Chromatography methods

Abstract

Disclosed herein are improved methods for quantifying monoclonal antibody (mAb) titer using Protein A affinity High-Performance Liquid Chromatography (HPLC). The methods address inaccuracies caused by mAb adsorption to sample vial surfaces by incorporating a surfactant into the mobile phase used for sample dilution. This approach enhances the accuracy, linearity, and robustness of the titer assay, making it suitable for process control and quality assessment in biopharmaceutical development and manufacturing. The methods can be applied as a platform approach for various mAb products.

Description

CHROMATOGRAPHY METHODSRELATED APPLICATIONS AND INCORPORATION BY REFERENCE

[0001] This application claims priority to U.S. Provisional Patent Application No. 63 / 643,170, filed May 6, 2024. The foregoing applications, and all documents cited therein or during their prosecution and all documents cited or referenced in the application cited documents, and all documents cited or referenced herein (herein cited documents), and all documents cited or referenced in herein cited documents, together with any manufacturer’s instructions, descriptions, product specifications, and product sheets for any products mentioned herein or in any document incorporated by reference herein, are hereby incorporated herein by reference, and may be employed in the practice of the invention. More specifically, all referenced documents are incorporated by reference to the same extent as if each individual document was specifically and individually indicated to be incorporated by reference.FIELD OF THE INVENTION

[0002] The present invention relates to the field of analytical chemistry, in particular to methods for quantifying monoclonal antibody titer using protein A affinity chromatography.BACKGROUND OF THE INVENTION

[0003] Citation or identification of any document in this application is not an admission that such document is available as prior art to the present invention.

[0004] Monoclonal antibody (mAb) therapeutics are an important class of biopharmaceuticals. Accurate measurement of mAb titer during production is critical for process control and ensuring product quality. Protein A affinity chromatography is commonly used for mAb titer analysis.SUMMARY OF THE INVENTION

[0005] Because traditional Protein A HPLC methods can suffer from issues like mAb losses due to adsorption to vial surfaces, impacting analytical performance, there is a need for more robust Protein A HPLC methods for mAb titer analysis that overcome these limitations.

[0006] The present disclosure relates to improved analytical methods using Protein A affinity chromatography, particularly HPLC, to analyze monoclonal antibodies (mAbs). Certain reference Protein A HPLC methods can be affected by the loss of mAb due to adsorption onto sample vial surfaces, which can compromise analytical accuracy.

[0007] The disclosure features methods addressing this adsorption issue by incorporating Polysorbate 20 (Tween 20) into the mobile phase buffer. In one aspect, the disclosure provides a method for separating a monoclonal antibody from a mixture using Protein A chromatography where the mobile phase comprises Tween 20.

[0008] In one aspect, this method uses an analytical column and involves gradient or isocratic elution, with run times between about 10 and 30 minutes. In one aspect, this separation method further includes detecting or isolating the mAb and, in some aspects, provides higher accuracy compared to methods performed without Tween 20, particularly for mAbs prone to adsorption like IgG isotypes (e.g., IgGl, IgG2) or relatively hydrophobic mAbs.

[0009] In some aspects, the Tween 20 is present in the mobile phase (e.g., used as a sample diluent for standards and samples) at a concentration sufficient to reduce non-specific adsorption to sample vials (e.g., glass or polypropylene), such as between about 0.01% v / v and 0.05% v / v, or about 0.02% v / v. In some aspects, the stationary phase comprises Protein A immobilized on a monolith support. In some embodiments, the presence of Tween 20 reduces y-intercept bias when generating a calibration curve for titer determination.

[0010] In some aspects, the disclosure provides a method for detecting and quantifying an IgG monoclonal antibody in a test sample. In some aspects, this involves preparing reference and test samples using a mobile phase containing Tween 20, separating them via Protein A HPLC to obtain reference and test chromatograms, comparing the chromatograms (based on retention time and peak profile) to detect the antibody's presence, and quantifying the antibody by comparing the test sample's detector response to a calibration curve generated from reference standards.

[0011] In an aspect, the disclosure provides a method for quality control of a pharmaceutical composition containing a mAb. In some aspects, this method involves separating the mAb using the Protein A HPLC method with Tween 20, comparing the separated mAb to a reference, quantifying the amount or concentration (e.g., via peak area integration against a calibration curve), and determining if the mAb or composition meets a selected character or quality standard. The methods disclosed herein are suitable for analytical titer determination during process development or quality control.

[0012]

[0013] It is noted that in this disclosure and particularly in the claims and / or paragraphs, terms such as comprises, comprised, comprising and the like can have the meaning attributed to it in U.S. Patent law; e.g., they can mean includes, included, including, and the like; and thatterms such as consisting essentially of and consists essentially of have the meaning ascribed to them in U.S. Patent law, e.g., they allow for elements not explicitly recited, but exclude elements that are found in the prior art or that affect a basic or novel characteristic of the invention.

[0014] These and other embodiments are disclosed or are obvious from and encompassed by the following Detailed Description.BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The following detailed description, given by way of example, but not intended to limit the invention solely to the specific embodiments described, may best be understood in conjunction with the accompanying drawings.

[0016] FIG. 1. Theory and mechanism of Protein A (ProA) affinity chromatography. (A) The cell culture sample is injected onto the column, and (B) the monoclonal antibody (mAb) binds to the ProA ligand via its crystallizable fragment (Fc) region, while all the other components or impurities in the mixture flow through the column. (C) The bound antibody is then eluted from the column by adjusting the mobile phase by altering the ionic strength, pH, or denaturant and subsequently quantified.

[0017] FIG. 2 illustrates a probability plot connecting accuracy and precision performance to a total analytical error requirement of being within ± 25 % of the true value with at least 90 % probability.

[0018] FIG. 3 displays Ishikawa diagram of identified risk factors for ProA titer HPLC procedure development.

[0019] FIG. 4 shows the risk assessment table combining elements of a risk heat map and CNX assessment. C: controlled, N: noise, X: experimental.

[0020] FIG. 5 displays the process capability analysis showing the distribution of the calculated intercept bias (%), of n=6 variable concentration and fixed injection volume calibration standards, of which n=3 were stored in glass vials and n=3 were stored in polypropylene vials, Lower (LSL) and upper specification limits (USL) corresponds to ATP criteria for linearity (intercept bias of ± 15%) and are denoted by lines. The dotted black curve shows the modeled normal distribution of the data

[0026] . The resulting Shapiro-Wilk value obtained (Prob > W) was 0.1071, indicating no significant lack-of-fit of these data to a normal distribution model.

[0021] FIG. 6 depicts plots showing the A: second-degree polynomial fit and B, C, and D: linear fits to the percent relative difference (%RD) of standard curve peak areas of variable concentration and fixed injection volume calibration standards versus fixed concentration and variable injection volume calibration standards. Variable concentration standard preparations were stored in either glass vials (solid circles) or polypropylene vials (empty rectangles). The confidence limits for model fits are indicated by the shaded regions about the solid lines. The horizontal dotted line indicates where there is no difference in peak area between injection strategies. Data in panel A were fit to a second-degree polynomial due to the presence of a significant quadratic term (p = 0.0011, p = 0.0003) for standard preparations stored in glass and polypropylene vials, respectively. Additional quadratic terms for data in panels B, C and D were not significant and thus excluded; B: (p = 0.4303, p = 0.3104), C: (p = 0.2509, p = 0.2243) and D: (p = 0.7949, p = 0.7041) for standard preparations stored in glass and polypropylene vials, respectively.

[0022] FIG. 7 displays a plot showing one-way analysis of variance (ANOVA) of glass vials (solid circles) and polypropylene vials (empty rectangles), to test for differences among the group means of conditions either with or without Tween® 20 added. The dotted diamonds represent the group means and their 95% confidence points shown by the horizontal lines in the center of the diamond and the top and bottom points on the mean diamonds, respectively. Overlap marks are represented by the horizontal dotted lines near the top and bottom of each mean diamond. In cases where the sample sizes of the groups are equal, the presence of overlapping marks signifies that the two group averages are not significantly different at the specified confidence level. The horizontal lines indicate the lower (LSL) and upper (USL) specification limits which correspond to the analytical target profile (ATP) criteria for linearity (intercept bias of ± 15%). The resulting p-value of 0.0006 can be considered as evidence that there are differences between the group means.

[0023] FIG. 8 displays a simulated performance of the intercept bias data. Panel A shows results from experiments using 0.00% Tween® 20. Panel B shows results from experiments using 0.01%, 0.02%, and 0.05% Tween® 20. The numbers and vertical error bars represent the confidence intervals for intercept bias at each factor setting. The horizontal lines indicate the lower and upper specification limits (LSL and USL, respectively) which correspond to the ATP criteria for an intercept bias of ± 15%. Simulated defect rates relative to ATP criteria are 96.6% and 0.0% for 0.00% and 0.02% Tween® 20, respectively. Residual error used to simulate expected values for intercept bias (histogram) simulations were calculated via using restrictedmaximum likelihood (REML) analysis and represent standard deviations of 7.04 (panel A) and 0.60 (panel B).

[0024] FIG. 9 displays linearity demonstrated by two different analysts on different days with different instruments and different column lots. R2for both analysts 1 and 2 were 1.00. Linear equations for the total integrated area (pV*sec) were 3122*X + 7780 and 3012*X + 4770, for analyst 1 and 2, respectively.

[0025] FIG. 10 shows a diagram illustrating the theory and mechanism of Protein A (Pro A) affinity chromatography for monoclonal antibody (mAb) titer analysis. The sample containing mAb (a Y-shaped molecule) is injected onto a column packed with ProA resin. The mAb binds to the ProA ligand (represented on the resin) immobilized on the resin via its Fc region, while impurities and other sample components flow through the column unbound. Bound mAb is then eluted from the column by switching to an elution buffer that disrupts the ProA-mAb interaction. Finally, the column is re-equilibrated with the initial binding buffer in preparation for the next sample injection. The method desired outcomes include: the mAb peak being well separated from the impurities peak (HCPs, cell culture supernatant, etc.); High-throughput and robust titer assay to support process development groups, column performance / reproducibility, and the applicability of final optimized method for a variety of mAb products.

[0026] FIG. 11 displays a representative calibration curve for mAb titer quantitation by ProA affinity chromatography. The calibration curve is generated by plotting the total integrated peak area (in pV.scc) as a function of known mAb concentration (in g / mL) for a series of calibration standards spanning a range of 200 to 2000 ug / mL, as detailed in the accompanying table. The data follow a linear relationship, with the best fit line equation ofy = 7780 + 3122X and a coefficient of determination (R2) of 1.00, indicating a strong linear correlation between mAb concentration and peak area across this range. Unknown sample mAb titers can be interpolated from this calibration curve based on their measured peak areas. The mobile phase used to dilute the standards is noted.

[0027] FIG. 12 shows a comparison of the distribution of percent deviation (% deviation) from the nominal or expected value for calibration curve data generated with and without Tween 20 surfactant added to the mobile phase A (MPA) diluent used to prepare the standards across several experiments, representing conditions without Tween 20 surfactant. The solid curve represents the overall distribution of all the data combined, while the dashed curve shows the within-experiment distribution. Without Tween 20, the overall distribution is negatively skewed, with a mean percent deviation around -15% and some values exceeding -20%. Thisindicates a systematic negative bias, likely due to mAb losses from adsorption to vial surfaces. In contrast, the within-experiment variability is more symmetrically distributed around 0% deviation and is much narrower, suggesting the bias is consistent within a given experiment but varies between experiments, possibly due to different degrees of adsorption. The accuracy should be established across the reportable range of an analytical procedure and is typically demonstrated through comparison of the measured results with an expected value. The accuracy should be demonstrated under regular test conditions of the analytical procedure (e.g., in the presence of sample matrix and using described sample preparation steps).

