Fourier transform infrared spectroscopy for quantification of mixtures

Abstract

The present invention relates to a method for quantitative analysis of a plurality of components in at least one aqueous sample comprising an active pharmaceutical ingredient (API), the method comprises the steps of (a) providing at least one aqueous sample comprising a plurality of components, wherein the sample comprises a biologic API; (b) measuring the extinction spectrum of the at least one aqueous sample using a Fourier transform infrared (FT IR) spectrometer comprising a transmission flow-through cell in a single measurement; (c) providing at least one reference extinction coefficient spectrum for each component of the plurality of components to be analyzed in the at least one aqueous sample; and (d) quantifying the concentration of each component in the plurality of components to be analyzed in the at least one aqueous sample based on the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b) by solving the classical least square (CLS) problem by an algorithm using a numerical linear algebra method for finding the approximate solution to an overdetermined system of linear equations characterized by the Beer-Lambert law for the extinction spectrum of the at least one aqueous sample. The invention further relates to methods for the production of an API or determining the purity of an API comprising using said method.

Description

[0001] 120333P1140PC

[0002] Fourier Transform Infrared Spectroscopy for Quantification of Mixtures

[0003] FIELD OF THE INVENTION

[0004]

[0001] The present invention relates to a method for quantitative analysis of a plurality of components in at least one aqueous sample comprising an active pharmaceutical ingredient (API), the method comprises the steps of (a) providing at least one aqueous sample comprising a plurality of components, wherein the sample comprises a biologic API; (b) measuring the extinction spectrum of the at least one aqueous sample using a Fourier transform infrared (FT IR) spectrometer comprising a transmission flow-through cell in a single measurement; (c) providing at least one reference extinction coefficient spectrum for each component of the plurality of components to be analyzed in the at least one aqueous sample; and (d) quantifying the concentration of each component in the plurality of components to be analyzed in the at least one aqueous sample based on the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b) by solving the classical least square (CLS) problem by an algorithm using a numerical linear algebra method for finding the approximate solution to an overdetermined system of linear equations characterized by the Beer-Lambert law for the extinction spectrum of the at least one aqueous sample. The invention further relates to methods for the production of an API or determining the purity of an API comprising using said method.

[0005] BACKGROUND

[0006]

[0002] Quantification of individual substances from a mixture is an omnipresent task in the development and production of pharmaceuticals, including New Biological Entities (NBEs), New Chemical Entities (NCEs), biosimilars and small molecule generics. Most of the currently used quantification techniques are target specific, such as in NMR, UV / Vis-, Solo- VPE, ELISA measurements of protein or peptide concentration or rely on supporting separation techniques, such as HPLC methods with UV / Vis detection. As several different methods have to be run to cover all relevant substances within a pharmaceutical sample, particularly a biopharmaceutical sample, it is time consuming, expensive and inefficient. Also, the sample volume available for analytics can be limiting for biopharmaceutical samples. For example, JP 2021 / 535374 A discloses methods for determining protein or peptide concentrations in a sample using nuclear magnetic resonance (NMR) spectroscopy using denaturation of the protein and a diffusion filter to eliminate resonance of water and other pharmaceutical ingredients and comparing the spectrum to a regular reference standard for determination of the concentration, allowing quantification of the protein or peptide only in 120333P1140PC the presence of other components, but not simultaneous quantifications of a plurality of components.

[0007]

[0003] Quantification of the different formulation components is necessary at various points during the development of biopharmaceuticals for example in in-process-control (I Postages, compounding steps (formulation) and the release analytics of the final drug. Particularly in later phases of the project, the quantification of more components, preferably all components, in the formulation is desirable and necessary. Thus, there is a need for a method to quantify all major components in a sample simultaneously in one very short and automated measurement, particularly for biopharmaceutical drugs at high concentrations.

[0008]

[0004] Fourier transform infrared (FTIR) spectroscopy has already been used for quantification, but most publications either only use single peaks for an evaluation with a calibration curve of spectra of known concentrations (Jasco Application Note “Quantitative Analysis of Powdered Solids with FTIR-ATR” or US 2009 / 242770 A1) or use a simple multivariate data analysis (MVDA) technique like partial least square regression (PLS) (Wasalathanthri et al., Biotechnology and Bioengineering, 2020; 117:3766-3774). However, PLS requires a large sample set for generating a suitable model. Also, this study uses a diamond attenuated total reflection (ATR) fiber optic probe, with an increased risk of surface contamination, due to, e.g., protein adsorption to the ATR crystal, which negatively affects quantification accuracy.

[0009]

[0005] Also Delbeck et al., (Anal. Bioanal. Chem. 2020, 412, 4647 - 4658, doi 10.1007 / s00216-020-02718-1), reports quantification of insulin in pharmaceutical formulations using an FT IR-ATR fiber optic diamond probe and pure excipient and protein spectra and a multivariate data analysis using partial least square (PLS) analysis for quantification.

[0010]

[0006] Mackie et al., (MethodsX. 2016, 3, 128 - 138, doi 10.1016 / j.mex.2016.02.002) reported that the standard methodologies for quantitative analysis of mixtures using Fourier transform infrared (FT IR) instruments have become complicated and the authors aim to provide a simpler methodology in an unrelated field (bio-hybrid fuel cells). More specifically, it is explained that previous quantitative analysis of mixtures using PLS require a software that uses the spectra of the mixtures to construct a quantitative analysis model and extra or other models are required if concentrations in the mixture vary or the composition of the mixture is altered. This results in extensive set up efforts that hinder the adoption of Fourier transform infrared quantitative analysis as standard technique for analysing aqueous samples. The authors describe a new fitting approach using a brute force (also referred to as “single mesh method” (SMM)) and an optimized brute force computing algorithm (also 120333P1140PC referred to as multi-pass “local adaptive mesh refinement” method (MPLM)) for the simultaneous quantification of multiple components in a mixture by fitting the spectra of the single components to the mixture spectrum. Yet, the evaluation of one spectrum for a mixture comprising 2 components using SMM still takes about 15 min due to the limited computation capacity of standard computers and time increases if the number of components (N) is increased. MPLM allows analysis of artificial mixtures reporting 1600s per fit for artificial mixtures comprising nine components, which is still impractical for high-throughput applications in the pharmaceutical field.

[0011]

[0007] WO 2021 / 171310 A reports that only limited literature exists on methods for analysing excipients in protein drug products, due to incompatibility of protein with organic solvents and the wide diversity of the nature of excipients and typically either measure excipients or protein. It discloses near-infrared spectroscopy (NIRS) for real-time measurement using NIRS or FT-IR probes that are dipped into the reservoir and also uses multivariate data analysis, such as PLS or OPLS, which requires building a model.

[0012]

[0008] Thus, there is still a need for a simple and fast method for quantification of multiple components in an aqueous sample during biopharmaceutical development (i.e., with limited sample volume available) and quality control that is easily adaptable to new or different samples and that can be easily used for high-throughput analysis in process development and quality control of biopharmaceuticals.

[0013] SUMMARY OF THE INVENTION

[0014]

[0009] The present invention provides a method that allows that all major components in an aqueous sample comprising a biopharmaceutical (biologic API) can be quantified in one very short and automated measurement by FT I R spectroscopy using a transmission flow-through cell combined with a least square fitting procedure fast and cost effectively with minimal effort for the operator while at the same time yielding highly accurate results. The method determines the concentration of each component of the plurality of components by comparing the measured extinction spectrum of the at least one aqueous sample with the expected / reference extinction coefficient spectra for each of the components of the sample by solving the classical least square problem. The method allows to account for all components in an aqueous sample (such as a formulation) at once and can also detect contaminations of the aqueous sample. There is no need for setting up a complex statistical model with hundreds of training spectra as would be necessary for PLS-based quantification methods, reducing the time needed to set up the method for a given formulation. The IR-extinction coefficient spectra of individual components can also be applied to other formulations than originally recorded for with an increasing database 120333P1140PC of I R-extinction coefficient spectra, bringing the setup time for the method down even further.

[0015]

[0010] The present invention relates in a first aspect to a method for quantitative analysis of a plurality of components in at least one aqueous sample comprising an active pharmaceutical ingredient (API), wherein the method comprises the steps of (a) providing at least one aqueous sample comprising a plurality of components, wherein the sample comprises an API; (b) measuring the extinction spectrum of the at least one aqueous sample using a Fourier transform infrared (FT IR) spectrometer comprising a transmission flow-through cell in a single measurement; (c) providing at least one reference extinction coefficient spectrum for each component of the plurality of components to be analyzed in the at least one aqueous sample; (d) quantifying the concentration of each component of the plurality of components to be analyzed in the aqueous sample of step (a) based on the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b) by solving the classical least square (CLS) problem by an algorithm using a numerical linear algebra method for finding the approximate solution to an overdetermined system of linear equations characterized by the Beer-Lambert law for the extinction spectrum of the at least one aqueous sample; and wherein the API in the at least one aqueous sample is a biologic pharmaceutical ingredient. The reference extinction coefficient spectrum of step (c) may be obtained by a step comprising (i) measuring the reference extinction coefficient spectrum for the component using a Fourier transform infrared (FT IR) spectrometer and a reference sample comprising said component; and / or (ii) providing from a spectral database the reference extinction coefficient spectrum for the component based on a measurement of a reference sample comprising said component using an FT I R spectrometer.

[0016]

[0011] In certain embodiments the Fourier transform infrared (FT IR) spectrometer comprises a transmission flow-through cell that has a high optical pathlength stability and / or an automated optical pathlength determination. Preferably the measuring in step (b) and / or step (c) according to the invention is automated. The spectra used for quantifying in step (d) may be (i) the original, non-derived spectra of the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b), or (ii) the first or higher order derivative spectra of the spectra of (i). The plurality of components to be analyzed according to the method comprises all major components present in the at least one aqueous sample; such as at least three, at least four, or at least five components comprised in the aqueous sample; and / or 2-20, preferably 3-15, more preferably 4-15, even more preferably 4-10, even more preferably 5-10 components comprised in the at least one aqueous sample. 120333P1140PC

[0017]

[0012] The API may be any biologic pharmaceutical ingredient, such as a biological pharmaceutical ingredient selected from the group consisting of a protein, a peptide, a nucleic acid and a virus and modified forms thereof. In certain embodiments, the at least one aqueous sample is an in-process-control (I PC) sample, an ultrafiltration / diafiltration sample (LIF / DF sample), a formulation (final or intermediate), a drug substance, or a drug product.

[0018]

[0013] According to the invention the reference extinction coefficient spectrum and / or the extinction spectrum of the at least one aqueous sample comprises a spectral range from at least 1000-1750 cm-1, preferably at least 950 - 3050 cm-1, preferably at least 930-3700 cm-1.

[0019]

[0014] The method according to the invention is a high-throughput method. Thus, the method preferably comprises analyzing at least 10 aqueous samples (or at least 50 or at least 100 samples) and the method is for high throughput quantitative analysis of a plurality of components in the at least 10 aqueous sample comprising an active pharmaceutical ingredient (API).

[0020]

[0015] In certain embodiments, the API is a protein or a peptide and the method quantifies the concentration of the API and optionally of at least two other components in the at least one aqueous sample, preferably wherein the at least two other components in the at least one aqueous sample are independently excipients, buffer components or contaminants.

[0021]

[0016] The present invention further relates to a method for quantitative analysis of batch variability of drug substance samples for drug release and / or defining product specification, wherein the method comprises steps (a) to (d) of the method according to the first aspect of the invention, wherein the at least one aqueous sample is a batch drug substance sample, and wherein the method further comprises for each batch drug substance sample or replicates thereof a step (e) comparing the concentrations of each component of the plurality of components in the at least one aqueous sample quantified in step (d) with pre-determined release concentration criteria and a step (f) release of the batch drug substance and / or defining product specification if the concentrations of the plurality of components in the aqueous sample quantified in step (d) meet the pre-determined release concentration criteria.

[0022]

[0017] The present invention further relates to a method for quantitative analysis of a plurality of components in at least one aqueous sample comprising an active pharmaceutical ingredient (API) during process development (such as purification or formulation), wherein the method comprises steps (a) to (d) of the method according to the first aspect of the invention, wherein the at least one aqueous sample (i) is an in-process-control (I PC) sample of different or modified processes or process steps or (ii) are samples of different or modified 120333P1140PC formulations and wherein the method further comprises a step (e) comparing the concentrations of the plurality of components in at least two aqueous samples quantified in step (d), and a step (f) evaluating (i) each process or process step or (ii) each formulation based on the comparison.

[0023]

[0018] In yet another aspect, the present invention relates to a method for characterizing an antibody sample, wherein the method comprises steps (a) to (d) of the method according to the first aspect of the invention, wherein the at least one aqueous sample comprises an antibody and wherein the method comprises quantifying the concentrations of the antibody, and optionally antibody variants, in the at least one aqueous sample quantified in step (d).

[0024]

[0019] In yet another aspect, the present invention relates to a method for determining purity of an API, wherein the method comprises steps (a) to (d) of the method according to the first aspect of the invention, wherein the at least one aqueous sample (i) is an in-process-control (I PC) sample or (ii) a formulation and wherein the method further comprises a step (e) detecting the presence of a variation of the extinction spectrum of the at least one aqueous sample, such as residual IR bands in the extinction spectra of the at least one aqueous sample and / or shifts of single or multiple IR bands of the extinction spectra of a component in the at least one aqueous sample and wherein the variation of the extinction spectrum of the at least one aqueous sample indicates a contamination or an otherwise faulty sample.

[0025]

[0020] In yet another aspect, the present invention relates to a method for the production of an active pharmaceutical ingredient (API) comprising a process of (i) generation of the API, wherein the API is a biologic pharmaceutical ingredient; (ii) purification of the API; (iii) ultrafiltration and diafiltration (LIF / DF) of the API into an aqueous solution; and (iv) optionally further formulating the API; and wherein the method comprises quality testing of at least one aqueous in-process-control sample of the process of steps (ii), (iii) and / or (iv), wherein the sample comprises the API, using the method according to the first aspect of the invention, and optionally wherein a variation of the concentration of a component in said at least one aqueous in-process-control sample from a pre-determined concentration range results in an adaptation of the purification of the API of step (ii), the LIF / DF of the API in step (iii) and / or the further formulating of the API of step (iv), thereby controlling and / or regulating the production of the API.

[0026] DESCRIPTION OF THE FIGURES

[0027]

[0021] FIGURE 1 : Mean infrared extinction spectra for the samples with varying (A) sucrose concentrations (Table 1), shown are representative extinction spectra for sucrose at 1001 , 500 and 101 mM and (B) varying protein (IgG) concentrations (Table 2) shown are representative extinction spectra for IgG at 13.8, 56.3 and 74.2 g / L). 120333P1140PC

[0028]

[0022] FIGURE 2: Linearity of the three highest maxima of the (A) sucrose spectra and protein (IgG) spectra (B) determined from the measured infrared extinction spectra (Table 1 and 2).

[0023] FIGURE 3: Mean IR-extinction coefficient spectra of (A) sucrose and (B) the protein (IgG) extracted from all replicate spectra measured at the different concentrations. Shown are IR-extinction coefficient spectra of three different representative concentrations.

[0029]

[0024] FIGURE 4: Correlation coefficients for the IR-extinction coefficient spectra from (A) sucrose and (B) the protein.

[0030]

[0025] FIGURE 5: Determined concentrations of the different components in the test mixture compared to their respective nominal concentration for (A) the protein (IgG), (B) sucrose, (C) histidine, (D) mannitol, (E) PS20. The nominal concentration (dashed line) and + / - 5% deviation (solid lines) for A-D and + / - 10% deviation (solid lines) for E are indicated.

[0031]

[0026] FIGURE 6: Residual first derivative spectra that were obtained after two representative evaluations by subtracting the fitted first derivative spectrum from the first derivative spectrum of the sample. A spectrum of pure water is shown for comparison.

[0032]

[0027] FIGURE 7: R2 scores in cross validation in the dependence of the number of PLS components (A) for histidine, sucrose, mannitol and the protein API, and (B) for PS20.

[0033]

[0028] FIGURE 8: Determined concentrations using the extinction coefficient method (“Ext. coef. method”; filled circles) and a PLS-based quantification approach (filled triangles) for (A) histidine, (B) sucrose, (C) mannitol, (D) PS20 and (E) the protein (IgG). The nominal concentration (dashed line) and + / - 10% deviation (solid lines) is indicated.

[0034]

[0029] FIGURE 9: Relative deviations of the determined concentrations from the nominal concentrations using the extinction coefficient method (“Ext. coef. method”; filled circles) and a PLS-based quantification approach (filled triangles) for (A) histidine, (B) sucrose, (C) mannitol, (D) PS20 and (E) the protein (IgG). The nominal concentration at 0% deviation (dashed line) and + / - 10% deviation (solid lines) is indicated.

[0035]

[0030] FIGURE 10: Determined concentrations of a peptide API in a verification sample comprising the peptide API in a phosphate buffer, pH 7.2, compared to the respective nominal concentration, (A) determined concentration against nominal concentration with percentage deviation (5% indicated by solid lines) of the measurement of the peptide API in phosphate buffer, and (B) relative deviation of the determined concentration from the nominal concentration of the peptide API (dashed line) in phosphate buffer.

[0036]

[0031] FIGURE 11 : Determined concentrations of sodium dihydrogen phosphate in a verification sample comprising the peptide API in a phosphate buffer, pH 7.2, compared to the respective nominal concentration, (A) determined concentration against nominal concentration with percentage deviation (5% indicated by solid lines) of the measurement of 120333P1140PC sodium dihydrogen phosphate, and (B) relative deviation of the determined concentration from the nominal concentration of sodium dihydrogen phosphate (dashed line).

[0037]

[0032] FIGURE 12: Determined concentrations of phenol in a verification sample comprising the peptide API in a phosphate buffer, pH 7.2, compared to the respective nominal concentration, (A) determined concentration against nominal concentration with percentage deviation (5% indicated by solid lines) of the measurement of phenol in phosphate buffer, and (B) relative deviation of the determined concentration from the nominal concentration of phenol in phosphate buffer (dashed line).

[0038]

[0033] FIGURE 13: Determined concentrations of propylene glycol in a verification sample comprising the peptide API in a phosphate buffer, pH 7.2, compared to the respective nominal concentration, (A) determined concentration against nominal concentration with percentage deviation (5% indicated by solid lines) of the measurement of propylene glycol in phosphate buffer, and (B) relative deviation of the determined concentration from the nominal concentration of propylene glycol in phosphate buffer (dashed line).

[0039]

[0034] FIGURE 14: Determined concentrations of another excipient (Excipient 1) in a verification sample comprising the peptide API in a phosphate buffer, pH 7.2, compared to the respective nominal concentration, (A) measured concentration against nominal concentration with percentage deviation (10% indicated by solid lines) of the measurement of Excipient 1 in phosphate buffer, and (B) relative deviation of the measured concentration from the nominal concentration of Excipient 1 in phosphate buffer (dashed line).

