Advertisement

Glass Leachables as a Nucleation Factor for Free Fatty Acid Particle Formation in Biopharmaceutical Formulations

Published:October 06, 2020DOI:https://doi.org/10.1016/j.xphs.2020.09.050

      Abstract

      Surfactants are essential components in protein formulations protecting them against interfacial stress. One of the current industry-wide challenges is enzymatic degradation of parenteral surfactants such as polysorbate 20 (PS20) and polysorbate 80, which leads to the accumulation of free fatty acids (FFAs) potentially forming visible particles over the drug product shelf-life. While the concentration of FFAs can be quantified, the time point of particle formation remains unpredictable. In this work, we studied the influence of glass leachables as nucleation factors for FFA particle formation. We demonstrate the feasibility of nucleation of FFA particles in the presence of inorganic salts like NaAlO2 and CaCl2 simulating relevant glass leachables. We further demonstrate FFA particle formation depending on relevant aluminum concentrations. FFA particle formation was subsequently confirmed with lauric/myristic acid in the presence of different quantities and compositions of glass leachables obtained by several sterilization cycles using different types of glass vials. We further verified the formation of particles in aged protein formulation containing degraded PS20 through the spiking of glass leachables. Particles were characterized as a complex of glass leachables, such as aluminum and FFAs. Based on our findings, we propose a likely pathway for FFA particle formation that considers specific nucleation factors.

      Keywords

      Introduction

      Surfactants are crucial excipients in protein formulations, as they protect the labile protein from interfacial stress that may lead to protein aggregation.
      • Kiese S.
      • Papppenberger A.
      • Friess W.
      • Mahler H.C.
      Shaken, not stirred: mechanical stress testing of an IgG1 antibody.
      ,
      • Khan T.A.
      • Mahler H.C.
      • Kishore R.S.
      Key interactions of surfactants in therapeutic protein formulations: a review.
      The parenteral administration of therapeutic proteins, such as monoclonal antibodies (mAbs), limits the choice of surfactants. Polysorbate 20 (PS20), polysorbate 80 (PS80), or poloxamer 188 are examples of surfactants used in parenteral products.
      • Kishore R.
      Part II: challenges with excipients - polysorbate degradation and quality.
      ,
      • Jones M.T.
      • Mahler H.C.
      • Yadav S.
      • et al.
      Considerations for the use of polysorbates in biopharmaceuticals.
      An increasing number of publications have reported surfactant related visible and sub-visible particle findings in biopharmaceutical formulations over recent years.
      • Jones M.T.
      • Mahler H.C.
      • Yadav S.
      • et al.
      Considerations for the use of polysorbates in biopharmaceuticals.
      • Labrenz S.R.
      Ester hydrolysis of polysorbate 80 in mAb drug product: evidence in support of the hypothesized risk after the observation of visible particulate in mAb formulations.
      • Grabarek A.D.
      • Bozic U.
      • Rousel J.
      • et al.
      What makes polysorbate functional? Impact of polysorbate 80 grade and quality on IgG stability during mechanical stress.
      • Dwivedi M.
      • Blech M.
      • Presser I.
      • Garidel P.
      Polysorbate degradation in biotherapeutic formulations: identification and discussion of current root causes.
      In those cases, particles were identified as FFAs and attributed to degradation of PS20 and PS80 over time.
      • Cao X.
      • Fesinmeyer R.M.
      • Pierini C.J.
      • et al.
      Free fatty acid particles in protein formulations, part 1: microspectroscopic identification.
      Polysorbates can undergo oxidative and/or hydrolytic degradation, and root causes have been thoroughly discussed in literature.
      • Kishore R.
      Part II: challenges with excipients - polysorbate degradation and quality.
      ,
      • Dwivedi M.
      • Blech M.
      • Presser I.
      • Garidel P.
      Polysorbate degradation in biotherapeutic formulations: identification and discussion of current root causes.
      ,
      • Kishore R.S.
      • Kiese S.
      • Fischer S.
      • Pappenberger A.
      • Grauschopf U.
      • Mahler H.C.
      The degradation of polysorbates 20 and 80 and its potential impact on the stability of biotherapeutics.
      • Saggu M.
      • Liu J.
      • Patel A.
      Identification of subvisible particles in biopharmaceutical formulations using Raman spectroscopy provides insight into polysorbate 20 degradation pathway.
      • Tomlinson A.
      • Demeule B.
      • Lin B.
      • Yadav S.
      Polysorbate 20 degradation in biopharmaceutical formulations: quantification of free fatty acids, characterization of particulates, and insights into the degradation mechanism.
      • Kerwin B.A.
      Polysorbates 20 and 80 used in the formulation of protein biotherapeutics: structure and degradation pathways.
      For the hydrolytic pathway, enzymatic degradation by residual host cell proteins was identified as a potential root cause.
      • Labrenz S.R.
      Ester hydrolysis of polysorbate 80 in mAb drug product: evidence in support of the hypothesized risk after the observation of visible particulate in mAb formulations.
      ,
      • Dwivedi M.
      • Blech M.
      • Presser I.
      • Garidel P.
      Polysorbate degradation in biotherapeutic formulations: identification and discussion of current root causes.
      ,
      • Hall T.
      • Sandefur S.L.
      • Frye C.C.
      • Tuley T.L.
      • Huang L.
      Polysorbates 20 and 80 degradation by Group XV lysosomal Phospholipase A2 isomer X1 in monoclonal antibody formulations.
      • Dixit N.
      • Salamat-Miller N.
      • Salinas P.A.
      • Taylor K.D.
      • Basu S.K.
      Residual host cell protein promotes polysorbate 20 degradation in a sulfatase drug product leading to free fatty acid particles.
      • Chiu J.
      • Valente K.N.
      • Levy N.E.
      • Min L.
      • Lenhoff A.M.
      • Lee K.H.
      Knockout of a difficult-to-remove CHO host cell protein, lipoprotein lipase, for improved polysorbate stability in monoclonal antibody formulations.
      For example, Hall et al.
      • Hall T.
      • Sandefur S.L.
      • Frye C.C.
      • Tuley T.L.
      • Huang L.
      Polysorbates 20 and 80 degradation by Group XV lysosomal Phospholipase A2 isomer X1 in monoclonal antibody formulations.
      reported that Lysosomal Phospholipase A2 Isomer X1 plays a key role in polysorbate degradation in some antibody preparations. Dixit et al.
      • Dixit N.
      • Salamat-Miller N.
      • Salinas P.A.
      • Taylor K.D.
      • Basu S.K.
      Residual host cell protein promotes polysorbate 20 degradation in a sulfatase drug product leading to free fatty acid particles.
      reported that residual host cell protein present in the drug substance was identified and experimentally shown to catalyze the breakdown of PS20 in a protein formulation over time, resulting in FFA particles and PS20 loss. Chiu et al.
      • Chiu J.
      • Valente K.N.
      • Levy N.E.
      • Min L.
      • Lenhoff A.M.
      • Lee K.H.
      Knockout of a difficult-to-remove CHO host cell protein, lipoprotein lipase, for improved polysorbate stability in monoclonal antibody formulations.
      have reported lipoprotein lipase knockout chinese ovary hamster cells using CRISPR and TALEN technologies, demonstrating subsequent significant reduction in polysorbate degradation. To decrease host cell protein content in drug substance processes, in particular lipases, different downstream process improvements were proposed.
      • Yigzaw Y.
      • Piper R.
      • Tran M.
      • Shukla A.A.
      Exploitation of the adsorptive properties of depth filters for host cell protein removal during monoclonal antibody purification.
      • Singh N.
      • Arunkumar A.
      • Peck M.
      • et al.
      Development of adsorptive hybrid filters to enable two-step purification of biologics.
      • Nguyen H.C.
      • Langland A.L.
      • Amara J.P.
      • Dullen M.
      • Kahn D.S.
      • Costanzo J.A.
      Improved HCP reduction using a new, all-synthetic depth filtration media within an antibody purification process.
      • Gilgunn S.
      • El-Sabbahy H.
      • Albrecht S.
      • et al.
      Identification and tracking of problematic host cell proteins removed by a synthetic, highly functionalized nonwoven media in downstream bioprocessing of monoclonal antibodies.
      As an example, Nguyen et al.
      • Nguyen H.C.
      • Langland A.L.
      • Amara J.P.
      • Dullen M.
      • Kahn D.S.
      • Costanzo J.A.
      Improved HCP reduction using a new, all-synthetic depth filtration media within an antibody purification process.
      tested a novel, all-synthetic depth filtration media and showed significant reduction in host cell proteins.
      Polysorbates are heterogeneous mixtures of different fatty acid esters and pegylated cyclic sorbitans.
      • Li Y.
      • Hewitt D.
      • Lentz Y.K.
      • Ji J.A.
      • Zhang T.Y.
      • Zhang K.
      Characterization and stability study of polysorbate 20 in therapeutic monoclonal antibody formulation by multidimensional ultrahigh-performance liquid chromatography-charged aerosol detection-mass spectrometry.
      • Brandner J.D.
      The composition of NF-defined emulsifiers: sorbitan monolaurate, monopalmitate, monostearate, monooleate, polysorbate 20, polysorbate 40, polysorbate 60, and polysorbate 80.
      • Borisov O.V.
      • Ji J.A.
      • Wang Y.J.
      • Vega F.
      • Ling V.T.
      Toward understanding molecular heterogeneity of polysorbates by application of liquid chromatography-mass spectrometry with computer-aided data analysis.
      • Hewitt D.
      • Alvarez M.
      • Robinson K.
      • et al.
      Mixed-mode and reversed-phase liquid chromatography-tandem mass spectrometry methodologies to study composition and base hydrolysis of polysorbate 20 and 80.
      Different grades of PS20 and PS80 with differing purities are commercially available and were recently tested. For example, all-laurate PS20 and all-oleate PS80 were compared to traditionally-used grades, which differ in their composition of FFA esters.
      • Grabarek A.D.
      • Bozic U.
      • Rousel J.
      • et al.
      What makes polysorbate functional? Impact of polysorbate 80 grade and quality on IgG stability during mechanical stress.
      ,
      • Kranz W.
      • Wuchner K.
      • Corradini E.
      • Berger M.
      • Hawe A.
      Factors influencing polysorbate's sensitivity against enzymatic hydrolysis and oxidative degradation.
      ,
      • Singh S.R.
      • Zhang J.
      • O'Dell C.
      • et al.
      Effect of polysorbate 80 quality on photostability of a monoclonal antibody.
      Siska et al.
      • Siska C.C.
      • Pierini C.J.
      • Lau H.R.
      • Latypov R.F.
      • Fesinmeyer R.M.
      • Litowski J.R.
      Free fatty acid particles in protein formulations, part 2: contribution of polysorbate raw material.
      emphasized the importance of control of raw material by showing the differences in FFA ester content of multiple PS20 and PS80 lots. These differences are of particular importance as fatty acids of different chain length exhibit different solubilities, implying different propensities to form particles when solubility limits are exceeded.
      • Tomlinson A.
      • Demeule B.
      • Lin B.
      • Yadav S.
      Polysorbate 20 degradation in biopharmaceutical formulations: quantification of free fatty acids, characterization of particulates, and insights into the degradation mechanism.
      ,
      • Kerwin B.A.
      Polysorbates 20 and 80 used in the formulation of protein biotherapeutics: structure and degradation pathways.
      ,
      • Doshi N.
      • Demeule B.
      • Yadav S.
      Understanding particle formation: solubility of free fatty acids as polysorbate 20 degradation byproducts in therapeutic monoclonal antibody formulations.

