cecal Seences 109 (2020) 585-602 Contents lists available at ScienceDirect Journal of Pharmaceutical Sciences journal homepage: www.jpharmscl.org Pharmaceutical Biotechnology Accelerated Aggregation Studies of Monoclonal Antibodies: Considerations for Storage Stability (check for upcatos] Ruben Walchli |, Pieter-Jan Vermeire *, Jan Massant *, Paolo Arosio '* * Department of Gems and Apple Bocenes, Institute fr Chemical and Beene ETH Zari Vink rl We 1-5/0 $09 neh Sizes Gc Pharma BioTech Sciences, Formation Development, Chemin ares, M20 Brsine-Al Begum ARTICLE INFO ABSTRACT [Aggtezation of mAbs is a crucial concer with respect to ther safety and efficacy. Among the various properties of protein aggregates, itis emerging that their size can potentially impact heir immunogenicity. ‘Therefore, stability studies of antibody formulations should not only evaluate the rate of monomer loss but also determine the size distribution ofthe protein aggregates, which in turn depends on the aggregation ‘mechanism. Here, we study the aggregation behavior of diferent 5 inthe tempe: aggregation kinetics of both antibodies follow aon-Arrhenius behavior and thatthe 3 nisms change between 40°C and 5° noglobulns Fewer onedlonalstabdyis) ure range from 5°C to 50°C ver 52 week of storage. We show thatthe regation mecha- leading to diferent rypes of aggregates. Specifically, fora given aby ‘monomer conversion, dimer formation dominates at low temperatures, while larger aggregates are net etn formed athigher temperatures. We further show thatthe stability ranking of diferent molecules as well at ecto medeling ofeiferent formulations i drastically different at 40"Cand 5°C while itcorrelates better between 30°C and Enos 5°C. Our findings have implieations forthe level of information provided by accelerated aggregation Seveopabty studies with respect to protein stablity under storage conditions. "© 2020 American Pharmacists Association”. Publis by Elsevier inc llrightsreserved. Introduction. The characterization and inhibition of protein aggregation is In the last decades, therapeutic proteins like mAbs have become an important class of pharmaceuticals.” With respect to small ‘molecule drugs, the complexity of their macromolecular structure poses greater challenges in terms of chemical and physical stability, which must be carefully guaranteed throughout storage and de- livery for drug safety and efficacy. Among the various possible in- stabilities, formation of protein aggregates can be a problem in the development of safe formulations because proteinaceous particles hhave been associated with an increased risk of immunogenicity. = Such adverse immune responses can induce serious complications and even fatalities and, in some cases, may lead to the retraction of biopharmaceuticals after marketing.” Current adder or Vermee: Laberstary for Bacrytallagraphy, Department of ria the Internet at peo aro 0a) phe 2015 0088 therefore a key topic in biopharmaceuticals development. John F Carpenter and Theodore W. Randolph have been pioneers in recos- nizing the importance ofthis problem." Starting from a first pub- Jication on the lyophilization of hemoglobin in 1996," their work has ‘tremendously contributed to the understanding of the sources and ‘mechanisms of protein aggregation during production, "formula tion," shipment and storage,” as well as handling by medical personnel and patients." Together with advances in experimental characterization, this progress led to a fundamental understanding of the aggregation process at the molecular level!” It has now become clear that protein aggregation is a complex rmulistep process, which involves a variety of possible microscopic reactions including protein confor- ‘mational changes, nucleation, and growth of the protein particles.” °° Different aggregation mechanisms lead to different types of ag ‘regates and therefore therapeutic proteins may form a broad variety of soluble and insoluble partiles, which can be reversible or ire versible?””” ‘The term “aggregation” should therefore not be used in a generic ‘way, in particular in connection to possible consequences such as immunogenicity." For instance, a recent study’ suggests that mAb aggregates in the range of 100-1000 nm could be more prob- Jematic for immunogenicity with respect to smaller or I 02020 Ameccan Pharmacists Assolton®. Published by Esevler Inc Al rights reserved596 Wah eo Jura of Pharmaceutical Sriencs 1082020 395-502 species. In other reports, micrometer-sized aggregates were found to be more immunogenic.” Moreover, the extent of chemical ‘modifications of the primary structure within the aggregates can be another important factor."""" The development of safe mAb for- ulations should therefore focus not only on the rate of monomer Joss but also on the size distribution of the protein aggregates that are formed, which, in turn, requires understanding the aggregation ‘mechanisms at the molecular level. This operation is very chal- Tenging because the set of microscopic steps composing the aggre sation reaction network is highly specific to molecule and solution conditions. Development of new therapeutic mAbs commonly relies on. forced degradation and accelerated stability studies performed at higher temperatutes relative to storage conditions, These studies are aimed, respectively, at generating degradation products and at ‘obtaining information about formulation stability within a shorter time frame.” However, the individual elementary reactions involved in the aggregation process typically exhibit diferent de- pendencies on temperature. Moreover, increasing temperature ‘migint not only raise the total faction of partially unfolded mono- ‘mers but also change their conformations, which probably exhibit different aggregation propensities. Consequently, extrapolating the aggregation mechanism to Jow temperature from forced degrada Lionlaccelerated stability studies is a challenging operation.” In addition, the identification of the relevant microscopic steps, in the aggregation mechanism requites the acquisition of a large amount of high-quality, time-resolved data, This operation is severely complicated at low temperatures due to the slow aggre- gation kinetics. Indeed, there are only few reports in literature presenting kinetic data for the aggregation of mAbs under rettig- erated conditions."""""" This observation underlines the need for further research on the temperature dependence of mAb aggre- gation in the temperature range that is relevant for standard sta- bility testing and product storage Here, we investigated the aggregation behavior of 2 mAbs at S°C, 30°C, 40°C, and 50°C over 52 weeks of incubation. For both ‘molecules, we studied formulations at 10 and 100 mg/ml. mAb at 5 different pH values in the absence and presence of 150 mM sodium, chloride, We show that the aggregation mechanism of those mAbs changes as a function of temperature. Consequently. the same de- agree of monomer loss at 5°C and 40°C corresponds to different protein aggregate populations. In particular, dimer formation dominates at low temperatures. This result demonstrates that a comprehensive picture of protein stability cannot rely on mea- surement of residual monomer only. In addition, we show that the ranking ofthe stability of different molecules as well as the optimal formulations that minimize aggregation are different at 5°C compared to 40°C. Materials and Methods Materials ‘mAb-1 and mAb-2 were, respectively. an IgG1 with experimental pl of £2 and an IgG with experimental pl of 26, Sodium chloride, tris(hydroxymethyl)aminomethane (Tris) and. sodium acetate anhydrous were purchased from Sigma-Aldrich (St Louis. MO). L- Histidine was purchased from Merck KGaA (Darmstadt, Germany), Al chemicals were of analytical grade. Ultrapure water was pre- pated using a Milli-Q system (Merck MillPore, Billerica, MA) Sample Preparation Tris-HCI and L-Histidine-HCl buffers were prepared by dissolv- ing the appropriate amount of buffer component in ulleapure water followed by pH adjustment through addition of 1 M HCl (Merck KGaA, Darmstadt, Germany), Acetate buller was prepared by mix- ing appropriate volumes of 200 mM sodium acetate and 200 mM acetic acid solutions to obtain the desited pH. The identical bufers containing 3 M sodium chloride were prepared in parallel. All buffers were filtered using a