[0028] FIG. 13 displays calibration curves for mAb titer analysis prepared by diluting the mAb standard in MPA without Tween 20 and storing aliquots in either plastic or glass HPLC vials prior to analysis. In both cases, the total peak area (in pVsec) is plotted against the nominal mAb concentration (in pg / mL). Although the data follow a linear relationship over the full concentration range (R2 > 0.99), the y-intercepts of the best fit lines are substantially negative, deviating from the theoretically expected y-intercept of 0. Specifically, the y-intercepts are - 121691 and -127125 for plastic and glass vials respectively. This indicates a negative bias in the measured peak areas compared to the nominal concentrations, consistent with mAb losses due to adsorption to the vial surfaces. The similarity in the y-intercepts for both vial materials suggests comparable degrees of adsorption in plastic and glass vials.

[0029] FIG. 14 shows a comparison of calibration curves for mAb titer analysis generated using three different mAb standard dilution and sample loading approaches: 1) preparing the standards in MPA without Tween 20 at the nominal concentrations and storing aliquots in either plastic or glass HPLC vials; 2) preparing a single concentrated mAb standard in MPA without Tween 20 and loading different volumes of this standard to achieve the target mass of mAb loaded on-column (variable volume injection approach). The total peak area (in pVsec) is plotted against the nominal mAb concentration (in pg / mL) for approaches 1, 2, and 3. As seen previously, the calibration curves generated using approach 1 show substantial negative deviations in the y-intercepts from the theoretical y-intercept of 0, indicating losses of mAb due to adsorption. In contrast, the variable volume injection approach (star symbols) yields a calibration curve with a y-intercept of -42051, much closer to 0, suggesting minimal mAb losses are occurring when the standards are prepared and loaded in this manner. This supports the hypothesis that mAb adsorption is occurring when standards are diluted to lower concentrations and stored in HPLC vials prior to analysis, and that this adsorptive loss can be mitigated by keeping the mAb at a high concentration and varying the injection volume instead.

[0030] FIG. 15 shows the percent relative deviation (%RD) of peak areas plotted against the calibration curve concentration ranging from 200 to 2000 pg / mL for various vial types, representing conditions without added Tween. The vial types shown include smooth glass vial (exp. 7), smooth plastic vial (exp. 7), smooth glass vial (exp. 6), smooth plastic vial (exp. 2), smooth glass vial (exp. 2), smooth thermo glass vial (exp. 6), smooth plastic vial (exp. 3), and smooth glass vial (exp. 3). The %RD values range from approximately -10% to 2%, with the smooth plastic vial (exp. 2) showing the largest negative deviation and the smooth glass vial (exp. 7) showing the largest positive deviation. A concentration dependent trend was observed with decreasing calibration curve concentrations. The graph illustrates the variability in peak area measurements across different vial types and concentrations when Tween is not present.

[0031] FIG. 16 displays the percent relative deviation (%RD) of peak areas plotted against the calibration curve concentration ranging from 200 to 1000 pg / mL. It displays the %RD of peak areas for smooth glass vial (exp. 7), smooth plastic vial (exp. 7), smooth glass vial (exp. 6), smooth plastic vial (exp. 2), smooth glass vial (exp. 2), smooth thermo glass vial (exp. 6), smooth plastic vial (exp. 3), smooth glass vial (exp. 3), and smooth (BSA). BSA (bovine serum albumin) was spiked into calibration curve. The %RD values span from around -8% to 2%, with the smooth plastic vial (exp. 2) exhibiting the most negative deviation and the smooth glass vial (exp. 7) showing the highest positive deviation. This graph allows for a more detailed view of the variability in peak area measurements at lower concentrations. Peak areas obtained from variable volume injection were used as the ‘nominal’ values to calculate %RD.

[0032] FIG. 17 shows the adsorption of a 0.01 mg / mL solution of mAb-1 to a glass surface (black line), investigated using TIRF, followed by the introduction of polysorbate (arrow) in PBS pH 7.4 as follows: Tween 20, 0.05 mM (half dash half dotted line) and 1 mM (half half dash line); Tween 80, 5 pM (dotted line) and 1 mM (dash line). The graph indicates that the addition of polysorbate reduces the adsorption of mAb-1 to the glass surface, with higher concentrations of polysorbate resulting in greater reduction of adsorption. Excipients are often added to protein formulations for stabilization and reducing aggregation and surface adsorption. Polysorbates compete with the protein for an interface and adsorb to exposed hydrophobic patches on the protein surface. Non-ionic surfactants including polysorbate 20 and 80 (Tween® 20 and 80) are commonly used excipients.

[0033] FIG. 18 shows the %RD of peak areas for various vial types across a wide calibration curve concentration range from 200 to 2000 pg / mL. The vial types include smooth glass vial (exp. 7), smooth plastic vial (exp. 7), smooth glass vial (exp. 6), smooth plastic vial(exp. 2), smooth glass vial (exp. 2), smooth plastic vial (exp. 3), smooth glass vial (exp. 3), smooth thermo glass vial (exp. 6), smooth plastic vial (0.02% tween), smooth glass vial (0.02% tween), and smooth glass vial (exp. 5). The %RD values range from approximately -8% to 2%, with the smooth plastic vial (exp. 2) showing the largest negative deviation and the smooth glass vial (exp. 7) exhibiting the highest positive deviation. Lines corresponding to conditions where Tween was added to the mobile phase (0.05%, 0.02%, 0.01%) are generally grouped in the upper portion of the data spread, while lines representing conditions where no Tween was added are generally in the lower portion. The addition of 0.02% Tween reduced the variability in peak area measurements for both smooth plastic and glass vials compared to other vial types. Peak areas obtained from variable volume injection were used as the ‘nominal’ values to calculate %RD.

[0034] FIG. 19 shows the impact of Tween 20 addition and assay design on accuracy metrics. Histograms for percent deviation (before Tween addition) and percent recovery (after Tween addition) illustrate differences in overall assay and within-assay performance. The overall assay has a wider spread in both percent deviation and percent recovery compared to the within-assay performance. The tables on the bottom provide the nonconformance data, including the portion observed and expected for Below LSL and Above USL. The image highlights the differences in assay performance between the overall assay and within-assay, as well as the nonconformance observed and expected for the lower and upper specification limits.

[0035] FIG. 20 shows a comparison of the Lower Specification Limit (LSL) and Upper Specification Limit (USL) recovery percentages. The upper histogram indicates the LSL is recovered between 95-100%, while the USL falls just above 100%. The lower histogram shows both the LSL and USL around 100% recovery. Nonconformance tables are shown on the bottom of each histogram. The top table shows 7.69% of data falls outside the total limits, with 0% below the LSL and 7.69% above the USL. The bottom table indicates 0% nonconformance, with all data falling within the limits.

[0036] FIG. 21 shows the evaluation of the main effects and determination whether there are any active effects among the parameters evaluated that include MPA salts, MPA pH, MPB salts, and MPB pH.

[0037] FIG. 22 shows the result of the robustness analysis using a prediction plot of the actual vs. predicted total peak area of Clarified Harvest. The effect summary table shows 4 'MPA' metrics: MPA salts, MPA pH, MPB salts, and MPB pH. No active effects observed, method is considered robust across the method operatable region (MODR).

[0038] FIG. 23 shows the slope, intercept, correlation coefficient, and integrated area for different concentration levels. The regression plot shows total integrated area increasing linearly with total concentration, with a correlation coefficient of 0.999991. The residual normal quantile plot shows the residuals fall close to a straight line, indicating a normal distribution. The target criteria for linearity expect that analytical procedure should meet linearity (r2> 0.99).

[0039] FIG. 24 shows the sample variability chart with the total area (pV sec) for different sample preparations / injections across multiple days / analysts / instruments / columns. Data points are randomly distributed about the group means. The average area is 3,092,708 uNsec. The table lists the IDs for the different days / analysts / instruments / columns used.DETAILED DESCRIPTION OF THE INVENTION

[0040] Although the present invention and its advantages have been described in detail, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined in the appended claims.

[0041] The present invention will be further illustrated in the following Examples which are given for illustration purposes only and are not intended to limit the invention in any way.

[0042] The present disclosure provides improved methods for the analysis of monoclonal antibody (mAb) titer using Protein A affinity High-Performance Liquid Chromatography (HPLC). These methods address inaccuracies encountered in conventional techniques due to the non-specific adsorption of mAbs onto sample vial surfaces. The methods featured herein utilize the addition of a surfactant to the sample diluent to mitigate such adsorption, thereby enhancing the accuracy, linearity, and robustness of the titer assay.

[0043] Protein A affinity HPLC is a standard technique for quantifying mAbs, leveraging the specific interaction between Protein A and the Pc region of immunoglobulins, such as IgG. Conventional methods typically involve binding the mAb from a sample (e.g., harvested cell culture fluid or a purified preparation) to a Protein A stationary phase, washing away unbound impurities, and subsequently eluting the bound mAb. The amount of eluted mAb is quantified, often by UV absorbance detection, and the concentration (titer) is determined by comparison to a calibration curve generated using standards of known mAb concentration.

[0044] A challenge observed with conventional Protein A HPLC titer methods is the potential for analyte loss due to non-specific binding or adsorption of the mAb to surfaces within the analytical system, particularly the sample vials used in autosamplers. Thisadsorption phenomenon can occur with various vial materials, including commonly used glass and polypropylene. Analyte loss due to adsorption is often more pronounced at lower mAh concentrations, such as those used to prepare the lower points of a calibration curve. This can lead to systematic errors, manifesting as inaccurate calibration curves (e.g., exhibiting significant negative y-intercept bias) and reduced accuracy in the quantification of mAb titer, especially for samples containing lower concentrations of the antibody. Such inaccuracies can negatively impact process monitoring and control in biopharmaceutical manufacturing.

[0045] The methods disclosed herein overcome the aforementioned limitations by incorporating a surfactant into the sample preparation process. In certain aspects, the method involves diluting the mAb calibration standards and unknown samples in a buffer that contains a suitable surfactant prior to loading them into sample vials for HPLC analysis. The presence of the surfactant minimizes the non-specific adsorption of the mAb to the vial surfaces. This mitigation of analyte loss results in more accurate calibration standards, improved linearity of the calibration curve (characterized, for example, by a y-intercept closer to the theoretical value of zero), and consequently, more reliable and accurate titer measurements across the analytical range.

[0046] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0047] The terms comprises, comprised, comprising and the like can have the meaning attributed to it in U.S. Patent law; e.g., they can mean includes, included, including, and the like. The terms consisting essentially of and consists essentially of have the meaning ascribed to them in U.S. Patent.

[0048] The articles a and an are used herein to refer to one or to more than one (i.e., to at least one) of the grammatical object of the article. By way of example, an element means one element or more than one element.

[0049] The term about or approximately means within an acceptable error range for the particular value as determined by one of ordinary skill in the art, which will depend in part on how the value is measured or determined, i.e., the limitations of the measurement system. For example, about can mean within an acceptable standard deviation, per the practice in the art. Alternatively, about means, in some embodiments, a range of up to + / -20%, up to + / - 10%, up to + / -5%, or up to + / -!% of a given value.