       Hypothesis

      It is well understood that the occurrence of polysorbate related visible particles is due to increased accumulation of FFAs and their solubility limits. Studies exploring the relative solubilities of various FFA homologues as a function of concentration, pH, and presence of residual amounts of polysorbate
      • Doshi N.
      • Demeule B.
      • Yadav S.
      Understanding particle formation: solubility of free fatty acids as polysorbate 20 degradation byproducts in therapeutic monoclonal antibody formulations.
      have led to a reasonable understanding of solubility limits of FFAs, which pose a risk to occurrence of visible particles. These visible particles when characterized by Fourier transformed infrared spectroscopy (FTIR) microscopy clearly show presence of FFAs or monovalent salts of FFAs.
      Interestingly though, the occurrence of visible particle formation is not always correlated to the solubility limits. In certain cases, we have observed that FFAs can precipitate below their theoretical solubility limits under conditions typically found in biopharmaceutical formulations (unpublished data). Characterization of these particles by FTIR point toward the presence of aluminum-fatty acid-tri-carboxylates or calcium-/magnesium-fatty acid-di-carboxylates in some particles, apart from FFAs. However, differentiation between the type of di-valent ions and fatty acid chain length remains challenging in FTIR. Further characterization of the particles by scanning electron microscopy associated with energy dispersive X-ray spectroscopy (SEM-EDX) demonstrated that FFA particles were a complex of FFAs with different elements, such as metals and earth alkali metals. In particular, we have identified aluminum, silicon, sodium, magnesium, and calcium in different products. Interestingly, recently presented SEM-EDX spectra of particles characterized in biopharmaceutical formulations as a result of surfactant degradation also showed elements such as aluminum and silicon, but these observations were discussed in a different background.
      • Wuchner W.K.
      • Corradini E.
      • Hawe A.
      Complexity of the Analytical Characterization of Polysorbate - Case Studies for Degradation Profiling.
      The composition of the identified ions was unique and indicative of elements dissolving from the glass container surface, that is for glass leachables and extractables. The distribution of these elements on the surface of the glass was previously characterized.
      • Brandner J.D.
      The composition of NF-defined emulsifiers: sorbitan monolaurate, monopalmitate, monostearate, monooleate, polysorbate 20, polysorbate 40, polysorbate 60, and polysorbate 80.
      • Borisov O.V.
      • Ji J.A.
      • Wang Y.J.
      • Vega F.
      • Ling V.T.
      Toward understanding molecular heterogeneity of polysorbates by application of liquid chromatography-mass spectrometry with computer-aided data analysis.
      • Hewitt D.
      • Alvarez M.
      • Robinson K.
      • et al.
      Mixed-mode and reversed-phase liquid chromatography-tandem mass spectrometry methodologies to study composition and base hydrolysis of polysorbate 20 and 80.
      • Kranz W.
      • Wuchner K.
      • Corradini E.
      • Berger M.
      • Hawe A.
      Factors influencing polysorbate's sensitivity against enzymatic hydrolysis and oxidative degradation.
      • Singh S.R.
      • Zhang J.
      • O'Dell C.
      • et al.
      Effect of polysorbate 80 quality on photostability of a monoclonal antibody.
      • Siska C.C.
      • Pierini C.J.
      • Lau H.R.
      • Latypov R.F.
      • Fesinmeyer R.M.
      • Litowski J.R.
      Free fatty acid particles in protein formulations, part 2: contribution of polysorbate raw material.
      • Doshi N.
      • Demeule B.
      • Yadav S.
      Understanding particle formation: solubility of free fatty acids as polysorbate 20 degradation byproducts in therapeutic monoclonal antibody formulations.
      • Wuchner W.K.
      • Corradini E.
      • Hawe A.
      Complexity of the Analytical Characterization of Polysorbate - Case Studies for Degradation Profiling.
      • Cailleteau C.
      • Angeli F.
      • Devreux F.
      • et al.
      Insight into silicate-glass corrosion mechanisms.
      Silicon dioxide, boron trioxide, and aluminum trioxide are typical glass network formers of type I borosilicate glass used for parenteral products.
      • Ditter D.
      • Mahler H.C.
      • Roehl H.
      • et al.
      Characterization of surface properties of glass vials used as primary packaging material for parenterals.
      Different glass network modifiers like alkali oxides (e.g. sodium, potassium), and oxides of alkaline earth metals (e.g. calcium) are added during the glass manufacturing process to decrease the melting temperature of the glass. The dissolution of glass elements is highly dependent on glass type and especially on the biopharmaceutical formulation properties.
      • Ditter D.
      • Nieto A.
      • Mahler H.C.
      • et al.
      Evaluation of glass delamination risk in pharmaceutical 10 mL/10R vials.
      Concentrations of glass leachables increase in the drug product over the storage period and are dependent on storage conditions such as temperature. Most importantly, we have previously characterized different type I borosilicate glass vials and reported heterogeneity of vial surfaces caused by the glass vial manufacturing process.
      • Ditter D.
      • Mahler H.C.
      • Roehl H.
      • et al.
      Characterization of surface properties of glass vials used as primary packaging material for parenterals.
      These heterogeneities can potentially translate into different concentrations of glass leachables.
      The observations and findings led us to hypothesize that inorganic elements leaching from the glass vials can act as nucleation seeds for FFA particle formation below the theoretical solubility limits of the FFAs. We further speculate that differences in glass leachable concentrations can affect particle formation, thus making its prediction highly challenging.
      In the present study, we used lauric and myristic acids as the main degradation products from enzymatic PS20 degradation. We targeted different glass leachables, as well as mixtures of the glass leachables, to test the hypothesis that the presence of glass leachables can induce the precipitation of FFAs below their solubility limits. To the best of our knowledge, this is the first report linking FFA particle formation in pharmaceutically-relevant, polysorbate-containing formulations to the presence of nucleation factors.