Stevicupe vacuum filtration system (Merck KGaA) and stored in the refrigerator before use Buffer exchange of the mAb starting material was performed using Vivaflow 50 eross flow cassettes (Sartorius Stedim Biotech GmbH, Goettingen, Germany) equipped with 30 kDa cutoff PES ‘membrane. The final permeate volume was approximately 5 times the volume of the mAb solution to ensure proper bufler exchange, The cassettes were cooled in a water bath during the entire pro- cedure. The bufferexchanged mAb solutions were concentrated to approximately 105 mg/mL after transfer into Amicon Ultra 15, centrifugal filter tubes (Merck KGaA, Darmstadt, HE, Germany) equipped with a 30 kDa PES membrane. The concentrated sol tions were sterile-filtered using Steriflip-GP. 0.22 um sterile centrifuge tube top filter units (Merck KGaA) and stored under refrigerated conditions before further use. mAb concentration of the solutions was determined by UV absorption at 280 nm on a SoloVPE variable path length UV-Vis spectrophotometer (C tech- nologies inc. Bridgewater, N)).A specific extinction coefficient ? of 1.58 mal. mg! em’! and 134 mL mg! em was used for mAb-1 and mAb-2, respectively. Stability Studies ‘The aggregation kinetics of the different mAb formulations was investigated during isothermal incubation at 5°C, 30°C, 40°C, and '50°C using climate chambers. Different formulations were consid cred by varying pH and sodium chloride concentration. More specifically, PH was selected relative to each mAD's isoelectric point and set equal to 0, 1, 2, 3, and 4 units below the pl For each pH value, one formulation without excipient and one with 150 mM. sodium chloride were prepared. Next, 200 ul of formulation was placed in HPLC vials with 300 ul fixed inserts (Thermo Scientific, Langerwehe, Germany). The vials had been sterilized in a Laboklav 80-V (Detzel Schloss, Haldensleben, Germany) autoclave before use, Those operations were performed inside a Herasafe HS 15 (Thermo Scientficr, Waltham, MA) laminar fow cabinet. In reg- ular time intervals, vials were withdrawn from the incubators and the samples were analyzed by size-exclusion chromatography (SEC)-multiangle light scattering and DLS to determine the mass fraction of aggregates, as well a to determine average molecular ‘weight and hydrodynamic size ofthe aggregate population. Size-Exclusion Chromatography Coupled to Multiangle Light Scattering ‘Chromatographic analyses of the mAb formulations were run on an Agilent 1260 Infinity (Agilent Technologies, Santa Clara, CA) system, Before injection, samples were kept at SC within the autosampler ofthe unit. AWyatt WIC-0305S analytical SEC column, (78 » 300 mm, 5 jm, 300 A) was used in combination with a Wyatt WTC.03085G guard column to separate the different species based ‘on their hydrodynamic size. For analysis, 250 yg of protein was injected and eluted at 1 ml/min with 100 mM sodium phosphate, 200 mM sodium sulfate at pH 70 as mobile phase. Chromatograms ‘were recorded in terms UV absorption at 280 nm, Furthermore, the chromatography system was coupled to a DAWN HELEOS multi- angle light scattering detector (Wyatt Technologies, Santa Barbara, CCA) to determine the molecular weight ofthe eluting species. The UV chromatograms were analyzed with the Empower 3 softwareWath se Figure 1. Measuremenfthe inital eerezton ae The letshowstbecme evalusen othe til mass acto of gzregates or 100 mgm AD 2 pH = pl and 30°C The Decline epresensalner eessonaalsisfthe tat determine (Wag -n (Waters, Milford, MA) and the light scattering data with the Astra 61.717 soltware (Wyatt, Santa Barbara, CA) Results and Discussion ‘The Aggregation Rate Shows Non-Arrhenius Behavior ‘We focused our attention on the effect of temperature on the initial tages of the aggregation process, corresponding to less than 10% monomer conversion, because typically only few percent of ‘monomeric mAbs are converted into aggregates during the storage of commercial products.”