[0040]

[0035] FIGURE 15: Determined concentrations of peptide API in a verification sample comprising the peptide API in a Tris buffer, pH 7.8, compared to the respective nominal concentration, (A) determined concentration against nominal concentration with percentage deviation (5% indicated by solid lines) of the measurement of the peptide API in Tris buffer, and (B) relative deviation of the determined concentration from the nominal concentration of the peptide API (dashed line) in Tris buffer.

[0041]

[0036] FIGURE 16: Determined concentrations of Tris buffer in a verification sample comprising the peptide API in a Tris buffer, pH 7.8, compared to the respective nominal concentration, (A) determined concentration against nominal concentration with percentage deviation (5% indicated by solid lines) of the measurement of Tris buffer, and (B) relative deviation of the determined concentration from the nominal concentration of Tris buffer (dashed line).

[0042]

[0037] FIGURE 17: Determined concentrations of phenol in a verification sample comprising the peptide API in a Tris buffer, pH 7.8, compared to the respective nominal concentration, (A) determined concentration against nominal concentration with percentage deviation (5% 120333P1140PC indicated by solid lines) of the measurement of phenol in T ris buffer, and (B) relative deviation of the measured concentration from the nominal concentration of phenol in Tris buffer (dashed line).

[0043]

[0038] FIGURE 18: Determined concentrations of propylene glycol in a verification sample comprising the peptide API in a Tris buffer, pH 7.8, compared to the respective nominal concentration, (A) determined concentration against nominal concentration with percentage deviation (5% indicated by solid lines) of the measurement of propylene glycol in Tris buffer, and (B) relative deviation of the determined concentration from the nominal concentration of propylene glycol in Tris buffer (dashed line).

[0044]

[0039] FIGURE 19: Determined concentrations of another excipient (Excipient 1) in a verification sample comprising the peptide API in a Tris buffer, pH 7.8, compared to the respective nominal concentration, (A) determined concentration against nominal concentration with percentage deviation (10% indicated by solid lines) of the measurement of Excipient 1 in Tris buffer, and (B) relative deviation of the determined concentration from the nominal concentration of Excipient 1 in Tris buffer (dashed line).

[0045]

[0040] FIGURE 20: Determined concentrations of the different components in the test mixture compared to their respective nominal concentration for (A) the protein (BSA), (B) histidine, (C) sucrose, (D) mannitol, (E) PS20. Top: Measured concentration against nominal concentration with percentage deviation (10% indicated by solid lines) of the measurement in the test mixture are indicated; Bottom: Relative (Rel.) deviation of the measured concentration from the nominal concentration of the compound (dashed line).

[0046]

[0041] FIGURE 21 : Determined concentrations of the different components in the test mixture compared to their respective nominal concentration for (A) the protein (IVIG), (B) histidine, (C) sucrose, (D) mannitol, (E) PS20. Top: Measured concentration against nominal concentration with percentage deviation (10% indicated by solid lines) of the measurement in the test mixture are indicated; Bottom: Relative deviation of the measured concentration from the nominal concentration of the compound (dashed line).

[0047]

[0042] FIGURE 22: Determined concentrations of the different components in the test mixture of a proteo-liposome formulation compared to their respective nominal concentration for (A) the protein (BSA), (B) phosphate, (C) lipid. Top: Measured concentration against nominal concentration with percentage deviation (10% indicated by solid lines) of the measurement in the test mixture are indicated; Bottom: Relative deviation of the measured concentration from the nominal concentration of the compound (dashed line). 120333P1140PC

[0048] DETAILED DESCRIPTION

[0049]

[0043] The term “comprises” or “comprising” means “including, but not limited to”. The term is intended to be open-ended, to specify the presence of any stated features, elements, integers, steps or components, but not to preclude the presence or addition of one or more other features, elements, integers, steps, components or groups thereof. The term “comprising” thus includes the more restrictive terms “consisting of” and “essentially consisting of’. Furthermore, singular and plural forms are not used in a limiting way. As used herein, the singular forms “a”, “an” and “the” designate both the singular and the plural, unless expressly stated to designate the singular only.

[0050]

[0044] The term “sample” as used herein refers to any sample comprising an active pharmaceutical ingredient (API), particularly wherein the API is a biologic pharmaceutical ingredient (biologic API). Typically, as used herein, the sample is an aqueous sample. The at least one sample may be an in-process control (I PC) sample, an ultrafiltration / diafiltration sample (LIF / DF sample), a drug substance sample (including a bulk drug substances sample) or a drug product sample, wherein the sample comprises a biologic pharmaceutical ingredient, such as a protein (including an antibody, an antibody fragment, an antibody derived molecule or a fusion protein, e.g., an Fc fusion protein), a peptide, a nucleic acid (DNA or RNA) or a virus and modified forms thereof.

[0051]

[0045] As used herein, the API is a biologic pharmaceutical ingredient (also referred to as a biologic API), such as a protein, including an antibody, an antibody fragment, an antibody derived molecule or a fusion protein (e.g., an Fc fusion protein), a peptide, a nucleic acid (DNA or RNA) or a virus and modified forms thereof. Modified forms include chemically and biologically modified forms of a protein, a peptide, a nucleic acid (DNA or RNA) or a virus. Similarly, the biologic API may be produced using biological means (e.g., production in prokaryotic or eukaryotic cells) or using chemical or enzymatic means (e.g., by chemical or enzymatic synthesis of peptides or nucleic acids). Further exemplary therapeutic proteins or proteins that are a biologic pharmaceutical ingredient are, without being limited thereto growth factors, cytokines or hormones.

[0052]

[0046] The term “aqueous sample” as used herein refers to a sample comprising a plurality of components in water as solvent. Preferably the water is distilled water or highly purified water. In case one or more of the components is poorly soluble in water, the aqueous sample may comprise up to 10%, preferably up 5%, more preferably up to 2% of an organic solvent (e.g., DMSO). In certain embodiments, the aqueous sample is free of organic solvent.

[0053]

[0047] The term “component” as used herein is a constituent part or ingredient that combines with other constituent parts or ingredients in water to make the aqueous sample. A component may be an API, an excipient, a buffer (or buffer component(s)) or contaminants. 120333P1140PC

[0054] The term “components” is to be understood to relate to the “major component” of an aqueous sample, such as a component at a certain detectable level, such as > 0.1 g / L. The opposite of a major component is a trace component. A component can be a single (chemical) substance, a single substance in its protonated or deprotonated form, or a mixture of substances. A chemical substance is a unique form of matter with constant chemical composition and characteristic properties and may take the form of a single element or chemical compound. If two or more chemical substances are combined without reacting, they form a mixture. A chemical substance relates to any organic or inorganic substance or a macromolecule (e.g., protein, peptide, polymer of a particular molecular entity, including any combination of substances occurring in whole or in part as a result of a chemical reaction or occurring in nature). A chemical element is a chemical substance made up of a particular kind of atom (all having the same number of protons, although they may be different isotopes). A chemical compound is a chemical that is composed of a particular set of atoms or ions. All compounds are substances, but not all substances are compounds.

[0055]

[0048] The term “contaminant” as used herein refers to the presence of an undesired and / or unintentional (or unexpected) component, such as endogenous proteins produced by the host cell also referred to as host cell proteins, a degradation or reaction product of any of the components within the aqueous sample, which is typically present only as a trace component in comparison to the API or the intended components in the aqueous sample, but may be or become a major component in an IPC sample, a LIF / DF sample, a drug substance or drug product before a certain purification step and / or after a faulty or ineffective purification step and / or after storage.

[0056]

[0049] The term “protein” as used herein refers to a biological macromolecule that comprises one or more chain(s) of amino acid residues (polypeptide(s)) of typically more than 30 amino acid residues. According to the FDA, a protein is a defined sequence that is greater than 40 amino acids in size (Federal Register 85 FR 10057, February 21 , 2020 [Docket No. FDA- 2018-N-2732]). Shorter polypeptides containing a backbone of equal or less than 30 amino acid residues are referred to as “peptides” herein. The individual amino acid residues are bonded together by peptide bonds. A therapeutic API has a therapeutic function and a peptide with a therapeutic function has a defined sequence of a certain length, such as comprising a backbone of 15-30 amino acids. The backbone is the longest linear amino acid chain in a peptide. In general, the genetic code specifies 20 standard amino acids and there are 22 proteinogenic amino acids. During or shortly after production, the residues in a protein or peptide are often chemically modified by post-translational modification (e.g. glycosylation, acetylation, phosphorylation, glycation, glycosylation or protein processing). Particularly, peptides may also be produced by chemical synthesis linking individual amino acid residues 120333P1140PC by peptide bonds. In chemically synthesized peptides also non-proteinogenic amino acids may be incorporated. Non-proteinogenic amino acids (also referred to as non-coded amino acids) are distinct from the 22 proteinogenic amino acids (21 in eukaryotic cells), which are naturally encoded in the genome of organisms for the assembly of proteins. However, non- proteinogenic amino acids also occur naturally in proteins, e.g, by post-translational modifications and further can be synthesized in the laboratory. Chemically synthesized amino acids may also be referred to as “unnatural amino acids” herein. Proteins and / or peptides may be further chemically modified, e.g., by chemically coupling polyethylene glycol (PEGylation) or fatty acids, typically via a linker, such as an amino acid or peptide linker. Particularly peptides may be coupled to fatty acids as a half-life extending principle by fatty acylation (promotes high affinity albumin binding).

[0057]

[0050] The term “recombinant protein” as used herein relates to a protein generated by recombinant techniques, such as molecular cloning. Recombinant techniques bring together genetic material from multiple sources or create sequences that do not naturally exist. A recombinant protein is typically based on a sequence from a different cell or organism or a different species from the recipient host cell used for production of the protein in cell culture, e.g., a CHO cell or a HEK 293 cell, or is based on an artificial sequence, such as a fusion protein. In the context of the present invention the recombinant protein is a therapeutic protein, such as an antibody, an antibody fragment, an antibody derived molecule (e.g., scFv, bi- or multi-specific antibodies) or a fusion protein (e.g., an Fc fusion protein). The term “therapeutic protein” as used herein refers to proteins that can be used in medical treatment of humans and / or animals and hence to a biologic API. These include, but are not limited to antibodies, growth factors, blood coagulation factors, vaccines, interferons, hormones and fusion proteins.

[0058]

[0051] The term “eukaryotic cell” as used herein refers to cells that have a nucleus within a nuclear envelop and include animal cells (such as mammalian cells and insect cells), human cells, plant cells and yeast cells. In the present invention an “eukaryotic cell” particularly encompasses mammalian cell, such as Chinese hamster ovary (CHO) cell or HEK293 cell derived cells, and yeast cells. Mammalian cells as used herein refer to cells, particularly cell lines, of mammalian origin. In the present invention a “mammalian cell” particularly encompasses human or rodent cells, and in most cases refers to a Chinese hamster ovary (CHO) cell or derivatives thereof. Cells as referred to herein are cells maintained in culture and do not relate to primary cells, but cell lines or cell line derived cells, i.e. , to immortalized cells. The term “prokaryotic cell” as used herein refers to cells that do not have a nucleus and include bacterial and archaea cells, preferably bacterial cells. 120333P1140PC

[0059]

[0052] The term “produced” as used herein in the context of a protein relates to the production of a therapeutic protein in cell culture in a prokaryotic or eukaryotic cell, preferably in an eukaryotic cell, such as a yeast cell or a mammalian cell. The person skilled in the art knows how to produce proteins in cells using fermentation. The production of proteins comprises cultivating the cell expressing the protein of interest in cell culture. Particularly, cultivating the cell expressing the protein in cell culture comprises maintaining the cells in a suitable medium and under conditions that allow growth and / or protein production / expression. In eukaryotic cells, the protein may be produced by fed-batch or continuous cell culture. Thus, the eukaryotic cells may be cultivated in a fed-batch or continuous cell culture or a combination thereof, preferably in a fed-batch cell culture. The term “expressing a protein” as used herein refers to a cell comprising a DNA sequence coding for the recombinant protein of interest, which is transcribed and translated into the protein sequence including post-translational modifications, i.e., resulting in the production of the protein in cell culture.

[0060]

[0053] The term “drug substance”, abbreviated as DS, as used herein refers to the active pharmaceutical ingredient (API) in a purified form and a DS sample typically further comprises excipients for, e.g., stability and / or tonicity, in a final or intermediate formulation. The API mediates the therapeutic effect in the body as opposed to the excipients, which assist with the delivery or stability of the API. In the case of a biologic pharmaceutical ingredient, the formulated API with excipients typically means the API in the final formulation buffer at a concentration of at least the highest concentration used in the final dosage form, also referred to as drug product. The final formulation is the formulation in the final marketed dosage form comprising the final formulation buffer and the final (target) API concentration.

[0061]

[0054] The term “drug product”, abbreviated as DP, as used herein refers to the final marketed dosage form of the drug substance for example a tablet or capsule or in the case of biologic pharmaceutical ingredient typically the solution for injection in the appropriate containment, such as a vial or syringe. The drug product may also be in a lyophilized form that is restored, typically in water or a buffer, prior to use.

[0062]

[0055] The term “polysorbate 20” as used herein refers to a non-ionic polysorbate-type detergent, which is a laureate ester of sorbitol and its anhydrides, copolymerized with approximately 20 moles of ethylene oxide for each mole of sorbitol and sorbitol anhydrides (polyoxyethylene (20) sorbitan monolaurate; CAS number: 9005-64-5). It is also known as Tween 20. Its stability and relative non-toxicity allow it to be used as a surfactant / detergent and emulsifier in a number of scientific analyses and applications. Polysorbate 20 can be used as washing agent in immunoassays, Western blots and ELISA. It can further be used in pharmacological applications, such as pharmaceutical formulations, particularly for 120333P1140PC biologies, such as antibodies and Fc-fusion proteins. Particularly it helps to prevent nonspecific protein binding.

[0063]

[0056] The term “about” or “approximately” as used herein refers to a variation of 10 % of the value specified or less, for example, about 50 % encompasses a variation from 45 to 55 %.

[0064]

[0057] The term “Fourier-transform infrared (FT IR) spectroscopy” as used herein refers to a technique used to obtain an infrared spectrum of absorption of a liquid. An FT IR spectrometer simultaneously collects high-resolution spectral data over a wide spectral range. Due to the significant IR absorption of water in aqueous samples, generating high quality FTIR spectra of aqueous samples can be difficult. At the same time, the significant water absorption limits the maximum usable pathlength for aqueous samples to a few microns in order to prevent total absorption of the incident radiation. Especially the amide I band of proteins overlaps with an absorption band of water, which may result in poorer signal- to-noise quality and deviations in this range if the water spectrum shifts due to slight interactions.

[0065]

[0058] The amount of light attenuated by a substance at a given wavelength 4 is described by the Beer-Lambert law:

[0066] EA = logw ( / < / / ) = £ALC, where EA is the wavelength dependent extinction, l0and I are, respectively, the intensity of incident and transmitted light, c is the concentration of the sample, L is the length of the light path through the sample (optical path length or sample path length), and EA is the extinction coefficient of the substance / compound at a specific wavelength.

[0067]

[0059] If a spectral range is observed using the Beer-Lambert law and not only one specific wavelength, derivatives of spectra (dE, / cM) can be calculated and have been known and used in spectroscopy for a long time. First derivatives (CIEA / CIA = (d£, / cM)Lc) and second derivatives (cPE cM2= (cpEA / dA^Lc) as well as higher order (nth) derivatives (cTEA / dAn= (dn£A / dAn)Lc) are in common use in modern spectroscopy. The mathematical properties and behavior of the derivative are independent of the particular spectroscopic technique to which it is applied.

[0068]

[0060] In the present invention using FT IR, where the spectra are commonly plotted over the wavenumber v instead of the wavelength, the Beer-Lambert law can be adapted as follows using the wavenumber dependent extinction E and the wavenumber dependent extinction coefficient Ev.

[0069] Ev = logw ( / < / / ) = SvLc. 120333P1140PC

[0070] Differentiating the Beer-Lambert law for FT IR with respect to the wavenumber results in the derivative spectrum (dEv / dv = (dSv / dv)Lc) and second derivatives (cPEv / dv2= (cfst / dv^Lc) as well as higher order (nth) derivatives (c Ev / dv" = (d^v / dv^Lc).

[0071]

[0061] When the Beer-Lambert law, is differentiated with respect to the wavenumber, only the extinction E and the extinction coefficient £v are dependent on the wavenumber, whereas the concentration c and the path length L are independent thereof. Therefore, the first derivative is given by the following Equation: dEv / dv = (d£v / dv)Lc.

[0072] Since the derivative extinction (dE / dv) is proportional to the concentration it can be used for quantitation. Differentiating the first derivative spectrum forms derivative spectra of higher order. The first and second or even higher order derivatives are often used for quantitation. The original, non-derived extinction spectrum may also be referred to as zero order derivative spectrum.

[0073]

[0062] The term “Partial Least Squares Regression (PLS)” as used herein is a statistical method for multivariate data analysis. It reduces the variables, which are used for the prediction, to a smaller set of predictors. These predictors are then used to perform a regression. PLS is often used in chemometrics for purposes such as quantification of individual components based on spectra of some sorts, e.g., infrared absorption spectra. The method requires the generation of a model with the help of a training data set consisting of spectra and the corresponding target values, e.g., the concentration of a component. The training data has a strong impact on the performance of the PLS model (output depends on the quality of the trainings data). More specifically, a large representative training dataset has to be generated for a PLS model, which can be challenging for applications in the pharmaceutical field with changing formulations and APIs, especially if e.g. API material is limited during early development. A new training data set has to be created for each formulation / combination of components that is to be quantified. Typically, tens to hundreds of exactly defined samples are necessary for a good model. Further, a PLS model has a blackbox character, meaning that the result and how it is obtained is not easily comprehensible for the analyst or authorities.

