manufacturing of high-concentration monoclonal …...research manufacturing of high-concentration...

RESEARCH

Manufacturing of High-Concentration Monoclonal AntibodyFormulations via Spray Drying—the Road to ManufacturingScaleBENSON GIKANGAa, ROBERT TUROKc, ADA HUIb, MAYUMI BOWENa, OLIVER B. STAUCHa,and YUH-FUN MAAa,*

aPharmaceutical Processing and Technology Development; bLate-Stage Pharmaceutical Development, Genentech, Inc., a memberof the Roche Group, South San Francisco, CA; and cSPX Flow Technology Systems, Inc., Elkridge, MD ©PDA, Inc. 2015

ABSTRACT: Spray-dried monoclonal antibody (mAb) powders may offer applications more versatile than the freeze-dried cake, including preparing high-concentration formulations for subcutaneous administration. Published studies on thistopic, however, are generally scarce. This study evaluates a pilot-scale spray dryer against a laboratory-scale dryer tospray-dry multiple mAbs in consideration of scale-up, impact on mAb stability, and feasibility of a high-concentrationpreparation. Under similar conditions, both dryers produced powders of similar properties—for example, water content,particle size and morphology, and mAb stability profile—despite a 4-fold faster output by the pilot-scale unit. Allformulations containing arginine salt or a combination of arginine salt and trehalose were able to be spray-dried with highpowder collection efficiency (�95%), but yield was adversely affected in formulations with high trehalose content due topowder sticking to the drying chamber. Spray-drying production output was dictated by the size of the dryer operated atan optimal liquid feed rate. Spray-dried powders could be reconstituted to high-viscosity liquids, �300 cP, substantiallybeyond what an ultrafiltration process can achieve. The molar ratio of trehalose to mAb needed to be reduced to 50:1 inconsideration of isotonicity of the formulation with mAb concentration at 250 mg/mL. Even with this low level of sugarprotection, long-term stability of spray-dried formulations remained superior to their liquid counterparts based on sizevariant and potency data. This study offers a commercially viable spray-drying process for biological bulk storage and anoption for high-concentration mAb manufacturing.

KEYWORDS: Spray drying, Monoclonal antibody (mAb), Powder collection yield, Powder reconstitution, Watercontent, Scale-up.

LAY ABSTRACT: This study evaluates a pilot-scale spray dryer against a laboratory-scale dryer to spray-dry multiplemonoclonal antibodies (mAbs) from the perspective of scale-up, impact on mAb stability, and feasibility of ahigh-concentration preparation. The data demonstrated that there is no process limitation in solution viscosity whenhigh-concentration mAb formulations are prepared from spray-dried powder reconstitution compared with concen-tration via the conventional ultrafiltration process. This study offers a commercially viable spray-drying process forbiological bulk storage and a high-concentration mAb manufacturing option for subcutaneous administration. Theoutcomes of this study will benefit scientists and engineers who develop high-concentration mAb products byproviding a viable manufacturing alternative.

Introduction

We previously reported that a laboratory-scale, high-efficiency spray dryer enables spray-drying multiplecommercial monoclonal antibodies (mAbs) with

�95% powder collection efficiency, and we demon-strated that mAbs are amenable to the transient, high-temperature stress (�200 °C) during spray drying (1).Despite these promising data, it is still premature toclaim that protein spray-drying technology is a viablemanufacturing process because the optimal liquidthroughput for the laboratory-scale dryer remains low(10 –15 mL/min, or 0.6 – 0.9 L/h). Therefore, the scal-ability question needs to be addressed and understood.“Manufacturing-scale” spray drying for mAb produc-tion is different from what is defined in the chemical

*Corresponding Author: Yuh-Fun Maa, Telephone: 650-225-3499; Fax: 650-742-1504; e-mail: [email protected]

doi: 10.5731/pdajpst.2015.01003

59Vol. 69, No. 1, January–February 2015

mailto:[email protected]

or food industry (tons per hour or per day). In bio-technology, a standard 12,000 L fermentation processcan produce the final bulk drug substance (DS) (e.g., at100 mg/mL) in hundreds of liters. Using 200 L as anexample, it will take 220 –330 h for laboratory-scaleequipment to spray-dry the whole batch. Thus, a spraydryer with a 5-fold increase in drying capacity canprocess 200 L of mAbs within 3 d, which seems to bea reasonable starting point for scale-up consideration.A pilot-scale spray dryer in the existing product linehas a maximum drying air rate of 154 kg/h, which is4.4 times higher than that of the laboratory-scaledryer.