* The aggregation rate was determined from the mass fraction of the aggregates measured via size- exclusion chromatography at the different time points. The total ‘mass fraction of aggregates waz at a generic incubation time ¢ was computed using the following formula Aneel) , Atet(0) ~ Ai Are) i} Wrgelt) Front) Here, Aagg denotes the combined peak area of all species eluting earlier than the monomer. Arg represents the total area under the ccurve and a (0) the initial value. This formula also accounts for the potential formation of insoluble aggregates, which would lead to incomplete mass recovery in SEC analysis. Such mass loss, however, was observed only for a very few formulations at higher temperatures. The initial rate of aggregate formation (dWagfdt), 0 was determined by linear regression in the range Whgg(t)© Wage(0), Wagg(0) +01) as shown in Figure 1. Finally, the initial aggregation fate rp was obtained by multiplying (dwags/40}0 by the total protein concentration co. Figure 2 presents the influence of temperature on the initia aggregation rate for 100 mgimL. mAb concentration, whereas the data at 10 mgimL are reported in Supplementary Figure SI. For all formulations, the initial aggregation rate showed nonlinear depen- dence on temperature according to the Arthenius framework. This result isin agreement with previous reports for other mAbs." The deviation from Arthenius behavior was more pronounced for formulations at lower pH, which might be attributed to the lower melting temperature of the mAbs under those conditions. ‘Our results are not surprising because for large proteins like mAbs, Arrhenius behavior is usually observed only over narrow temperature ranges."* Moreover, for mAbs, this temperate range is typically above 45°C, which significantly exceeds the temperature ‘window relevant for formulation development and storage." In agreement with our results, a bend in the Arvenius plots typically ‘observed between 30°C and 40°C, which challenges the predictions of stability at low temperatures based on data generated under thermal stress, Joural of Pherae Scenes 108 (2020) 595.692 ser The Apparent Reaction Order Increases With Decreasing Temperature The possibility to measure the time evolution of the mass loss allows us to estimate the apparent reaction order », which is an important parameter that quantifies the dependence of the ag- sgregation rate on the monomer concentration [Ml and is indicative ‘of the dominant microscopic aggregation mechanism. We deter- ined r by analyzing the dependence of the intial aggregation rate 9 on the initial monomer concentration [Mlo rom tin (SG) — hast w ez. les denotes the observed at constant the aggregation rate Ths expression can be linearized as, Intro) =I) + (M9) @) which cn be applied to estimate from the slope of the linea lot I(r) against nM) igure 3, we show the dependence of Gn temperature, For both molecules and all formulation cond tions the reaction order changed 36a function of temperature Specialy» increased from values closer fo one toward values lose to 2 with decreasing temperature. This important est Sndieates that temperature afets both the overall angregation rate (see Fig. 2) and the aggregation mechanism (see Fig. 3), thereby leading not only to different amounts but also to different types of protein aggregates. ‘We note that a 5° data could ony be acquired for mAb-1 for :nultons without excipient whereas fra other formulations the Eggreation rate at lower TmAb concentration Was to0 sow 1D Jebus este the reaction ordet Moreover, formation of so ble ageregates was observed ina feve formulations at 50°C (see Sopoletenay tole st}, which could expan the inease of + bee ‘tween 40°C and 50°C in those cases.”** Finally, the kinetic data for DI pitt sodium eNloridewece too nosy or prope nas Lowering Temperature Arrests Growth and Coagulation of mAb Aggregates ‘The apparent reaction order conveys important information about the underlying microscopic aggregation mechanism because the individual elementary steps exhibit different reaction, orders with respect to the monomer concentration. For instance, dimer nucleation is second order, whereas addition of monomers to preexisting aggregates is frst order in terms of monomer concentration.” The apparent order will be a convolution of the 2 contributions and » is expected to assume values closer to 2, when dimer nucleation becomes the dominating