[0074]

[0063] The term “IR-extinction coefficient spectrum method” as used herein is a method based on the determination of extinction coefficient spectra and linear algebra for quantitative analysis of the components in an aqueous sample. It comprises that predetermined extinction coefficient spectra for all major components in an aqueous sample are multiplied by the respective concentrations to be determined and the sum of these spectra yields a calculated (also referred to as fitted or approximated) spectrum, which 120333P1140PC closely approximates the sample spectrum for a successful quantification. Therefore, the results can easily be understood later on. The method of the present invention does not comprise machine-learning or artificial intelligence. Troubleshooting of the invented method is easy, as the residual spectra provide a visual clue as to the numerical approximation to the spectra and thus to the quality of the quantification in question. These residuals can also alert the analyst in case something went wrong during the measurement or if the sample itself contains unexpected components, which is an important potential safety issue in pharmaceutical samples. PLS, by contrast, offers no easy way to detect such events, which is why the IR-extinction coefficient spectrum method is superior over the PLS method, particularly for analysis of pharmaceutical samples. For samples with highly defined matrices - which the samples for analysis in the pharmaceutical field typically are (e.g., for biologies downstream of the harvested cell culture fluid) -, the IR-extinction coefficient spectrum method delivers at least comparable quantification performance to a PLS, but requires considerably less effort to set up. The effort taken to create the PLS model may be reduced, however, this leads to a loss in performance. If the same effort is put in for PLS and the IR-extinction coefficient spectrum method, namely n training samples for n components, the invented methods vastly outperforms PLS. In the context of the present invention the terms “extinction coefficient spectrum” and “IR-extinction coefficient spectrum” are used synonymously.

[0075]

[0064] The term “reference sample” as used herein refers to a sample comprising a component of the plurality of components in the at least one aqueous sample for measuring the reference extinction coefficient spectrum using a FT IR spectrometer, wherein the reference sample may comprise a single component or more than one components in a mixture at a known concentration. The resulting reference extinction coefficient spectrum (also referred to as IR-extinction coefficient spectrum) can be saved in or added to one or several spectral database(s). Such a database is preferably device, time, user and location independent. Preferably the FT IR spectrometer used for generating the reference extinction coefficient spectrum is compatible with FT IR spectrometer used for measuring the extinction spectrum of the at least one aqueous sample. Two FT IR spectrometers are, e.g., typically compatible with each other if they are, without being limited thereto, the same, of the same type and / or of the same manufacturer. A suitable FT IR spectrometer, such as the MIRA Analyzer (CLADE GmbH, Esslingen, Germany), has the ability to record precise mid-IR spectra of aqueous samples in transmission mode, provides automated correction of atmospheric influences and automated determination of optical path length and preferably has automated 120333P1140PC sample and reference sample injection and automated blank subtraction. Automated correction of atmospheric influences and automated determination of optical path length are beneficial as they allow comparability of the resulting spectra in a device, in a time, user and location independent manner. Moreover, automated blank subtraction means recording a spectrum of a blank solution, such as water or water with CaF2 (e.g., AquaSolv CF, CLADE, Esslingen, Germany) and subtracting it from the sample spectrum. Doing this alongside each sample measurement not only avoids any solvent bands, but also minimizes carry-overs that are visible in the spectrum.

[0076]

[0065] The term “extinction coefficient” as used herein refers to a spectral characteristic of the compound being analyzed. According to the Beer-Lambert law, it describes the light attenuation capability at a specific wavenumber of a compound normalized to the optical path length and compound quantity. It can be determined for a certain component by preparing either a pure aqueous solution of said compound with known concentration or by preparing an aqueous solution of said compound with known concentration in a defined buffer / excipient solution.

[0077]

[0066] The term “verification samples” as used herein refers to a set of samples consisting of a mixture of the same components as the reference samples, representing variations of the at least one aqueous sample to be measured, e.g. the formulation, in- process control (IPC) sample or UF / DF sample. The verification samples are used for the method validation.

[0078]

[0067] The term “plurality of components to be analyzed” as used herein refers to all major components contained in the at least one aqueous sample that provide a signal and therefore need to be analyzed or subtracted via reference samples using the method of the present invention. Thus, typically the “plurality of components to be analyzed” is the same as the “plurality of components in the at least one aqueous sample”, and encompasses all major components contained in the at least one aqueous sample. Particularly it comprises the API. Thus, in a preferred embodiment, at least one of the plurality of components to be analyzed is the API, i.e., the biologic pharmaceutical ingredient.

[0079]

[0068] In a first aspect the invention provides a method for quantitative analysis of a plurality of components in at least one aqueous sample comprising an active pharmaceutical ingredient (API), wherein the method comprises the steps of (a) providing at least one aqueous sample comprising a plurality of components, wherein the sample comprises an API; (b) measuring the extinction spectrum of the at least one aqueous sample using a Fourier transform infrared (FT IR) spectrometer comprising a transmission flow-through cell, 120333P1140PC preferably in a single measurement; (c) providing at least one reference extinction coefficient spectrum for each component of the plurality of components to be analyzed in the at least one aqueous sample; (d) quantifying the concentration of each component of the plurality of components to be analyzed in the at least one aqueous sample of step (a) based on the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b) by solving the classical least square (CLS) problem by an algorithm using a numerical linear algebra method for finding the approximate solution to an overdetermined system of linear equations characterized by the Beer-Lambert law for the extinction spectrum of the at least one aqueous sample, wherein the API in the at least one aqueous sample is a biologic pharmaceutical ingredient.

[0080]

[0069] The method according to the invention (first aspect) is a high-throughput method that allows quantitative analysis of a plurality of components in an aqueous sample comprising an active pharmaceutical ingredient (API), wherein a plurality of aqueous samples are analyzed. Thus, the at least one aqueous sample may be 10 or more, 20 or more, 50 or more, 100 or more, or even 300 or more aqueous samples. In step (b) the extinction spectrum of the at least one aqueous sample using a Fourier transform infrared (FT IR) spectrometer comprising a transmission flow-through cell is measured preferably in a single measurement. Typically, the at least one sample is measured in a single measurement, preferably in replicates (n), such as duplicates (n=2) or triplicates (n=3). The at least one sample may be measured directly without any prior sample treatment, such as dilution. In order to allow high- throughput analysis of the plurality of aqueous samples, the Fourier transform infrared (FT IR) spectrometer preferably comprises automatic sample handling. This reduces human error and handling-time and / or increases accuracy and reproducibility of the method. In certain embodiments of the method according to the invention the measuring in step (b) and / or step (c) (if applicable) is automated. In a preferred embodiment, the samples are provided in a microtiter plate and the Fourier transform infrared (FT IR) spectrometer allows measuring the extinction spectrum of the at least one sample provided in a microtiter plate, such as a 24 well plate, a 48 well plate, a 96 well plate or a 384 well plate, preferably a 96 well plate or a 384 well plate. The replicates may be taken from the same well (about 60 - 80 pl are sufficient per measurement) or replicate wells. Thus, even with triplicates this allows measuring up to 96 aqueous samples per 96 well microtiter plate or up to 384 single measurements with a 384 well plate.

[0081]

[0070] The FT IR spectrometer is a mid-infrared FT IR spectrometer with a transmission cell for aqueous solutions, preferably with an integrated autosampler, automated optical pathlength determination and automated blank subtraction. An exemplary suitable FT IR spectrometer is the MIRA Analyzer (CLADE GmbH, Esslingen, Germany) together with an 120333P1140PC

[0082] MIRA eCell flow-through transmission cell with a pathlength of approximately 7-9 pm that is optimized for the measurement of aqueous solutions and a minimal sample volume of about 60-80 pL. The windows of the cell are made from CaF2. A suitable flow-through transmission cell combines a minimized sample volume, high pressure stability and a high pathlength precision and is preferably a CaF2 transmission cell. A CaF2 transmission cell can in principle be also used in other FT IR spectrometers (e.g. within the CONFOCHECK system, Bruker Corporation, Billerica US) but the MIRA Analyzer is able to achieve superior accuracy and repeatability in the resulting spectra which is essential for a reliable quantification.

[0083]

[0071] The term “transmission flow-through cell” is used synonymously with “flow-through transmission cell” or “transmission cell” and refers to a cell suitable for aqueous samples, including water solved protein samples, which allows a small sample volume and has a high pathlength precision (high optical pathlength stability).

[0084]

[0072] In contrast to an attenuated total reflection probe (ATR probe), where the optical path length depends on refractive indices, wavenumber and angle of light incidence, a transmission flow-through cell can provide extinction spectra, where the absolute values depend less on the above-mentioned parameters, allowing good reproducibility across devices. Transmission measurements are based on the attenuation of IR radiation passing through the sample, while attenuated total reflection (ATR) is based on total internal reflection of the IR radiation inside the ATR crystal, where the IR radiation interacts with the sample only in the immediate vicinity where IR radiation is reflected and not in the bulk sample. ATR measurements are therefore very surface sensitive, resulting in a large impact of surface contamination to the measurements. The method of the present invention makes use of FT IR transmission spectra of aqueous samples, which — compared to ATR-FTIR spectra or solid phase transmission spectra (e.g. utilizing a KBr pellet) — can be recorded much faster and in an automated fashion, significantly increasing throughput and reducing operator time.

[0073] The FT IR spectrometer in combination with the transmission flow-through cell should have a high optical pathlength stability and / or an automated optical pathlength determination, preferably a high optical pathlength stability and an automated optical pathlength determination. This allows comparability of spectra obtained with the FT IR spectrometer. The method according to the present invention can be used to quantify nearly all substances in aqueous solution in the right concentration range (usually > 0.5 g L'1or even > 0.1 g L'1).

[0085]

[0074] There are several difficulties when measuring liquids using FT IR, particularly strong water absorption and overlapping spectra. Water molecules have strong absorption bands in the mid-IR region, particularly in the range of 1,500 to 4,000 cm-1. Thus, the water 120333P1140PC absorption bands overlap with the absorption bands of other components in a sample, particularly proteins and peptides, making it challenging to distinguish and quantify specific analytes accurately. The transmission flow-through cell used together with the FT IR spectrometer in the method according to the present invention therefore has an optical pathlength of less than 15 pm, preferably of less than 10 pm, more preferably of 7-9 pm.

[0086]

[0075] The range of wavenumbers used in the method according to the invention comprises at least from about 1000 to 1750 cm-1, preferably at least from about 950 to 3050 cm-1, more preferably at least from about 930 to 3700 cm-1, even more preferably from about 500 to 4000 cm-1. Thus, in certain embodiments of the invention the reference extinction coefficient spectrum and / or the extinction spectrum of the at least one aqueous sample comprises a spectral range from at least 1000-1750 cm-1, preferably at least 950 - 3050 cm-1, more preferably at least 930-3700 cm-1, even more preferably from about 500 to 4000 cm-1.

[0087]

[0076] The at least one aqueous sample comprising a plurality of components according to step (a) comprises an API. The API in the at least one aqueous sample is a biologic pharmaceutical ingredient (biologic API). The person skilled in the art will understand that the API is one of the plurality of components in the at least one aqueous sample and preferably of the plurality of components to be analyzed according to steps (b) and (c).

[0088]

[0077] In certain embodiments the biologic pharmaceutical ingredient is selected from the group consisting of a protein (including an antibody, an antibody fragment, an antibody derived molecule or a fusion protein, e.g., an Fc fusion protein), a peptide, a nucleic acid (DNA or RNA) and a virus and modified forms thereof. Preferably, the biological pharmaceutical ingredient is a protein or a peptide. It is explicitly noted that the protein or peptide may also be packaged or complexed in a liposome or other lipophilic carriers. Modified forms include chemically and biologically modified forms of a protein, a peptide, a nucleic acid (DNA or RNA) or a virus. Suitable aqueous samples may be any aqueous samples comprising a biologic API, particularly samples during or after purification (downstream processes). For example, the method may be used to analyze samples in downstream process development and / or formulation development and / or the method may be used to analyze samples for quality control in an established and finalized downstream process and / or formulation process and / or drug release. In certain embodiments, the at least one aqueous sample is therefore an in-process control (I PC) sample, an ultrafiltration / diafiltration sample (LIF / DF sample), or a formulation, e.g., a drug substance sample (including a bulk drug substances sample) or a drug product sample. 120333P1140PC

[0089]

[0078] A component in the at least one aqueous sample is a constituent part or ingredient of the at least one aqueous sample that combines with other constituent parts or ingredients in water to make the aqueous sample. A component may be an API, an excipient, a buffer (or buffer component(s)) or a contaminant. A component can be in addition to a biologic API a single (chemical) substance, a single substance in its protonated or deprotonated form, or a mixture of substances and a chemical substance may be a single element or chemical compound. Thus, a component comprising a mixture of substances or protonated and unprotonated forms thereof may be, e.g., a phosphate buffer or a histidine / histidine HCI buffer, but also other mixtures of substances that are present at the same ratio in the plurality of samples to be analyzed. In certain embodiments, the method comprises analyzing at least 10 aqueous samples, at least 20, at least 50, at least 100, or at least 300 aqueous samples. The extinction spectra of the samples comprising an active pharmaceutical ingredient (API) are measured sequentially (in quick succession). The reference extinction coefficient spectra for each component of the plurality of components are the same for all aqueous samples analyzed. Thus, steps (a), (b) and (d) are performed in quick succession. In certain embodiments the API is a protein or a peptide and the method quantifies the concentration of the API and of at least two other components (e.g., excipients or buffer components) in the at least one aqueous sample.

[0090]

[0079] For the protein or peptide quantification, the main advantage over the commonly used UV-absorption-based quantification at 280 nm is that the protein concentration can be determined also in the presence of molecules like aromatics that also absorb light at 280 nm and therefore make the UV-quantification impossible. Also, proteins and peptides that do not contain any aromatic amino acids and do not absorb at 280 nm can be quantified accurately.

[0080] Suitable components tested for FT IR based quantifications are without being limited thereto APIs: Proteins (BSA, lysozyme, IgG, new antibody formats, including bispecific antibodies), peptides; buffering substances (e.g. acetate, citrate, phosphate, Tris, succinate); amino acids (e.g., histidine, arginine, methionine, lysine etc., some of which can also be used as buffering substance); various other salts including small inorganic salts such as NaCI (with slightly reduced accuracy as aqueous NaCI itself is not IR-active - its extinction coefficient spectrum shows mainly the influence the ions have on the surrounding water molecules and the change in the water I R-vibrations compared to pure water) or sulfate salts; tonicity agents: sugars (e.g., sucrose, mannitol, trehalose, sorbitol), other tonicity agents (e.g., propylene glycol, glycerol, polypropylene glycol or polyethylene glycol); preservatives: phenol, m- cresol, further excipients; lipids: liposomes (e.g., in the nanometer range, such as with a diameter of 20 to 500 nm, preferably 50 to 250 nm, more preferably about 100 nm) and other 120333P1140PC lipophilic carriers; and (with slightly reduced accuracy due to the small amounts in the solution) detergents (e.g., PS20, PS80, poloxamer 188, SDS).

[0091]

[0081] Liposomes are nanoparticles with diameters in the nanometer range of, e.g., about 100 nm, and therefore scatter significant amounts of UV- and visible light resulting in slightly opaque to even turbid solutions. Conventional methods for the determination of the protein API content, like the UV absorption at 280 nm, face significant problems for liposome-based formulations due to the interference of the scattered light on the absorption spectra. The scattered light results in an increase of the measured signal in UV / vis measurements necessitating sophisticated methods to account for the scattering background. Due to the larger wavelength, IR radiation is scattered much less by nanoparticles below 1 pm and FTIR can be used to simultaneously quantify the API and other components in proteo-liposome formulations. The samples may be measured using dynamic light scattering (DLS) to confirm monomodal particle size distribution and / or particle size. Particularly, quantification is not hindered by the opacity of the sample. Thus, in certain embodiments, the aqueous sample comprising a plurality of components, wherein the sample comprises an API to be analyzed according to the method of the invention is a liposome-containing formulation and / or one component of the plurality of components to be analyzed according to the method of the invention is a liposome or another lipophilic carrier.

[0092]

[0082] Most components can be measured with deviations of + / - 15% or lower, preferably + / - 10%, more preferably + / - 7.5%, even more preferably + / - 5% or lower. For certain components, particularly at low concentrations, the deviation may be higher, such as for Tris buffer at concentrations below 4 mM and PS20 or PS80 below 0.8 g / L, at a deviation of + / - 25%, preferably + / - 22 %, more preferably + / - 20% or lower.

[0093]

[0083] Step (c) requires the provision of at least one reference extinction coefficient spectrum for each component of the plurality of components to be analyzed in the at least one aqueous sample. In certain embodiments the method comprises (i) measuring the reference extinction coefficient spectrum for the component using a Fourier transform infrared (FT IR) spectrometer and a reference sample comprising said component; and / or (ii) providing from a spectral database the reference extinction coefficient spectrum for the component based on a measurement of a reference sample using an FT I R spectrometer. The method for providing the at least one reference extinction coefficient spectrum may be independently selected for each component, e.g., the at least one reference extinction coefficient spectrum for one component may be provided from a spectral database, while the at least one reference extinction coefficient spectra for the remaining components may be measured (i.e., newly determined). Thus, any combination of alternative methods (i) and (ii) for providing the at least one reference extinction coefficient for the different components is suitable. 120333P1140PC

[0094] Preferably, the at least one reference extinction coefficient spectrum is available for most components and the at least one reference extinction coefficient spectrum needs to be measured only for a component that does not yet have a reference extinction coefficient spectrum in the spectral database. While the reference sample preferably comprises a single component comprising a single substance, e.g., in water, the reference sample may also comprise a mixture of substances or protonated and unprotonated forms thereof. Thus, in certain embodiments, the reference sample comprises a single substance or a mixture of substances in an aqueous solution, preferably in water.

[0095]

[0084] The reference extinction coefficient for each component can be obtained by measuring a sample of the pure component in a defined concentration or via the measurement of samples consisting of mixtures of the individual components or of several samples consisting of mixtures having different concentration ratios of the individual components. Generally, a single sample with a defined concentration for each component (one sample per component as pure components or the number of components for mixtures of the individual components) is sufficient. Either way, the infrared extinction coefficient spectra may be deduced based on the matrix Equation 1. Equation 1 also takes different replicate spectra of the different components into account.

[0096]

[0085] Considered are k components in the range [1 ,..i,...k], I samples measurements [1 and n wavenumbers [1 ,...m,...n], With the measured pathlength-normalized infrared extinction spectra and the concentrations c7, being known, the IR-extinction coefficient spectra for the individual species can be deduced via a numerical leastsquare solver (solving Ax=B with matrix A and B known). The data evaluation program may be implemented in any programming language, e.g., in a Python script.

[0097]

[0086] In order to deduce extinction coefficient information on e.g. k components at least k different samples with varying concentration ratios of the k-components (or k samples of pure components) need to be analyzed to allow for numerically solving Equation 1 for &^im. On the other hand, measuring more than k samples of varying concentration ratios or using additional replicates of samples leads to the retrieval of the best fitting solution for ^im. Depending on the given unit for the concentration, the IR-extinction coefficient spectra can be related, e.g., to the molar concentration or the mass concentration. For each 120333P1140PC component, a different unit can be attributed to the concentration which then yields the respective unit also in the quantification.

[0098]

[0087] In certain embodiments the API is a protein or a peptide and the method quantifies the concentration of the API and of at least two other components in the at least one aqueous sample. The at least two other components (e.g., 2-14, 3-14, 3-9 or 4-9 other components in addition to the API) in the at least one aqueous sample may be independently excipients, buffer components or contaminants.