      Materials and Methods

       FFA Solutions (Stock Solutions)

      Aqueous stock solutions were prepared as previously described by Doshi et al.,
      • Doshi N.
      • Demeule B.
      • Yadav S.
      Understanding particle formation: solubility of free fatty acids as polysorbate 20 degradation byproducts in therapeutic monoclonal antibody formulations.
      with slight modifications. In brief, lauric acid (Sigma-Aldrich/Merck, Darmstadt, DE) and myristic acid (Sigma-Aldrich/Merck, Darmstadt, Germany) were suspended in PS20 HP (Croda, Edison, NJ, USA) in concentrations as summarized in Table 1. The solution was stirred at 150 rpm for 30 min at 60°C until the FFAs were fully dissolved. Solutions were diluted 1:5 with pre-warmed (60°C) water for injection (WFI): Solutions were either directly filtered through 0.22 μm PVDF Steriflip filters (Merck Millipore, Darmstadt, Germany) before dilution, followed by mixing on an orbital shaker at 100 rpm for 1 h at 25°C (Table 1 A, single FFA solution), or solutions were mixed at 150 rpm for 30 min at 60°C after dilution before filtration (Table 1 B, FFA mixtures). Stock solutions were visually inspected for absence of particles.
      Table 1FFA Concentration in Stock Solutions and Study Samples.
      #Stock Solution
      in 20% (w/v) PS20 in water for injection (WFI).
      Study Samples
      in matrix of spiking solutions resulting in 0.04% (w/v) PS20.
      Purpose of Solution
      FFA Concentration (mg/mL)FFA Target Concentration (μg/mL)
      Lauric AcidMyristic AcidLauric AcidMyristic Acid
      (A) Single FFA solution
      solubility limits of lauric and myristic acids at room temperature in WFI (∼pH 6) were determined to be 15 μg/mL and 7.5 μg/mL, respectively.
      #15-10-Spiking, negative control
      #212.5-25-Positive control
      #3-1.5-3Spiking, negative control
      #4-5-10Positive control
      (B) FFA mixtures
      solubility of mixtures of lauric and myristic acids in 20 mM histidine acetate buffer at pH 5.5 was studied based on visible and sub-visible particle counts as provided in Table 1.
      #5104208Spiking, negative control
      #67.53156
      #752104
      a in 20% (w/v) PS20 in water for injection (WFI).
      b in matrix of spiking solutions resulting in 0.04% (w/v) PS20.
      c solubility limits of lauric and myristic acids at room temperature in WFI (∼pH 6) were determined to be 15 μg/mL and 7.5 μg/mL, respectively.
      d solubility of mixtures of lauric and myristic acids in 20 mM histidine acetate buffer at pH 5.5 was studied based on visible and sub-visible particle counts as provided in Table 1.
      Stock solutions were used for spiking experiments (1:500 dilution) in the respective matrix, yielding final FFA concentrations as summarized in Table 1, with a final PS20 concentration of 0.04%. Samples were mixed on an orbital shaker at 100 rpm for 1 h at 25°C and subsequently stored at 5°C. Study samples with FFA concentrations above their solubility limits as summarized in Table 1 (single FFA per solution) or in Table S-1 in the supporting information (mixtures of FFA) serve as positive control. FFA concentration of stock solutions and final FFA concentrations of selected study samples were verified by LC-MS according to Honemann et al.
      • Honemann M.N.
      • Wendler J.
      • Graf T.
      • Bathke A.
      • Bell C.H.
      Monitoring polysorbate hydrolysis in biopharmaceuticals using a QC-ready free fatty acid quantification method.

       Inorganic Salt Solutions

      Aqueous stock solutions of different salts were prepared at concentrations between 0.25 mg/mL and 1000 mg/mL and used for spiking experiments. NaCl, NaAlO2, NaBO2, B2O3, and CaCl2 (Sigma-Aldrich/Merck, Darmstadt, Germany) were selected as their dissolution products (ions), representing typical glass components that can dissolve from Type I borosilicate glass.
      • Ditter D.
      • Nieto A.
      • Mahler H.C.
      • et al.
      Evaluation of glass delamination risk in pharmaceutical 10 mL/10R vials.
      ,
      • Ditter D.
      • Mahler H.C.
      • Gohlke L.
      • et al.
      Impact of vial washing and depyrogenation on surface properties and delamination risk of glass vials.
      The NaAlO2 and NaBO2 stock solutions were adjusted to pH 6 using HCl, and subsequently filtered using 0.22 μm PVDF Sterivex filters (Merck Millipore, Darmstadt, Germany). Solutions were visually inspected for absence of visible particles (negative control) and their elemental concentrations were determined by inductively-coupled plasma mass spectrometry (ICP-MS). Elemental concentrations were 0.048 μg/mL aluminum and 295 μg/mL sodium (NaAlO2), and 78 μg/mL boron and 168 μg/mL sodium (NaBO2). Spiking experiments with FFAs were performed in duplicate.

       Aluminum Spiking Solutions

      An aqueous 100 ppm (100 μg/mL) aluminum (Al3+) stock solution was prepared from aluminum chloride hexahydrate in 20 mM histidine acetate at pH 5.5. The actual concentration of aluminum was determined by ICP-MS. The stock solution was diluted 1:10 and sterile-filtered through a 0.22 μm porosity filter cartridge (Sterivex-GV, Millipore). Further dilutions to 10—250 ppb Al3+ (0.01—0.25 μg/mL) were prepared aseptically under laminar air flow and visually inspected for absence of visible particles. Solutions were used for spiking experiments with FFA combinations and were performed in triplicate.

       Glass Leachables Solutions

      Representative mixtures of glass leachables were obtained from two different types of glass vials in the 6 mL format, that is Duran® and Fiolax® vials (Schott AG, Müllheim, Germany, and Schott North America Inc., NY, USA). The vials are referred to as Vial Type 1 (Duran®) and Vial Type 2 (Fiolax®) and differ in chemical composition.
      • Ditter D.
      • Mahler H.C.
      • Roehl H.
      • et al.
      Characterization of surface properties of glass vials used as primary packaging material for parenterals.
      The vials underwent three autoclave cycles (121°C, 20 min) representing accelerated aging conditions. Vials were filled with 6 mL of either water for injection (WFI) or a 20 mM glycine solution at pH 10 and stoppered with D777-1 serum stoppers (DAIKYO Seiko Ltd., Tokyo, Japan). The glycine solutions were adjusted to pH 6.0 with HCl after autoclaving, and filtered through a 0.22 μm PVDF Sterivex filter (Merck Millipore, Darmstadt, Germany). Solutions were visually inspected for the absence of visible particles (negative control) and leachable concentrations were verified by ICP-MS (Table 2) as described by Ditter et al.
      • Ditter D.
      • Nieto A.
      • Mahler H.C.
      • et al.
      Evaluation of glass delamination risk in pharmaceutical 10 mL/10R vials.
      Spiking experiments with FFAs were performed in triplicate.
      Table 2Concentration of Selected Glass Leachables in Spiking Solutions. 3xTS = Three Times Terminal Sterilized.
      Vial TypeMatrixElemental Concentration (μg/mL)
      AlBSiNaCaK
      Vial Type 13xTS WFI<LOQ10.311.10<LOQ0.05
      3xTS Glycine pH 60.564.733361<LOQ1.1
      Vial Type 23xTS WFI0.080.121.40.800.06<LOQ
      3xTS Glycine pH 60.021.9193970.280.71
      LOQ
      LOQ = Limit of quantification depending on matrix.
      3xTS WFI0.050.050.050.10.050.05
      3xTS Glycine pH 60.010.10.50.50.10.1
      a LOQ = Limit of quantification depending on matrix.

       mAb Formulations

      mAb1 (IgG1, Mw = 145.5 kDa) and mAb2 (IgG1, Mw = 148 kDa) were obtained from F. Hoffmann-La Roche and formulated with 1000 U/mL hyaluronidase in 20mM HisHCl buffer at pH 5.5, 105 mM Trehalose, 100 mM Sucrose, 10 mM Methionine, and 0.4 mg/mL Polysorbate 20. The corresponding placebo was the same formulation without mAb and hyaluronidase. Formulations were stored at 5°C and 25°C for 24 and 12 months, respectively. Spiking experiments, including respective controls with placebo, were performed in triplicate using either inorganic salt solutions (CaCl2 and NaAlO2 in 10 + 1 dilution) or glass leachables from stock solutions in 10 + 1 or 100 + 1 dilution.