aggregation route, Thus, the observed » of 2 at lower temperatures suggests that aggregate growth by monomer addition is negligible for those conditions. ‘We verified this prediction by measuring the time evolution of the size of the aggtegates at different temperatuses, reported as the weight-average number of mAb monomers per aggregate (see Fig. 4). Indeed, at 5°C and 30°C, the sizeof the aggregates remained constant over time and equal to ng~2 indicating that the aggregate population at low temperatures consisted of dimers that did not grow over time. By contrast, at higher temperatures (Le. 40°Cand 50°C), the size ofthe aggregates rapidly increased for both _mAbs within few weeks, both in the absence and presence of so- dium chloride as excipient. ‘An important consequence of this behavior is that the same extent of monomer loss can lead to dillerent aggregate sizeses Wah ea Jura of Pharmaceutical Srencs 1082020 595-502 ECCT) ECL) oor [1 10007 (1K) roar (1K) 1004 (1) igure 2 Ares pot fhe inti geresation rate t 100 mlm mAb oncenation and pH =p le qu tangles} pl pHs (oange downvaré anges an pl pH 4 (re amends) (a) mAb whan exci ‘lone () mAb-2 with 150 mM sod chord pH = 2 yelow upward Ab witht exci mAb eth 190 mM sor distributions, which could potentially be associated with different Aggregation Under Storage Conditions Correlates Poorly With risks of immunogenicity.”»”To illustrate this concept, in Figure 5, Forced Degradation Studies at 40°C we plot the average aggregate size ng against the monomer conversion 1 ~ mfor the different temperatures (with m =(Mli[Mjo Overall, the results discussed previously indicate that higher representing the fraction of unreacted monomer). At any given temperature not only increases the overall aggregation rate but also conversion, nf¥ increased as function of temperature, clearly changes the relative contribution of the different microscopic steps indicating that aggregate growth becomes more important relative composing the aggregation reaction network, which has deep im- a b 7 i S Raed d d sow so c d 2 , ype iv 1 SSE aaa ae SSE a aa to00-r UK} {000K Figure 3. Tenperatce dependence ofthe appatet reuton ode: ra he nl ageregation ate with respect othe monomer concent. p— pe suas) ppt blue cesh pt-pt 2 yellow spwacd tangles ppl ferange downward tangles) and pl pitted duende s) mab without expen (0) D2 without etpient ) mbWat ural of Phere Sceners 108 (2020) 585.602 sso xf y Y Bitese—o—s "owes gure. Time evolition ofthe weight average nuber of mAb manom ‘Fonth 150 mV sock chloride will have different effects on the individual microscopic steps of the aggregation process. For instance, sugars strongly impact protein, conformational changes, whereas salts typically affect nucleation and growth events by modulating electrostatic interactions be- ‘tween protein molecules. Consequently. formulation chosen based ‘on forced degradation studies at 40°C may not representthe optimal formulation for storage at refrigerated temperatures of 2°C-8°C. ‘To illustrate this point, we plotted the aggregation rate of both molecules at 5°C against the rate at 40°C for all investigated 5 agente 2100 gik mAb and pl p= 2 ‘war trang nd SU [red daomsdtangles } mA! whet ado ari, (b MAb? who todiom clade. (cmb wh 150m rad hei. % * d © Y | Y wold, [eee 64On—n—o 5 (ue squares 30°C yew formulations (see Fig. 6), which differ in terms of pH values and sodium chloride concentration, The 2 aggregation rates correlated poorly (Pearson correlation coefficient of R — 0.01, in particular for mAb-2 and for typical pH values of commercial formulation (orange symbols in Fig. 6b), Moreover, itis worth noting that most formulations of mAb-2 showed slower aggregation compared to formulations of mAb-1 at 5°, while this trend was much less pronounced at 40°C, as high- lighted in Figure Sb, In fact, the Spearman rank correlation a ~ Bea v ‘teeter Stoo a as vat c E - v A A ‘ A a a bt Figure 5, Weight-average number of mtb rnonomers per agreate as function of nett conversion fe pl — rigs a) ADT wide expen, (b) mab? without excl, Award lng) and 50° (red dowtnad to {HAD wi 130 nM sel cle (6) mdb wih 50Wale our of har 940°C [eck] PI pit=T (ight