[0099]

[0088] The difficulty for the reference sample of the API may be the limited amount of API and / or its concentration. As for the other components the reference extinction coefficient spectrum of the API is based on the extinction spectrum of a reference sample comprising the API (API reference sample) at a given concentration. This can either be achieved by measuring a sample of the (pure) component in a defined concentration or via the measurements of one or more samples consisting of mixtures having different concentration ratios of the individual components. In order to deduce extinction coefficient information on e.g. k components at least k different samples with varying concentration ratios of the k-components (or k samples of pure components) need to be analyzed to allow for numerically solving Equation 1 as shown in the Examples for

[0100] On the other hand, measuring more than k samples of varying concentration ratios or using additional replicates of samples leads to the retrieval of the best fitting solution for £Viim.

[0101]

[0089] Depending on the given unit for the concentration, the IR-extinction coefficient spectra can be related to the molar concentration or the mass concentration. For each component, a different unit can be attributed to the concentration which then yields the respective unit also in the quantification.

[0102]

[0090] Thus, the concentration of the API in the reference sample may be based on the initial dry weight of the API dissolved in the API reference sample or may be determined in solution in the API reference sample (e.g., from a process step, such as following a LIF / DF step) following a. Further, the API reference sample may comprise the API (i) in water or (ii) in a buffer, wherein the buffer spectrum is subtracted from the reference extinction coefficient spectrum of the API. Methods for determining the concentration of the API in an aqueous API reference sample are known in the art and include UV absorption using the experimental determined (UV-) extinction coefficient of the API (e.g., the protein) at 280 nm in water. For the API, particularly proteins with residual contaminations of small molecules, the original protein sample may be dialyzed against water to remove the contaminants. The content of the dialyzed solution (or original solution without dialysis) may be determined via UV-Vis spectroscopy. By unfolding of 120333P1140PC the protein in 6 M guanidinium hydrochloride solution, the content can be directly calculated from the UV Vis spectrum for a known sequence of the protein using the Edelhoch method. For projects with an already determined UV Vis extinction coefficient, the concentration may be determined from the protein solution in order to ensure identical concentration values in the later on determined samples compared to the already established UV Vis measurements. For pure solid or liquid samples, the amount of the substance necessary for the different dilution steps may be weighed in and the substance dissolved in a small amount of water. For compounds which can be protonated or deprotonated with the pH, the pH needs to be adjusted to the target pH of the later measurement by the addition of HCI or NaOH or other suitable acids or bases until the correct pH is reached. Alternatively, two extinction coefficient spectra of the substance at a higher and a lower pH than the target pH or the completely protonated and deprotonated species can be used for the quantification (total concentration equals the sum of the two determined concentrations). For compounds which cannot be protonated or deprotonated, the solution may be used without further pH adjustment. The solution is further diluted with water to an exactly defined volume, e.g., via a volumetric flask.

[0103]

[0091] The aqueous API reference sample, if not generated from weighing in may be obtained from a downstream process step, e.g., the UF / DF sample. It may be required that the UF / DF sample is dialyzed against water or a buffer prior to the use as an aqueous API reference sample. Typically, all components present in the at least one aqueous sample are analyzed. In this context all components present in the at least one aqueous sample means all major components in the at least one aqueous sample, which means all components of > 0.5 g / L, preferably even of > 0.1 g / L, and / or that provide a signal. This further allows the detection of unexpected or additional bands, which is indicative of an unexpected component, particularly a contamination or not yet removed component. Particularly, components at a concentration of less than about 0.1 g / L do not need to be analyzed and are considered as trace components in the context of the present invention. Thus, if for example the at least one aqueous sample comprises 5 (major) components (the plurality of components in the at least one aqueous sample is 5 components), preferably a reference extinction coefficient spectrum is provided for each of the 5 components (the plurality of components to be analyzed is 5 components) and the concentration of each of the 5 components is quantified. In a preferred embodiment the plurality of components in the at least one aqueous sample therefore equals the plurality of components to be analyzed. However, theoretically not all components need to be analyzed (plurality of components in the at least one aqueous sample > plurality of components to be analyzed), such as if the concentration of said compound is low (e.g., < 0.1 g / L) or if said compound is subtracted by a reference sample. In certain embodiments 120333P1140PC the plurality of components to be analyzed is at least three, at least four or at least five components. Preferably the plurality of components in the at least one aqueous sample equals the plurality of components to be analyzed. In certain embodiments the plurality of components to be analyzed is 2-20, preferably 3-15, more preferably 4-15, even more preferably 4-10 or 5-10 components and preferably the plurality of components in the at least one aqueous sample equals the plurality of components to be analyzed. The plurality of components to be analyzed typically comprises the API.

[0104]

[0092] The reference extinction coefficient spectrum for each component of the plurality of components to be analyzed in the at least one aqueous sample of step (c) are preferably measured using the FT IR spectrometer of step (b), wherein FT IR spectrometer means the same FT IR spectrometer type of step (b) (i.e., from the same manufacturer) or any FT IR spectrometer compatible therewith. This particularly also applies to the transmission flow- through cell used together with or rather as part of the FT IR spectrometer. Moreover, the reference sample preferably comprises each component in water (such as distilled water or highly purified water). If needed, the reference sample may comprise small amounts of organic solvent (e.g., up to 10% organic solvent, preferably less, such as DMSO) if required for solubilizing the component in the reference sample or in a stock solution prepared for generating the reference sample. However, if the organic solvent is part of the at least one aqueous sample to be measured in step (b), the organic solvent (such as DMSO, propylene glycol or ethanol) may also be a component to be analyzed in the aqueous sample.

[0105]

[0093] The Fourier transform infrared (FT IR) spectrometer comprising a transmission flow- through cell preferably has a high optical pathlength stability and / or an automated optical pathlength determination. Thus, the FT IR spectrometer generates spectra normalized to a specific pathlength. This allows comparability of spectra obtained with the FT IR spectrometer. Theoretically, the spectra may be generated using a fixed specific optical pathlength, however, in most cases the transmission flow-through cell has a high optical pathlength stability and the FT IR spectrometer an automated optical pathlength determination. With an automated optical pathlength determination the spectrometer normalizes the spectra to a specific pathlength by computation.

[0106]

[0094] The spectra used for quantifying in step (d) are the original, non-derived spectra or first or higher order derivative spectra of the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b). However, it is mandatory that the extinction spectrum of step (b) and the reference extinction coefficient spectra of step (c) are used as the same order derivative spectra, i.e., both original, nonderived spectra (zero order derivative spectra), both first order derivative spectra or both higher order derivative spectra (such as second or third order). In certain embodiments the 120333P1140PC spectra used for quantifying in step (d) are (i) the original, non-derived spectra of the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b), or (ii) the first or higher order derivative spectra of the spectra of (i), i.e., the first or higher order derivative spectra of the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b). Preferably the spectra are first order derivative spectra of the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b), or second order derivative spectra of the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b), more preferably first order derivative spectra of the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b).

[0107]

[0095] Step (d) requires quantifying the concentration of each component of the plurality of components to be analyzed in the at least one aqueous sample of step (a) based on the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b) by solving the classical least square (CLS) problem by an algorithm using a numerical linear algebra method for finding the approximate solution to an overdetermined system of linear equations characterized by the Beer-Lambert law for the extinction spectrum of the at least one aqueous sample. The method according to the present invention does not rely on brute force or partial least square (PLS). In certain embodiments, the algorithm comprises the numerical linear algebra method Gauss-Jordan elimination, Gaussian elimination with back substitution, LU decomposition, QR decomposition, calculating the pseudoinverse of the matrix (for example the Moore-Penrose-inverse), single-value decomposition (SVD), a method for solving sparse matrix equations, such as the Golub-Kahan bidiagonalization (for example LSMR or LSQR), or a similar method, preferably LSMR, LSQR or SVD. The advantage of an algorithm comprising a numerical linear algebra method over e.g., brute force, is the severely reduced time for quantifying the concentration and the reduced energy consumption of the computer performing the evaluation which results from this reduced time. Pure brute force algorithms determine the best approximation of the solution by iteratively multiplying the whole spectra of the single components with various different concentrations and then calculating the deviation of the summed calculated spectrum to the measured spectrum. The set of concentrations with the smallest deviation are then considered the solution. Improved brute force algorithms as described by Mackie et al. for example (MethodsX. 2016, 3, 128 - 138, doi 10.1016 / j.mex.2016.02.002) usually use a coarser mesh in the first iteration, select the best set of concentrations and then compute a second slightly finer mesh around this 120333P1140PC set of concentrations in the second iteration. This iterative approach is continued for a certain number of iterations or until a certain threshold, e.g. less than a certain deviation of the calculated spectrum from the measured spectrum, is reached. Usually, the concentrations are evenly spaced in a certain concentration interval for each component. Therefore, the maximum achievable accuracy is given by the spacing of the mesh. Choosing a smaller mesh spacing will increase accuracy but will also lead to an increase in computational steps, time and energy. The use of brute force algorithms therefore always comes with a trade-off decision between accuracy and computation time and energy. Moreover, the computational time can be reduced by selecting a smaller concentration range for each component, but this approach increases the chance that the brute force method might only yield a local minimum as the solution and not the global minimum. In contrast, calculating the concentrations using linear algebra methods has the advantage that the matrices containing the spectral data can usually be simplified in the process and the solution of the matrix equation is usually analytical. Therefore, iterative approximation is usually not needed and the resulting concentrations are exact (ignoring machine precision which is negligible for the use cases described herein). In addition to the already much longer computation time of brute force algorithms, the computation time also scales much stronger with the number of components that need to be quantified compared to linear algebra methods. According to the method of the present invention, the quantification of the concentration in step d) takes less than a minute, preferably less than a second for the components in the at least one aqueous sample.

[0108]

[0096] Conventional Beer's Law based approaches for the quantification of individual formulation components often rely on just a couple spectral features, e.g. the height, width or area of a specific peak. The method of the present invention utilizes the complete available spectral range of the spectrometer to make full use of the information contained in a given components mid-infrared extinction spectrum. This is a key advantage for the correct separation of the concentration dependent contributions of the different components to the sample spectrum and thereby for the correct quantification of the components in a multicomponent system. Additionally, it is a key advantage for the detection of possible contaminants (components with small concentrations in the sample), as these might not have significant spectral features at the measurement points selected in state-of-the-art approaches. The method of the present invention does not involve machine-learning, as compared, e.g., to the CLADE™ QA Scan, and hence problems with the quantification of the components can be easily detected and the method requires no training samples for 120333P1140PC measuring a broad concentration range (e.g., for proteins at least between 1 g / L and 200 g / L) without systematic deviation.

[0109]

[0097] For example, the simultaneous quantification of all components in the sample is achieved by approximating the first derivative spectra of the IR-extinction coefficient spectra of all components to the first derivative sample spectrum of the at least one aqueous sample using a least-squares based algorithm. This can either be done for each separate spectrum yielding result concentrations only for one spectrum or for multiple spectra at once yielding the concentrations of all spectra. During the method development, using the first derivative spectra showed the highest accuracy of the results, followed by the original, non-derived spectra and then the higher derivative spectra. Therefore, usually the first derivative spectra are used.

[0110]

[0098] For one spectrum: With known IR-extinction coefficient spectra £^imover the whole wavenumber range of the spectrum for each component and the pathlength- normalized extinction spectra Ev,moriginating from each sample measurement, the respective concentrations are determined based on the matrix Equation 2. Simultaneously, k components are considered in the range [1 ,..i,...k] and n wavenumbers [1 ,...m,...n], t)99] Quantification of the components in an aqueous sample is necessary at various points during the development of biopharmaceuticals (biologic API) for example in in- process-control (IPC)-stages, compounding steps (formulation) and the release analytics of the final drug. Thus, in certain embodiments, the method according to the invention is used for quantitative analysis of a plurality of components in at least one aqueous sample of an in-process analytical procedure, of a compounding step or for drug release. The person skilled in the art will understand that the quantitative analysis of a plurality of components quantifies the concentration of a plurality of individual components, preferably of all (major) components in the at least one aqueous sample. This explicitly includes the biologic pharmaceutical ingredient.

[0111]

[0100] The increased speed of the method due to the use of a transmission flow-through cell and the faster analysis in step (d) also allows for rapid and informed decisions regarding the adjustment of the following process steps based on the determined aqueous sample composition during the manufacturing process of the biopharmaceutical product. Applications of the method according to the invention are also described in the following aspects. 120333P1140PC

[0112]

[0101] The method according to the invention comprising an FT I R extinction coefficient spectra-based quantification allows a fast and user-friendly quantification of all major components in a drug substance or drug product sample or another aqueous sample comprising a biologic API, such as during purification, in a single measurement resulting in a huge gain in efficiency. At the same time, the evaluation of the FT IR data with the IR-extinction coefficient spectrum method has the advantage of an easier troubleshooting and less potential for unrecognized errors compared to competing data evaluation techniques like more complex, statistical data analysis models (e.g. partial least-squares method (PLS)), where the different components are quantified not by fitting their respective singular spectra but by the decomposition in artificially calculated spectra of virtual components derived from the used training data. At the same time, no extensive model training needing hundreds of training spectra is necessary compared to PLS models. In comparison to existing literature using specific features of the FT IR spectra of the single components, where only smaller wavenumber intervals or even single peaks are used for the quantification, the method of the present invention has the advantage that the numerical approximation is much more robust in multicomponent mixtures, especially in the presence of compounds with similar FT IR spectra. Also, unknown contaminants or erroneously added components in the samples in suitable concentrations usually become apparent in the results, particularly the residual spectra. Although FT IR has been used previously, multicomponent measurements including biological pharmaceutical ingredients in aqueous solutions has been a challenge, particularly due to the overlapping spectra of water, excipients and protein or peptide containing compounds.

[0113]

[0102] In contrast to the UV-absorption-based quantification of proteins and peptides in formulations, the FT IR-based quantification can also be used if UV-absorbing molecules are present in the formulation, e.g., preservatives like phenol or m-cresol (compounds with aromatic groups) or otherwise UV-absorbing substances like parts of the cells in solutions from cell culture processes. The invention is particularly advantageous for aqueous samples comprising a biologic API without a viable alternative for the fast peptide or protein quantification in formulations, because UV-absorbing substances are present in the designated formulation.

[0114]

[0103] The invention also offers a procedure for the determination of an IR-extinction coefficient spectrum of the pure protein or peptide. Particularly, this procedure allows the determination without having to rely on weighing in of the pure, solid substance. The protein concentration of the reference sample is determined by an independent reference method. UV-absorption measurements at 280 nm are currently the standard 120333P1140PC method for the protein concentration determination. In the first approach, for a protein with a known UV-extinction coefficient at 280 nm, the invention allows the protein quantification via an absorption measurement at 280 nm, wherein the protein concentration can be calculated from the UV-absorption via the experimentally determined UV-extinction coefficient at 280 nm. This is favored to ensure the comparability of concentrations that are later obtained via FTIR and via UV measurements of a sample. If the UV extinction coefficient at 280 nm is not known for a protein, the so called Edelhoch method (Edelhoch H. Biochemistry, 1967, 6(7):1948- 1954) can be used. In the Edelhoch method, the theoretical UV-extinction coefficient of a protein at 280 nm in a 6 M guanidine hydrochloride solution (where the protein usually is completely unfolded) is determined through the number and respective extinction coefficient of tyrosines, tryptophanes and cystines (two S-S-linked cysteines) in the molecule and the respective molar mass. Alternative reference methods without relying on UV measurements could be the amino acid analysis (Benson et al., J. Biol. Chem., 1975, 250(1): 276-280), the Kjeldahl nitrogen determination (Jaenicke, Anal. Biochem., 1974, 61(2): 623-627) or the dry weight method (Hunter, The Journal of Physical Chemistry, 1966, 70(10):3285-3292; Kupke & Dorrier, Methods Enzymol., 1978, 48:155- 162; Nozaki, Arch Biochem Biophys., 1986, 249(2):437-446).

[0115]

[0104] Since nearly all chemical and biochemical compounds have a specific IR spectrum, the method can be used identically for NBEs or biopharmaceuticals as well as excipients or buffer components in aqueous solutions.

[0116]

[0105] IR spectra of different components present in the same solution that do not react or interact with each other usually show no or very little deviations from the spectra obtained from pure solutions. Additionally, without interactions like reactions or selfassembly taking place, no effects are known which cause the IR-extinction coefficient of biopharmaceutical relevant substances to increase or decrease in the presence of another compound since the IR signals correspond to vibrations of the molecules that are unaffected from other molecules as long as no binding interactions, long range interactions or indirect influence via e.g. the solution pH take place between the molecules. Therefore, the spectrum of a plurality of components usually is the direct superposition of all the spectra of the single components in their respective concentrations. This allows the quantification of all the substances of a mixture by fitting their respective IR-extinction coefficient spectra to the mixture spectrum.

[0117]

[0106] For the method according to the invention, FT IR-transmission spectra of all components in water of a given aqueous sample (e.g., a formulation) are recorded 120333P1140PC individually using samples with defined prepared concentrations. The IR-extinction coefficient spectra are obtained by solving, e.g., Equation 1 , which means in principle dividing all obtained spectral values by the concentration of the respective component.

[0118]

[0107] For a given aqueous sample, the transmission-FT IR extinction spectrum is then recorded and the aforementioned IR-extinction coefficient spectra are fitted to the sample spectrum with a least-squares algorithm. The obtained scaling factors for each of the components are the concentrations of the respective components in the sample.

[0119]

[0108] When all of the components that should be in the sample are accounted for, the residual spectrum obtained after the least squares approximation can be used to check for possible contaminants of the sample or otherwise mismatched components. Possible interactions between formulation components could also be detected this way. Additionally, the residual spectra can be used for the ID testing of components in the mixtures. Once acceptable deviations of the residual spectra from zero because of measurement deviations are elucidated, analyses with larger deviations become obvious. This can be used in combination with a database of spectra of different substances to identify the best fitting known substance where the residuum is minimal. Without knowing the spectra of the single components, the differences in the spectrum compared to a known spectrum can be used to at least verify the identity of the mixture and show possible differences and deviations. Starting from the difference spectra, contaminating substances can often be identified from databases.

[0120]

[0109] In the following, further aspects using the method according to the invention (first aspect) are specified, for which the embodiments and descriptions provided above similarly apply.