       Analytical Characterization

       Visual Inspection and Particle Identification

      Samples were visually inspected using a black/white panel according to European Pharmacopeia (Ph. Eur.) 2.9.20.
      European Directorate for the Quality of Medicines
      Particulate contamination: visible particles [2.9.20].
      and/or analyzed by enhanced visual inspection on a Seidenader V 90-T instrument (Seidenader Maschinenbau GmbH, Markt Schwaben, Germany). During enhanced visual inspection, samples were illuminated from the back, rotated, and magnified as described previously.
      • Ditter D.
      • Nieto A.
      • Mahler H.C.
      • et al.
      Evaluation of glass delamination risk in pharmaceutical 10 mL/10R vials.
      Negative and positive controls were all evaluated based on enhanced visual inspection results. The number of particles per container was reported.
      Samples were visually inspected immediately after spiking and 1 hour homogenization at 25°C (d0), and regularly thereafter for a maximum of 28 days (d) when equilibrated to room temperature, that is inspection at d1, d2, d7, d14, and d28.
      Particles >20 μm were identified by FTIR using a Nicolet™ iNTM10 Infrared Microscope (Thermo Fisher Scientific, Reinach, Switzerland) by comparison to reference spectra. Samples using the complete vial were filtered under laminar air flow through gold-coated polycarbonate filters (pore size 0.8 μm, diameter 13 mm). Filter conditioning included a few droplets of ethanol followed by 1 mL of particle-free water. After filtering the samples, approximately 1 mL of cooled, particle-free water was used as a final wash step before analysis.
      Chemical compositions of selected particles were verified by SEM-EDX using a Phenom XL instrument from LOT Quantum Design GmbH (Darmstadt, Germany).

       Sub-visible Particle Analysis

      Sub-visible particles (SVP) were determined by light obscuration according to Ph. Eur.
      European Directorate for the Quality of Medicines
      Particulate contamination: subvisible particles [2.9.19].
      using a HIAC/ROYCO 9703 Liquid Syringe Sampler 3000A with a HRLD-150 sensor (Skan AG, Allschwil, Switzerland) as previously described by Ditter et al.
      • Ditter D.
      • Nieto A.
      • Mahler H.C.
      • et al.
      Evaluation of glass delamination risk in pharmaceutical 10 mL/10R vials.
      Particle morphology was characterized by a flow imaging technique using a FlowCam 8000 instrument (300 μm flow cell). Prior to each measurement, the system was pre-rinsed with filtered sample solution. Samples were analyzed with a sampling efficiency of 100% and a flow rate of 2 mL/min.

       Physico-Chemical Analysis

      Turbidity was determined as outlined in Ph. Eur. 2.2.1
      European Directorate for the Quality of Medicines
      Clarity and degree of opalescence of liquids [2.2.1].
      using a Hach 2100AN turbidimeter (Hach Company, Loveland, CO, USA) in the ratio mode.
      The pH of all solutions was verified.
      Polysorbate 20 content was determined by mixed-mode high-performance liquid chromatography (HPLC) with evaporative light scattering detection (ELSD), as previously described by Lippold et al.
      • Lippold S.
      • Koshari S.H.S.
      • Kopf R.
      • et al.
      Impact of mono- and poly-ester fractions on polysorbate quantitation using mixed-mode HPLC-CAD/ELSD and the fluorescence micelle assay.

      Results and Discussion

       Artificially Prepared Glass Leachables (Salts) Lead to FFA Particle Formation

      We prepared different inorganic salt solutions of CaCl2, NaAlO2, NaBO2, B2O3, and NaCl, as their dissolution products (ions) represent typical glass leachables from Type I borosilicate glass. Myristic acid and lauric acid, as main degradation products from hydrolytic PS20 degradation, were added at concentrations below their solubility limit, that is 3 μg/mL and 10 μg/mL, respectively. Solutions were prepared using FFA stock solutions as described in the methods section. The samples were visually inspected immediately after spiking (d0) and at day 1, 2, and 7 after spiking. Particles were subsequently identified by FTIR. All samples were analyzed after being equilibrated to room temperature. Samples were compared against positive controls with FFA concentrations above solubility limits, and the pH was verified to be pH 6.
      We observed immediate particle formation analyzed by enhanced visual inspection (Seidenader, d0) directly after spiking with myristic acid and CaCl2 for all salt concentrations tested (Table 3). Immediate formation of particles after spiking (d0) was generally seen for both myristic acid and lauric acid with both CaCl2 and NaAlO2, but to a different extent. However, absolute aluminum concentrations in the final solution were lower than calcium concentrations due to pH adjustment of the spiking solution, as discussed below. Thus, a direct comparison of particle formation between the salts from this set of experiments is not possible. In general, particle formation was positively correlated with incubation time for both fatty acid solutions and increasing NaAlO2 and CaCl2 concentrations (Table 3). Also, particle formation was more pronounced for both metal ions in combination with myristic acid (Table 3). Visible particles determined according to Ph. Eur. follow trends similar to those of enhanced visual inspection, as summarized in Table 3. Visible particle formation was dependent on the time point of inspection and on salt concentration. We subsequently characterized the particles by FTIR and identified them as FFA (day 1 after spiking).
      Table 3Particle Count From Visual Inspection According Ph. Eur. and From Enhanced Visual Inspection (Seidenader). (A) Myristic Acid (3 μg/mL) and (B) Lauric Acid (10 μg/mL) Were Spiked at Different Concentrations Below Their Solubility Limits Into CaCl2 and NaAlO3 Solutions. Data From Duplicates are Presented, Which Were Verified Against the Negative Controls (Without Salt) and Positive Controls (FFAs Above Solubility Limit, i.e. 7 μg/mL Myristic Acid and 25 μg/mL Lauric Acid). d = day of inspection. d0 = directly after spiking.
      (A) Myristic Acid
      CaCl2 Spiking SolutionVisual Inspection
      Ph. Eur.Seidenader
      Salt Conc. (μg/mL)Ca2+ Conc. (μg/mL)Sampled0d1d2d7d0d1d2d7
      100036010>7>7>7>10>10>10>10
      20>7>7>7>10>10>10>10
      5001801075>7>10>10>10>10
      2053>7>10>10>10>10
      250901000>7>10>10>10>10
      2000>7>10>10>10>10
      NaAlO2 Spiking SolutionVisual Inspection
      Ph. Eur.Seidenader
      Salt Conc.
      Nominal salt concentration before pH adjustment and filtration.
      (μg/mL)
      Al3+ Conc. (μg/mL)Sampled0d1d2d7d0d1d2d7
      10000.048100>7>71-5>10>10>10
      2021>71-5>101-5>10
      5000.024120151-5>101-5>10
      2012301-51-5>10
      2500.012102141-51-51-5>10
      2002201-51-5>10
      (B) Lauric Acid
      CaCl2 Spiking SolutionVisual Inspection
      Ph. Eur.Seidenader
      Salt Conc. (μg/mL)Ca2+ Conc. (μg/mL)Sampled0d1d2d7d0d1d2d7
      100036010>7>701-5>10>10>10
      2056006-10>106-10
      50018010>7300>10>10>10
      2031006-10>10>10
      2509010>7200>10>10>10
      2032006-10>10>10
      NaAlO2 Spiking SolutionVisual Inspection
      Ph. Eur.Seidenader
      Salt Conc.
      Nominal salt concentration before pH adjustment and filtration.
      (μg/mL)
      Al3+ Conc. (μg/mL)Sampled0d1d2d7d0d1d2d7
      10000.048122001-51-51-5>10
      2010001-50>10
      5000.024120001-51-51-51-5
      200001-51-500
      2500.0121000001-51-51-5
      2000001-51-50
      a Nominal salt concentration before pH adjustment and filtration.
      For NaBO2 and B2O3 solutions, we detected particles by enhanced visual inspection for both myristic acid and lauric acid dependent on the salt concentration over time, but to a much lesser extent compared to CaCl2 and NaAlO2 (data not shown). No particle formation (Ph. Eur., Seidenader) was observed upon addition of NaCl up to a salt concentration of 1 mg/mL.
      Further characterization for sub-visible particles (SVP) and turbidity was performed on day 1 and day 7. No significant differences in turbidity were observed between samples. For SVP analysis, we detected higher sub-visible particle counts for myristic acid-spiked samples at increasing calcium concentrations compared to other samples. However, we did not observe a clear trend for other samples dependent on elemental concentrations or time. SVP counts are provided in the supporting information in Table S-2 for both calcium- and aluminum-spiked samples with lauric and myristic acids. Further studies are needed to link SVP counts to visible particle formation and to investigate the kinetics of SVP formation depending on relevant metal concentrations. Light obscuration was used as the method of choice to study SVP, as outlined by Eur. Pharm./USP. In this regard, we used micro-flow imaging as an orthogonal method to light obscuration in the next study in order to capture translucent (FFA) particles.
      With the present spiking experiments, we demonstrate feasibility of FFA particle formation in the visible range in the presence of salts simulating relevant glass leachables from borosilicate glass typically used as primary packaging for parenteral products. We report that particle formation was dependent on type and concentration of the ion/salt, like Ca2+ or Al3+, FFA type, and incubation time.
      Aside from adequate controls, the time point of inspection and equilibration of the samples/temperature of the solution are crucial for these experiments: FFA solubility has been shown to be dependent on temperature,
      • Khuwijitjaru P.
      • Adachi S.
      • Matsuno R.
      Solubility of saturated fatty acids in water at elevated temperatures.
      and samples were equilibrated to room temperature before for analysis in our studies.
      The solubility limit of FFAs is also dependent on pH
      • Doshi N.
      • Demeule B.
      • Yadav S.
      Understanding particle formation: solubility of free fatty acids as polysorbate 20 degradation byproducts in therapeutic monoclonal antibody formulations.
      ; this set of experiments was performed at pH 6. As NaAlO2 and NaBO2 initially form hydroxides in solution (~pH 10), the pH of the spiking solutions was adjusted and the remaining ion concentrations (after filtration) were subsequently checked by ICP-MS as displayed in Table 3. This implies that for this first screening to test our hypothesis, absolute elemental concentrations in the final solutions were different between the selected salts. In particular, calcium concentrations were higher compared to aluminum concentrations. Thus, a direct comparison of particle formation between the salts from this set of experiments is not possible. However, we identified aluminum as an element to induce particle formation in the ppb range (0.01—0.05 μg/mL) compared to other ions/salts. The tested aluminum concentration is representative for relevant leachable concentrations of a parenteral product, as also outlined in the supporting information (Table S-3), that is 0.02—0.24 μg/mL for relevant placebo formulations over shelf life. As a consequence, we selected aluminum for further experiments to obtain a detailed understanding of particle formation kinetics both in the visible and sub-visible range dependent on FFA and relevant metal concentration.