[0050] The term antibody as used herein refers to an immunoglobulin molecule that recognizes and specifically binds to a target. In certain aspects, antibodies are comprised of four polypeptide chains, two heavy (H) chains and two light (L) chains inter-connected by disulfide bonds. Each heavy chain comprises a heavy chain variable region (VH) and a heavy chain constant region (CH). Each light chain comprises a light chain variable region (VL) and a light chain constant region (CL). The VH and VL regions can be further subdivided into regions of hypervariability, termed complementarity determining regions (CDRs), interspersed with regions that are more conserved, termed framework regions (FR). An antibody is, in some embodiments, a full-length antibody or an antigen-binding fragment thereof that retains the ability to specifically bind to an antigen.

[0051] The term monoclonal antibody or mAb as used herein refers to an antibody obtained from a population of substantially homogeneous antibodies, such that the individual antibodies comprising the population are identical except for possible naturally occurring mutations that may be present in minor amounts. Monoclonal antibodies are highly specific, being directed against a single antigenic site. Examples include, but are not limited to, human antibodies, humanized antibodies, and chimeric antibodies. In some aspects, the monoclonal antibody is an IgG, such as an IgGl, IgG2, IgG3, or IgG4 subtype.

[0052] The term Fc region refers to the C-terminal region of an immunoglobulin heavy chain that interacts with Fc receptors and complement. It typically comprises the CH2 and CH3 domains. Protein A binds primarily to the Fc region of certain immunoglobulins, particularly IgG.

[0053] The term “relatively hydrophobic mAb” as used herein, refers to a monoclonal antibody characterized by a higher proportion, presence, or surface exposure of hydrophobic amino acid residues compared to hydrophilic amino acid residues, particularly within regions accessible for interaction with surfaces, such as variable domains or complementarity determining regions (CDRs). In some aspects, hydrophobic amino acid residues include Alanine (A), Valine (V), Isoleucine (I), Leucine (L), Methionine (M), Phenylalanine (F), Tryptophan (W), Proline (P), and Glycine (G). In certain aspects, hydrophilic amino acid residues include Arginine (R), Asparagine (N), Aspartate (D), Glutamine (Q), Glutamate (E), Lysine (K), Serine (S), Threonine (T), and Histidine (H). In some embodiments, relatively hydrophobic mAbs exhibit a propensity for non-specific adsorption to surfaces, such as glass or polypropylene sample vials, under certain buffer or concentration conditions.

[0054] The term mixture as used herein comprises a substance of interest (such as a monoclonal antibody) and one or more contaminants or impurities. In some aspects, the mixture is obtained from or produced by living cells that express the substance of interest (either naturally or recombinantly). Such mixtures include, for example, cell culture fluid, harvested cell culture fluid, cell lysates, clarified bulk (e.g., clarified cell culture supernatant), or partially purified preparations.

[0055] The term impurity or contaminant refers to any undesired component or compound within a mixture containing the substance of interest. Examples include, but are not limited to, host cell proteins (HCPs), host cell nucleic acids (e.g., DNA), host cell metabolites, host cell constitutive proteins, endotoxins, viruses, product-related substances such as aggregates or fragments, lipids, media additives, media derivatives, or leachates from process materials.

[0056] The term chromatography as used herein refers to the process by which a solute of interest, e.g., a protein of interest, in a mixture is separated from other solutes in the mixture by percolation of the mixture through an adsorbent or stationary phase.

[0057] The term High-Performance Liquid Chromatography or HPLC as used herein refers to a type of column chromatography generally used in analytical chemistry to separate, identify, and quantify components in a mixture. It relies on pumps to pass a pressurized liquid solvent containing the sample mixture through a column filled with a solid adsorbent material.

[0058] The term Affinity Chromatography as used herein refers to a type of chromatography that separates molecules based on a highly specific interaction between the molecule of interest and an immobilized ligand (bound to the stationary phase), such as the interaction between an antibody's Fc region and Protein A.

[0059] The term Protein A or ProA as used herein refers to a surface protein originally found in the cell wall of the bacterium Staphylococcus aureus, or a recombinant or synthetic variant thereof, which has high affinity for the Fc region of certain immunoglobulins, particularly IgG. Protein A Affinity Chromatography utilizes Protein A immobilized on a stationary phase to capture and purify such immunoglobulins.

[0060] The term stationary phase as used herein refers to the solid adsorbent material in a chromatography column to which target molecules or impurities bind or interact. Examples relevant herein include Protein A immobilized on a support matrix, such as a resin (e.g., agarose beads) or a monolith. Resin refers to the solid support material, typically porous beads, used in column chromatography. A monolith is a continuous porous structure used as a stationaryphase. The Bio-Monolith Recombinant Protein A refers to a specific commercially available monolithic column functionalized with Protein A.

[0061] The term mobile phase as used herein refers to the liquid solvent that moves the sample mixture through the chromatography column. In gradient elution, the composition of the mobile phase changes over time. In isocratic elution, the composition remains constant. MPA refers to Mobile Phase A and MPB refers to Mobile Phase B, typically representing different buffer compositions mixed in varying proportions during a gradient elution.

[0062] The term gradient elution as used herein, refers to a chromatography procedure where the composition of the mobile phase is changed during the separation process, for example, by varying the concentration of salts or changing the pH.

[0063] The term isocratic elution as used herein, refers to a chromatography procedure where the composition of the mobile phase remains constant throughout the separation.

[0064] The term titer as used herein, refers to the concentration of a substance, such as a monoclonal antibody, in a solution (e.g., a cell culture supernatant or purified sample). Accurate titer measurement is relevant in biopharmaceutical production.

[0065] The term Host Cell Protein or HCP as used herein, refers to proteins originating from the host cell line (e.g., CHO cells) used for recombinant production of the desired protein (e.g., mAb) that remain in the product mixture. These are considered impurities.

[0066] The term Harvested Cell Culture Fluid or HCCF as used herein, refers to the liquid medium collected from a cell culture after the cells have produced the protein of interest, typically containing the protein along with cells, cell debris, and other impurities like HCPs.

[0067] The term surfactant as used herein, refers to a compound that lowers the surface tension between two liquids, between a gas and a liquid, or between a liquid and a solid. Surfactants may act as detergents, wetting agents, emulsifiers, foaming agents, and dispersants. Polysorbate 20 (commercially known as Tween® 20) is a specific non-ionic surfactant commonly used in biopharmaceutical formulations and analytical methods.

[0068] The term adsorption or non-specific binding as used herein, refers to the adhesion of molecules (like proteins) from a solution onto a surface (like a vial wall or chromatography resin). This can be mediated by interactions such as electrostatic or hydrophobic interactions. In the context featured herein, it refers particularly to the undesirable loss of mAb to vial surfaces, impacting analytical accuracy.

[0069] The term excipient as used herein, refers to an inactive substance formulated alongside the active ingredient of a medication or used in an analytical process, included forthe purpose of stabilization, bulking, or modifying properties like adsorption. Tween® 20 is used as an excipient in the methods disclosed herein.

[0070] The term standard curve or calibration curve as used herein, refers to a graph plotting the measured response (e.g., HPLC peak area) versus known concentrations of a substance. It is used to determine the concentration of the substance in unknown samples by interpolation.

[0071] The term linear regression as used herein, refers to a statistical method for modeling the relationship between a dependent variable (e.g., peak area) and an independent variable (e.g., concentration) by fitting a linear equation to the observed data.

[0072] The term y-intercept as used herein, refers to the point where the fitted regression line crosses the y-axis (i.e., the predicted response when the concentration is zero). In the context disclosed herein, a non-zero y-intercept in a calibration curve can indicate systematic bias. Intercept bias quantifies this deviation.

[0073] The term peak area or total integrated peak area as used herein, refers to the area under a peak in a chromatogram, typically measured in units like pV * sec which is generally proportional to the amount of the substance detected.

[0074] The term accuracy as used herein, refers to the closeness of agreement between a measured value and an accepted reference value or the true value. In the context disclosed herein, it is often expressed as Percent Recovery (% Recovery) of a known amount of spiked analyte or as Percent Relative Deviation (%RD) from a nominal value.

[0075] The term precision as used herein, refers to the closeness of agreement between independent test results obtained under stipulated conditions. It reflects the random error of a method. It is often expressed as Percent Relative Standard Deviation (%RSD). Repeatability as used herein, refers to precision under conditions where independent test results are obtained with the same method on identical test items in the same laboratory by the same operator using the same equipment within short intervals of time. Intermediate Precision expresses within- laboratory variations (different days, analysts, equipment, etc.).

[0076] The term linearity as used herein, refers to the ability of an analytical procedure to obtain test results which are directly proportional to the concentration (amount) of analyte in the sample within a given range. It is often assessed by the coefficient of determination (r2or R2) of the calibration curve.

[0077] The term specificity as used herein, refers to the ability of an analytical procedure to assess unequivocally the analyte in the presence of components which may be expected tobe present (e.g., impurities, matrix components). It can be assessed by analyzing blank matrices or placebos. Percent Interference (% Interference) may quantify the signal from interfering components relative to the analyte signal at a specific level (e.g., the lowest standard).

[0078] The term working range or analytical range as used herein, refers to the interval between the upper and lower concentration (amount) of analyte in the sample for which the analytical procedure has a suitable level of precision, accuracy, and linearity. Column load as used herein, refers to the mass of analyte applied to the chromatography column.

[0079] The term Analytical Quality by Design or AQbD as used herein, refers to a systematic, science- and risk-based approach to analytical method development that begins with predefined objectives and emphasizes product and process understanding and process control.

[0080] The term Analytical Target Profile or ATP, as used herein, defines the performance requirements for an analytical method to ensure it is fit for its intended purpose.

[0081] The term Total Analytical Error or TAE as used herein, refers to the combination of systematic error (bias / accuracy) and random error (precision) of an analytical method.

[0082] The term Risk Assessment in the context of AQbD as used herein, refers to the systematic evaluation of potential sources of variation in an analytical procedure and their impact on method performance and data quality. Tools like Ishikawa (fishbone) diagrams and CNX (Controlled / Noise / Experimental) analysis may be used.

[0083] The term Design of Experiments or DoE as used herein, refers to a structured, statistical approach to experimentation used to efficiently evaluate the effect of multiple factors (variables) and their interactions on a response. D-optimal design is a type of experimental design optimized for fitting a specified model.

[0084] The term Robustness as used herein, refers to the capacity of an analytical procedure to remain unaffected by small, but deliberate variations in method parameters and provides an indication of its reliability during normal usage. The Method Operable Design Region or MODR defines the multi-dimensional space of method parameters within which the method has been shown to perform reliably.

[0085] The term International Conference on Harmonisation or ICH as used herein, refers to an international body that develops guidelines for pharmaceutical development and regulation. Guidelines such as ICH Q2 (Validation of Analytical Procedures) and ICH Q14 (Analytical Procedure Development) are relevant herein.

[0086] In certain embodiments, the method for analyzing the titer of a mAb comprises preparing mAb standards and samples in a sample diluent containing a non-ionic surfactant, placing these preparations into sample vials, and analyzing them using Protein A affinity HPLC.

[0087] In some aspects, the surfactant employed is a non-ionic surfactant. Non-limiting examples of suitable non-ionic surfactants include polysorbates. In certain embodiments, the non-ionic surfactant is Polysorbate 20 (Tween® 20). In other embodiments, Polysorbate 80 is used.