[0121]

[0110] In another aspect, the invention provides a method for quantitative analysis of batch variability of drug substance samples for drug release and / or defining product specification, wherein the method comprises steps (a) to (d) of the method according to the first aspect of the invention, wherein the at least one aqueous sample is a batch drug substance sample, and wherein the method further comprises for each batch drug substance sample or replicates thereof a step (e) comparing the concentrations of each component of the plurality of components in the at least one aqueous sample quantified in step (d) with pre-determined release concentration criteria and a step (f) release of the batch drug substance and / or defining product specification if the concentrations of the plurality of components in the aqueous sample quantified in step (d) meet the pre-determined release concentration criteria. A reference to steps (a) to (d) of the method according to the first aspect of the invention means that the entire method of the first aspect of the invention is part of the method 120333P1140PC according to a further aspect of the invention and that the disclosure, description and embodiments with regard to said method of the first aspect similarly applies to the method according to a further aspect of the invention.

[0122]

[0111] In yet another aspect, the present invention provides a method for quantitative analysis of a plurality of components in at least one aqueous sample comprising an active pharmaceutical ingredient (API) during process development (such as purification or formulation), wherein the method comprises steps (a) to (d) of the method according to the invention, wherein the at least one aqueous sample (i) is an in-process-control (I PC) sample of different or modified processes or process steps or (ii) are samples of different or modified formulations and wherein the method further comprises a step (e) comparing the concentrations of the plurality of components in at least two aqueous samples quantified in step (d), and a step (f) evaluating (i) each process or process step or (ii) each formulation based on the comparison.

[0123]

[0112] In yet another aspect, the present invention provides a method for characterizing a biologic pharmaceutical ingredient sample, wherein the method comprises steps (a) to (d) of the method according to the invention, wherein the at least one aqueous sample comprises a biologic pharmaceutical ingredient and wherein the method comprises quantifying the concentrations of the a biologic pharmaceutical ingredient, and optionally of a biologic pharmaceutical ingredient variant, in the at least one aqueous samples quantified in step (d). In certain embodiments the method is for characterizing an antibody sample, wherein the method comprises steps (a) to (d) of the method according to the invention, wherein the at least one aqueous sample comprises an antibody and wherein the method comprises quantifying the concentrations of the antibody, and optionally antibody variants, in the at least one aqueous sample quantified in step (d).

[0124]

[0113] In yet another aspect, the present invention provides a method for determining the purity of an API, wherein the method comprises the steps (a) to (d) of the method according to the invention (first aspect), wherein the at least one aqueous sample (i) is an in-process- control (I PC) sample or (ii) a formulation, wherein all components in the at least one aqueous sample are analyzed, and wherein the method further comprises a step (e) detecting the presence of a variation of the extinction spectrum of the at least one aqueous sample, such as residual IR bands in the extinction spectra of the at least one aqueous sample and / or shifts of single or multiple IR bands of the extinction spectra of a component in the at least one aqueous sample and wherein the variation of the extinction spectrum of the at least one aqueous sample indicates a contamination or an otherwise faulty sample. More specifically the method according to the first aspect comprises the steps of (a) providing at least one 120333P1140PC aqueous sample comprising a plurality of components, wherein the sample comprises an API, wherein the API is a biologic pharmaceutical ingredient; (b) measuring the extinction spectrum of the at least one aqueous sample using a Fourier transform infrared (FT IR) spectrometer comprising a transmission flow-through cell in a single measurement; (c) providing at least one reference extinction coefficient spectrum for each component of the plurality of components in the at least one aqueous sample; and (d) quantifying the concentration of each component of the plurality of components in the at least one aqueous sample of step (a) based on the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b) by solving the classical least square (CLS) problem by an algorithm using a numerical linear algebra method for finding the approximate solution to an overdetermined system of linear equations characterized by the Beer-Lambert law for the extinction spectrum of the at least one aqueous sample, wherein preferably the algorithm comprises single-value decomposition (SVD) or a method for solving sparse matrix equations, such as the Golub-Kahan bidiagonalization (for example LSMR or LSQR).

[0125]

[0114] In yet another aspect, the present invention provides a method for the production of an active pharmaceutical ingredient (API) comprising a process of (i) generation of the API, wherein the API is a biologic pharmaceutical ingredient; (ii) purification of the API; (iii) ultrafiltration and diafiltration (LIF / DF) of the API into an aqueous solution; and (iv) optionally further formulating the API; and wherein the method comprises quality testing of at least one aqueous in-process-control sample of the process of steps (ii) (iii) and / or (iv), wherein the sample comprises the API, using the method according to the invention. Any variation of the concentration of a component in said at least one aqueous in-process-control sample from a pre-determined concentration range may results in an adaptation of the purification of the API of step (ii), the LIF / DF of the API in step (iii) and / or the further formulating of the API of step (iv), thereby controlling and / or regulating the production of an API. A production method typically comprises an ultrafiltration and diafiltration (LIF / DF) step at the end of step (ii), e.g., to generate the drug substance, such as an intermediate formulation or the final formulation. The person skilled in the art will understand that in the context of the present invention this means ultrafiltration and diafiltration of the API into an aqueous solution. Diafiltration is used to buffer exchange the protein from its initial composition into the final formulation and pH or into an intermediate formulation for the preparation of the final formulation, while ultrafiltration is used to concentrate the solution to the desired protein concentration, such as the desired protein concentration of the intermediate or final formulation. The term intermediate formulation refers to any formulation prepared from the (purified) API that is not the final formulation, e.g., a formulation prior to adding polysorbate (PS20 or PS80), or a formulation 120333P1140PC lacking an excipient compared to the final formulation, or a different formulation comprising different excipients, excipients and / or API at different concentrations, and / or a different pH compared to the final formulation. The final formulation as used herein refers to the formulation in the final marketed dosage form. This LIF / DF step is therefore suitable to adjust the API concentration and / or the composition of the drug substance.

[0126]

[0115] The disclosure, description and embodiments with regard to the first aspect similarly apply to the further aspects referred to herein and vice versa.

[0127] The biologic API

[0128]

[0116] According to the methods of the invention the active pharmaceutical ingredient (API) is a biologic pharmaceutical ingredient (biologic API). In certain embodiments the biologic pharmaceutical ingredient is selected from the group consisting of a protein, a peptide, a nucleic acid and a virus and modified forms thereof. Preferably the biologic API is a protein or a peptide.

[0129]

[0117] The protein may be any therapeutic protein, such as an antibody, an antibody fragment, an antibody derived molecule (e.g., scFv, bi- or multi-specific antibodies) or a fusion protein (e.g., an Fc fusion protein), growth factors, blood coagulation factors, vaccines, cytokines, interferons, hormones and fusion proteins. The peptide may be modified by post-translational modification (e.g. glycosylation, acetylation, phosphorylation, glycation, glycosylation or peptide processing) or chemically. Proteins are typically produced using biological means, such as genetic expression, including production in prokaryotic and eukaryotic cells.

[0130]

[0118] Peptide may be any therapeutic peptide (e.g., a peptide containing less than 30-40 amino acid residues or a peptide containing a backbone of equal or less than 30 amino acids). The peptide may be modified by post-translational modification (e.g. glycosylation, acetylation, phosphorylation, glycation, glycosylation or peptide processing) or chemically. Therapeutic peptides may be produced by (i) expression in host cells using recombinant technologies, (ii) chemical or chemo-enzymatic synthesis or (iii) enzymatic degradation or fermentation of proteins. Particularly, peptides may be produced by chemical synthesis linking individual amino acid residues via peptide bonds. In chemically synthesized peptide also non-proteinogenic amino acids may be incorporated. Peptides may be further chemically modified, e.g., by chemically coupling polyethylene glycol (PEGylation) or fatty acids. Particularly peptides may be coupled to fatty acids as a half-life extending principle by fatty acylation (promotes high affinity albumin binding). In certain embodiments the peptide comprises non-proteinogenic amino acid residues and / or comprises a half-life extending principle, such as being coupled to a fatty acid by fatty acylation 120333P1140PC

[0131]

[0119] Nucleic acid may be any therapeutic DNA or RNA, such as a DNA or a RNA coding for a therapeutic protein (DNA and RNA) or a non-coding RNA (e.g., shRNA, miRNA). The nucleic acid may be provided as naked DNA or RNA or packaged as lipid nanoparticle complexes or lipid nanoparticles (LNPs) also packaged LNPs etc. Particulary, therapeutic RNAs often contain modified bases (NTPs) to increase stability and to minimize immunogenicity.

[0132]

[0120] Virus may be any therapeutic virus, such as oncolytic viruses and viruses used for delivery in gene therapy and / or vaccination. Suitable viruses include without being limited thereto adeno associated virus (AAV) and vesicular stomatitis virus (VSV).

[0133]

[0121] The term “therapeutic” as used herein in the context of the API (protein, peptide, nucleic acid or virus) refers to an API that elicits a therapeutic effect when administered to a subject. The therapeutic effect as referred to herein also includes a prophylactic effect, such as for a vaccine (protein, peptide, nucleic acid or virus) that is to be administered to prevent a disease or an infection.

[0134]

[0122] Thus, the method according to the invention can be advantageously used for quantitative analysis of aqueous samples comprising a therapeutic protein, such as an antibody (e.g., an IgG antibody), an antibody fragment, an antibody derived molecule or a fusion protein (e.g., an Fc fusion protein). Typically, an antibody is mono-specific, but an antibody may also be multi-specific. Thus, the method according to the invention may be used for samples comprising mono-specific antibodies, multi-specific antibodies, or fragments thereof, preferably of antibodies (mono-specific), bispecific antibodies, trispecific antibodies or fragments thereof, preferably antigen-binding fragments thereof. Exemplary antibodies within the scope of the present invention include but are not limited to anti-CD2, anti-CD3, anti-CD20, anti-CD22, anti-CD30, anti-CD33, anti-CD37, anti-CD40, anti-CD44, anti-CD44v6, anti-CD49d, anti-CD52, anti-EGFR1 (HER1), anti-EGFR2 (HER2), anti-GD3, anti-IGF, anti-VEGF, anti-TNFalpha, anti-IL2, anti-IL-5R, anti-IL-36R or anti-lgE antibodies, and are preferably selected from the group consisting of anti-CD20, anti-CD33, anti-CD37, anti-CD40, anti-CD44, anti-CD52, anti-HER2 / neu (erbB2), anti-EGFR, anti-IGF, anti-VEGF, anti-TNFalpha, anti-l L2, anti-IL-36R and anti-lgE antibodies.

[0135]

[0123] The term “antibody”, "antibodies", or "immunoglobulin(s)" as used herein relates to proteins selected from among the globulins, which are naturally formed as a reaction of the host organism to a foreign substance (=antigen) from differentiated B-lymphocytes (plasma cells). There are various classes of immunoglobulins: IgA, IgD, IgE, IgG, IgM, IgY, IgW. Preferably the antibody is an IgG antibody, more preferably an lgG1 or an lgG4 antibody. The terms immunoglobulin and antibody are used interchangeably herein. Antibody includes monoclonal, monospecific and multi-specific (such as bispecific or trispecific) antibodies, a 120333P1140PC single chain antibody, an antigen-binding fragment of an antibody (e.g., an Fab or F(ab')2 fragment), a disulfide-linked Fv, etc. Antibodies can be of any species and include chimeric and humanized antibodies. “Chimeric” antibodies are molecules in which antibody domains or regions are derived from different species. For example, the variable region of heavy and light chain can be derived from rat or mouse antibody and the constant regions from a human antibody. In “humanized” antibodies only minimal sequences are derived from a non-human species. Often only the CDR amino acid residues of a human antibody are replaced with the CDR amino acid residues of a non-human species such as mouse, rat, rabbit or llama. Sometimes a few key framework amino acid residues with impact on antigen binding specificity and affinity are also replaced by non-human amino acid residues. Antibodies may be produced through chemical synthesis, via recombinant or transgenic means, via cell (e.g., hybridoma) culture, or by other means.

[0136]

[0124] Typically antibodies are tetrameric polypeptides composed of two pairs of a heterodimer each formed by a heavy and a light chain. Stabilization of both the heterodimers as well as the tetrameric polypeptide structure occurs via interchain disulfide bridges. Each chain is composed of structural domains called “immunoglobulin domains” or “immunoglobulin regions” whereby the terms “domain” or “region” are used interchangeably. Each domain contains about 70 - 110 amino acids and forms a compact three-dimensional structure. Both heavy and light chain contain at their N-terminal end a “variable domain” or “variable region” with less conserved sequences which is responsible for antigen recognition and binding. The variable region of the light chain is also referred to as “VL” and the variable region of the heavy chain as “VH”.

[0137]

[0125] Antigen-binding fragments include without being limited thereto e.g. “Fab fragments” (Fragment antigen-binding = Fab). Fab fragments consist of the variable regions of both chains, which are held together by the adjacent constant region. These may be formed by protease digestion, e.g. with papain, from conventional antibodies, but similarly Fab fragments may also be produced by genetic engineering. Further antibody fragments include F(ab‘)2 fragments, which may be prepared by proteolytic cleavage with pepsin.

[0138]

[0126] Using genetic engineering methods it is possible to produce shortened antibody fragments which consist only of the variable regions of the heavy (VH) and of the light chain (VL). These are referred to as Fv fragments (Fragment variable = fragment of the variable part). Since these Fv-fragments lack the covalent bonding of the two chains by the cysteines of the constant chains, the Fv fragments are often stabilized. It is advantageous to link the variable regions of the heavy and of the light chain by a short peptide fragment, e.g. of 10 to 30 amino acids, preferably 15 amino acids. In this way a single peptide strand is obtained consisting of VH and VL, linked by a peptide linker. An antibody protein of this kind is known 120333P1140PC as a single-chain-Fv (scFv). Examples of scFv-antibody proteins are known to the person skilled in the art. Thus, antibody fragments and antigen-binding fragments further include Fv- fragments and particularly scFv.

[0139]

[0127] In recent years, various strategies have been developed for preparing scFv as a multimeric derivative. This is intended to lead, in particular, to recombinant antibodies with improved pharmacokinetic and biodistribution properties as well as with increased binding avidity. In order to achieve multimerisation of the scFv, scFv were prepared as fusion proteins with multimerisation domains. The multimerisation domains may be, e.g. the CH3 region of an IgG or coiled coil structure (helix structures) such as Leucine-zipper domains. However, there are also strategies in which the interaction between the VH / VL regions of the scFv is used for the multimerisation (e.g. dia-, tri- and pentabodies). By diabody the skilled person means a bivalent homodimeric scFv derivative. The shortening of the linker in a scFv molecule to 5 - 10 amino acids leads to the formation of homodimers in which an inter-chain VH / VL-superimposition takes place. Diabodies may additionally be stabilized by the incorporation of disulphide bridges. Examples of diabody-antibody proteins are known from the prior art.

[0140]

[0128] By minibody the skilled person means a bivalent, homodimeric scFv derivative. It consists of a fusion protein which contains the CH3 region of an immunoglobulin, preferably IgG, most preferably lgG1 as the dimerisation region which is connected to the scFv via a Hinge region (e.g. also from lgG1) and a linker region. Examples of minibody-antibody proteins are known from the prior art.

[0141]

[0129] By triabody the skilled person means a: trivalent homotrimeric scFv derivative. ScFv derivatives wherein VH-VL is fused directly without a linker sequence lead to the formation of trimers.

[0142]

[0130] The skilled person will also be familiar with so-called miniantibodies which have a bi-, tri- or tetravalent structure and are derived from scFv. The multimerisation is carried out by di-, tri- or tetrameric coiled coil structures. In a preferred embodiment of the present invention, the gene of interest is encoded for any of those desired polypeptides mentioned above, preferably for a monoclonal antibody, a derivative or fragment thereof.

[0143]

[0131] The immunoglobulin fragments composed of the CH2 and CH3 domains of the antibody heavy chain are called “Fc fragments”, “Fc region” or “Fc” because of their crystallization propensity (Fc = fragment crystallizable). These may be formed by protease digestion, e.g. with papain or pepsin from conventional antibodies but may also be produced by genetic engineering. The N-terminal part of the Fc fragment might vary depending on how many amino acids of the hinge region are still present. 120333P1140PC

[0144]

[0132] Antibodies comprising an antigen-binding fragment and an Fc region may also be referred to as full-length antibody. Full-length antibody may be mono-specific and multispecific antibodies, such as bispecific or trispecific antibodies.

[0145]

[0133] Preferred therapeutic antibodies according to the invention are multispecific antibodies, particularly bispecific or trispecific antibodies. Bispecific antibodies typically combine antigenbinding specificities for target cells (e.g., malignant B cells) and effector cells (e.g., T cells, NK cells or macrophages) in one molecule. Exemplary bispecific antibodies, without being limited thereto are diabodies, BiTE (Bi-specific T-cell Engager) formats and DART (DualAffinity Re-Targeting) formats. The diabody format separates cognate variable domains of heavy and light chains of the two antigen binding specificities on two separate polypeptide chains, with the two polypeptide chains being associated non-covalently. The DART format is based on the diabody format, but it provides additional stabilization through a C-terminal disulfide bridge. Trispecific antibodies are monoclonal antibodies which combine three antigen-binding specificities. They may be build on bispecific-antibody technology that reconfigures the antigen-recognition domain of two different antibodies into one bispecific molecule. For example, trispecific antibodies have been generated that target CD38 on cancer cells and CD3 and CD28 on T cells. Multispecific antibodies are particularly difficult to product with high product quality.

[0146]

[0134] Another preferred therapeutic protein is a fusion protein, such as an Fc-fusion protein. Thus, the invention can be advantageously used for production of fusion proteins, such as Fc-fusion proteins. Furthermore, the method of increasing protein producing according to the invention can be advantageously used for production of fusion proteins, such as Fc-fusion proteins.

[0147]

[0135] The effector part of the fusion protein can be the complete sequence or any part of the sequence of a natural or modified heterologous protein. The immunoglobulin constant domain sequences may be obtained from any immunoglobulin subtypes, such as I gG 1 , 1 gG2, lgG3, lgG4, lgA1 or lgA2 subtypes or classes such as IgA, IgE, IgD or IgM. Preferentially they are derived from human immunoglobulin, more preferred from human IgG and even more preferred from human IgG 1 and lgG2. Non-limiting examples of Fc-fusion proteins are MCP1-Fc, ICAM-Fc, EPO-Fc and scFv fragments or the like coupled to the CH2 domain of the heavy chain immunoglobulin constant region comprising the N-linked glycosylation site. Fc-fusion proteins can be constructed by genetic engineering approaches by introducing the CH2 domain of the heavy chain immunoglobulin constant region comprising the N-linked glycosylation site into another expression construct comprising for example other immunoglobulin domains, enzymatically active protein portions, or effector domains. Thus, an Fc-fusion protein according to the present invention comprises also a single chain Fv 120333P1140PC fragment linked to the CH2 domain of the heavy chain immunoglobulin constant region comprising e.g. the N-linked glycosylation site.

