNN model significantly outperforms the other models, as indicated by the improved fit between the values seen in the 1:1 line. Figures 4A-4D The evaluation criteria are analyzed as shown in Table 1, but Taylor diagrams (TD) are also used to compare the methods used as described in this paper. The basic principle behind TD is to depict the closest prediction model using the real correlation observations on a two-dimensional scale, which includes the correlation coefficient on the polar axis and the standard deviation on the radial axis. Therefore, according to Figures 5A-5D It can be observed that the CNN model performs better than other current process-level techniques. As shown in Figure 6, the estimated values related to charge variant concentrations have been plotted for each charge variant and total protein measurements using the four AI-ML models. Figures 6A-6D Represents the actual and predicted values of different prediction models regarding acidic, major, basic, and total protein concentrations. 2. Deployment of Raman spectroscopy for real-time analysis of chromatographic process materials

[0035] Build Figure 7A-7B The experimental setup shown in FIG is for integrating Raman spectroscopy with preparative chromatography. The sampling location for the Raman sample can be in the feed tank for CEX, i.e., the buffer tank where the neutralized Protein A eluate is stored after virus inactivation ( Figure 7A ). During continuous chromatography of the mAb, Protein A-purified mAb is collected in a buffer tank, and a peristaltic pump circulates a fixed amount of sample to a 96-well plate from which Raman spectra are acquired. After the spectra are acquired, another peristaltic pump recirculates the sample from the 96-well plate back to the buffer tank. Both pumps are operated so that 200 μl of sample is continuously held in the 96-well plate. Another option is to use a similar pump instrument at the outlet of the CEX column by diverting a small stream of material from the CEX elution stream into the 96-well plate and then remixing it into the CEX elution stream from the buffer tank ( Figure 7B (Setup B in Figure 1). The Raman spectrophotometer is focused on a 200 μl sample in a 96-well plate, allowing sampling from any desired location in a continuous process. For real-time analysis, the calibrated model is exported to a method file and then input into a real-time spectrum collection and processing workflow. As demonstrated in the next section, this workflow is initiated at the same time as the CEX elution and automatically transfers the real-time Raman spectrum and the corresponding stored background spectrum to the calibration model. Numerical concentration measurements are reported in a time-stamped Excel array that can be accessed using Python scripts and used as input for control actions. 3. Case Study for Monitoring and Controlling Charge Variants in mAb Purification

[0036] The monitoring tool was tested in CEX chromatography using an Eshmuno CPX packed column with an elution gradient from 0 to 100% buffer B in 20 CV and a loading of 15 mg mAb / mL resin. Feed samples as well as elution fractions were also collected for offline analysis using analytical CEX HPLC. Figure 8 The result of the charge isomer elution curve compared with the Raman spectrometer measurement of CNN framework and with analytical HPLC is shown.It is obvious that the charge isomer curve is successfully tracked by Raman spectrometer, wherein acidic substance, main substance, alkaline substance and total protein concentration are measured once every 200 seconds.The mean absolute error between all Raman and HPLC measurements is 0.15g / L.Therefore, this method is to be able to integrate the effective online (on-line) PAT tool of the chromatographic control strategy based on model and experience as published in recent literature.The first method is to use Raman to quantify the charge isomer in CEX chromatogram loading and run mechanical chromatographic model for predicting elution curve and regulating pooling standard on the basis of circulation to circulation before loading.Another option is to monitor the elution charge isomer in real time and set up the empirical model for triggering pooling control action, and also can use online (at-line) HPLC rather than Raman spectroscopy as the method that monitoring tool performs.

[0037] When being deployed for continuous monitoring and control of charge isomers, these two methods have advantages and disadvantages. In the case of the former, a delay of 200 seconds is insignificant because the total elution time of CEX is typically 0.5-2 hours long. Therefore, a Raman method can be used once per cycle to quantify the charge isomer distribution in the feed and to regulate pooling during the elution of that particular cycle. However, additional complexity is introduced because additional model-based prediction tools are required to simulate the elution curve based on a given charge isomer load, which requires calibration of several mechanical parameters depending on column size, resin type, elution gradient characteristics and mAb binding isotherms. Therefore, if a robust mechanical model has been developed and verified, the method is preferred. On the other hand, using Raman to directly quantify the composition of the elution stream is a simpler method, but when attempting to achieve real-time pooling decisions, a time delay of 200 seconds may be less desirable because a 3-minute delay may be an important part of the elution block that may be about 30 minutes long.

[0038] Unless otherwise indicated herein or clearly contradicted by context, the use of the terms "a" and "an" and "the" and "at least one" and similar referents in the context of describing the present invention (especially in the context of the following claims) are to be interpreted as encompassing both the singular and the plural. Unless otherwise indicated herein or clearly contradicted by context, the use of the term "at least one" followed by a list of one or more items (e.g., "at least one of A and B") is to be interpreted as referring to one item (A or B) selected from the listed items or any combination of two or more of the listed items (A and B). Unless otherwise indicated herein, the terms "comprising," "having," "including," and "containing" are to be interpreted as open-ended terms (i.e., meaning "including, but not limited to"), unless otherwise indicated herein. The recitation of ranges of values herein is merely intended to serve as a shorthand method of individually referring to each independent value falling within the range, and each independent value is incorporated into the specification as if it were individually recited herein. Unless otherwise indicated herein or clearly contradicted by context, all methods described herein can be performed in any suitable order. Unless otherwise stated, the use of any and all examples or exemplary language (e.g., "such as") provided herein is intended merely to better illuminate the invention and does not limit the scope of the invention. No language in the specification should be construed as indicating any non-claimed element as essential to the practice of the invention.

[0039] The preferred aspects of the present invention are described herein, including the best mode known to the inventor for carrying out the present invention. Upon reading the foregoing description, variations in those preferred aspects will become apparent to those of ordinary skill in the art. The inventors expect that the skilled person will appropriately adopt such variations, and the inventors intend to implement the present invention in a manner different from that specifically described herein. Therefore, the present invention includes all modifications and equivalents of the subject matter recited in the claims appended hereto as permitted by applicable law. In addition, unless otherwise indicated herein or clearly contradicted by the context, the present invention includes any combination of the elements described above in all possible variations thereof.

Claims

1. A method for determining charge variants in a monoclonal antibody, the method comprising: (a) continuously passing a fluid containing a monoclonal antibody through a chromatography column, comprising sampling the fluid containing the monoclonal antibody before passing the fluid containing the monoclonal antibody through the chromatography column, and binding the monoclonal antibody in the fluid containing the monoclonal antibody to the chromatography column; (b) eluting the bound monoclonal antibody from the chromatography column to provide an eluted monoclonal antibody-containing fluid, and sampling the eluted monoclonal antibody-containing fluid; (c) analyzing a sample of the monoclonal antibody-containing fluid before passing the monoclonal antibody-containing fluid through the chromatography column and a sample of the eluted monoclonal antibody-containing fluid by means of Raman spectroscopy; and (d) quantifying the charge variant distribution present in a sampled monoclonal antibody-containing fluid prior to passing the monoclonal antibody-containing fluid through the chromatography column.

2. The method of claim 1, comprising adjusting the elution rate of the monoclonal antibody from the chromatography column based on the charge variant distribution quantified in (d).

3. The method according to claim 1 or 2, wherein (d) comprises quantifying the charge variant distribution present in the sampled eluted monoclonal antibody-containing fluid.

Images