       Kinetics of Particle Formation Depend on Aluminum Content and FFA Concentration

      We further studied the formation of visible and sub-visible particles dependent on FFA concentration and dependent on metal concentration with aluminum as an example. We tested mixtures of myristic acid (5—10 μg/mL) and lauric acid (2—4 μg/mL) in concentrations below their solubility limit in combination with aluminum concentrations between 0 to 0.25 μg/mL (250 ppb). Solubility limits were assessed based on visible and sub-visible particle counts as reported in Table S-1 in the supporting information.
      Aluminum concentrations in the ppb range are comparable to real-time (aluminum) leachable concentrations as typically found over a drug product’s shelf life of biological formulations. As a comparison, Table S-3 in the supporting information summarizes historical real-time glass leachable data from Vial Type 2 generated from three different placebo solutions over storage time at 2—8°C in different vial formats. Aluminum concentrations were reported between 0.02 and 0.24 μg/mL depending on storage time.
      Table 4 summarizes the results from visual inspection using the compendial visual inspection method and enhanced visual inspection after equilibration of the samples to room temperature. Particle formation was dependent on incubation time, FFA concentration, and aluminum concentration for both inspection methods. We studied particle formation over 28 days and detected an increasing numbers of particles. Aluminum concentrations as low as 0.01 μg/mL led to particle formation after 1 week for all three tested combinations and levels of lauric acid and myristic acid, as detected by enhanced visual inspection. Sub-visible particles were studied by light obscuration over 28 days and cumulative counts per mL are presented in the supporting information. In contrast to visible particle formation, no clear trend in SVP formation was observed depending on aluminum or FFA concentration. For this set of experiments, light obscuration was performed at 2—8°C. Even for these conditions at a lower inspection temperature, we did not observe a clear trend in SVP formation. In addition, we performed flow imaging analysis (at room temperature) to characterize SVP, and representative diagrams are presented in Table 5. Cumulative counts per mL for light obscuration and flow imaging are reported in Table 5 as well. Although direct comparison was not possible between the methods due to analysis at different temperatures, sub-visible particle counts were even higher for flow imaging analysis, also capturing translucent particles. Visible and sub-visible particles were subsequently characterized by FTIR and identified as lauric/myristic acid (samples without aluminum), or FFA salts (aluminum spiked samples).
      Table 4Particle Count From Visual Inspection According Ph. Eur. and From Enhanced Visual Inspection (Seidenader). Mixtures of Lauric Acid (LA) and Myristic Acid (MA) at Concentrations Below Their Solubility Limits Were Added Into 20 mM Histidine Acetate Buffer at pH 5.5 Containing Increasing Amounts of Aluminum (0—0.250 μg/mL). Mean Values From Triplicates are Presented, Which Were Verified Against the Negative Controls (Without Salt) and Positive Controls (FFAs Above Solubility Limit, i.e. 10 μg/mL Myristic Acid and 25 μg/mL Lauric Acid). d = day of inspection. d0 = directly after spiking. nd = not determined.
      FFA Concentration (μg/mL)Aluminum Concentration (μg/mL)Ph. Eur.Seidenader
      LAMAd0d7d14d28d0d7d14d28
      1040.25010>7>7>10>10>10>10
      0.1250000>10>10>10>10
      0.1001100>10>10>10>10
      0.0751210>10>10>10>10
      0.05001208>10>10>10
      0.02500002>10>10>10
      0.010nd2001>10>10>10
      000000112
      7.530.2503>7>7>7>10>10>10>10
      0.1250>710>10>10>10>10
      0.1001010>10>10>10>10
      0.0750000>10>10>10>10
      0.0501110>10>10>10>10
      0.02500103>10>10>10
      0.010nd0000>1086
      000000645
      520.2502>7>7>7>10>10>10>10
      0.12500>70>10>10>10>10
      0.1000>7>7>7>10>10>10>10
      0.075021>73>10>10>10
      0.05000000300
      0.02501000>10>10>10
      0.010nd0001>10>10>10
      000000110
      000.250nd0000>1050
      0.125nd0001101
      0.100nd0000111
      0.075nd1002100
      0.050nd0002121
      0.025nd0000122
      0.010nd0100010
      001003241
      251000>7>7>7>10>10>10>10
      Table 5Sub-visible Particle Analysis After 28 d of Storage. Mixtures of Lauric Acid (LA) and Myristic Acid (MA) in 20 mM Histidine Acetate Buffer at pH 5.5 Containing 0.04% PS20 Were Spiked With Aluminum and Compared to Relevant Controls. Representative Pictures ≥5 μm From Flow Imaging Analysis, and Cumulative Sub-visible Particle Counts per mL for Both Flow Imaging (Measured at Room Temperature) and for Light Obscuration (Measured at 2—8°C) are Reported.
      Concentration (μg/mL)Example Pictures Flow Imaging

      50μm¯
      (a) Flow Imaging
      ≥3μm≥5μm≥10μm≥25μm≥100μm
      LAMAAl3+(b) Light obscuration
      ≥2μm≥5μm≥10μm≥25μm≥50μm
      ---(a)12861810
      (b)236881900
      2510-(a)2932178163118328
      (b)33,03719,309474250
      1040.25(a)68537194190
      (b)23015200
      7.530.25(a)1512809160181
      (b)12262081710
      520.25(a)1634994296784
      (b)12672172100
      The present experiment revealed that aluminum in the ppb range can serve as a nucleation factor. Aluminum can induce particle formation, which was dependent on FFA concentration and dependent on maturation time.