[0148]

[0136] The biologic pharmaceutical ingredient may be produced by chemical or chemo- enzymatic synthesis (e.g. protein, peptide or nucleic acid). Alternatively, the biologic pharmaceutical ingredient (e.g. protein, peptide, nucleic acid or virus) may be produced in a prokaryotic or eukaryotic cell in cell culture. Particularly peptides and small non-glycosylated proteins may be produced in prokaryotic cells, such as Escherichia coli. Larger proteins and particularly viruses are preferably produced in eukaryotic cells. Preferably, the eukaryotic cell used for producing proteins is a yeast cell (e.g., Saccharomyces, Klyveromyces) or a mammalian cell (e.g., hamster or human cells). Yeast cells can be, without being limited thereto Saccharomyces cerevisiae, Pichia pastoris, Klyveromyces lactis or marxianus. The mammalian cell is preferably a CHO cell, a HEK 293 cell or a derivative thereof. HEK293 cells include without being limited thereto HEK293 cells, HEK293T cells, HEK293F cells, Expi293F cells or derivatives thereof. Commonly used CHO cells for large-scale industrial production are often engineered to improve their characteristics in the production process, or to facilitate selection of recombinant cells. Such engineering includes, but is not limited to increasing apoptosis resistance, reducing autophagy, increasing cell proliferation, altered expression of cell-cycle regulating proteins, chaperone engineering, engineering of the unfolded protein response (UPR), engineering of secretion pathways and metabolic engineering.

[0149]

[0137] Preferably, CHO cells that allow for efficient cell line development processes are metabolically engineered, such as by glutamine synthetase (GS) knockout and / or dihydrofolate reductase (DHFR) knockout to facilitate selection with methionine sulfoximine (MSX) or methotrexate, respectively.

[0150]

[0138] Preferably, the CHO cell used for producing the recombinant protein of interest is a CHO-DG44 cell, a CHO-K1 cell, a CHO-DXB11 cell, a CHO-S cell, a CHO glutamine synthetase (GS)-deficient cell or a derivative of any of these cells.

[0151]

[0139] Non-limiting examples of mammalian cells which can be used in the meaning of this invention are also summarized in Table A. However, derivatives / progenies of those cells, other mammalian cells, including but not limited to human, mice, rat, monkey, and rodent cell lines, can also be used in the present invention, particularly for the production of biopharmaceutical proteins, such as the recombinant protein of interest. 120333P1140PC

[0152] Table A: Exemplary mammalian production cell lines

[0153] 1CAP (CEVEC's Amniocyte Production) cells are an immortalized cell line based on primary human amniocytes. They were generated by transfection of these primary cells with a vector containing the functions E1 and pIX of adenovirus 5. CAP cells allow for competitive stable production of recombinant proteins with excellent biologic activity and therapeutic efficacy as a result of authentic human posttranslational modification. 120333P1140PC

[0154]

[0140] Mammalian cells are most preferred, when being established, adapted, and completely cultivated under serum free conditions, and optionally in media, which are free of any protein / peptide of animal origin. Commercially available media such as Ham's F12 (Sigma, Deisenhofen, Germany), RPMI-1640 (Sigma), Dulbecco's Modified Eagle's Medium (DMEM; Sigma), Minimal Essential Medium (MEM; Sigma), Iscove's Modified Dulbecco's Medium (IMDM; Sigma), CD-CHO (Invitrogen, Carlsbad, CA), serum-free CHO Medium (Sigma), and protein-free CHO Medium (Sigma) are exemplary appropriate nutrient solutions. Any of the media may be supplemented as necessary with a variety of compounds, non-limiting examples of which are recombinant hormones and / or other recombinant growth factors (such as insulin, transferrin, epidermal growth factor, insulin like growth factor), salts (such as sodium chloride, calcium, magnesium, phosphate), buffers (such as HEPES), nucleosides (such as adenosine, thymidine), glutamine, glucose or other equivalent energy sources, antibiotics and trace elements. Any other necessary supplements may also be included at appropriate concentrations that would be known to those skilled in the art. For the growth and selection of genetically modified cells expressing a selectable gene a suitable selection agent is added to the culture medium.

[0155]

[0141] A recombinant protein of interest may be produced in eukaryotic cells in cell culture. Following expression, the recombinant protein is harvested and further purified. The recombinant protein of interest may be recovered from the culture medium as a secreted protein in the harvested cell culture fluid (HCCF) or from a cell lysate (i.e. , the fluid containing the content of a cell lysed by any means, including without being limited thereto enzymatic, chemical, osmotic, mechanical and / or physical disruption of the cell membrane and optionally cell wall) and purified using techniques well known in the art. The samples obtained and / or analyzed at the various steps of purification are also referred to as in-process control (I PC) samples or process intermediates. The harvest typically includes centrifugation and / or filtration, such as to produce a harvested cell culture fluid or cell lysate, preferably harvested cell culture fluid. Thus, the harvested cell culture fluid or the cell lysate may also be referred to as clarified harvested cell culture fluid or clarified cell lysate. It does not contain living cells and cell debris as well as most cell components have been removed. Clarified typically means centrifugation or filtration, preferably filtration. Further process steps may include affinity chromatography, particularly Protein A column chromatography for antibodies or Fc- containing proteins, to separate the product from contaminants. Further process steps may include acid treatment to inactivate viruses, clarifying the product pool by depth filtration, preferably following acid treatment, to remove cell contaminants, such as HCPs and DNA. Further process steps may include in this order or any other order as may be appropriate in the individual case: ion exchange chromatography, particularly anion exchange 120333P1140PC chromatography to further remove contaminating cell components and / or cation exchange chromatography to remove product related contaminants, such as aggregates. Alternatively mixed mode chromatography and / or hydrophobic interaction chromatography may be used. Further, preferably following process steps may include nanofiltration to further remove viruses and ultrafiltration and diafiltration to concentrate the recombinant protein of interest and to exchange buffer, respectively.

[0156]

[0142] The method according to the present invention may be particularly useful for analyzing process intermediates after (preferably before and after) purification steps that remove HCPs and other contaminants in order to adapt the relevant step to more efficiently purify the protein during process intermediates, such as before and after affinity chromatography, before and after depth filtration in combination with acid treatment and / or before and after ion exchange chromatography (anion exchange chromatography and / or cation exchange chromatography), multi-mode chromatography and / or hydrophobic interaction chromatography. In some embodiments the method comprises (e.g., for antibody or antibody related proteins) obtaining at least one sample after affinity chromatography, and / or after depth filtration in combination with acid treatment (or after acid treatment and / or after depth filtration) and / or after ion exchange chromatography (such as anion exchange chromatography and / or cation exchange chromatography), preferably anion exchange chromatography. In some embodiments the method comprises obtaining at least one sample before and after affinity chromatography and / or before and after depth filtration in combination with acid treatment (or before and after acid treatment and / or before and after depth filtration) and / or before and after ion exchange chromatography, such as anion exchange chromatography and / or cation exchange chromatography, preferably anion exchange chromatography. The person skilled in the art will know that the sample obtained after a certain method step may be the same as the sample obtained before the following method step, such as the sample obtained after affinity chromatography (e.g., Protein A chromatography) may be the same sample as the sample before acid treatment (or before depth filtration in combination with, i.e., following, acid treatment). Other samples that may be analyzed using the method according to the invention are samples after ultrafiltration / diafiltration (LIF / DF samples) drug substance or drug product samples. Drug substance or drug product samples comprise formulation buffer and therefore often contain polysorbate, at typical concentration of 0.2 to 0.8 mg / ml polysorbate.

[0157]

[0143] The method of manufacturing a recombinant protein of interest typically comprises an ultrafiltration diafiltration step using tangential flow filtration (TTF) to provide the purified protein (e.g., antibody) product pool. The purified recombinant protein of interest is buffer exchanged and further concentrated by ultrafiltration / diafiltration. Polysorbate and / or other 120333P1140PC excipients are then added to the concentrated recombinant protein (e.g., antibody), or the recombinant protein comprising other excipients of the formulation.

[0158]

[0144] In certain embodiments the method for the production of an active pharmaceutical ingredient according to the invention comprises obtaining at least one sample comprising a therapeutic protein (e.g., an antibody or antibody related protein), e.g., in a step of harvesting the recombinant protein of interest (in step (i), wherein the sample is a harvested cell culture fluid (HCCF) or a cell lysate; in a step of purifying the recombinant protein of interest (in step (ii)), in a step of ultrafiltration and diafiltration (LIF / DF) of the recombinant protein of interest into an aqueous solution; and in an optional step of further formulating the recombinant protein of interest into a pharmaceutically acceptable formulation suitable for administration (in step (iv)) wherein the method comprises quality testing of at least one aqueous in-process control (IPC) sample of the process of steps (ii), (iii) and / or (iv), wherein the sample comprises the recombinant protein of interest using the method according to the first aspect of the invention. Preferably, the method for the production of an active pharmaceutical ingredient according to the invention comprises obtaining at least one sample comprising the protein of interest in step (ii), wherein the sample is an in-process control (IPC) sample, such as comprising obtaining at least one sample after affinity chromatography, after depth filtration following acid treatment (or after acid treatment and / or after acid treatment), and / or after a polishing step, such as selected from ion exchange chromatography, preferably anion exchange chromatography or cation exchange chromatography, multi-mode chromatography (MMC) and hydrophobic interaction chromatography (HIC) or combinations thereof. In certain embodiments the polishing step comprises (a) ion exchange chromatography or (b) MMC and / or HIC. Other samples that may be analyzed using the method according to the invention are samples after ultrafiltration / diafiltration (LIF / DF samples), drug substance or drug product samples. A variation of the concentration of a component in said at least one aqueous in-process-control sample from a pre-determined concentration range results in an adaptation of the purification of the recombinant protein of interest of step (ii), the LIF / DF of the recombinant protein of interest of step (iii) and / or the further formulating of the recombinant protein of interest of step (iv), thereby controlling and / or regulating the production of the recombinant protein of interest.

[0159] EXAMPLES

[0160] 1. FT IR extinction coefficient-based quantification

[0161] 1.1 FT IR extinction coefficient-based quantification workflow

[0162]

[0145] The workflow of the FT IR extinction coefficient-based quantification comprises three main steps. First, the IR-extinction coefficient spectra of each component in the mixture were collected from samples containing known amounts of the respective compounds. The 120333P1140PC developed data analysis method can use spectra with mixtures of multiple components as well as spectra of single compounds. Because of the easier exchangeability of the spectra to different projects and in order to reduce the likelihood of mistakes, usually spectra of single compounds with a known concentration dissolved in water were used to calculate the IR- extinction coefficient spectra. In the second step, the determined extinction coefficients were verified using mixtures of known concentrations. In the third step, the IR-extinction coefficient spectra were used to calculate the concentrations in the spectrum of an unknown sample through least-squares-fitting of the data.

[0163] 1.2 Determination of the excipient IR-extinction coefficient spectrum

[0164]

[0146] The determination of the excipient IR-extinction coefficient spectrum has been done for each compound using the following procedure. In addition to protein formulations, this procedure can also be used directly for the IR-extinction coefficient spectrum determination of water-soluble small molecule APIs like NCEs or peptides which are obtained in a pure, solid, or liquid form where the concentration in the solution can be determined via weighing.

[0165]

[0147] A dilution series of each component with exactly known concentrations was prepared. The highest and lowest concentrations were chosen to at least cover the relevant measurement range for each component. The highest concentration should yield a good signal-to-noise ratio in the spectrum. This step is optional, as a single sample of defined concentration yielding good signal-to-noise ratio in the spectrum is sufficient to determine the extinction coefficient of a compound, if the FT I R extinction spectra of the compound follow the Beer-Lambert law. Limitations of extinction to concentration linearity can occur e.g. due to high solution viscosity of the injected solution, due to interactions between the molecules at very high concentrations or due micellar formation

[0166]

[0148] For the API, particularly proteins with residual contaminations of small molecules, the original protein sample was dialyzed against water to remove the contaminants. The content of the dialyzed solution (or original solution without dialysis) was determined via UV-Vis spectroscopy. For APIs with unknown UV Vis extinction coefficients, the content may be directly calculated from the UV Vis spectrum for a known sequence of the protein using the Edelhoch method by unfolding of the protein in 6 M guanidinium hydrochloride solution. In principle, this works without the need of a previously determined UV-Vis extinction coefficient in water, which may add additional uncertainty in the content determination. For active projects with an already determined UV Vis extinction coefficient, the concentration was instead determined from the protein solution in order to ensure identical concentration values in the later on determined 120333P1140PC samples compared to the established UV Vis measurements of this project. Usually, the pH of the sample was not adjusted.

[0167]

[0149] The API solution was then directly used for the determination of the IR-extinction coefficient spectrum of the most concentrated API solution. The solution was additionally diluted to the further desired concentrations of the dilution series.

[0168]

[0150] For pure solid or liquid samples, the amount of the substance necessary for the different dilution steps was weighed in and the substance was dissolved in a small amount of water. For compounds which can be protonated or deprotonated with the pH, the pH was adjusted to the target pH of the later measurement by the addition of HCI or NaOH or other suitable acids or bases until the correct pH was reached. Alternatively, two solutions, one with the fully or partially protonated and one with the fully or partially deprotonated compound can be prepared. For compounds which cannot be protonated or deprotonated, the solution was used without further pH adjustment. The solution was diluted with water to an exactly defined volume via a volumetric flask.

[0169]

[0151] FT IR measurement of each solution (multiple replicates were recorded) was performed using an autosampler. The autosampler reduced measurement time for the operator.

[0170]

[0152] The different IR-extinction coefficient spectra were determined by least-squares- fitting of the replicate spectra of each concentration for the respective concentrations. Comparison between the different concentrations for each component yielded the optimal concentration for the extinction coefficient determination for the desired measurement range.

[0171]

[0153] Guidelines for the selection of the optimal concentration are: (A) typically the highest concentration to optimize the signal-to-noise ratio, (B) no concentration dependent shifts in the signals should be detected in the spectra over the measurement range (which was only observed very rarely), (C) The IR-extinction coefficient spectra should be comparable to the other, lower concentrations in the measurement range, which is equivalent to the linearity of the measurement (due to the high viscosity, sometimes the highest concentrations cannot be measured accurately and the range should be reduced accordingly).

[0172] 1.3 Verification of the determined IR-extinction coefficient spectra

[0173]

[0154] Test mixtures with exactly determined concentrations were prepared using different combinations of higher and lower concentrations for the different components. 120333P1140PC

[0174] FT IR measurement of each mixture (multiple replicates were recorded) were performed using an autosampler, which reduces measurement time for the operator.

[0175]

[0155] The concentration was determined by approximation of the product of the concentrations and the first derivative spectra of the IR-extinction coefficient spectra to the first derivative sample spectra for all recorded spectra via a least-squares-algorithm. Comparison of the measured concentrations to the nominal concentrations gives the measurement accuracies at the different tested concentrations.

[0176] 1.4 Simultaneous quantification with the previously determined IR-extinction coefficient spectra

[0177]

[0156] Samples were filled into GC-vials. FT IR measurement of each sample (multiple replicates were recorded) were performed using an autosampler, which reduces measurement time for the operator.

[0178]

[0157] Determination of the concentrations by least-squares-fitting of the first derivative spectra of the IR-extinction coefficient spectra to the first derivative sample spectra for all recorded spectra.

[0179] 1.5 Used / suitable measurement devices

[0180]

[0158] The FT IR spectra were recorded on a MIRA Analyzer, a mid-infrared FT IR- spectrometer manufactured by Clade GmbH using a MIRA eCell flow-through transmission cell transmission cell for aqueous solutions with a pathlength of about 8 pm and an integrated autosampler. The eCell transmission cell can in principle be also used in other FT IR spectrometers, but the MIRA Analyzer is able to achieve an up to now unprecedented accuracy and repeatability in the resulting spectra which is essential for a reliable quantification. The developed method can be used to quantify nearly all substances in aqueous solution in the right concentration range (usually > 0.5 g L'1).

[0181] 1.6 Components tested for the FT IR based quantification

[0182]

[0159] Without being limited thereto, components tested and suitable for the FT IR- based quantification are:

[0183] APIs:

[0184] • Proteins (e.g. BSA, Lysozyme, IgGs, new antibody formats)

[0185] • Peptides (e.g., comprising a backbone of 20-30 amino acids, including non- proteinogenic amino acids and acylated peptide chemically modified with a fatty acid via a small peptide linker)

[0186] • Liquid formulations of small molecules

[0187] Buffering substances:

[0188] • Amino acids (e.g. histidine, arginine, methionine, lysine etc.) 120333P1140PC

[0189] • Salts (e.g. acetate, citrate, phosphate, Tris / TrisHCI etc.)

[0190] Tonicity agents:

[0191] • Sugars (e.g. sucrose, mannitol, trehalose, sorbitol)

[0192] • Other tonicity agents (e.g. propylene glycole)

[0193] Preservatives:

[0194] • e.g. phenol, m-cresol

[0195] Detergents (with slightly reduced accuracy, due to the small amount in the solution):

[0196] • e.g. PS20, PS80, Poloxamer 188, SDS

[0197] Further excipients:

[0198] • e.g. Chloride salts (with slightly reduced accuracy, since the IR signal is very weak), sulfate salts

[0199] Lipids:

[0200] • Liposomes (about 100 nm)

[0201] Example 1 : Step I - Determination of the IR-extinction coefficient spectra for two exemplary excipients

[0202]

[0160] In order to further verify the method, an evaluation was performed for the spectra of all the excipients in an antibody formulation to generate the respective IR-extinction coefficient spectra and results are exemplarily demonstrated for one excipient (sucrose) and a protein (monoclonal IgG) (Figures 1-4). The same experiment was performed for all the tested excipients mentioned above, as well as for further APIs (new formats, peptides) in the same manner.

[0203] 1. Preparation of a dilution series

[0204]

[0161] For the measurement of the IR-extinction coefficient spectra of the excipients, e.g. sucrose, approximately the necessary amounts of sucrose needed for the target concentrations in 25 mL were weighed in and the exact concentrations in the sample given in Table 1 were determined. Afterwards, the sample was dissolved in a 25 mL volumetric flask and diluted with water to the desired volume.

[0205] Table 1 : Concentrations of different sucrose solutions 120333P1140PC

[0206]

[0162] The protein dilution series was prepared in 5 mL samples using a 5 mL volumetric flask. The amounts of the protein dialyzed against water needed for the target concentrations in 5 mL were taken from the dialysed IgG sample (IgG dia 2) and added via pipette. The exact amount was measured via weighing. Afterwards, the sample was diluted with MilliQ water to 5 mL in a volumetric flask. The determined amounts, including calculated and measured concentrations, of the samples are given in Table 2. The calculated concentrations were then calculated from the weight, density and concentration of the added dialyzed protein solution. The concentration of the dialyzed protein stock solution was determined via UV absorption using the experimentally determined extinction coefficient of the protein at 280 nm in water. Additionally, the concentration of each diluted sample was also determined via the UV absorption at 280 nm. During the method development, it was shown that it is also possible to reduce the amount of protein solution to 1mL or even lower for each diluted sample by pipetting the solution and afterwards using the UV concentration or by weighing in the stock solution with a density correction for the calculation of the concentration without losing accuracy of the measurement.