Thus, the objective of this study was twofold. Firstly,the performance of the pilot-scale dryer in spray dry-ing multiple mAbs was evaluated against the labora-tory-scale unit in terms of production output, powdercollection efficiency, and powder properties such aswater content, particle distribution, and particle mor-phology. The impact of spray drying on short-term andlong-term mAb stability was assessed by protein sizevariant quantitation and potency assay. The secondobjective was to assess the viability of manufacturinghigh-concentration/high-viscosity mAb formulationsby reconstituting spray-dried powders. Normally, finalprotein/mAb bulk DS is prepared at the end of thepurification process using ultrafiltration/diafiltration(UF/DF). UF/DF may encounter significant challengeswith high-viscosity bulk, such as prolonged pro-cessing time, decreased recovery, and possible ad-verse changes in product quality attributes.

Formulation is known to play a critical role in spraydrying performance and mAb stability (1– 4). Carbo-hydrate sugars, such as trehalose and sucrose, havedemonstrated their effectiveness in stabilizing mAb inthe dehydrated state via spray drying (5– 8) and lyoph-ilization (9 –12), and the 300:1 molar ratio of trehaloseto mAb (or proteins)—a 0.7:1 weight ratio—is typi-cally considered sufficient to protect the proteins. Un-fortunately, at this level, powder collection efficiencyis diminished because of the tackiness of trehalose,and its amount should be decreased to a 0.5:1 weightratio (220:1 molar ratio) in order to maintain powdercollection efficiency at �95% (1). Although a 220:1molar ratio is still effective in protecting mAbs, otherrestrictions may require the amount of trehalose to befurther reduced: The isotonicity of the liquid formu-lation for subcutaneous injection and the need to re-duce the viscosity of high-concentration mAb liquidformulations must be considered. Thus, the formula-

tions investigated herein contained trehalose, an or-ganic salt, or a combination of both. The inclusion ofan organic salt, for example, arginine succinate, was toreduce the viscosity of the high-concentration mAbliquid upon powder reconstitution. The amount ofarginine succinate and trehalose, however, needs to beconfined in consideration of the formulation’s tonicity.Overall, this study reported a scaled-up spray-dryingprocess with mAb formulations tailored to a high-concentration/high-viscosity drug product (DP) man-ufacturing process.

Materials and Methods

Two recombinant humanized mAbs of the IgG1 sub-class, mAb A and mAb B, were manufactured byGenentech (South San Francisco, CA). These mAbswere expressed in Chinese hamster ovary (CHO) celllines. All mAbs were prepared into comparable levelsof mAb concentrations using a tangential-flow filtra-tion unit (Pellicon3 10kD, Millipore, Billerica, MA)and formulated with arginine succinate with or with-out trehalose dihydrate (Table I). All bulks were buff-ered to a pH of �6.0.

Spray Drying

Two types of spray dryers were used in this study, alaboratory-scale unit (Anhydro MicraSpray35, SPXFlow Technology Systems, Inc., Elkridge, MD) (MS-35; Figure 1a) and a pilot/manufacturing-scale unit(Anhydro MicraSpray150, also from SPX) (MS-150;Figure 1b). The capacity of the MS-150 dryer is ap-proximately 4 times larger than that of the MS-35dryer (154 vs 38 kg/h drying air rate). The laboratory-scale unit was constructed mostly of stainless steelwith surface heating and insulation (drying chamber,cyclone, etc.) plus a glass vision panel, while thepilot/manufacturing-scale unit was made of stainlesssteel without surface heating.

Particle Size Analysis

Particle size distribution was measured using a laserdiffraction analyzer (LA-950, HORIBA, Ltd., Kyoto,Japan). Based on light scattering, the particle sizedistribution of a sample can be analyzed and calcu-lated using the Mie theory. For analysis, several mil-ligrams of the spray-dried powders were dispersed in50 mL of isopropyl alcohol in the instrument’s samplehandler and sonicated for �1 min to disperse theparticles prior to analysis.