[0088] The concentration of the surfactant in the sample diluent is selected to be effective in reducing or preventing significant mAb adsorption to the vial surfaces under the conditions of the analysis. In some aspects, the concentration of the surfactant, such as Polysorbate 20, is in the range of about 0.01% v / v to about 0.1% v / v. In certain embodiments, the concentration is in the range of about 0.02% v / v to about 0.05% v / v. In a particular embodiment, the concentration of Polysorbate 20 used is about 0.02% v / v. The optimal surfactant concentration can be determined through routine experimentation based on the specific mAb, analytical system, and vial materials used.

[0089] The Protein A affinity HPLC analysis is performed using conditions suitable for mAb titer quantification.

[0090] In certain embodiments, the stationary phase comprises Protein A immobilized on a support. The support may be, for example, porous beads (resin) or a monolith. In one embodiment, the stationary phase is a Protein A monolith column, such as a Bio-Monolith Recombinant Protein A column (e.g., having dimensions of about 4.95 x 5.2 mm). An inline filter (e.g., 0.2 pm pore size) may be used upstream of the analytical column.

[0091] In some aspects, the analysis utilizes a mobile phase system consisting of at least two buffers, Mobile Phase A (MPA) and Mobile Phase B (MPB), used in a gradient elution mode. In certain embodiments, MPA serves as the equilibration and initial binding buffer and also as the sample diluent containing the surfactant. In one aspect, MPA comprises an aqueous buffered solution at a substantially neutral pH, containing salts. For example, MPA comprises about 50 mM sodium phosphate and about 150 mM sodium chloride, at a pH of about 7.0, further comprising the selected surfactant (e.g., about 0.02% Polysorbate 20). In certain embodiments, MPB serves as the elution buffer. In one aspect, MPB comprises an aqueous buffer at an acidic pH, containing salts. For example, MPB comprises about 10 mM phosphoric acid and about 150 mM sodium chloride, at a pH of about 2.2.

[0092] The separation is achieved by applying a gradient program that modifies the proportion of MPA and MPB over time. An exemplary gradient may involve an initial isocratic hold in high %MPA, followed by a rapid switch or steep gradient to high %MPB to elute the bound mAb, and subsequent re-equilibration with high %MPA. In one embodiment, the gradient involves steps including equilibration / loading at 90% MPA / 10% MPB, elution at 0% MPA / 100% MPB, and re-equilibration at 90% MPA / 10% MPB.

[0093] The flow rate is maintained at a level suitable for the column dimensions and particle / pore size. In one embodiment, using a 4.95 x 5.2 mm monolith column, the flow rate is about 1.5 mL / min.

[0094] The analysis is typically performed at a controlled temperature. In some aspects, the column compartment temperature is maintained at about ambient temperature, for instance, about 23°C.

[0095] Detection of the eluted mAb is typically performed using UV absorbance. In certain embodiments, detection is at a wavelength of about 280 nm. A specific detector bandwidth (e.g., about 4 nm) and data acquisition rate or sampling frequency (e.g., about 5 Hz) may be employed.

[0096] The methods disclosed herein may be developed and characterized using principles of Analytical Quality by Design (AQbD). This approach involves defining an Analytical Target Profile (ATP) that outlines the required performance characteristics (e.g., accuracy, precision, linearity, range, specificity, robustness) for the titer method to be fit for its intended purpose.

[0097] Risk assessment methodologies may be employed to identify factors that could potentially impact method performance. Non-specific adsorption of the analyte to vial surfaces is one such factor that can be identified as a risk impacting accuracy and linearity.

[0098] Experimental studies, potentially including Design of Experiments (DoE), can be used to investigate the effect of adding surfactants, optimize the surfactant type and concentration, and define appropriate operating conditions. Such studies can demonstrate the effectiveness of the surfactant in mitigating adsorption-related bias, for example, by comparing calibration curves generated using different sample preparation or injection strategies (e.g., variable concentration vs. variable volume) with and without the surfactant.

[0099] Method robustness can be assessed by evaluating the impact of small, deliberate variations in method parameters (e.g., buffer pH, buffer component concentrations, flow rate, temperature) on the analytical results. This helps define a Method Operable Design Region (MODR) within which the method consistently delivers reliable results. The inclusion of thesurfactant in the sample diluent contributes to the overall robustness of the method against variations related to sample handling and vial interactions.

[0100] Performance verification or validation studies are conducted according to established guidelines (e.g., ICH Q2) to confirm that the finalized method meets the predefined ATP criteria for accuracy, precision, linearity, specificity, range, and robustness. Successful verification demonstrates the method's suitability for routine use in quantifying mAb titer.

[0101] The methods disclosed herein are applicable to the titer analysis of various monoclonal antibodies, particularly those of the IgG isotype, which bind to Protein A. They can be used for analyzing samples from different stages of the manufacturing process, including cell culture supernatants (e.g., HCCF) and purified antibody preparations. The methods are compatible with standard HPLC equipment and common sample vial materials like glass and polypropylene.

[0102] The primary advantage of incorporating a surfactant as disclosed herein is the significant improvement in analytical accuracy and reliability, particularly at lower analyte concentrations, by mitigating analyte loss due to adsorption. This leads to more linear calibration curves with reduced intercept bias, enhancing confidence in the obtained titer results. The resulting methods are robust and suitable for routine application in a biopharmaceutical quality control or process development setting. Furthermore, addressing the adsorption issue analytically can contribute to more efficient overall process development timelines.

[0103] Implementation of the methods typically involves standard analytical grade reagents and laboratory equipment. Reagents may include buffer salts (e.g., sodium phosphate monobasic monohydrate, sodium phosphate dibasic heptahydrate), salts for adjusting ionic strength (e.g., sodium chloride), acids / bases for pH adjustment (e.g., phosphoric acid, hydrochloric acid, sodium hydroxide), the selected non-ionic surfactant (e.g., Polysorbate 20), and high-purity water.

[0104] Chromatography consumables include the Protein A affinity column (e.g., resinbased or monolithic), appropriate inline filters, and HPLC sample vials (glass or polypropylene) with suitable closures. Low retention microcentrifuge tubes may be beneficial for preparing initial dilutions.

[0105] The HPLC instrumentation includes a pump capable of delivering accurate gradients (e.g., a quaternary pump), an autosampler capable of precise injections from vials, a temperature-controlled column compartment, and a UV detector.

[0106] Data handling involves a chromatography data system for instrument control, data acquisition, and peak integration, potentially supplemented by statistical software for experimental design and data analysis.

[0107] HPLC encompasses various separation modes based on the interaction between the analyte, the stationary phase, and the mobile phase. Besides affinity chromatography, other common HPLC modes used in biopharmaceutical analysis include, but are not limited to, Reversed-Phase HPLC (RP-HPLC), Ion-Exchange Chromatography (IEX-HPLC), Size- Exclusion Chromatography (SEC-HPLC), and Hydrophobic Interaction Chromatography (HIC-HPLC). In some aspects, RP-HPLC separates molecules based on hydrophobicity, typically using a non-polar stationary phase and a polar mobile phase, often with an organic modifier gradient. In some aspects, IEX-HPLC separates molecules based on their net surface charge using stationary phases with charged functional groups (anion or cation exchangers) and mobile phases where elution is controlled by ionic strength or pH gradients. In some aspects, SEC-HPLC separates molecules based on their hydrodynamic size, using porous stationary phases where larger molecules elute earlier than smaller molecules that penetrate the pores. In some aspects, HIC-HPLC separates molecules based on their surface hydrophobicity using stationary phases with hydrophobic ligands (e.g., phenyl, butyl, octyl) and mobile phases typically involving high salt concentrations for binding and decreasing salt gradients for elution.

[0108] The methods disclosed herein are suitable for use in quality control (QC) testing during the manufacturing of pharmaceutical compositions containing monoclonal antibodies. By providing accurate and robust titer measurements, the methods support in-process control, batch release testing, and stability studies. For example, the method can be used to compare the titer of a production batch sample against a reference standard or pre-defined specification limits.

[0109] A method of quality control of a pharmaceutical composition comprising a monoclonal antibody, comprises, in some aspects: a. separating a monoclonal antibody from a mixture comprising one or more additional impurities by the methods described herein; b. comparing the measured titer or chromatographic profile of the separated monoclonal antibody with that of a reference monoclonal antibody or a specification; and c. determining whether the pharmaceutical composition meets a quality control standard based on said comparison.ExamplesExample 1: A case study application of AQbD to the re-development and validation of an affinity chromatography analytical procedure for mAh titer quantitation.

[0110] Protein A (ProA) high-performance liquid chromatography (HPLC) is a common analytical procedure for measuring monoclonal antibody (mAb) titers due to its high specificity and efficiency. Accurate and reliable results of this procedure are imperative, as the quantitation of the total mAb present for in-process samples directly impacts downstream purification steps related to the removal of process-related impurities. This study aimed to improve a platform ProA HPLC analytical procedure which was previously developed using traditional approaches and was not always reliable. By retrospectively applying Analytical Quality by Design (AQbD) principles and statistical assessments of performance, a bias in the calibration standard due to protein-adsorption to common sample vial materials was identified. The inclusion of Tween® 20 into the mobile phase used as sample diluent was optimized to ensure procedure performance and improve analytical range. The resulting procedure robustness was evaluated using Design of Experiment (DoE) approaches and performance was verified against Analytical Target Profile (ATP) criteria as recommended by regulatory agencies. Application of this new platform procedure has since reduced development timelines for new mAb products by 50% and allowed for accurate titer determination to support early process decisions without requiring extensive product- specific analytical procedure development. This work demonstrates the utility and relative ease of adopting AQbD concepts, even for established procedures, and supporting them with a lifecycle approach to managing procedure performance.

[0111] Monoclonal antibodies (mAbs) are an important class of biopharmaceutical products due to their widespread use in treating various cancers, autoimmune diseases, and other clinical applications [1]. They are typically produced using mammalian cell line expression systems which have been genetically engineered to produce mAbs under controlled growth conditions in bioreactors or flasks [1]. Accurate measurement of mAb concentration, or titer, is allows for controlling product quality during production, ultimately ensuring the final product meets specifications [2]. Furthermore, titer values are used to inform strategic decisions throughout the development and manufacturing of these products, including clonal selection, feed strategies to increase cell line expression, process scale-up conditions, and the sizing of purification columns following bioreactor harvest [3]. Therefore, analyticalprocedures for titer analysis should be rapid, robust, and require minimal sample pretreatment prior to analysis to meet

[0112] time- sensitive decision deadlines on a batch-to-batch basis [4].

[0113] Protein A (ProA) affinity high-performance liquid chromatography (HPLC) is an analytical technique for measuring mAb titers that is based on the specific binding of antibodies to Staphylococcus aureus ProA [5]. ProA binds to the crystallizable fragment (Fc) region of immunoglobulins (Ig), such as IgGs, with high affinity and specificity [6]. In a ProA titer analytical procedure, mAbs selectively bind to ProA ligands covalently immobilized onto a chromatographic support, such as a resin or monolith, while impurities and other cell culture components are removed [7]. The bound mAb can then be eluted from the ProA ligand by altering the ionic strength, pH, or presence of denaturant in the mobile phase [7]. Titer quantitation occurs by comparing the area of the sample elution peak to a calibration curve comprised of standards with known concentrations [8]. This process is summarized in Fig. 1.