ise) pit 2(yelow) pt pit 3 eangeh and 9 pt coefficient p for the 2 aggregation rates is equal to 0.26, Thu, tem- perature can change not only the ranking of different formulations fof the same molecule but also the overall relative stability of for- ulations of several candidates. This observation underlines the inherent limitations of thermal stress as a mean to accelerate pro- tein aggregation and obtain information about stability under con- ditions that are relevant for long-term storage." ‘These limitations are strongly associated with the interplay between protein conformational stability and aggregation pro- pensity. The outcome of forced degradation studies at 40°C could be influenced by the thermal stability of the mAb conformation, while this biophysical parameter might have little relevance for stability at lower temperatures, Because antibodies generally possess differing melting temperatures, the same increase in tem- perature may indice different conformational changes in different ‘molecules. In our study, the midpoint temperature of the first transition Tyyx measured by differential scanning calorimetry was lower for most of the formulations of mAb-2 compared to those of ‘mAb-1, that is, an overall average of 62°C relative to one of 68°C (ee Supplementary Fig, $2). Thus, at 40°C, mAb-2 may undergo larger conformational changes compared to mAb-1. By contrast, those structural changes could be much less relevant at storage temperatures, which are far from the Ty of both mAbs. This ‘observation is confirmed by the additional analysis of the correla- tion between the aggregation rate at 5°C and 40°C for the 2 mAbs separately, which is presented in Supplementary Figuve $5. Indi= vidually, mAb-1 (which exhibits higher Tra1) showed better corre- lation relative to mAb-2, For mAb-2, the correlation was very poor, even excluding the outlier in Figure Ga, Moreover, the correlation ‘was better for formulations with pH close to or at the mAbs’ iso electric point, which is of limited practical relevance, because tia Secs 1082020 595-602 0 Rank @ 40°C 15 30 1 s (a) Relationship between the agrepstion ate at Cand a 40°C {Plato the stabity pond 00 mg fran fib eee abd mAb-2 (cleat pt ~ pb eed Fal symbol ctrespond to rmulssins witout excipent sn hliled sybel fo antibodies are normally formulated at pH values lower than the pl (ee Supplementary Fig. $5). In agreement with these consider- ations, we observed that the correlation between the kinetics of aggregation and the ranking of different formulations was much better between 5°C and 20°C (see Fig. 7) The interplay between thermal stability and aggregation is probably also responsible forthe observed reversal in the ranking of the aggregation rates as a function of formulation pH between, high and low temperatures: At 50°C, the aggregation rate increased with decreasing pH value (see Figs. 2-2¢ and 6). The exception vere formulations of mAb-1 without salt, where the formulation at DpH = pl was the least stable at 50°C. By contrast, at §°C, lower pH corresponded to slower aggregation in most cases, Decreasing pH simultaneously reduces the conformational stability (ie, lower Tr) fof the mAbs and increases their colloidal stability by increasing electrostatic repulsion between antibody molecules. At low tem= perature, the later stabilizing effect appears to dominate, whereas at high temperature, the abundance of unfolded species and attractive interactions among them become more important. ‘The role of conformational stability at different temperatures is also relevant inthe context of the different microscopic aggregation mechanisms observed in forced degradation studies and under storage conditions, as summarized in Figure 8, Under thermal sess, mAb aggregates increase their size either by growth via ‘monomer addition or through aggregate-aggregate coagulation, By contrast, at refrigerated temperatures at comparable monomer conversion, only dimers are observed, The growth of dimers into oligomers requires the establishment of a link between one of the monomers engaged in the dimer and a third mAb monomer. Probably, the most flexible domain of a mAb will be the one involved in