[0207] Table 2: Protein concentrations of different protein dilutions in the 5 mL scale determined from the weights of the used protein solutions and the mean concentrations determined via UV absorption. A dialysed IgG antibody was used as the stock protein solution.

[0208] 2. Measurement of the FT IR spectra at different concentrations

[0209]

[0163] Figure 1 shows the mean spectra obtained from the measured replicates for each sucrose and protein concentration. The deviation of the different single spectra is very small and therefore not represented separately. The extinction values have the unit pm-1, because these are pathlength normalized extinction values. Since these spectra are the raw spectra obtained from the MIRA Analyzer, the term extinction is used in this document for simplicity reasons.

[0210] 3. Evaluation of the linearity of the extinction

[0211]

[0164] In order for the FT IR quantification to work, the IR extinction of the different components in the solution over the whole wavenumber range has to be strictly linear 120333P1140PC over the concentration range. The linearity of the extinction with the concentration is characterized through the respective extinction values at the three highest maxima of the original, non-derived spectra obtained for each concentration. The values were plotted in Figure 2 for sucrose and the protein. The datapoints obtained from different replicate spectra of the same concentration are nearly identical and overlap in the figures.

[0212]

[0165] The linear fits of the datapoints are in very good agreement with the measured datapoints which can be seen by the Revalues of nearly 1. In Table 3 the fit parameters and Revalues for the different maxima of the sucrose spectra are given.

[0213] Table 3: Linear fits for the linearities at different maxima of the sucrose spectra from Figure 2 and their Revalues.

[0214] In Table 4 the fit parameters and Revalues for the different maxima of the protein spectra are given.

[0215] Table 4: Linear fits for the linearities at different maxima of the protein spectra from Figure 2 and their Revalues.

[0216] 4. Calculation of the IR-extinction coefficient spectra at different concentrations

[0217]

[0166] The first step of the simultaneous quantification of excipients or proteins from a mixture via FT I R is the determination of the infrared extinction coefficient spectra of the individual components. This can either be achieved by measuring a sample of the pure component in a defined concentration or via the measurements of several samples consisting of mixtures having different concentration ratios of the individual components. In either way, the infrared extinction coefficient spectra can be deduced based on the matrix Equation 1. Equation 1 also takes different replicate spectra of the different components into account. 120333P1140PC

[0218]

[0167] Considered were k components in the range [1 ,..i,...k], I samples measurements [1 and n wavenumbers [1 ,...m,...n], With the measured pathlength-normalized infrared extinction spectra and the concentrations c7, being known, the IR-extinction coefficient spectra for the individual species can be deduced via a numerical leastsquare solver (solving Ax=B with matrix A and B known). In the present case the data evaluation program was implemented in a Python script.

[0219]

[0168] It should be noted that in order to deduce extinction coefficient information on e.g. k components at least k different samples with varying concentration ratios of the k- components (or k samples of pure components) need to be analyzed to allow for numerically solving Equation 1 for spim. On the other hand, measuring more than k samples of varying concentration ratios or using additional replicates of samples leads to the retrieval of the best fitting solution for £Viim.

[0220]

[0169] Depending on the given unit for the concentration, the IR-extinction coefficient spectra can be related to the molar concentration or the mass concentration. For each component, a different unit can be attributed to the concentration which then yields the respective unit also in the quantification.

[0221]

[0170] The respective IR-extinction coefficient spectra of sucrose and the protein were calculated for each concentration from all replicates of this concentration in Figure 3. The agreement between the different IR-extinction coefficient spectra is very good. Only very small differences can be seen mainly at low concentrations where the measurement noise of the analyzer is larger compared to the signal.

[0222] 5. Correlation of the IR-extinction coefficient spectra

[0223]

[0171] To evaluate the correlation of the IR-extinction coefficient spectra obtained from different concentrations, the Pearson correlation coefficient rxyof the IR-extinction coefficient spectra originating from different concentrations was calculated, wherein 1 indicates a perfect match and 0 indicates no correlation. Figure 4 shows a heatmap of those correlation coefficients for the spectra of sucrose and the protein. The correlation is high for all concentrations. The correlation is slightly lower for the highest concentration of sucrose because of the high viscosity of the sample and for the lowest concentration due to the larger influence of the measurement noise. For sucrose, correlation coefficients > 0.9994 are observed with a Pearson correlation coefficient > 0.9998 for concentrations in the range of 100-500 mM and > 0.9999 in the most common concentration range of 100 — 250 mM. For the Protein, an even better Pearson correlation coefficient of > 0.999990 was observed over the entire concentration range of 13.74 g / L to 73.52 g / L. 120333P1140PC

[0224] 6. Selection of the best concentration for the extinction coefficient determination

[0225]

[0172] In the present case a dilution serious was measured, although a single sample at a determined concentration is typically sufficient. The range of the dilution row measured for the determination of the IR-extinction coefficient spectrum preferably should be chosen to cover the necessary measurement range. The highest concentration should be at least 10 - 15 g / L in order to obtain good IR-extinction coefficient spectra. Typically, the spectra of the highest concentration are chosen for the IR-extinction coefficient spectra calculation to maximize signal-to-noise ratio. At the same time, no concentration dependent shifts in the signals should be detected over the measurement range, because that would have an impact on the quantification. Such shifts were only observed very rarely and were not very large. Finally, the IR-extinction coefficient spectra should be comparable to the other, lower concentrations in the measurement range. Due to the high viscosity, sometimes the highest concentrations cannot be injected accurately into the analyzer anymore resulting in smaller signal intensities and the measurement range should then be reduced accordingly. Finally, only a single concentration is sufficient to determine the extinction coefficient spectra. Therefore, the dilution row is not strictly necessary, but helps identifying suitable conditions for extinction coefficient spectra determination.

[0226] Example 2: Validation of the quantification method for a given antibody formulation

[0227]

[0173] To validate whether the invention works as intended for a given formulation, in this case an IgG formulation, the IR-extinction coefficient spectra of all respective components are used to quantify a set of verification samples with known composition.

[0228] 1. Preparation of the verification samples

[0229] Method 1 :

[0230]

[0174] The verification samples can be created akin to the solutions used for the determination of the IR-extinction coefficient spectra of the individual components, except that all components in the formulation are weighed into the same vessel. Again, a protein sample dialyzed against water was used as the stock solution. For components that are not pure, solid substances, e.g. the protein solution, the mass is inferred through the density and the concentration of the stock solution determined via UV absorption at 280 nm. Alternatively, it has been shown that equally good results can be obtained if the protein concentration was determined afterwards by directly measuring the concentration via UV absorption at 280 nm for each sample, if no UV-absorbing 120333P1140PC excipients like phenol are used. The pH of the verification samples is set similarly to the procedure for individual components described above. The same goes for the following dilution with water to an exactly defined volume using a volumetric flask. For this method, a minimum volume of 5 ml has to be prepared per sample in order to achieve a suitable accuracy in the concentration of all substances in the sample using 20 - 30 ml of highly concentrated UF / DF-material for eight test samples.

[0231] Method 2:

[0232]

[0175] Since the necessary amount of the protein solution is quite high, the effort for exactly weighing in the mg range is very high for the operator and errors can occur more easily, an alternative method was established as well, which reduces the protein and time consumption. For this method, four different excipient mixtures with varying concentrations of the different components were prepared and pipetted together with varying amounts of a stock protein solution. For each excipient stock, four 1 mL test samples are created by automatically pipetting 200 pl : 800 pl, 400 pl : 600 pl, 600pl : 400 pl and 800 pl : 200 pl (protein stock : excipient stock) using a Tecan Fluent robot. The rather large, pipetted volumes ensure a high accuracy even for new protein and excipient systems without prior optimization of the pipetting procedure. The 16 test samples consume only up to 6 - 8 ml of highly concentrated UF / DF-material. The protein concentration can be verified by measuring the concentration via UV absorption at 280 nm for each sample.

[0233] Method 3:

[0234]

[0176] The protein and time consumption can be further reduced via a density corrected spiking approach. For this method, a highly concentrated excipient mixture is prepared for each sample with varying concentrations of the different components similarly to Method 1 but without the addition of API solution and the density is determined using a suitable device. The maximum achievable concentration of the sometimes dialyzed API stock solution and the therefore necessary very high concentrations and solubility problems of the excipient stock solutions can be a limiting factor of this approach. The density of the API solution is determined similarly to the excipient stock solutions. The excipient stock solution, the API stock solution and sometimes water for dilution are then pipetted into a suitable vial while separately determining the weight of each added volume. The final concentration of the different components is calculated separately using the weighed in masses and the density correction in Equation 3 where cAfinaiis the concentration of component A in the final sample, p are the respective densities and nA is the amount of component A either the molar amount or the mass, depending on the desired 120333P1140PC unit. Depending on the volume of the spiked samples (even only 300 pL or less is possible), less than 6 - 8 ml of highly concentrated UF / DF-material can be necessary for the 16 test samples.

[0235]

[0177] The nominal concentrations of the proteins and excipients are given in Table 5. For the protein the concentration determined from the weight and the UV-measurement are listed for comparison. For the evaluation, the UV-concentration was used in the present case.

[0236] Table 5: Protein and excipient concentrations of different test mixtures in the 5 ml_ scale. Dialysed IgG was used as the stock protein solution. The protein concentration was determined from the weight of the used protein stock solution and the UV measurement at 280 nm.

[0237] 2. Measurement of the FT IR spectra of the test samples

[0238]

[0178] The test samples were placed into the autosampler. The measurement was started and ran automatically.

[0239] 3. Quantification of all components through least-squares fitting

[0240]

[0179] The simultaneous quantification of all components in the sample was achieved by approximating the product of concentration and first derivative spectra of the IR- extinction coefficient spectra of all components to the first derivative sample spectrum of the mixture using a least-squares based algorithm. This can either be done for each separate spectrum yielding result concentrations only for one spectrum or for multiple spectra at once yielding the concentrations of all spectra. During the method development, using the first derivative spectra showed the highest accuracy of the results, followed by the original, non-derived spectra and then the higher derivative spectra. Therefore, usually the first derivative spectra were used.

[0241]

[0180] For one spectrum: With known IR-extinction coefficient spectra £Viimover the whole wavenumber range of the spectrum for each component and the pathlength- normalized extinction spectra EVimoriginating from each sample measurement, the 120333P1140PC respective concentrations were determined based on the matrix Equation 2. Simultaneously considered were k components in the range [1 ,..i,...k] and n wavenumbers [1 ,...m,...n],

[0242] The least-squares solver of the self-developed Python script (solving Ax=B / l with matrix A and B and pathlength I known) optimizes Equation 2 depending on the input data yielding the concentrations of all components in each spectrum.

[0243] 4. Evaluation of the quantification accuracy

[0244]

[0181] The determined concentrations of the test mixtures were compared to the respective nominal concentrations and the performance of the method was evaluated. Figure 5 shows the obtained results compared to the nominal concentration in the test samples. For each test mixture, a separate datapoint is obtained in each spectrum for the respective component. For clarity, ± 5 % of the nominal concentrations were also plotted. For large concentration ranges, the accuracy is better than ± 5 % of the nominal concentration for all components. In this test, only PS20 showed larger deviations but this is due to the small concentrations in which PS20 is present in the samples. Whether or not the measured accuracy is deemed sufficient for a given component in a specific formulation and concentration range is dependent on the requirements that are set for the respective component and must be decided accordingly.

[0245] Example 3: Quantification of unknown sample of a given IgG formulation

[0246]

[0182] The quantification of an unknown sample of a given IgG formulation is akin to the quantification of the verification samples above. After measurement using the autosampler and fitting of the IR-extinction coefficient spectra to the sample spectrum, the residuals are also looked at in order to identify possible contaminations, shifts in the IR-spectra of the components in the mixture or errors in the substances selected for quantification which become obvious due to their larger than normal residual extinction and specific spectral features in the residual FT IR-spectra.

[0247]

[0183] Representative residual first derivative extinction spectra after two fits of the test mixtures are shown in Figure 6 compared to an exemplary water spectrum which should in principle be zero and gives an estimation of instrument inherent deviations. No distinct features of different excipients can be seen anymore in any of the residual spectra and the overall magnitude is not very large compared to the water spectrum. This means that most of the signal was accounted for with the fit. 120333P1140PC

[0248] Example 4: Analysis according to the extinction coefficient method in comparison to PLS

[0249] In order to demonstrate the efficiency and accuracy of the extinction coefficient method the same samples were analysed in parallel using the commonly applied PLS model. The test samples used for the analysis comprises histidine, sucrose, mannitol, PS20 and the same protein API (IgG) as used in Examples 1 , 2 and 3.

[0250] 1. General features of a PLS model

[0251]

[0184] Partial Least Squares Regression (PLS) is a multivariate data analysis method often used in chemometrics for purposes such as quantification of individual components based on spectra of some sorts, e.g., infrared absorption spectra. As a powerful multivariate data analysis technique, PLS has the theoretical potential to be on par or more accurate and precise than the extinction coefficient method described herein. However, it comes with several drawbacks that are overcome by the extinction coefficient method.

[0252]

[0185] A PLS requires the generation of a model with the help of a training data set consisting of spectra and the corresponding target values, e.g. the concentration of a component. The training data has a huge impact on the performance of the PLS model. A large representative training dataset has to be generated for a PLS model, which can be challenging, especially if e.g. API material is limited during early development. The performance of a PLS is hugely dependent on this training data set which has to be carefully selected. A new training data set has to be created for each formulation / combination of components that is to be quantified. Usually, tens to hundreds of exactly defined samples are necessary for a good model.

[0253]

[0186] In addition, a PLS model has a blackbox character, meaning that the result and how it is obtained is not easily comprehensible for the analyst or authorities.

[0254] 2. General features of the extinction coefficient method

[0255]

[0187] The method described herein requires only the recording of one spectrum for each individual substance. More importantly, these spectra can be reused for other formulations than they were originally intended for, leading to a drastic advantage of the invented method over a PLS model in terms of time and work that has to be done before the quantification method can be used.

[0256]

[0188] The IR-extinction coefficient spectrum method is easy to comprehend even without extensive knowledge of spectroscopy and data analysis techniques. The extinction coefficient spectra can simply be multiplied by the respective determined 120333P1140PC concentrations and the sum of these spectra yields the approximated spectrum, which closely fits the sample spectrum for a successful quantification. Therefore, the results can easily be understood later on even using only Excel for example. Troubleshooting of the IR-extinction coefficient spectrum method is easy, as the residual spectra provide a visual clue as to the fit quality of the quantification in question. These residuals can also alert the analyst in case something went wrong during the measurement or if the sample itself contains unexpected components, which could be an important potential safety issue. PLS offers no easy way to detect such events, which is why the IR- extinction coefficient spectrum is preferable.

[0257]

[0189] For samples with highly defined matrices - which the samples for uses as described herein are -, the IR-extinction coefficient spectrum method delivers at least comparable quantification performance to a PLS, but requires considerably less effort to set up. The effort taken to create the PLS model can be reduced, however, this will lead to a loss in performance. If the same effort is put in for PLS and the IR-extinction coefficient spectrum method, namely n training samples for n components, the invented methods vastly outperforms PLS.

[0258] 3. Training data generation for the PLS model

[0259]

[0190] Training data should be representative of real-life samples. For highly concentrated API samples, this can be problematic, since the high concentrations often cannot be reached with a spiking approach due to the upper stock solution limitation for the API solution.

[0260]

[0191] The most accurate sample preparation would be to weigh in all components for every single training sample, which is very time consuming for large sample sets and requires large amounts of protein and is therefore not a realistic option. Instead, stock solutions of the individual excipients and the protein were prepared. The samples were pipetted using a pipetting robot to save time and reduce pipetting errors (in terms of accuracy and precision as well as in terms of not accidently pipetting the wrong volume into the wrong well).

[0261]

[0192] Samples were pipetted in such a way that the final concentrations of all components don’t / barely correlate with one another. For example, the final concentrations of the substances in the various samples were chosen with a random number generator and the pipetted volumes were then calculated from these concentrations.

[0262]

[0193] Each pipetting approach has its drawbacks, mainly the introduction of statistical and systematic errors due to the pipetting robot and the limitation of the concentrations 120333P1140PC that can be achieved. For example, for 5 components and thus 5 pipetting steps plus one step for water the stock solutions are usually diluted heavily. Therefore, the final protein concentration in samples often cannot reach the nominal concentration of e.g. the final drug product. The use of multiple stock solutions per component, to account for batch-to-batch variations, further increases the workload. Additionally, spectra of the stock solutions of the single excipients and the protein as well as water were included in the training data.

[0263]

[0194] All samples were measured with n=3 in the FT IR, which is time-consuming although automated. In total, 307 training samples were measured for this PLS model resulting in 921 spectra used for the generation of the model.

[0264] 4. Reference measurements for the extinction coefficient spectrum method

[0265]

[0195] One pure sample of each component of the sample in water was prepared using a sufficiently high concentration. The reduced number of reference samples compared to the training samples in the PLS model tremendously reduced the time, effort and resources required, particularly when determining multiple components in a sample. The samples were measured with n=3 in the FT IR (automated). In total, 5 reference samples were measured for the determination of the extinction coefficient spectra resulting in 15 spectra in total.

[0266] 5. PLS model generation:

[0267]

[0196] One PLS1 model is generated for each excipient / API as concentrations are not correlated. There are multiple ways of choosing the number of fitted PLS components. Here, the optimization of the number of PLS components was done in a five-fold cross- validation with the scoring parameter R2(Figure 7).

[0268]

[0197] The final PLS models were then trained on the complete training data set (307 samples, 921 spectra)

[0269] 6. Extinction coefficient determination

[0270]

[0198] The extinction coefficient spectra were calculated for each component in the sample as described in the general description of the FT IR extinction coefficient-based quantification under 1.2 above using the three replicate spectra of each component.

[0271] 7. Determination of the PLS and extinction coefficient model accuracy

[0272]

[0199] The PLS model was tested with a test data set consisting of samples that were created independently of the training data with weighing in of all components. More explicitly, the spectra obtained from the verification samples described in Example 2 120333P1140PC were used. The same spectra were evaluated using the extinction coefficient method and the results are shown as determined concentrations in Figure 8 and as relative deviation in concentration in Figure 9.

[0273]

[0200] For sucrose, mannitol and PS20, both methods showed comparable results (Figures 8 B-D and 9B-D). Whilst sucrose and mannitol were quantified accurately, PS20 showed a systematic error in the concentrations of some test samples with both evaluation methods (Figures 8D and 9D).