[0114]

[0115] Recent regulatory guidelines have been authored which place specific emphasis on the development and maintenance of analytical procedures according to Analytical Quality by Design (AQbD) principles [9]. These principles can extend and improve on the analytical method development performed by experienced analytical scientists by supporting a lifecycle approach in the development of fit-for-purpose analytical procedures to assess and ensure the quality and reliability of biopharmaceutical products

[0010] . This approach involves defining the procedure goals, conducting appropriate risk assessments, and implementing continuous improvement through control strategies [9]. AQbD offers several advantages over traditional approaches including improved understanding and control of the sources of variability in procedures as linked to the analytical target profile (ATP); more robust procedures leading to fewer failures; detection and proactive remediation of adverse trends; and a net reduction in resources needed to develop a suitable analytical procedure

[0013] .

[0116] An ATP consists of a set of specific performance characteristics that define the analytical measurement to ensure that the analytical procedure is fit-for-purpose [10,14]. The typical procedure performance parameters and their target values included within an ATP include specificity, working range, and total analytical error (TAE), which is inclusive of accuracy and precision

[0013] . By establishing an effective ATP early in the development process, optimization to meet those performance targets can ensure that the resulting analytical data are reliable during its routine use throughout its lifecycle [ 12, 14] . For procedures that wereestablished by traditional development practices, an ATP can be defined and applied retrospectively to interpret historical performance data and provide a foundation for any future analytical procedure updates and ensure continuous improvement [12,14]. Furthermore, through routine monitoring or the invention of new technologies, an ATP may be updated throughout the analytical lifecycle

[0010] .

[0117] Following the establishment of an ATP, risk assessments are conducted to provide a systematic evaluation of the potential risks resulting from sources of variation associated with an analytical procedure, including procedure parameters, materials, and personnel [11,15]. The objective is to identify these potential sources of variation and determine both the likelihood and relative impact of these risks on the quality of the final analytical results

[0013] . Using quality risk management (QRM) principles, risk assessment tools can then be deployed for quantitative and qualitative analysis, such as Ishikawa (fishbone) diagrams to identify contributing factors and heat maps to qualitatively represent the relative risk associated with factors that have low, moderate, or high impact on the performance of the procedure [10,11,16]. Factors with associated risks can then be categorized using a CNX risk analysis tool which categorizes parameters according to whether they will be controlled (C), treated as a source of random noise (N), or experimentally assessed and set to an optimal and robust range (X)

[0015] .

[0118] Design of Experiments (DoE) applications are fundamental to the QRM process and are becoming increasingly recommended by regulatory agencies for AQbD workflows [13,17]. Once factors have been identified for experimental investigation by risk assessment, DoE offers a structured approach to identifying and optimizing interactions between multiple factors, as shown by previous studies

[0018] , and to subsequently evaluating the robustness of the optimized analytical procedure as recommended by regulatory guidelines [13,17]. Robustness studies involve intentionally making small variations to the procedure parameters, similar to what may occur during the routine execution of the method. These development studies are necessary to ensure that the procedure is reliable and can routinely meet the expected performance criteria defined within the ATP [17,19]. DoE application to these studies provides enhanced understanding of procedure parameters, identification of interactions between procedure parameters, and quantitation of the impact of procedure parameter variations on the procedure’s performance.

[0119] Validation of the analytical procedure can also be conducted to determine whether the ATP requirements are met

[0010] . Outlined by the International Conference on Harmonization (ICH), typical characteristics which should be considered during validationinclude accuracy, precision, specificity, linearity, and working range

[0020] . The AQbD strategy offers a significant benefit by reducing the likelihood of failing to meet the validation criteria since the ATP is based on these same characteristics and is referenced throughout analytical procedure development efforts

[0010] .

[0120] Described herein is the use of an AQbD approach to improve a ProA titer procedure previously developed via traditional approaches which was used for platform application to mAbs. During application to a variety of mAb products, limited accuracy was observed at the bottom of the working range for some mAbs. A platform analytical procedure which can be utilized across multiple products without significant adjustments to measurement conditions was desired

[0017] . Platform improvements to the ProA titer procedure were implemented via a lifecycle approach informed by AQbD concepts. An ATP was retroactively defined, and upon application of quality risk management tools it became apparent that protein adsorption to commonly used HPLC vials was a potential risk to method performance. A development study was executed to identify an optimal level of Tween® 20 in the mobile phase A (MPA) to mitigate the issue, and DoE was applied to evaluate the procedure’s robustness. Once determined to be robust, verification of procedure performance was performed as recommended by the ICH.

[0121] The results of this study demonstrate the potential for AQbD to support method performance throughout the product lifecycle. In particular, the retrospective application of AQbD to existing procedures was used to identify limitations and guide re-optimization efforts to ensure accurate and robust measurements of mAb titer during routine applications. Additionally, this work highlights that pragmatic applications of AQbD need not be overly complicated or resource intense, supporting widespread adoption in the pharmaceutical industry for efficient lifecycle management.

[0122] Results and discussion

[0123] Analytical target profile and procedure risk assessment

[0124] An ATP (Table 4) was drafted to guide the re-development of a fit-for-purpose ProA platform procedure. The method was targeted for the determination of mAb titer in cell culture matrices to support quantitation of purified and in-process samples. To be considered fit-for-purpose, procedure accuracy, representing the mean % recovery of the procedure, should fall between 90-110% for all tested levels of the working range and procedure precision should fall within <10% RSD. These target criteria for accuracy and precision were chosen to align with traditional procedure validation criteria and correlate to a combined TAE of ±25%of the true value with at least 90% probability. FIG. 2 demonstrates how accuracy and precision relate to total analytical error.

[0125] Table 4. Analytical target profile (ATP) describing the performance criteria for aPro A titer analytical procedure as linked to the Critical Quality Attribute (CQA).Intended PurposeDetermination of the concentration (titer) of a mAbsolutions and cell culture media.Link to CQATiter determination may indirectly impact CQAs, but the analytical procedure should allow for the quantitation of the total mAb present within in-process samples as it impacts downstream purification steps related to process impurity removal.Performance Characteristics of the Reportable Titer ResultsCharacteristic Acceptance Criteria RationaleAnalytical procedure accuracy should fall Target criteria for accuracy and precision Accuracy between 90.0-110.0 % recovery for all were chosen as they correlate to a combined tested levels. _ total analytical error (TAE) of + 25 % of theAnalytical procedure precision should fall true value with at least 90 % probability. See Precision within < 10.0 % relative standard Fig. 2 for an example of how accuracy and deviation. precision relate to TAE.The % interference for the cell cultureAcceptance criteria ensures that sample media and mobile phase blank injectionsSpecificity matrices will not impact accuracy or TAE are each < 10% of the 200 pg / mL standard acceptance criteria. peak area. _The analytical procedure should meet linearity (r2> 0.99), bias of y-interceptAcceptance criteria ensures reliable titerWorking relative to lowest standard (± 15%), and quantitation across a working range Range accuracy and precision criteria across a spanning one order of magnitude. column load working range of 10 to 100

[0126] Following establishment of the ATP, possible sources of procedure risk and the specific factors that contribute to those risks were identified. To enhance understanding and visualize these risks, an Ishikawa diagram was constructed (Fig. 3). A secondary risk assessment was created to capture all factors and their associated risks, and each parameter was then assigned a risk score via heat map scoring (Fig. 4). Each risk was assigned a risk score of low, medium, or high based on its potential impact on the TAE of the reportable result, as linked to the ATP. This was done by utilizing extensive analytical procedure knowledge and the expertise of subject matter experts. The risks were then categorized into controlled (C), noise (N), or experimental (X) factors. Control strategies were identified for each risk according to their CNX classification. Risks identified to be of moderate and high impact to the TAE further informed the optimization, robustness, and procedure verification studies below. Of note, vial material and Tween® 20 concentration were high risk factors that guided analytical procedure optimization. Mobile phase compositions, including concentration andpH, were a reoccurring risk determined to have a moderate level impact, therefore these factors and their impact(s) on the analytical procedure were investigated in the robustness DoE in this work. Further, the factors that were determined to have a moderate risk and subsequent impact to the analytical procedure’s ability to meet the target ATP criteria, including performance variability between days within the same laboratory, analyst experience, sample concentration, sample and standard preparation, performance variability between sample preparation equipment, mobile phase lot, column lot, instrument manufacturer and model, and autoinjector precision were identified, and their impact was assessed via the performance verification study according to ICH recommendations.

[0127] Procedure

[0128] Prior to this work, a ProA titer procedure was developed as a site platform by traditional one-factor-at-a-time experiments, and subsequent routine application of the platform procedure to different molecules and upstream processes frequently exhibited variable negative y-intercept biases for the calibration regression, which was directly related to the inconsistent accuracy observed. Upon internal discussion with subject matter experts, it was suggested that these issues originated due to non-specific binding, and subsequent loss, of the protein at some step in the procedure for calibration standards. The most likely step which could contribute to protein loss via adsorption were identified as the vialing of standards and samples. Thus, initial experiments investigated standard curves prepared from serially diluted mAbs that were transferred into either glass or polypropylene vials at 200 pL. These vial materials are commonly used for routine method execution, and as such were used for comparative analysis. Process capability analysis (Fig. 5) of the obtained distribution for intercept bias values indicate that 100% of the observed data, and 96.6 % of the expected data based on normal distribution modeling, is below the lower specification limit (ESE). These data are summarized in Table 5. Thus, the ProA platform procedure was demonstrably not meeting performance criteria as defined by the new ATP. Interestingly, the goodness-of-fit test for these data fit to a normal distribution returned a Shapiro-Wilk p-value of 0.1071, which is above our selected significance criterion of 0.05. This indicates that there is no major discrepancy between vial types; storage in both glass and polypropylene vials indicated similar issues with intercept bias.

[0129] To further interrogate this observation, the %RD of peak areas for variable concentration standards and constant injection volume stored in either glass or polypropylene vials were compared relative to a fixed concentration calibration standard loaded onto thecolumn by variable volume injections (Fig. 6 (A)). Variable volume injection data was used as the nominal values for comparison as it was suspected that product loss to the vial was occurring when diluted to lower concentrations. Upon visual inspection of the data, large negative deviations were increasingly observed at lower prepared standard concentrations. This concentration-dependent phenomena supports the hypothesis that glass and polypropylene vials are adsorbing a fixed amount of protein which becomes increasingly apparent as the total concentration is lowered. These results also corroborate earlier observations in Fig. 6 (A) that both types of vials have similar adsorption issues, as the 95% confidence interval (CI) regions for each second-order polynomial fit are largely overlapping. It is also worth noting that adsorption may be occurring to other materials in the preparation procedure, including microcentrifuge tubes used for dilutions and pipette tips used to transfer the samples to the analysis vials.

[0130] There are several mechanisms that can contribute to protein-surface adsorption, including electrostatic and hydrophobic interactions [24,25]. In biological formulations, excipients such as surfactants are often added to reduce surface adsorption. Commonly used non-ionic surfactants include polysorbate 20 (commercially known as Tween® 20) for its ability to bind protein interfaces, thereby preferentially adsorbing to any exposed hydrophobic patches present on the surface of the molecule

[0025] .