       ‘Real’ Glass Leachables (Mixtures) Lead to FFA Particle Formation

      In a next set of experiments, we generated ‘real’ glass leachables from different types of glass vials with different matrix solutions, including WFI and a glycine solution adjusted to pH 6. Concentrations of glass leachables were quantified by ICP-MS and are displayed in Table 3. As an example, aluminum concentration were between 0.02 and 0.56 μg/mL. We added defined amounts of myristic and lauric acid below their solubility limits to the solutions/mixtures of glass leachables and analyzed the samples up to 7 days. Samples were equilibrated to room temperature before analysis. Positive controls were prepared with FFAs at concentrations above their solubility limits, and glass leachable and spiking solutions were visually controlled for the absence of particles (negative controls).
      Particles detected by enhanced visual inspection are summarized in Table 6 and were present for all samples in contrast to various controls. Particle formation was dependent on the glass leachable solution and incubation time. We did not observe differences in turbidity between the samples (inspected at day 1). No clear trends were determined for SVP analysis by light obscuration, as also reported in previous experiments (d1, d7), and cumulative counts per mL are shown for a size range ≥2 μm in Table S-5 in the supporting information.
      Table 6Particle Count From Enhanced Visual Inspection (Seidenader), and Particle Characterization of Selected Samples by FTIR and SEM-EDX. Particles are reported After spiking of Myristic Acid (3 μg/mL) and Lauric Acid (10 μg/mL) Below Their Solubility Limits Into Different Glass Leachable Containing Solutions Generated by Three Autoclave Cycles in Different Glass Vials. Particle Counts From Triplicates Studies are Reported. The Dependency of Particle Formation on Incubation Time is Shown for Myristic Acid in Vial Type 2/Glycine Matrix. n.t. = not tested, d = day of inspection, 3xTS = three times terminal sterilized.
      Vial TypeMatrixFFAsParticlesFTIR (d1)Glass Leachables by SEM-EDX (d1)
      the surface of selected particles was analyzed; trace elements are limited to glass leachables in this list; the order of elements does not reflect the amount of element present.
      d1d7
      13xTS WFIMyristic acid1-51-5n.t.n.t.
      Lauric acid1-51-5FFAsAl, Si, Ca
      3xTS Glycine pH 6Myristic acid1-51-5FFAsAl, Ca, Mg, Na
      Lauric acid>10>10FFAsAl
      23xTS WFIMyristic acid1-51-5n.t.n.t.
      Lauric acid01-5n.t.n.t.
      3xTS Glycine pH 6Myristic acid1-5>10FFAsCa, Mg
      Lauric acid>10>10FFAsSi, Na
      Example for dependency on incubation time:Particles
      Sample no.d0d1d5d7
      Vial type 23xTS Glycine pH 6, Myristic acid100>10>10
      21-51-5>10>10
      301-5>10>10
      a the surface of selected particles was analyzed; trace elements are limited to glass leachables in this list; the order of elements does not reflect the amount of element present.
      An example of the dependency of particle formation on incubation time, as detected by enhanced visual inspection, is provided in Table 6 for myristic acid in glass leachable solution from Vial Type 2/glycine solution. The example highlights the kinetics of particle formation with no particles detected directly after spiking, and more than 10 particles detected at incubation days 5 and 7. For selected samples, particles were further characterized by FTIR, confirming the presence of FFAs (Table 6). FFAs were not tested by FTIR for all glass leachable solutions generated with WFI, which is linked to the time point of FTIR analysis at day 1 and the late onset of particle formation for these samples. We confirmed the presence of glass leachables, such as aluminum, on the surface of the FFA particles by high-resolution SEM-EDX. The qualitative analysis of the chemical composition also revealed other glass leachables such as silicon, calcium, or magnesium (Table 6). Figure 1 displays a typical portrayal of a gold filter after FTIR analysis, highlighting a few FFA particles of different sizes and a representative FFA spectrum. The spiking study highlights that mixtures of ‘real’ glass leachables lead to precipitation of FFAs and particle formation in the visible range. The onset and extent of particle formation is dependent on the mixture/composition and amount of glass leachables, as well as incubation time.
      Figure thumbnail gr1
      Figure 1Particle characterization by FTIR analysis. Visible particles were identified as FFAs after spiking with myristic acid to glass leachables solution (Vial Type 2/20 mM glycine). Only a small selection of FFA particles are highlighted.

       Case Study: FFA Particle Formation in Presence of Glass Leachables in Aged Proteinaceous Matrix

      We further studied the precipitation of FFA particles in a representative protein matrix, that is in aged mAb1 and mAb2 solutions (22M, 5°C). We added different concentrations of ‘real’ glass leachables to the mAb formulations, that is 50 or 500 μL of glass leachables extracted from Vial types 1 and 2. In this experiment, the presence of FFAs resulted from PS20 degradation over shelf-life of the drug product. mAb1 and mAb2 were formulated in the same matrix but differ in the identity of the mAb and the drug substance processes. For mAb1 and mAb2, the PS20 degradation rates at 5°C were found to be similar (Fig. 2a), but the type and concentrations of FFAs were different (Fig. 2b and c). In particular, the concentration of myristic acid (C14) was higher for mAb2 than mAb1 for storage temperatures of 5°C, 25°C, and 40°C, while the opposite was observed for lauric acid (C12). Interestingly, after 12 months storage at 25°C, mAb2 showed visible particles characterized as FFAs and aluminum, whereas mAb1 did not display any visible particles. After 12 months storage at 25°C plus 10 months storage at 5°C, both formulations showed visible particles identified as a complex of FFAs and different glass leachables.
      Figure thumbnail gr2
      Figure 2(a) PS20 concentration of mAb1 (filled symbols) and mAb2 (empty symbols) at release and over storage time at different temperatures. (b, c) Lauric acid and myristic acid concentrations in mAb1 and mAb2 at release and over storage time at different temperatures. The presence of visible particles (Ph. Eur.) is indicated by dashed boxes. The samples used for the spiking experiments are indicated by gray arrows.
      The formulated materials stored for 22 months at 5°C used for the spiking experiments were characterized as free of particles by enhanced visual inspection before the experiment. For both formulations, we observed particle formation by enhanced visual inspection for ten out of 12 samples after incubation with either 50 or 500 μL of different mixtures of glass leachables directly after spiking or on day 1 (Table 7). Results were compared to various controls, such as the initial time point and a spiked placebo solution (negative control), which remained free of particles, as well as positive controls with FFAs at concentrations above their solubility limits. As with the previous studies, sub-visible particle analysis was not conclusive (Table S-6 in the supporting information). Samples showed a slight increase in ≥2 μm particles on the first day of inspection; however no further increase up to the day 7 incubation time point was observed. To this end, flow imaging as an orthogonal method to count SVP is recommended for future studies in order to capture translucent particles.
      Table 7Particle Count From Enhanced Visual Inspection (Seidenader) and Chemical Composition of Selected Particles by SEM-EDX Analysis. Aged mAb1 and mAb2 Formulations (22 Mo, 5°C) Were Analyzed in Comparison to Placebo After Spiking of Either 50 μL or 500 μL of Different Mixtures and Amounts of Glass Leachables. The Mean Values of Triplicates are Presented for Particle Counts. 3xTS = 3 times terminal sterilized, d = day of inspection, n.t. = not tested.
      ProductGlass Leachable SolutionVisual Inspection by SeidenaderSEM-EDX
      TypeSpiking VolumeBefore Spiking (Initial)After Spiking (d0)d1Glass Leachables
      The surface of selected particles was analyzed; trace elements are limited to glass leachables in this list; the order of elements does not reflect the amount of element present.
      500 μL50 μL
      PlaceboGlycin 3xTS, Vial Type 1x-000n.t.
      Glycin 3xTS, Vial Type 1-x000n.t.
      WFI 3xTS, Vial Type 1x-000n.t.
      WFI 3xTS, Vial Type 1-x000n.t.
      Glycin 3xTS, Vial Type 2x-000n.t.
      Glycin 3xTS, Vial Type 2-x000n.t.
      CaCl2 1 mg/mLx-000n.t.
      NaAlO2 pH 6x-000n.t.
      mAb1Glycin 3xTS, Vial Type 1x-01-51-5Si, Al
      Glycin 3xTS, Vial Type 1-x01-51-5Si, Al, Mg, Na
      WFI 3xTS, Vial Type 1x-01-51-5Al
      WFI 3xTS, Vial Type 1-x000Si, Al, Ca, Mg, Na
      Glycin 3xTS, Vial Type 2x-01-51-5Al, Ca, Mg
      Glycin 3xTS, Vial Type 2-x01-51-5Si, Al, Ca, Mg, Na
      CaCl2 1 mg/mLx-01-51-5n.t.
      NaAlO2 pH 6x-01-51-5n.t.
      mAb2Glycin 3xTS, Vial Type 1x-000n.t.
      Glycin 3xTS, Vial Type 1-x01-51-5n.t.
      WFI 3xTS, Vial Type 1x-01-51-5Si, Al, Ca, Mg, Na
      WFI 3xTS, Vial Type 1-x01-51-5n.t.
      Glycin 3xTS, Vial Type 2x-01-51-5Si, Al, Ca, Mg, Na
      Glycin 3xTS, Vial Type 2-x01-51-5n.t.
      CaCl2 1 mg/mLx-01-51-5n.t.
      NaAlO2 pH 6x-01-51-5n.t.
      a The surface of selected particles was analyzed; trace elements are limited to glass leachables in this list; the order of elements does not reflect the amount of element present.
      We identified selected particles as FFAs (FTIR) in combination with a mixture of inorganic ions detected at the surface of the particles (SEM-EDX). The qualitative chemical composition analyzed by SEM-EDX revealed the presence of a variety of glass leachables, such as aluminum, silicon, sodium, calcium, or magnesium (Table 7). Further elements such as chromium, titanium, nickel, chlorine or sulfur were detected sporadically but their correlation to the fatty acid containing particles is not fully understood. Figure 3 shows a representative SEM diagram of a FFA particle, highlighting the presence of aluminum. This suggests the precipitation of FFAs in the presence of the spiked glass leachables acting as nucleation factors.
      Figure thumbnail gr3
      Figure 3SEM-EDX analysis of selected particle. Representative FFA particle on gold filter with aluminum highlighted in green. The chemical composition of the particle surface is summarized in the inserted table. The particle was characterized after spiking of glass leachables (generated from Vial Type 1 vials with WFI) into aged mAb2 solution (22 months at 5°C) containing degraded PS20 and associated mixtures of FFAs.