[0274]

[0201] Overall the extinction coefficient approach consistently provided equal or slightly better results for the tested samples (Figure 9A-D). This may be due to a possibly larger systematic error of the automatized PLS training sample spiking compared to the weighing in of the test samples and reference samples for the extinction coefficient method.

[0275]

[0202] For histidine and the API, the extinction coefficient method clearly outperforms the PLS method, with the latter showing a systematic deviation from the nominal values (Figures 8A, E and 9A, E). A possible explanation could be the selection of the training samples or the preparation thereof. The large amount of training samples needed are typically prepared (and are only really feasible to prepare) by pipetting together stock solutions of the individual components, which is prone to introducing systematic deviations from the intended target concentrations. This is particularly problematic if, e.g., highly viscous stock solutions are used (as is often the case for the API) that cannot be pipetted as accurately as other solutions. Then weighing in of the stock solutions and a density correction has to be applied which further complicates the sample preparation.

[0276] Example 5: Verification of the FT IR extinction coefficient-based quantification method for a given peptide formulation

[0277]

[0203] The quantification method was further tested for two formulations comprising a different API, in this case peptide formulations. Two different formulations of this API were evaluated. The peptide has a backbone of 20-30 amino acids containing acylated peptide chemically modified with a fatty acid via a small peptide linker and comprises non-proteinogenic amino acids with a molecular mass between about 4000 and 4500 Da. The extinction coefficient spectra were measured as described above for the IgG formulation. For the reference samples, all components were weighed in. The Tris solution was adjusted to pH 7.8 and the phosphate solution was adjusted to pH 7.2 using a dilute HCI solution. The phenol and Excipient 1 solutions were adjusted to pH 7.5. The solutions of the other components were used without pH adjustment. 120333P1140PC

[0278] 1. Preparation of the verification samples, phosphate formulation pH 7.2

[0279]

[0204] The verification samples listed in Table 6 were prepared via weighing in of the excipient and peptide stock solutions and subsequent spiking. In Figures 10B-14B, the deviations of the measured values from the nominal values are depicted for all components.

[0280] Table 6: Concentrations of the verification samples in Phosphate Buffer

[0281] 2. Evaluation of the quantification accuracy, phosphate buffer 7.2

[0282]

[0205] The determined concentrations of the verification sample were compared to the respective nominal concentrations and the performance of the method was evaluated. Figures 10-14 show the obtained results compared to the nominal concentration in the verification samples. For each test mixture, a separate datapoint is obtained in each spectrum for the respective component. For clarity, ± 5 % of the nominal concentrations were also plotted in Figures 10A-13A and ± 10 % of the nominal concentrations were also plotted in Figure 14A.

[0283]

[0206] Figure 10 shows the measured concentrations (A) and the relative deviations from the nominal concentrations (B) for the peptide. No larger deviations than 10 % were found. At a sample concentration of more than 2 g L-1the measured deviations were less than 2 %.

[0284]

[0207] For sodium phosphate, deviations of less than 10 % of the nominal concentration were observed at concentrations above 10 mM (Figure 11A and B). The deviation of the 120333P1140PC measured concentration compared to the nominal concentration of phenol is small for the whole measured range with less than or equal to ± 5 % (Figure 12A and B). Similarly, the measured concentration of propylene glycol deviates by a maximum of only ± 3 % from the concentration calculated by weight (Figures 13A and B). For Excipient 1 (other excipient) the measured values deviated significantly from the nominal concentrations, however, for a concentration of 0.25 g / L an acceptable deviation of up to ± 10 % was observed (Figure 14A and B). The large deviation can be explained by the small concentration of this component of sometimes even less than the usually minimal lower quantification limit of approximately 0.1 g L’1.

[0285] 3. Preparation of the verification samples, Tris formulation at pH 7.8

[0286]

[0208] The verification samples listed in Table 7 were prepared via weighing in of the excipient and peptide stock solutions and subsequent spiking. In Figures 15B-19B, the deviations of the measured values from the nominal values are depicted for all components.

[0287] Table 7: concentrations of the verification samples in Tris Buffer

[0288]

[0209] The determined concentrations of the verification samples were compared to the respective nominal concentrations and the performance of the method was evaluated. Figures 15-19 show the obtained results compared to the nominal concentration in the verification samples. For each test mixture, a separate datapoint is obtained in each spectrum for the respective component. For clarity, ± 5 % of the nominal concentrations 120333P1140PC were also plotted in Figures 15A-18A and ± 10 % of the nominal concentrations were also plotted in Figure 19A.

[0289]

[0210] Figure 15 shows the measured concentrations (A) and the relative deviations from the nominal concentrations (B) for the peptide. No larger deviations than 8 % were found. At a sample concentration of more than 2 g L'1the measured deviations were less than 3 %.

[0290]

[0211] For small concentrations (less than 2 mM) of Tris buffer the measurement deviated by up to 22%. However, for concentrations of 4 mM and above, the measurements are within a ± 10 % deviation range or less than 10 % of the nominal concentration were observed at concentrations above 10 mM (Figure 16A and B). The deviation of the measured concentration compared to the nominal concentration of Phenol was less than or equal to ± 5 % over the whole measured range (Figure 17A and B). Similarly, the measured concentration of propylene glycol deviates by a maximum of only ± 3 % from the concentration calculated by weight (Figures 18A and B). For Excipient 1 (other excipient) the measured values deviated significantly from the nominal concentrations, however, for a concentration of 0.25 g / L an acceptable deviation of up to ± 15% was observed (Figure 19A and B). The large deviation can be explained by the small concentration of this component of sometimes even less than the usually minimal lower quantification limit of approximately 0.1 g L’1.

[0291] Example 6: Verification of the FT IR extinction coefficient-based quantification method for a given BSA formulation and a given MG formulation

[0292]

[0212] The quantification method was further tested for two formulations comprising two different APIs, in this case (1) BSA as a protein API surrogate and (2) intravenous immunoglobulin (I VI G) . The extinction coefficient spectra of the proteins were measured by similar procedures as described above for the IgG formulation. In brief, the BSA stock solution in water was measured for the extinction coefficient determination. An IVIG preparation was purchased as a drug product containing excipients without detergent. The preparation was dialyzed against water to obtain a pure stock solution of the polyclonal IVIG IgGs and the extinction coefficient spectrum was determined. The nominal concentrations of both protein stock solutions were determined via UV absorption at 280 nm using extinction coefficients deduced from literature. For the excipients, the previously determined extinction coefficient spectra of the IgG formulation were directly used. This reduced the effort to set up the quantification method. For histidine, extinction coefficient spectra of histidine and histidine 120333P1140PC hydrochloride were used, and the determined histidine concentration is the sum of both determined concentrations.

[0293] 1. Preparation of the verification samples

[0294]

[0213] Verification samples covering the concentration ranges of the different components listed in Table 8 were prepared to test the performance of the quantification. The measured concentrations and relative deviations from the nominal values for all components are depicted in Figures 20A-E for the BSA formulations and in Figures 21 A- E for the IVIG formulations.

[0295] Table 8: Concentration ranges of the verification samples of the BSA and IVIG formulations

[0296]

[0214] For the preparation of the verification samples, stock solutions containing the excipients were prepared in addition to the stock solutions of the proteins and mixed using density-corrected spiking. The final concentrations of the different components in the samples were calculated as described above in Example 3 for the density-corrected spiking of the IgG.

[0297] 2. Evaluation of the quantification accuracy

[0298]

[0215] The determined concentrations of the verification samples were compared to the respective nominal concentrations in Figures 20A-E for the BSA formulation and the performance of the method was evaluated. For every replicate spectrum of the test mixtures, a separate datapoint is obtained in each of the Figures 20A-E representing the concentration of the respective component in that replicate spectrum. BSA showed a very good performance and no larger deviations than 3 % were found (Figure 20A). No larger deviations than 7.5 % were found for histidine (Figure 20B). For sucrose and mannitol, no deviations of more than 5 % were found (Figure 20C and D, respectively). Only the detergent PS20, which was present in many samples in only small amounts, 120333P1140PC showed deviations of up to 20 % below 0.8 g L-1above that concentration, deviations no larger than 7 % were found (Figure 20E).

[0299]

[0216] Figures 21 A-E showed comparable results for the verification samples of the I VI G formulation. Upon evaluating the performance of the method, again no larger deviations than 7 % were found for the API (Figure 21A), deviations no larger than 10 % were found for histidine (Figure 21 B) and no deviations above 7 % were found for sucrose and mannitol (Figure 20C and D, respectively). In this case, PS20 showed deviations of up to 24 % below 0.8 g L'1and above that concentration deviations no larger than 5 % were found (Figure 21 E).

[0300] Example 7: Verification of the FT IR extinction coefficient-based quantification method for a given proteo-Liposome formulation

[0301]

[0217] The quantification method was further tested for a proteo-liposome formulation comprising BSA as a protein API surrogate, an E. coli lipid mix (E. coli Polar Lipid Extract; Avanti Research, US) for the preparation of liposomes and sodium phosphate as buffer. Proteo-liposomes can be used to transport APIs into cells resulting in a higher bioavailability and a more targeted delivery into cells which may in some cases reduce toxicity and increase efficacy of the treatment. Liposomes are nanoparticles with diameters of for example 100 nm and therefore scatter significant amounts of UV- and visible light resulting in slightly opaque to even turbid solutions. Conventional methods for the determination of the protein API content, like the UV absorption at 280 nm, face significant problems for liposome based formulations due to the interference of the scattered light on the absorption spectra. The scattered light results in an increase of the measured signal in UV / vis measurements necessitating sophisticated methods to account for the scattering background. Due to the larger wavelength, IR radiation is scattered much less by nanoparticles below 1 pm and FT IR can be used to simultaneously quantify the API and other components in proteo-liposome formulations.

[0302]

[0218] The extinction coefficient spectrum of BSA measured in Example 6 was used for the evaluation. For the phosphate quantification, extinction coefficient spectra of sodium dihydrogenphosphate and disodium hydrogenphosphate were prepared. For the liposome extinction coefficient spectrum, a stock solution of the E. coli lipid mix of defined concentration was passed through an extruder with a pore size of 100 nm. The sample was measured using dynamic light scattering (DLS) which showed monomodal particle size distribution with a mean particle size of slightly below 100 nm indicating the successful formation of liposomes of rather similar size. The sample was measured in the FTIR and the extinction coefficient spectrum was calculated. 120333P1140PC

[0303] 1. Preparation of the verification samples

[0304]

[0219] Verification samples covering the ranges for the different components listed in Table 9 were prepared to test the performance of the proteo-liposome quantification. The BSA and lipid concentrations were varied but the same sodium phosphate buffer was used for all samples. The measured concentrations and relative deviations from the nominal values for all components of the proteo-liposome formulation verification samples are depicted Figures 22A-C.

[0305] Table 9: Concentration ranges of the verification samples of the proteo-liposome formulation

[0306]

[0220] For the preparation of the verification samples, two stock solutions containing the E. coli lipid mix and sodium phosphate at pH 7.3 respectively were prepared in addition to the BSA stock solution and mixed using density-corrected spiking. The final concentrations of the different components in the samples were calculated as described above in Example 3 for the density-corrected spiking of the IgG. The mixtures were passed through an extruder featuring a membrane with a pore size of 100 nm to generate the proteo-liposomes. The samples were measured using dynamic light scattering (DLS) which showed monomodal particle size distributions with a mean particle sizes of slightly below 100 nm indicating the successful formation of liposomes of rather similar size.

[0307] 2. Evaluation of the quantification accuracy

[0308]

[0221] The determined concentrations of the verification samples were compared to the respective nominal concentrations in Figures 22A-C for the proteo-liposome formulation and the performance of the method was evaluated. For every replicate spectrum of the test mixtures, a separate datapoint is obtained in each of the Figures 22A-C representing the concentration of the respective component in that spectrum. BSA showed a very good quantification performance and no larger deviations than 5 % were found above 2 g L'1(Figure 22A). Below 1 g L’1, increasing deviations were observed. Phosphate showed deviations of less than 3 % at 50 mM (Figure 22B). For the lipid, deviations of less than 6 % with a systematic deviation of roughly + 3 % were observed (Figure 22C). The systematic deviation might be due to the extrusion process where a small part of 120333P1140PC the lipid mix might have been held back by the 100 nm filter during the extinction coefficient spectra determination. Nevertheless, the quantification performed well overall and was not hindered by the opacity of the samples.

Claims

120333P1140PCCLAIMS1. A method for quantitative analysis of a plurality of components in at least one aqueous sample comprising an active pharmaceutical ingredient (API), wherein the method comprises the steps of(a) providing at least one aqueous sample comprising a plurality of components, wherein the sample comprises an API;(b) measuring the extinction spectrum of the at least one aqueous sample using a Fourier transform infrared (FT IR) spectrometer comprising a transmission flow- through cell in a single measurement;(c) providing at least one reference extinction coefficient spectrum for each component of the plurality of components to be analyzed in the at least one aqueous sample;(d) quantifying the concentration of each component of the plurality of components to be analyzed in the aqueous sample of step (a) based on the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b) by solving the classical least square (CLS) problem by an algorithm using a numerical linear algebra method for finding the approximate solution to an overdetermined system of linear equations characterized by the Beer-Lambert law for the extinction spectrum of the at least one aqueous sample; and wherein the API in the at least one aqueous sample is a biologic pharmaceutical ingredient.

2. The method of claim 1 , wherein step (c) comprises(i) measuring the reference extinction coefficient spectrum for the component using a Fourier transform infrared (FT IR) spectrometer and a reference sample comprising said component; and / or(ii) providing from a spectral database the reference extinction coefficient spectrum for the component based on a measurement of a reference sample comprising said component using an FT I R spectrometer.

3. The method of claim 1 or 2, wherein(a) the Fourier transform infrared (FT IR) spectrometer comprising a transmission flow-through cell has a high optical pathlength stability and / or an automated optical pathlength determination; and / or67120333P1140PC(b) the spectra used for quantifying in step (d) are (i) the original, non-derived spectra of the reference extinction coefficient spectra of step (c) and the extinction spectrum of the at least one aqueous sample of step (b), or (ii) the first or higher order derivative spectra of the spectra of (i), preferably the first or second order derivative spectra, more preferably the first order derivative spectra.

4. The method of any one of the preceding claims, wherein the plurality of components to be analyzed comprises(a) all major components present in the at least one aqueous sample; and / or(b) at least three, at least four, or at least five components comprised in the aqueous sample; and / or(c) 2-20, preferably 3-15, more preferably 4-15, even more preferably 4-10, even more preferably 5-10 components comprised in the at least one aqueous sample.

5. The method of any one of the preceding claims, wherein the API is a biologic pharmaceutical ingredient selected from the group consisting of a protein, a peptide, a nucleic acid and a virus and modified forms thereof.

6. The method of any one of the preceding claims, wherein the at least one aqueous sample is an in-process-control (I PC) sample, an ultrafiltration / diafiltration sample (LIF / DF sample), a formulation, a drug substance, or a drug product.

7. The method of any one of the preceding claims, wherein the reference extinction coefficient spectrum and / or the extinction spectrum of the at least one aqueous sample comprises a spectral range from at least 1000-1750 cm-1, preferably at least 950 - 3050 cm-1, preferably at least 930-3700 cm-1.

8. The method of any one of the preceding claims, wherein the measuring in step (b) and / or step (c) is automated.

9. The method of any one of the preceding claims, wherein the method comprises analyzing at least 10 aqueous samples and wherein the method is for high throughput quantitative analysis of a plurality of components in the at least 10 aqueous sample comprising an active pharmaceutical ingredient (API).

10. The method of any one of the preceding claims, wherein the API is a protein or a peptide and the method quantifies the concentration of the API and optionally of at68120333P1140PC least two other components in the at least one aqueous sample, preferably wherein the at least two other components in the at least one aqueous sample are independently excipients, buffer components or contaminants.

11. A method for quantitative analysis of batch variability of drug substance samples for drug release and / or defining product specification, wherein the method comprises steps (a) to (d) of any one of claims 1-10, wherein the at least one aqueous sample is a batch drug substance sample, and wherein the method further comprises for each batch drug substance sample or replicates thereof a step (e) comparing the concentrations of each component of the plurality of components in the at least one aqueous sample quantified in step (d) with pre-determined release concentration criteria and a step (f) release of the batch drug substance and / or defining product specification if the concentrations of the plurality of components in the aqueous sample quantified in step (d) meet the pre-determined release concentration criteria.

12. A method for quantitative analysis of a plurality of components in at least one aqueous sample comprising an active pharmaceutical ingredient (API) during process development, wherein the method comprises steps (a) to (d) of any one of claims 1- 10, wherein the at least one aqueous sample (i) is an in-process-control (IPC) sample of different or modified processes or process steps or (ii) are samples of different or modified formulations and wherein the method further comprises a step (e) comparing the concentrations of the plurality of components in at least two aqueous samples quantified in step (d), and a step (f) evaluating (i) each process or process step or (ii) each formulation based on the comparison.

13. A method for characterizing an antibody sample, wherein the method comprises steps (a) to (d) of any one of claims 1-10, wherein the at least one aqueous sample comprises an antibody and wherein the method comprises quantifying the concentrations of the antibody, and optionally antibody variants, in the at least one aqueous sample quantified in step (d).

14. A method for determining purity of an API, wherein the method comprises steps (a) to (d) of any one of claims 1-10, wherein the at least one aqueous sample (i) is an in- process-control (IPC) sample or (ii) a formulation and wherein the method further comprises a step (e) detecting the presence of a variation of the extinction spectrum of the at least one aqueous sample, such as residual IR bands in the extinction spectra of the at least one aqueous sample and / or shifts of single or multiple IR bands of the extinction spectra of a component in the at least one aqueous sample69120333P1140PC and wherein the variation of the extinction spectrum of the at least one aqueous sample indicates a contamination or an otherwise faulty sample.

15. A method for the production of an active pharmaceutical ingredient (API) comprising a process of(i) generation of the API, wherein the API is a biologic pharmaceutical ingredient;(ii) purification of the API;(iii) ultrafiltration and diafiltration (LIF / DF) of the API into an aqueous solution; and(iv) optionally further formulating the API; and wherein the method comprises quality testing of at least one aqueous in-process- control sample of the process of steps (ii), (iii) and / or (iv), wherein the sample comprises the API, using the method of any one of claims 1-10, and optionally wherein a variation of the concentration of a component in said at least one aqueous in-process-control sample from a pre-determined concentration range results in an adaptation of the purification of the API of step (ii), the LIF / DF of the API in step (iii) and / or the further formulating of the API of step (iv), thereby controlling and / or regulating the production of the API.

Images