[0131] To investigate whether addition of surfactants could attenuate loss of product due to adsorption of the product to the vials, Tween® 20 was added to MPA at increasing concentrations, which was then used as diluent for the standards prior to vialing. Fig. 6 (B), (C), (D) compares the %RD for standard peak areas of fixed volume injection calibration curves to variable volume injection calibration curves. A second-order polynomial was fit to the resulting data, but p-values for the quadratic term were non-significant relative to a selected significance criterion of 0.05 (data not shown). Thus, these data were instead fit with to a simple linear model. P-values for the modeled linear slopes were also determined to be non-significant for all samples tested in the presence of additional Tween® 20. These model statistics suggest no explicit evidence that the slopes of these linear fits are different from zero, indicating that the addition of Tween® 20 has prevented adsorption to either type of vial, even at the lowest standard tested.

[0132] Table 5. Nonconformance statistics calculated via process capability analysis of the calculated intercept bias (%), for n=6 variable concentration and fixed injection volume calibration standards, of which n=3 were stored in glass vials and n=3 were stored inpolypropylene vials showing the observed and expected percent of data below the lower specification limit (LSL), above the upper specification limit (USL), and the total outside the specification range.. ExpectedPortionS^,rVeOverall54%Below LSL 100.0 96.6Above USL 0.0 0.0Total outside 100.0 96.6

[0133] One-way ANOVA was conducted to further confirm the differences among the group means of the standard curve intercept biases generated with and without Tween® 20. Data from standards vialed in either glass or polypropylene vials were pooled prior to analysis, based on the previous data that confirmed the two vial types behave similarly both with and without Tween® 20. This analysis is visually depicted in Fig. 7. The resulting p-value for this analysis is 0.0006, which is below our selected significance criterion of 0.05, and can be considered as evidence that there are clear differences between the group means

[0027] . Visual inspection of the mean diamonds for each condition in Fig. 7 further indicates that this difference in group means can be attributed to the condition without Tween® 20 added, as its means diamond does not have CI overlap with the conditions tested in the presence of Tween® 20. Furthermore, all conditions tested in the presence of Tween® 20 have their entire CI regions without the specification range indicated by the ATP, providing evidence that supplementing Tween® 20 into MPA can generate a modified procedure that is fit-for-purpose. A two-sided F test for equality of two variances was performed comparing the intercept bias in the absence of Tween® 20 to the intercept bias resulting from combining all Tween® 20 concentration levels into a single grouped variable

[0028] . The resulting p-value was 0.0060, supporting that the variance of the intercept bias depends on whether or not Tween® 20 is added to MPA.

[0134] To build further confidence that adding Tween® 20 will generate a fit-for- purpose analytical procedure, the simulation tool within the prediction profiler platform in JMP® was used to simulate the defect rates of the intercept bias with and without Tween® 20, corresponding to the likelihood of it falling outside of specification during routine application of the procedure. Due to the overlapping confidence intervals for the conditions containing Tween® 20, a single concentration level would be representative as the nominal concentration. The 0.02% concentration was in the near middle of the Tween® 20 ranges tested, buildingadditional robustness, with regard to Tween® 20 concentration, into the final procedure to prevent disruptions to the final procedure performance during routine use.

[0135] Tween® 20 (0.02%) was used for a simulated comparison to the condition without Tween® 20. This simulation, shown in Fig. 8, highlights that by adding Tween® 20 to MPA prior to standard and sample dilution, the projected defect rate can be reduced from 96.6% to 0.0%.

[0136] Combined with the data above, the re-development of this procedure was concluded, and the final procedure conditions identified for further robustness testing and performance verification were defined with MPA composition of 0.02% Tween® 20 in 50 mM sodium phosphate, 150 mM sodium chloride, pH 7.0. Due to the acceptable performance observed by adding Tween® 20 to MPA, the composition of MPB was left unchanged at 10 mM phosphoric acid, 150 mM sodium chloride, pH 2.2.

[0137] Robustness

[0138] Robustness studies are recommended by regulatory bodies as they help to establish method operable design regions (MODRs) which, when generated using DoE approaches, demonstrate that a procedure is fit-for-purpose across a multi-dimensional combination of procedure parameter ranges possible during routine use [13,17]. The robustness study in this work was designed to detect if there are significant changes to the reportable result of the analytical procedure by intentional variation introduced to the mobile phase sodium phosphate and sodium chloride concentrations and their pH. The range of this intentional variation was chosen to reflect common-cause variation during routine application of the procedure. The DoE design selected allowed for the systematic investigation and modeling of these procedure parameters. Only the main effects for the MPA and MPB salt concentrations and pH values were studied, as an assumption was made that any factor interactions or curvature are not present in the response surface across the range of procedure parameters tested. Additional center point repeats were included in the final model to help test these assumptions.

[0139] The analysis of the robustness data set resulted in a final model consisting of all terms having associated p-values > 0.05, indicating that the response surface studied is highly robust. Additionally, the presence of additional center point repeats enabled testing for a lack- of-fit, which returned a non- significant p-value of 0.5114. These results are summarized in Table 6. A commonly applied principle of DoE is the effect hierarchy principle, which states that higher-order terms are not likely to be active if their constituent main effects are not statistically significant

[0030] . Thus, since no main effects are active in the final robustnessmodel, the hierarchy principle postulates that any higher-order interaction or quadratic curvature terms can likely be ignored. This assumption is further supported by no detectable lack-of-fit in our model to the collected data.

[0140] These data support that the revised procedure is considered sufficiently robust across the MODR tested. For MPA, this corresponds to 47.5 - 52.5 mM sodium phosphate and 142.5 - 157.5 mM NaCl at pH 6.9 - 7.1 and for MPB this corresponds to 142.5 - 157.5 mM NaCl at pH 2.1 - 2.3.

[0141] Table 6. Robustness model summary which lists all the model effects sorted by increasing p-values from top to bottom, along with the resulting lack of fit test. P-values > 0.05 indicate model statistics are not statistically significant.Procedure Parameter P-valueMobile phase A pH 0.20909Mobile phase B pH 0.32020Mobile phase B salt 0.40233Mobile phase A salt 0.90728Lack-of-fit test 0.5114

[0142] Performance verification

[0143] With the conditions optimized and considered robust, the resulting final procedure was subject to performance verification executed in alignment with ICH recommendations for analytical procedure validation

[0019] . This included assessments of linearity, accuracy, specificity, and precision, along with both intermediate precision and repeatability.

[0144] Linearity for analyst 1 and analyst 2 displayed R2 values of 1.00 and intercept biases of 1.2% and 0.8%, respectively (Fig. 9). These intercept bias values are drastically improved compared to the procedure prior to re-development according to AQbD principles. This improvement further supports the simulated defect rate of falling outside of specification for the intercept bias, as previously discussed in the procedure optimization section. Additionally, the specificity of the production media generated a peak area that was 3.1% of the lowest calibration level tested (data not shown). Accuracy across all levels was reported at 99.2 % recovery (Table 7). Precision of the HCCF sample was assessed using REML, enabling variance component analysis to assess where major sources of random variation occur during use of this procedure (Table 8). The variance component attributed to different instruments, column lots, and analysts on different days, or inter-day precision, was determined to be 77571.3, while the variance component for the procedure repeatability was 10979.2. Thus, the validation data suggest that the major source of variation came from the procedure inter-dayprecision. The intermediate precision is calculated as the sum of these two variance components and is thus equal to the repeatability in this work, or 3.0% RSD.

[0145] Every metric assessed during procedure performance verification met the ATP criteria with respect to their reportable estimates. These data build support that the procedure performance is fit-for-purpose following re-development, and while the procedure verification data in this work only represents the variability between two sources of variation, application of this procedure to several different products since its re-development has indicated stable performance thus far. Now, as a component of an analytical lifecycle, the performance of this procedure will continue to be monitored and iterated upon as needed.

[0146] Taken together, the AQbD approach executed in this work to ensure procedure robustness and performance verification adds rigor into the understanding of the procedure and its performance. This further supports its broad application across mAb products, adding efficiency in analytical procedure development and supporting expedited development timelines.

[0147] Table 7. Assay accuracy, analyzed at 5 different levels, n = 3 preparations at each level.Level „ „ Average % Recovery z ( iis / / mTL x) % RecoveryJPer-Level All Levels200 98.3%200 99.1% 99.0%200 99.6%500 97.6%500 97.4% 97.6%500 97.9%1000 98.8%1000 99.6% 99.4% 99.2%1000 99.7%1500 99.8%1500 100.1% 100.1%1500 100.5%2000 100.2%2000 99.7% 99.8%2000 99.4%

[0148] Table. 8. Restricted maximum likelihood (REML) analysis of precision data.Standard Mean Total TareetPrecision Level Variance Deviation Integrated %RSD %RSDArea (gg / mL)Inter-day Precision 77571.3 278.5 2.8 NA Repeatability 10979.2 104.8 10066.6 1.0 NA Intermediate Precision 88550.5 297.6 3.0 < 10.0

[0149] Materials and methods

[0150] Columns, reagents, and vials

[0151] The columns used in this study, the Bio-Monolith Recombinant Protein A, 4.95 x 5.2 mm (p / n 5190-6903) including the InfinityLab Quick Change inline filter assembly, 2.1 mm id, 0.2 pm pore size with 90 mm flexible capillary (p / n 5067-1603), were purchased from Agilent (Agilent Technologies, Santa Clara, CA, USA).

[0152] Sodium phosphate monobasic monohydrate (p / n 3821-01), sodium phosphate dibasic heptahydrate (p / n 3803-01), sodium hydroxide (p / n 5000-03), and hydrochloric acid (p / n 0327-02) were purchased from Avantor™ (Radnor, PA USA). Phosphoric acid (p / n 49685) was purchased from Sigma Aldrich. Tween® 20 Surfact-Amps® Detergent solution (p / n 28320) and sodium chloride (p / n S271-500) were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Water was obtained from a Milli-Q® Purification System (Millipore Sigma, Burlington, MA, USA).

[0153] Snap cap low retention microcentrifuge tubes (Catalog #3448) were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Screw top 2 mL glass vials and caps (p / n 5182-0714), 250 uL glass inserts (p / n 5181-1270), and snap top 250 uL polypropylene vials (p / n 5190-3155) were purchased from Agilent (Agilent Technologies, Santa Clara, CA, USA).

[0154] Equipment and software

[0155] Agilent 1260 Infinity II Bio-inert LC and Agilent 1260 Infinity Stainless Steel LC systems were used in this study. Both systems used were equipped with a quaternary pump (GF611 A- 1260 BIO QUAT PUMP, G7111B-1260-QUAT PUMP), an autosampler (GF667A- 1260 HiP Bio ALS, G7167A-1260 MULTISAMPLER), and detector (G4212B-1260 DAD, G7165A-1260 MWD), respectively. The autosamplers utilized a 100 pL sample loop. Instrument operation and data acquisition was performed using Empower 3 software (Waters™). Statistical analysis, DoE, and visualization of data were conducted using JMP® version 16.0.0 (JMP® Statistical Discovery, Cary, NC, USA). Graphic illustrations were created using BioRender (Toronto, Ontario, Canada). Calculation of total analytical error probabilities was performed in Microsoft® Excel® (Version 2309 Build 16.0.16827.20278) using the following function:

[0156] = norm. dist(USL, p, a, True) — norm. dist LSL, p, a, True) (1)

[0157] where p is the procedure accuracy in % relative deviation (%RD), c is the procedure precision in % relative standard deviation (%RSD), and USL and LSL are the upper and lower limits for total analytical error, respectively. The true arguments in the norm.dist functions above return the cumulative distribution function. The final value was taken to be the probability of meeting a given total analytical error target criterion. The resulting analysis was then visualized in the Graph Builder platform in JMP®.