the establishment of the initial bond = ons| oan 000 035 _@src 050 rck] Figure 7. Corelason betwee stability under storage conisans ad in acletate 0 Rane @ 30°C is was es} laos between the aggregation cate 215°C and 30°C (bla the sabiayWael ea Jura of Pharmaceutical Secs 1082020) 595-52 oon a foc Mice Ae SC Cte “ee Diericaion | | 9 + x D Ginen Memumer ‘Comeran ve + weet iN Nockaton —_Growih at ao, oe eh Figure Sherali istration highlighting the diferences in mA ageeglion under thermal ses and under stersge codons, (2) AL igh temperature, the sizeof mAb segregates increases due to growth by mopereradaion and aggresategsretate coagulation By conta, the aggrezsion process atest fe cme lormason at lower Temperatures (b) Cansequrily he sare extent of enor ean Teas lo aggregate populations ehaaciezed by dasa dierent se dsb, with an, all lgemers formed a lw temperate a ewe, ge agefezaesgeecled at gh emperaare within the dimer, Thus, formation of additional bonds with other ‘mAb molecules will require flexibility also in other parts of the smAb. Increasing temperature may render other sections af the mAb structure also available for intermolecular contacts, therefore increasing the likeibood of growth and coagulation of aggregates. ‘Overall, our results suggest that thermal stress also accelerates, aggregation pathways that are not relevant for storage conditions, leading tothe formation ofan aggregate population that may not be representative of the aggregates formed over the shelf-life of anti- body drugs. Although these findings require further validation with a larger number of molecules and formulation conditions, these results suggest that under storage conditions, dimer formation is an important process. Future directions should therefore focus on the characterization of mAb dimers, which can be very different in terms ofboth intermolecular linkages (te. covalent vs. noncovalent) and iG domains involved in the connections." Inthis context it is important to combine modeling activities with advances in analytical methods, The characterization of dimers and other aggregate end products involves a variety of orthogonal tech- niques’ and ean benefit from emerging approaches based on microfluidic technology"**° We refer to recent reviews and reports, for discussion about the strengths and shortcomings ofthese tech- niques Finally, the design of accelerated stability studies would benefit from the establishment of rules (possibly based on Ta) to determine the temperature range in which the predominant aggregation ‘mechanism corresponds to the pathway under storage conditions. Conclusion We investigated the aggregation kinetics of 2 mAbs in the temperature range from $°C to 50°C to access the window relevant for drug product storage as well as accelerated stability and forced degradation studies. Our results highlight several challenges for shelflife predictions based on thermal stress studies. First, the aggregation rate of mAb formulations shows non-Arthenius Dehavior as a function of temperature. Second, changes in tem- perature also affect the underlying aggregation mechanism and the corresponding reaction order. Specifically, aggregation at 5°C and 30°C is dominated by dimerization, while at the same monomer conversion, larger aggregates are formed at higher temperatures. Consequently, the same extent of monomer loss leads to the formation of different aggregate populations, which may poten- ually be associated with different risks of immunogenicity Moreover, our results show that there is very poor correlation, between the stability ranking of different molecules as well as of different formulations under storage conditions and under thermal sess at 40°C, in particular under typical pH values of commercial ormulation, ‘Overall, our results demonstrate that, in addition to the rate of ‘monomer loss, characterizing the microscopic aggregation mech- anisms and the end aggregate products is crucial to design suitable accelerated stability studies. Acknowledgments ‘The authors are grateful to UCB Pharma (Braine-t'Alleud, Belgium) for supply of material and financial support References 1. Maison € Res tothe bth pipeline 2017. Nr tei 