       Mechanism of Particle Formation

      The current study highlights that metal ions like aluminum leaching out of glass act as nucleation factors for visible particle formation in the presence of myristic acid or lauric acid at levels below their theoretical solubility limits. The onset and extent of particle formation was dependent on the type, mixture, and concentration of the nucleation factor, on the type and concentration of FFAs, and on incubation time. The studied elements mimic leachables from type I borosilicate glass typically used for parenteral products, and were investigated using relevant concentrations.
      Based on these findings, we propose a potential mechanism of particle formation based on nucleation theory (Fig. 4).,
      • Karthika S.
      • Radhakrishnan T.K.
      • Kalaichelvi P.
      A review of classical and nonclassical nucleation theories.
      FFAs exist in equilibrium of their protonated and deprotonated forms at relevant pH values for biopharmaceuticals.
      • Doshi N.
      • Demeule B.
      • Yadav S.
      Understanding particle formation: solubility of free fatty acids as polysorbate 20 degradation byproducts in therapeutic monoclonal antibody formulations.
      Taking the example of Al3+, triple charged aluminum ions may react with a deprotonated FFA and form highly insoluble aluminum-fatty acid-tri-carboxylates, which would act as nucleating seed. The hydrophobic chain of FFAs may further interact by hydrophobic interaction, fostering seed growing, which needs further investigation. As displayed in Figure 4, the proposed mechanism is shown for myristic acid in the presence of aluminum. Finally, particles may precipitate due to increasing hydrophobicity. The proposed mechanism may also explain why aluminum may not always be resolvable in FTIR analysis, that is FFAs are detected in some cases instead of FFA salts. In such cases, the presence of metals can be characterized by an orthogonal method like SEM-EDX. We showed that low leachable concentrations are needed for particle formation, as demonstrated by relevant aluminum levels. Additional work is needed to fully and holistically validate the proposed mechanism.
      Figure thumbnail gr4
      Figure 4Proposed mechanism of FFA particle formation dependent on nucleation factors is shown for myristic acid and aluminum.

      Conclusion

      Formation of visible particles composed of FFAs as a result of PS20 or PS80 hydrolytic degradation represents a major challenge because these polysorbates are the surfactant of choice in the biopharmaceutical industry. FFAs can precipitate when their solubility limits are exceeded in the absence of specific nucleation factors. Based on our studies, we propose that FFAs can also precipitate below their solubility limit in the presence of nucleation factors. The presence of FFA particles in both situations can limit the shelf life of a product.
      We conclude that FFAs can precipitate and form particles in the presence of mixtures of different metal ions such as aluminum, and in particular, glass leachables. However, we would like to highlight the complex interplay between (1) the presence of different FFAs in different concentrations; (2) the presence of different concentrations of intact PS and their degradation products (esters), potentially solubilizing FFA particles; (3) the absolute concentration and combination of glass leachables; and (4) the kinetics and temperature dependence of particle formation. In particular, the relationship between (3) and (4) is currently unclear and warrants further investigation.
      In the event of occurrence of visible particles during drug product stability with FFA levels within solubility limits, it is likely that nucleation factors as described in the current work are playing a role in seeding the precipitation of the FFAs. The source of metal ions for such nucleation can be (but is not limited to) glass leachables. In such cases, apart from mitigation of polysorbate degradation, assessment and control of material attributes of raw material should also be considered.

      Acknowledgments

      The authors would like to thank Jasmin John (Pharmaceutical Development, F. Hoffmann-La Roche) for supporting the experimental work, Jan Wendler and Anja Bathke for supporting the free fatty quantification analysis (Analytical Development, F. Hoffmann-La Roche), Matthias Braun, Anacelia Rios (Analytical Development, F. Hoffmann-La Roche), Christiane Hess, Sophie Hochenauer, and Jeremy Duboeuf (Pharmaceutical Development, F. Hoffmann-La Roche) for supporting the particle characterization, as well as Eva Roedel (Solid State Science, F. Hoffmann-La Roche), Martin Rempt and Kay Hagedorn (Instrumental Analytics, Roche Diagnsotics, Penzberg) for supporting the SEM-EDX analysis. The authors would like to further acknowledge the scientific input from Tarik Khan, Michael Adler, Holger Roehl, and Philip Lester (Pharmaceutical Development, F. Hoffmann-La Roche), Inn Yuk, Barthelemy Demeule, Karen Rutherford, Nidhi Doshi, and Samir Sane (Pharmaceutical Development, F. Hoffmann-La Roche/Genentech), as well as from Christian Bell (Analytical Development, F. Hoffmann-La Roche).