60 PDA Journal of Pharmaceutical Science and Technology

Water Content

Water content in the spray-dried samples was deter-mined using a volumetric Karl Fischer titration an-alyzer (DL31, Mettler-Toledo, Columbus, OH). Ap-proximately 100 mg of each sample were injectedinto the titration cell that contained anhydrousmethanol. Hydranal�-Composite 2 volumetric re-agent (Riedel-de Haen, Heidelberg, Germany) wasused as a titrant.

Size Exclusion Chromatography (SEC)

The quantitation of size variants was determined bySEC. The analysis utilized a G3000SWXL column, 7.8mm inner diameter (ID) � 30 cm, 5 �m (TOSOHBioScience, King of Prussia, PA) run on a high-performance liquid chromatography (HPLC) system(model 1200, Agilent Technologies, Santa Clara, CA).The mobile phases were 0.2 M potassium phosphateand 0.25 M potassium chloride at pH 6.2 for mAb Aand 0.1 M potassium phosphate at pH 6.8 for mAb B.The chromatography was run isocratically at a flowrate of 0.5 mL/min for 30 min. The column tempera-ture was maintained at ambient for both mAb A andmAb B, and the eluent absorbance was monitored at280 nm. Each mAb was diluted with its respectiveformulation buffer to 25 mg/mL for mAb A and 10mg/mL for mAb B. The injection volumes were 10 �Lfor mAb A and 20 �L for mAb B.

Osmolality Measurement

Osmolality was determined by the freezing-pointdepression method using an Advanced� 2020-BIOMulti-Sample Osmometer (Advanced Instrument, Inc.,Norwood, MA). Each sample (20 �L) was transferredinto disposable tubes and loaded onto the osmometer.The osmolality of each sample was determined basedon the freezing point of the solution. For the samplesthat did not freeze, Vapro Osmometer 5520 (AdvancedInstrument, Inc.) was used. Each sample (10 �L) wastransferred onto a solute-free paper disc in the sampleholder. The osmolality of each solution was deter-mined based on the dew-temperature depression.

Viscosity Measurement

Viscosity was measured using a Physica MCR501controlled stress rheometer (Anton Paar, Ashland,VA) equipped with Rheoplus software. A 75 cm IDcone plate was used for all the analyses. The sample(70 �L) was transferred on the cone plate and sub-jected to constant shear of 1000/s at 25 °C for 10 s.Ten readings were taken, and an average viscosity wasdetermined.

In Vitro Potency Assay

An in vitro potency assay was used to determine theactivity of mAb A by measuring its ability to inhibitvascular endothelial growth factor–induced human

TABLE ISpray-Drying Formulations of mAb A and mAb B

Formulation

Concentration (mg/mL)Osmolality

(mOsmol/kg)

mAb

NaPhosphate

BufferHisHClBuffer

NaCitrateBuffer Trehalose

mAb:Trehalose(molar ratio)

ArginineSuccinate NaCl

TotalSolid Average

mAb A

A-1 25 6.29 - - 54 0.5 (1:947) 0 0 86 280

A-2 109 6.29 - - 54 2 (1:217) 0 0 170 304

A-3 109 5.59 - - 209 0.5 (1:841) 0 0 323 1245

A-4 98 - - - 0 NA 43.5 0 142 206

A-5 100 - - - 8.6 12 (1:38) 43.5 0 152 233

A-6 100 - - - 34.3 3 (1:150) 43.5 0 178 325

mAb B

B-1 25 - 0.87 - 20 1 (1:350) 0 0 46 87

B-2 104 - 0.84 - 48 2 (1:202) 0 0 153 200

B-3 117 - 0.77 - 220 0.5 (1:825) 0 0 338 1153

B-4 99 - - - 0 NA 43.5 0 143 196

B-5 100 - - - 8.6 0.5 (1:37) 43.5 0 153 211

B-6 93 - - - 34.3 3 (1:162) 43.5 0 171 299


umbilical vein endothelial cell (HUVEC) proliferation(13). For determining the activity of mAb B, themethod is based on measuring its ability to inhibitproliferation of BT-474 cells using alamarBlue� (LifeTechnologies Corporation, Grand Island, NY) (14).Samples were reconstituted to 100 mg/mL and dilutedwith formulation buffer to 0.5 mg/mL before submis-sion for analysis.