[0158] Samples

[0159] An internally-developed IgGl mAb, purified and formulated in 10 mM sodium phosphate, 140 mM NaCl, pH 7.2, along with upstream harvested cell culture fluid (HCCF) was used. Cell culture media, G12.1 (p / n 98945), was purchased from FUJIFILM Irvine Scientific (Santa Ana, CA, USA).

[0160] Chromatographic conditions

[0161] All experiments were conducted using the chromatographic conditions of the platform method which was previously developed via traditional approaches. A flow rate of 1.5 mL / min was applied, utilizing the gradient method shown in Table 1, with MPA of 50 mM sodium phosphate and 150 mM sodium chloride, pH 7.0 and mobile phase B (MPB) of 10 mM phosphoric acid and 150 mM sodium chloride, pH 2.2. A column compartment temperature of 23°C was used for analysis and detection was performed at 280 nm with a 4 nm bandwidth and 5 Hz sampling frequency.

[0162] Table 1. Gradient method timetable using mobile phase A (MPA) and mobile phase B (MPB), executed using a 1.5 mL / min flow rate.Time MPA MPB(min) _ [%] _ [%]0.0 90 100.4 90 100.5 0 1002.0 0 1002.1 90 103.0 90 10

[0163] Procedure optimization

[0164] Procedure optimization experiments applied standard curves at 9 concentrations across a range of 200 - 2000 pg / mL. Using the purified mAb material, all standards were prepared in polypropylene microcentrifuge tubes diluted in MPA comprised of varying concentrations of Tween® 20 and transferred immediately into polypropylene and glass HPLC vials and analyzed as two separate standard curves. The analyses were executed within 24hours, and the samples were stored at 5 °C for the duration of the analysis. See Table 2 for experimental details.

[0165] The first standard preparation design, utilizing variable concentrations and fixed volume, (VC-FV) included the preparation of standards at 200, 300, 400, 800, 1000, 1200, 1400, 1600, and 2000 pg / mL. Each of these two curves were injected at 50 pL for each concentration. In addition, a fixed-concentration and variable volume (FC-VV) design was assessed for a third curve with a single 2000 pg / mL preparation in a glass vial injected at 5 - 50 pL. All three curves resulted in equivalent column loads of 10 - 100 pg of purified mAb product.

[0166] Across all experiments, the standard curve data were fit by linear regression in JMP®, using the total integrated peak area (pV*sec at 280 nm) as a function of mAb concentration, and the y-intercepts were obtained. Intercept bias was calculated using the following function:

[0167] = (Y — intercept / total integrated peak area) x 100 (2)

[0168] where the y-intercept was obtained from the linear regressions as noted above, and the total integrated peak area is of the 200 pg / mL standard, in units of pV*sec.

[0169] Further, the percent relative difference (%RD) of the total integrated peak areas of each of the standard curve peak areas for VC-FV versus FC-VV was calculated using the following function:

[0170] 100 (3)

[0171] where peak areavc is the peak area obtained from each of the standards of VC-FV calibration standards and peak areafc is the peak area obtained from each of the standards of FC-VV calibration standards.

[0172] Statistical analysis, including analysis of variance (ANOVA) and process capability were analyzed using the JMP® Fit Y by X and Distribution platforms, respectively. The data were also fit via standard least squares regression using Tween® 20 concentration as a categorical variable. The defect rate of the intercept bias falling outside of specification was then simulated in JMP® using the Simulator tool within the Prediction Profiler.

[0173] Table 2. Experimental design conducted for standard curve evaluations. Tween® 20 concentration in MPA standard diluent 0.00 % 0.00 % 0.00 %0.01 %0.02 %0.05 %

[0174] Robustness

[0175] The robustness DoE was built using the Custom Design platform in JMP® and represents a D-optimal design containing 8 unique conditions and 3 center point repeats. D- optimality criterion in a DoE design is useful for estimating effects or identification of active factors by reducing the variance of their estimates, suitable for robustness studies

[0021] . The final D-optimal model search converged to a resolution IV fractional factorial design, where the resolution is a measure of the degree of confounding, and in IV designs all main effects are not confounded with second order interactions

[0022] . This design was generated to evaluate main effect impacts of procedure parameters on procedure performance and for the detection of potential model curvature. Evaluation of only the main effects in robustness studies is widely adopted based on the assumption that interactions and model curvature from non-linear effects are not significant across narrow factor ranges typically explored during robustness studies. This assumption allows for a reduction in the total experimental conditions required

[0023] .

[0176] The procedure parameters included all mobile phase components, covering a range of ± 5% of their nominal values. Relative salt concentrations were tested at low, mid and high values and are denoted as arbitrary values -1, 0, and 1, respectively, in the final robustness design summarized in Table 3. MPA salts, which were sodium phosphate and sodium chloride, were grouped together to further simplify the robustness response surface. This corresponded to 47.5, 50, and 52.5 mM sodium phosphate and 142.5, 150, and 157.5 mM sodium chloride. In contrast, MPB contained sodium chloride as the only salt, which corresponded to 142.5, 150, and 157.5 mM sodium chloride. The pH values for both mobile phases were also investigated across a range of ± 0.1 of their nominal values, resulting in pH 6.9, 7, and 7.1 for MPA and pH 2.1, 2.2, and 2.3 for MPB.

[0177] Table 3. D-optimal custom design executed for the assessment of procedure robustness. The low, middle, and high values of each of the experimental factors tested are denoted using arbitrary values of -1, 0, and 1, respectively. For mobile phase A (MPA), these values correspond to 47.5, 50, and 52.5 mM sodium phosphate, 142.5, 150, and 157.5 mM sodium chloride, and pH 6.9, 7, and 7.1. For mobile phase B (MPB), these values correspond to 142.5, 150, and 157.5 mM sodium chloride and pH 2.1, 2.2, and 2.3.„ . MPA MPA MPB MPBCondition Salt pH Salt pH1 1 -1 1 -12 -1 -1 1 13 -1 1 -1 14 0 0 0 05 1 -1 -1 16 1 1 1 17 -1 -1 -1 -18 0 0 0 09 1 1 -1 -110 -1 1 1 -111 0 0 0 0

[0178] The standards were prepared by diluting the purified mAb to 6 levels spanning a range of 200 - 2000 pg / mE using the nominal MPA condition. The HCCF sample was diluted 10-fold using the nominal MPA condition to achieve a concentration that would be within the area range of the standard curve. Each DoE condition was run using the following sequence: single injections of the MPB blank, MPA (diluent) blank, standards, diluent blank to assess carryover, the HCCF sample, and a bracketing standard. All preparations were injected at 50 pL. The same preparations of the standards and HCCF sample were used for each condition to eliminate any differences that could arise due to sample preparation.

[0179] For each DoE condition tested, a linear fit was applied to the standards and the HCCF sample was quantitated based on its area (pV*sec). The resulting mAb peak areas were used as the model response and were fit via standard least squares using the Fit Model platform in JMP®.

[0180] Performance verification

[0181] Analytical procedure performance verification was executed in alignment with traditional procedure validation criteria. Briefly, analyst 1 assessed analytical procedure linearity via the standard curve of the purified mAb at 6 concentrations spanning a range of 200 - 2000 pg / mE, diluted in MPA. Triplicate injections of MPA followed the standard injections for carryover assessment. A cell culture media blank was injected to assess specificity. For precision assessment, 6 replicates of 10-fold dilutions of HCCF in MPA were prepared and each replicate was injected once. Accuracy solutions were prepared by spiking the purified mAb material into cell culture media at 5 different concentrations (200, 500, 1000, 1500, and 2000 pg / mE). Each concentration was prepared in triplicate and injected once. Analyst 2 repeated linearity and precision assessments in the same way as analyst 1, while varying days, instruments, column lots, reagent preparations, and sample preparations.Precision performance was estimated using restricted maximum likelihood (REML) in the Fit Model platform of JMP®, where the residual and total variance were taken to reflect the procedure repeatability and intermediate precision, respectively.

[0182] Conclusion

[0183] This work highlights the benefits and pragmatism of adopting AQbD and lifecycle approaches when re-developing analytical procedures. Further, there is great value in the adoption of AQbD regardless of the current stage within the procedure lifecycle. Overall, AQbD provides a valuable approach for efficiently developing analytical procedures through a data-driven lens. Eifecycle applications place focus on the continuous monitoring of analytical performance in order to ensure product safety, quality, and efficacy.

[0184] Additionally, this work highlights issues relating directly to the assessment of mAb titer demonstrating protein adsorption to two common types of HPEC vial materials. The simple addition of the commercially available surfactant Tween® 20 at low concentrations during sample dilution is observed to ameliorate these issues and restore procedure performance.

[0185] References

[0186] [1] S. Javanmard, An Overview of the Various Appropriate Types of Cell Fines for the Production of Monoclonal Antibodies, Int. J. Pharm. Investig. 13 (2023) 381-392. https: / / doi.Org / 10.5530 / ijpi.13.3.049.

[0187] [2] E. Paradina Fernandez, E. Calvo, E. Vina, Development and Validation of anAffinity Chromatography-Protein G Method for IgG Quantification, Int. Sch. Res. Not. 2014 (2014) 1-7. https: / / doi.org / 10.1155 / 2014 / 487101.

[0188] [3] compendium-titer-advancebio-5994-0038EN-us-agilent.pdf, (n.d.).

[0189] [4] P. Satzer, A. Jungbauer, High-capacity protein A affinity chromatography for the fast quantification of antibodies: Two-wavelength detection expands linear range, J. Sep. Sci. 41 (2018) 1791-1797. https: / / doi.org / 10.1002 / jssc.201701481.

[0190] [5] C. Zhang, E. Rodriguez, C. Bi, X. Zheng, D. Suresh, K. Suh, Z. Li, F. Elsebaei,D.S. Hage, High performance affinity chromatography and related separation methods for the analysis of biological and pharmaceutical agents, The Analyst 143 (2018) 374-391. https: / / doi.org / 10.1039 / C7AN01469D.

[0191] [6] P. Leblebici, M.E. Leblebici, F. Ferreira-da-Silva, A.E. Rodrigues, L.S. Pais,Separation of human immunoglobulin G subclasses on a protein A monolith column, J. Chromatogr. B 962 (2014) 89-93. https: / / doi.Org / 10.1016 / j.jchromb.2014.05.029.

[0192] [7] C. Reed, S.M. Koza, W. Chen, Protein A for Monoclonal Antibody TiterMeasurements with ACQUITY Arc Bio, Alliance, and ACQUITY UPLC H-Class Bio Chromatography Systems, (n.d.).

[0193] [8] T.-W. Chu, A. Coffey, A. Technologies, Robust, Reliable Recombinant ProteinA Monolith Column for Antibody Titer Determination, (n.d.).

[0194] [9] Q14 Analytical Procedure Development Guidance for Industry, (2024). https: / / www.fda.gov / media / 161202 / download.

[0195]

[0010] M.C. Breitkreitz, H. Goicoechea, eds., Introduction to Quality by Design in Pharmaceutical Manufacturing and Analytical Development, Springer International Publishing, Cham, 2023. https: / / doi.org / 10.1007 / 978-3-031-31505-3.

[0196]

[0011] M.K. Parr, A.H. Schmidt, Life cycle management of analytical methods,J. Pharm. Biomed. Anal. 147 (2018) 506-517. https: / / doi.Org / 10.1016 / j.jpba.2017.06.020.