      Appendix A. Supplementary Data

      References

        • Kiese S.
        • Papppenberger A.
        • Friess W.
        • Mahler H.C.
        Shaken, not stirred: mechanical stress testing of an IgG1 antibody.
        J Pharm Sci. 2008; 97: 4347-4366
        • Khan T.A.
        • Mahler H.C.
        • Kishore R.S.
        Key interactions of surfactants in therapeutic protein formulations: a review.
        Eur J Pharm Biopharm. 2015; 97: 60-67
        • Kishore R.
        Part II: challenges with excipients - polysorbate degradation and quality.
        in: Warne N.W. Mahler H.C. Challenges in Protein Product Development, AAPS Advances in the Pharmaceutical Sciences Series 38. 1st ed. Springer, Switzerland2018: 25-62
        • Jones M.T.
        • Mahler H.C.
        • Yadav S.
        • et al.
        Considerations for the use of polysorbates in biopharmaceuticals.
        Pharm Res. 2018; 35: 148
        • Labrenz S.R.
        Ester hydrolysis of polysorbate 80 in mAb drug product: evidence in support of the hypothesized risk after the observation of visible particulate in mAb formulations.
        J Pharm Sci. 2014; 103: 2268-2277
        • Grabarek A.D.
        • Bozic U.
        • Rousel J.
        • et al.
        What makes polysorbate functional? Impact of polysorbate 80 grade and quality on IgG stability during mechanical stress.
        J Pharm Sci. 2020; 109: 871-880
        • Dwivedi M.
        • Blech M.
        • Presser I.
        • Garidel P.
        Polysorbate degradation in biotherapeutic formulations: identification and discussion of current root causes.
        Int J Pharm. 2018; 552: 422-436
        • Cao X.
        • Fesinmeyer R.M.
        • Pierini C.J.
        • et al.
        Free fatty acid particles in protein formulations, part 1: microspectroscopic identification.
        J Pharm Sci. 2015; 104: 433-446
        • Kishore R.S.
        • Kiese S.
        • Fischer S.
        • Pappenberger A.
        • Grauschopf U.
        • Mahler H.C.
        The degradation of polysorbates 20 and 80 and its potential impact on the stability of biotherapeutics.
        Pharm Res. 2011; 28: 1194-1210
        • Saggu M.
        • Liu J.
        • Patel A.
        Identification of subvisible particles in biopharmaceutical formulations using Raman spectroscopy provides insight into polysorbate 20 degradation pathway.
        Pharm Res. 2015; 32: 2877-2888
        • Tomlinson A.
        • Demeule B.
        • Lin B.
        • Yadav S.
        Polysorbate 20 degradation in biopharmaceutical formulations: quantification of free fatty acids, characterization of particulates, and insights into the degradation mechanism.
        Mol Pharm. 2015; 12: 3805-3815
        • Kerwin B.A.
        Polysorbates 20 and 80 used in the formulation of protein biotherapeutics: structure and degradation pathways.
        J Pharm Sci. 2008; 97: 2924-2935
        • Hall T.
        • Sandefur S.L.
        • Frye C.C.
        • Tuley T.L.
        • Huang L.
        Polysorbates 20 and 80 degradation by Group XV lysosomal Phospholipase A2 isomer X1 in monoclonal antibody formulations.
        J Pharm Sci. 2016; 105: 1633-1642
        • Dixit N.
        • Salamat-Miller N.
        • Salinas P.A.
        • Taylor K.D.
        • Basu S.K.
        Residual host cell protein promotes polysorbate 20 degradation in a sulfatase drug product leading to free fatty acid particles.
        J Pharm Sci. 2016; 105: 1657-1666
        • Chiu J.
        • Valente K.N.
        • Levy N.E.
        • Min L.
        • Lenhoff A.M.
        • Lee K.H.
        Knockout of a difficult-to-remove CHO host cell protein, lipoprotein lipase, for improved polysorbate stability in monoclonal antibody formulations.
        Biotechnol Bioeng. 2017; 114: 1006-1015
        • Yigzaw Y.
        • Piper R.
        • Tran M.
        • Shukla A.A.
        Exploitation of the adsorptive properties of depth filters for host cell protein removal during monoclonal antibody purification.
        Biotechnol Prog. 2006; 22: 288-296
        • Singh N.
        • Arunkumar A.
        • Peck M.
        • et al.
        Development of adsorptive hybrid filters to enable two-step purification of biologics.
        MAbs. 2017; 9: 350-363
        • Nguyen H.C.
        • Langland A.L.
        • Amara J.P.
        • Dullen M.
        • Kahn D.S.
        • Costanzo J.A.
        Improved HCP reduction using a new, all-synthetic depth filtration media within an antibody purification process.
        Biotechnol J. 2019; 14: e1700771
        • Gilgunn S.
        • El-Sabbahy H.
        • Albrecht S.
        • et al.
        Identification and tracking of problematic host cell proteins removed by a synthetic, highly functionalized nonwoven media in downstream bioprocessing of monoclonal antibodies.
        J Chromatogr A. 2019; 1595: 28-38
        • Li Y.
        • Hewitt D.
        • Lentz Y.K.
        • Ji J.A.
        • Zhang T.Y.
        • Zhang K.
        Characterization and stability study of polysorbate 20 in therapeutic monoclonal antibody formulation by multidimensional ultrahigh-performance liquid chromatography-charged aerosol detection-mass spectrometry.
        Anal Chem. 2014; 86: 5150-5157
        • Brandner J.D.
        The composition of NF-defined emulsifiers: sorbitan monolaurate, monopalmitate, monostearate, monooleate, polysorbate 20, polysorbate 40, polysorbate 60, and polysorbate 80.
        Drug Dev Ind Pharm. 1998; 24: 1049-1054
        • Borisov O.V.
        • Ji J.A.
        • Wang Y.J.
        • Vega F.
        • Ling V.T.
        Toward understanding molecular heterogeneity of polysorbates by application of liquid chromatography-mass spectrometry with computer-aided data analysis.
        Anal Chem. 2011; 83: 3934-3942
        • Hewitt D.
        • Alvarez M.
        • Robinson K.
        • et al.
        Mixed-mode and reversed-phase liquid chromatography-tandem mass spectrometry methodologies to study composition and base hydrolysis of polysorbate 20 and 80.
        J Chromatogr A. 2011; 1218: 2138-2145
        • Kranz W.
        • Wuchner K.
        • Corradini E.
        • Berger M.
        • Hawe A.
        Factors influencing polysorbate's sensitivity against enzymatic hydrolysis and oxidative degradation.
        J Pharm Sci. 2019; 108: 2022-2032
        • Singh S.R.
        • Zhang J.
        • O'Dell C.
        • et al.
        Effect of polysorbate 80 quality on photostability of a monoclonal antibody.
        AAPS PharmSciTech. 2012; 13: 422-430
        • Siska C.C.
        • Pierini C.J.
        • Lau H.R.
        • Latypov R.F.
        • Fesinmeyer R.M.
        • Litowski J.R.
        Free fatty acid particles in protein formulations, part 2: contribution of polysorbate raw material.
        J Pharm Sci. 2015; 104: 447-456
        • Doshi N.
        • Demeule B.
        • Yadav S.
        Understanding particle formation: solubility of free fatty acids as polysorbate 20 degradation byproducts in therapeutic monoclonal antibody formulations.
        Mol Pharm. 2015; 12: 3792-3804
        • Wuchner W.K.
        • Corradini E.
        • Hawe A.
        Complexity of the Analytical Characterization of Polysorbate - Case Studies for Degradation Profiling.
        Presentation at the CASSS’ AT Europe, Barcelona2018
        • Cailleteau C.
        • Angeli F.
        • Devreux F.
        • et al.
        Insight into silicate-glass corrosion mechanisms.
        Nat Mater. 2008; 7: 978-983
        • Ditter D.
        • Mahler H.C.
        • Roehl H.
        • et al.
        Characterization of surface properties of glass vials used as primary packaging material for parenterals.
        Eur J Pharm Biopharm. 2018; 125: 58-67
        • Ditter D.
        • Nieto A.
        • Mahler H.C.
        • et al.
        Evaluation of glass delamination risk in pharmaceutical 10 mL/10R vials.
        J Pharm Sci. 2018; 107: 624-637
        • Honemann M.N.
        • Wendler J.
        • Graf T.
        • Bathke A.
        • Bell C.H.
        Monitoring polysorbate hydrolysis in biopharmaceuticals using a QC-ready free fatty acid quantification method.
        J Chromatogr B Analyt Technol Biomed Life Sci. 2019; 1116: 1-8
        • Ditter D.
        • Mahler H.C.
        • Gohlke L.
        • et al.
        Impact of vial washing and depyrogenation on surface properties and delamination risk of glass vials.
        Pharm Res. 2018; 35: 146
        • European Directorate for the Quality of Medicines
        Particulate contamination: visible particles [2.9.20].
        European Pharmacopeia 9. 4. Council of Europe, Strassbourg2020
        • European Directorate for the Quality of Medicines
        Particulate contamination: subvisible particles [2.9.19].
        European Pharmacopeia 9. 4. Council of Europe, Strassbourg2020
        • European Directorate for the Quality of Medicines
        Clarity and degree of opalescence of liquids [2.2.1].
        European Pharmacopeia 9. 4. Council of Europe, Strassbourg2020
        • Lippold S.
        • Koshari S.H.S.
        • Kopf R.
        • et al.
        Impact of mono- and poly-ester fractions on polysorbate quantitation using mixed-mode HPLC-CAD/ELSD and the fluorescence micelle assay.
        J Pharm Biomed Anal. 2017; 132: 24-34
        • Khuwijitjaru P.
        • Adachi S.
        • Matsuno R.
        Solubility of saturated fatty acids in water at elevated temperatures.
        Biosci Biotechnol Biochem. 2002; 66: 1723-1726
        • Karthika S.
        • Radhakrishnan T.K.
        • Kalaichelvi P.
        A review of classical and nonclassical nucleation theories.
        Cryst Growth Des. 2016; 16: 6663-6681