Reconstitution Time

Each of the spray-dried powders was reconstitutedwith 1 mL of purified water to make a protein solution

of 25 mg/mL in a 2 mL glass vial. The vial wasagitated at 450 rpm on a shaker (model 120, Glas-Col,Terre Haute, IN) at ambient temperature. The timerequired to completely dissolve the powder was re-corded.

Turbidity of Reconstituted Solution

Opalescence of the reconstituted solutions at a proteinconcentration of 25 mg/mL was measured in a 1 cmpath-length cuvette using a UV spectrophotometer(model 8453, Agilent Technologies). The test sampleswere blanked against purified water. Absorbance was

Figure 1

Spray dryer designs: (a) MS-35, (b) MS-150 with dual cyclone design, (c) original MS-150 cyclone, (d) designof dual cyclone for MS-150.


recorded at 340, 345, 350, 355, and 360 nm, and theaverage of the absorbance readings was reported as theturbidity of the samples.

Results and Discussion

High-Concentration/High-Viscosity DPProcess—UF/DF versus Spray Drying

A typical biologics production process features DSand DP manufactured at different sites. The DS siteproduces the final bulk DS through a series of up-stream/downstream unit operations concluded by UF/DF. The bulk is then filled into DS storage containersand is typically frozen for storage and shipping to aDP site where bulk DS can be further compounded(e.g., pooling, dilution, reformulation, etc.) prior tobeing filled into a primary container/device, (e.g.,vials or syringes; Figure 2a). To produce high-concen-tration mAb bulk DS, (e.g., 200 mg/mL or higher), theconventional single UF/DF process may not be appro-priate and can be replaced by a two-stage UF1/DF/UF2 (15) operation as shown in Figure 2b. However,at such a high concentration, gelation can occur at thefilter membrane surface and dramatically reduce thepermeate flux. Normally, to overcome this challenge,the UF/DF processing temperature is elevated (�25 °C)to reduce viscosity. Unfortunately, this high temperaturecan raise stability concerns, especially for sensitive pro-teins. Low DS bulk recovery (i.e., yield) at high viscosityremains a drawback.

An alternative approach is to replace the UF2 processwith spray drying of the DS bulk derived from UF/DF(Figure 2c). In this single UF/DF process, the proteinconcentration can be increased to a reasonable level(e.g., �100 mg/mL) to reduce the bulk volume forspray drying. The resulting powder can be stored in adisposable bag. The storage and shipping of dry pow-der bulk can offer advantages over transporting DSbulk in the liquid or frozen state. At the DP site, thepowder can be reconstituted with a diluent into a finalhigh-concentration DS bulk. Our study was based onthis concept and focused on developing a scaled-upspray-drying process on multiple mAb formulationstailored to satisfy long-term mAb stability and recon-stitution properties.

Formulation Considerations

Formulation strategies are set based on the applica-tion of spray-dried powders. In a previous studyintended for the application of biologic bulk stor-age, formulations containing only different levels oftrehalose were assessed to evaluate the impact onpowder collection efficiency and mAb stabilizationduring and after spray drying (1). For the currentapplication in powder reconstitution into a high-concentration mAb liquid (�200 mg/mL) forsubcutaneous administration—for example, via apre-filled syringe—two additional constraints weregiven to formulation consideration: viscosity reduc-

Figure 2

Process flow for producing high-concentration mAb formulations: (a) conventional UF/DF process, (b)UF1/DF/UF2 process, (c) reconstitution of spray-dried powder.


tion at high mAb concentration and maintainingacceptable formulation osmolality.