2018250) 2. Agrawal RS. Whats fueling the biotech engine-2012 to 2012, Net techno 3 FeV, Hawe A Sehellekens steo: W. Aggrezation and immunogen of ‘tapes tes I greg of Tape Peters Habe No 4 Mosse EM, Fanchl P- Moorthy 8S, «¢ a. mmanagenciy of therapestie poten aggregates | Pharm Se 2018 03(2)417-430, 5 esc i andi 7, Va D8 a oxtedaton Dann an {6 Wang W, Singh 5K LIN Toler MK. King KR. Nema Smmunoze cers and resist] Pharm, 2012.41, 1~2) 9-1, fC, De Pal SH ea Subse parale conten, oul, hd dove ofan enythvopoitin peptide mimetic prodset se siete th Severe adverse postnarteing eves] Phar 20161053) 1023-1027, 4, Gescesll Atabodies apne uso: native and recombinant Nephi awplon. 20021790003} -42-1 9 Mating ME Maura Kei 35, ta Approaches the 40, Kendrick 85. Clelane Il, arn Xe al-Aggegstion of recombinant human Incerferon gamma: kinetics and structural fanscons Phorm. 7988;87(3) 11 Reni B, Carpets, lad J. Randy A tein expansion of ammna Proc Nal Ad Sa USA f9b-95 24h tala2- 1a 4b va. il Caen a El fh eating ses on ity of protein» 16, 1%. 19. 2», 2, a, 2, 2, os, ¥. 3s, 40 Wace Jura of Pharmaceutical Sriencs 1082020 595-502 Ui A Meyer, Manning MC Aeson of uma en thay dior teal wal ata ep | Mao eka hese np Tupelo the ert let saveznon snd pare fort Pam Sa Greare MeN Seba DK Be 5. Crete J, Randa 1 Ft ateaten sparc nan nee fas syne Pm dt CW Les ER ad 1 Capen ede abe pc 1 oan pple nwa nog sen ok thing aa homens else ica cnn prt treatin apres Para Bis oaome Spe duces: andeph TW. Fare mat and aeien ete pc absense) Erbe ian Kenie Chan BS, Caen Rnd Roe of coloatenl aby ae clea! aby nthe aeerton ease fie hman sen cosnysany fat PS 00.08) frend 5. apne, andsgh 1, eplaton ance mang station hnete embian man mech eer aa Sh fr Sos aera Aah pit ester: mechanisms ds. an cil Mang Wi hse ten sreiton-techanas detect cena Inj nao 0 a Robes) Kes of urevecble tn ae: ays of extended ion ma anno cen es Ps new Mc tober CA meng eed pabezatin mie! Di eaten nee Actegston wh precutated wlan Pec a0 a0 ae thease gpgton charm enon sine ‘ro ina Lande. Mell Poplin base ode of fede ior eth a a Tae {foo abode low om ane vg sees Phe Rr sa ashe Jesu i iQ Nusedsanuel yy Saf 0. cseaton and Chieti hcgee ted sept el he as) Ne Set tees Sharma. Csieaton beans Jarra. biter Randi Tal ba rt wb an nonoge forsttine eave tare a0 ois abe ‘Shand ee Kaman Se bien se putes of « mune Donnas embody se mereanegme ene eganeomen Seep op vss sistance Pom 3 {eels Siep HE Tapa Sa Fysal araceneaon ain vive Disapcl impo neh acre stbees eaten nace poplin by cena el sag Ph 2085 00 Ina Mak Main a hy teat andy ape Cirencester repose Bo Ginn aonsanvaouastee ase ey aie et a mange fanny ages in Salm ten fuer env biel medic he imaty brn scr of Chase presse nce fr beng Mg jae Bares Aa Shes Bk Vine sve A weenom van ge Wert Cae HO Mae Hho feted upon o enpeate pa) or Sus eH than. acd eration dies: ds te pcg lerprtensne therapeu ey tem 20 13) step ovate cheung Je Beate Mt eed dean fame es VW vung TM Reber ices spresces nd chaengs pricing pte Spenser sei | Pa 3 DOO *. 6. *. so. 3 2, 6, ‘ln et note Stars SEAPANON ie at sc rh uci in in oe Rane ste aes ete fe ean cana Soe atte ites mcs re ty weeste asa [re Se Pi ae Aah, Nata iN Hes of temperance an eases {Sepals routs mn ge abe Par caren et ad os Neer eer osc nh Eur J Pharm Se. 2015;77:170-178, SAE et pen arse. M2035) Inara et Ae an cite aan sen ete ee a, Feet t ienae ish cto Seagate SCENES eM i petro Pref mch eer, soo a TERRE leat ons tt pte el of in Sit sites a gh tens newt manne poe ns Stowe mse Mood Pha 0 soon T- STUDS RUSS eAmina lgn sce fies tes eet ep eran abd eT tr ny ty shih noah ing ae i ee cn as tote ik raped TW. Corel MEM Ries an temadymaics ge ne Sel Soa Re Bens as Remmele RL, Callahan W, Krishnan 5, eal Active dimer Sosa ith Kage See hare rh ine a Peto ec a epee ea SNe 9) ett ett hin osc ns Sr saceny ta Snes Bete atte hoe phn mang aeen Celt pe oreo a ms a Cea Ce hpi Sas Sr te Pte es! pe py JM Kapp WRG. Rove Mibu hes fc he q otc air) mse PRS SATAN ea Sete: br she user of Meena aman pale Scop hs CGrabarek AD, Weinbuch D, Jskoot W, Hawe A, Critical evaluation of micro Ms BS Tae Seman eh Ba ‘A-comprehensive evaluation of nanoparticle tracking ahalysis (NanoSight) for