[0197]

[0012] T. Verch, C. Campa, C.C. Chery, R. Frenkel, T. Graul, N. Jaya, B.Nakhle, J. Springall, J. Starkey, J. Wypych, T. Ranheim, Analytical Quality by Design, Life Cycle Management, and Method Control, AAPS J. 24 (2022) 34. https: / / doi.org / 10.1208 / sl2248-022-00685-2.

[0198]

[0013] ( 1220) Analytical Procedure Life Cycle, n.d. https: / / doi.org / 10.31003 / USPNF_M10975_02_01.

[0199]

[0014] P. Jackson, P. Borman, C. Campa, M. Chatfield, M. Godfrey, P.Hamilton, W. Hoyer, F. Norelli, R. Orr, T. Schofield, Using the Analytical Target Profile to Drive the Analytical Method Lifecycle, Anal. Chem. 91 (2019) 2577-2585. http s : / / doi . org / 10.1021 / ac s . analchem.8b04596.

[0200]

[0015] P.J. Borman, M.J. Chatfield, I. Damjanov, P. Jackson, Method ruggedness studies incorporating a risk based approach: A tutorial, Anal. Chim. Acta 703 (2011) 101-113. https: / / doi.Org / 10.1016 / j.aca.2011.07.008.

[0201]

[0016] ICH guideline Q9 on quality risk management, (n.d.). https: / / www.ema.europa.eu / en / documents / scientific-guideline / international-conference- harmonisation-technical-requirements-registration-pharmaceuticals-human-use-ich-guideline- q9-quality-risk-management-step-5-first-version_en.pdf (accessed January 2, 2024).

[0202]

[0017] ICH Q14 Analytical Procedure Development, International Conference on Harmonisation, (2023), (n.d.). https: / / database.ich.org / sites / default / files / ICH_Q14_Guideline_2023_l 116.pdf (accessed January 2, 2024).

[0203]

[0018] T. Cemosek, N. Jain, M. Dalphin, S. Behrens, P. Wunderli, Accelerated development of a SEC-HPLC procedure for purity analysis of monoclonal antibodies using design of experiments, J. Chromatogr. B 1235 (2024) 124037. https: / / doi.org / 10.1016 / j.jchromb.2024.124037.

[0204]

[0019] ICH Q2(R2) Validation of Analytical Procedures, InternationalConference on Harmonisation, (2023), (n.d.). https: / / database.ich.org / sites / default / files / ICH_Q2%28R2%29_Guideline_2023_1130.pdf (accessed January 2, 2024).

[0205]

[0020] J. Abraham, International Conference On Harmonisation Of TechnicalRequirements For Registration Of Pharmaceuticals For Human Use, in: C. Tietje, A. Brouder (Eds.), Handb. Transnatl. Econ. Gov. Regimes, Brill | Nijhoff, 2010: pp. 1041-1053. https: / / doi.org / 10.1163 / ej.9789004163300.i-1081.897.

[0206]

[0021] Optimality Criteria, (n.d.). https: / / www.jmp.com / support / help / en / 17.2 / index.shtml#page / jmp / optimality- criteria.shtml#ww600943 (accessed March 21, 2024).

[0207]

[0022] Resolution in Screening Designs, (n.d.). https : / / w w w .jmp .com / support / help / en / 17.2 / index. shtml#page / jmp / resolution-in- screeningdesigns. shtml (accessed March 21, 2024).

[0208]

[0023] S.L.C. Ferreira, A.O. Caires, T.D.S. Borges, A.M.D.S. Lima, L.O.B.Silva, W.N.L. Dos Santos, Robustness evaluation in analytical methods optimized using experimental designs, Microchem. J. 131 (2017) 163-169. http s : / / doi . org / 10.1016 / j . microc .2016.12.004.

[0209]

[0024] M. Duncan, Influence of surfactants upon protein / peptide adsorption to glass and polypropylene, Int. J. Pharm. 120 (1995) 179-188. https: / / doi.org / 10.1016 / 0378- 5173(94)00402-Q.

[0210]

[0025] R.G. Couston, M.W. Skoda, S. Uddin, C.F. van der Walle, Adsorption behavior of a human monoclonal antibody at hydrophilic and hydrophobic surfaces, mAbs 5 (2013) 126-139. https: / / doi.org / 10.4161 / mabs.22522.

[0211]

[0026] Individual Detail Reports, (n.d.). https : / / www .jmp .com / support / help / en / 16.0 / #page / jmp / individual-detail- reports.shtml?os=win&source=application#ww324052 (accessed November 15, 2023).

[0212]

[0027] The Analysis of Variance Report, (n.d.). https: / / www.jmp.eom / support / help / en / 16.0 / #page / jmp / the-analysis-of-vari ance- report.shtml#ww 178230 (accessed December 7, 2023).

[0213]

[0028] 1.3.5.9. F-Test for Equality of Two Variances, (n.d.). https: / / www.itl.nist.gov / div898 / handbook / eda / section3 / eda359.htm (accessed January 3, 2024).

[0214]

[0029] Mean Diamonds and X-Axis Proportional, (n.d.). https: / / www.jmp.eom / support / help / en / 16.0 / index. shtml#page / jmp / mean-diamonds-and-x- axis-proportional.shtml#ww66701 (accessed January 4, 2024).

[0215]

[0030] Effect Hierarchy, (n.d.). https : / / w w w .jmp .com / support / help / en / 17.2 / index. shtml#page / jmp / effect-hierarchy . shtml# (accessed January 4, 2024).

[0216] Example 2: Optimization and validation of a Protein A HPLC method for mAh titer analysis using QbD principles

[0217] The application of Quality by Design (QbD) principles was employed to optimize and validate a Protein A HPLC method for mAb titer analysis. The basic principle and workflow of Protein A affinity chromatography for mAb quantitation is illustrated in FIG. 10.

[0218] Initial experiments revealed issues with mAb adsorption to plastic and glass sample vials, leading to negative bias in the measured mAb concentrations. This is evident from the representative calibration curve in FIG. 11, which shows a large negative y-intercept despite good linearity (R2 = 1.00). Figures 12-14 further demonstrate the systematic negative bias due to adsorption, with deviations of up to -20% from nominal concentrations, and comparable adsorption to both plastic and glass vials.

[0219] To address this issue, the surfactant Tween 20 was added to the sample diluent (mobile phase A). Figures 15-18 show that Tween 20 effectively mitigates mAb adsorption, reducing the variability in peak area measurements across different vial types and concentrations. The optimal Tween 20 concentration was determined to be 0.02%.

[0220] The impact of assay design and control strategy on method performance was evaluated. FIG. 19 compares the accuracy of the overall assay vs. the within-assay performance, highlighting the tighter accuracy and precision of the latter. FIG. 20 shows the effect of implementing appropriate system suitability controls, reducing nonconformance and ensuring the method meets the target specification limits.

[0221] Method robustness was assessed by varying key chromatographic parameters like mobile phase composition and pH using a DoE approach (FIG. 21). The optimized method was then validated per ICH guidelines. Excellent linearity was demonstrated over the 200-2000 pg / mL range, with R2 > 0.999 and normally distributed residuals (FIG. 23). Good intermediate precision was shown across multiple days, analysts, instruments and columns, with an average total peak area of 3,092,708 pVsec (FIG. 23).* * *

[0222] Having thus described in detail preferred embodiments of the present invention, it is to be understood that the invention defined by the above paragraphs is not to be limited to particular details set forth in the above description as many apparent variations thereof are possible without departing from the spirit or scope of the present invention.

Claims

CLAIMSWHAT IS CLAIMED IS:

1. A method for separating a monoclonal antibody from a mixture comprising said antibody and one or more impurities, the method comprising: a. contacting a stationary phase of a Protein A chromatography column with a mixture; and b. eluting the monoclonal antibody with a mobile phase, wherein the mobile phase comprises a Tween 20, such that the monoclonal antibody traverses the column with a retention time that is different than the one or more impurities.

2. The method of claim 1, wherein the column is an analytical column.

3. The method of any one of claims 1 or 2, wherein the eluting is gradient with respect to mobile phase solvent composition.

4. The method of claim 1, wherein the eluting is isocratic with respect to the concentration of Tween 20 in the mobile phase.

5. The method of claim 1, wherein the method has a run time of between about 10 minutes and about 30 minutes.

6. The method of claim 1, wherein the method further comprises the step of detecting or isolating the monoclonal antibody.

7. The method of claim 1, wherein the accuracy of the method is higher than the accuracy of a corresponding method performed with a mobile phase that does not contain Tween 20.

8. A method of quality control of a pharmaceutical composition comprising a monoclonal antibody, the method comprising: a. separating a monoclonal antibody from a mixture comprising one or more additional impurities by the method of claim 1 ; b. comparing the separated monoclonal antibody with a reference monoclonal antibody; c. quantifying an amount or concentration of the monoclonal antibody based on the elution profile; and d. determining the monoclonal antibody has a selected character (e.g., binding ability to Protein A solid phase) based on a comparison of the separated and quantified monoclonal antibody with the reference monoclonal antibody.

9. The method of claim 8, wherein quantifying the amount or concentration comprises integrating a peak area corresponding to the separated monoclonal antibody in an HPLC chromatogram obtained from step (b) and comparing the peak area to a calibration curve.

10. The method of claim 1 or 8, wherein the monoclonal antibody is an IgG isotype.

11. The method of claim 10, wherein the monoclonal antibody is an IgGl or an IgG2 isotype.

12. The method of claim 1 or 8, wherein the monoclonal antibody is a relatively hydrophobic mAb.

13. The method of claim 1, wherein the monoclonal antibody exhibits reduced accuracy or linearity in a corresponding analytical Protein A HPLC titer method performed without Tween 20 due to adsorption to sample vial surfaces.

14. The method of claim 1, wherein the Tween 20 is present in the mobile phase used to prepare calibration standards and samples for the monoclonal antibody at a concentration sufficient to reduce non-specific adsorption of the monoclonal antibody to a sample vial.

15. The method of claim 14, wherein the concentration of Tween 20 is between about 0.01% v / v and about 0.05% v / v.

16. The method of claim 15, wherein the concentration of Tween 20 is about 0.02% v / v.

17. The method of claim 1, wherein the stationary phase comprises Protein A immobilized on a monolith support.

18. The method of claim 1, wherein determining the titer of the monoclonal antibody involves generating a calibration curve and wherein the presence of Tween 20 reduces y-intercept bias of the calibration curve compared to a corresponding method performed without Tween 20.

19. The method of claim 1, wherein the method is used for analytical titer determination during process development or quality control of monoclonal antibody production.

20. A method of detecting an IgG monoclonal antibody in a test sample, comprising:(a) preparing a reference sample, comprising:(i) obtaining a reference sample comprising the IgG monoclonal antibody;(ii) preparing the reference sample in a mobile phase comprising Tween 20;(iii) separating the reference sample by Protein A HPLC; and(iv) processing the separated reference sample through a detector to obtain a reference chromatogram showing a reference peak profile and retention time for the IgG monoclonal antibody;(b) preparing a test sample, comprising:(i) obtaining a test sample comprising the IgG monoclonal antibody;(ii) preparing the test sample in the mobile phase comprising Tween 20;(iii) separating the test sample by Protein A HPLC; and(iv) processing the separated test sample through the detector to obtain a test chromatogram;(c) comparing the test chromatogram to the reference chromatogram to detect the presence of the IgG monoclonal antibody in the test sample based on comparison of retention time and peak profile; and(d) quantifying the IgG monoclonal antibody in the test sample by comparing the test detector response to the calibration curve.

Images