A potential challenge in the development of high-protein concentration formulations is concentration-dependent solution viscosity (16 –20). For subcutane-ous injection using a pre-filled syringe, injection force(or glide force) is a complex factor influenced bysolution viscosity and the size (gauge) of the needle(21). Smaller needles (e.g., �26 gauge) will pose lesspain sensation to the patient but require more force toinject a high-viscosity drug. Based on a viscosity–glide force relationship calculated by the Hagen-Poi-seuille equation as a function of needle gauge, with a27 gauge thin-walled needle (ID, minimum: 0.241mm), the liquid viscosity should be maintained below20 cP in order not to exceed a glide force of 20 N (21).Unfortunately, formulation scientists are constantlychallenged by the conflicting reality of a high mAbconcentration and high solution viscosity (14 –18).Different formulation strategies have been proposed toreduce the viscosity of liquid solutions at high mAbconcentrations, including formulating with organicand inorganic salts to balance repulsive and attractiveforces through intermediate ionic strengths (16 –19,22, 23). Of these, arginine salt has been proven effec-tive (17, 18, 20). The impact of arginine salt on spraydrying is, however, unproven and was thus investi-gated in this study. In addition, the concentration ofthe excipients should be confined to meet the isoto-nicity requirement of �300 mOsmol/kg (24). How-ever, maintaining this level of osmolality for subcuta-neous administration remains debatable and may notbe necessary (17). A recent randomized, placebo-con-trolled, double-blind, cross-over trial assessed the ef-fect of osmolality (300, 600, 850, and 1100 mOsmol/kg) administered intramuscularly on local tolerance.The results did not show any dose– effect relationshipbetween burning and pain sensations and the differentosmolalities tested (24, 25, 26). As an osmotic effectcan be influenced by the volume of injection, which isnormally the case for high-dose mAb formulations,our formulation strategy was still to try to keep withina threshold, for example, 600 mOsmol/kg.

With these considerations, all formulations tested forspray drying, as listed in Table I, included four majorcompounds: mAb, a buffer, trehalose, and/or argininesuccinate. Formulations containing arginine salt (A-4,A-5, A-6, B-4, B-5, and B-6) were the focus of thisstudy. The osmolality of each formulation was mea-sured; all formulations prior to spray drying were

hypotonic to isotonic (i.e., �300 mOsmol/kg) with theexception of A3, B3, and C3, which contained thehighest amount of trehalose (�200 mg/mL) andheightened osmolality to �700 mOsmol/kg. When thespray-dried powders of the arginine-containing formu-lations were reconstituted with water to a mAb con-centration of 200 mg/mL, their osmolality would staybelow 800 mOsmol/kg.

The laboratory-scale spray dryer MS-35 was equippedwith a high-efficiency cyclone, to which the enhancedpowder collection efficiency was attributed. The pilot-scale unit MS-150 originally came with a single, largercyclone (Figure 1c), which was replaced and testedwith a dual cyclone unit (Figure 1d). Each of the twocyclones was fabricated according to the cyclone de-sign associated with MS-75 (SPX Flow TechnologySystems, Inc., Elkridge, MD; detailed drawing notshown). Two cyclones were coupled to accommodatea 2-fold increase in drying air flow rate. The batch sizeof each run was also increased per the size of the spraydryer, from 0.10 – 0.25 L (MS-35) to 0.5–1.0 L (MS-150). To calculate the yield of powder collection, onlythe powder collected in the receiver was considered.The default drying air inlet and outlet temperatures, Tiat 180 –190 °C and To at 80 –90 °C, respectively, wereapplied to both MS-35 and MS-150. Spray dryingconditions and the characteristics of the dry powdersproduced using each spray dryer are listed in Table II.

TABLE IISpecification Comparison and Residence TimeCalculation of MS-35 versus MS-150 SprayDryers

SpecificationsLaboratory-Scale

MS-35Pilot/Manufacturing-Scale

MS-150

Chamber diameter (m) 0.3 0.9

Chamber straight sideheight (m)

0.64 0.83

Chamber cone height (m) 0.322 0.965

Chamber cross-sectionalarea (m2)

0.0707 0.636

Chamber volume (m3)(cylinder � conevolume)

0.05 0.73

Drying air rate (kg/h) 38 154

Drying air rate (m3/s)a 0.0106 0.043

Drying air velocity inchamber (m/s)

0.15 0.07

Theoretical air residencetime (s)b

4.72 16.98

a Assuming air density � 1 kg/m3.b Theoretical air residence time � chamber volume/airrate (in m3/s).


Spray Dryer Scale-Up: Laboratory-Scale (MS-35)versus Pilot-Scale (MS-150)

MS-35 was previously assessed and optimized forspray-drying mAb formulations (1). In that study, highpowder-collection efficiency (�95% yield) was re-ported for three mAb models. Trehalose was effectivein mAb stabilization, but powder yield deterioratedwith increasing trehalose concentration. In general,the 2:1 weight ratio of mAb:trehalose (i.e., 1:220molar ratio) offered acceptable stability and goodpowder yield. A standard drying condition, Ti of 180 –190 °C and To of 80 –90 °C could produce powders ofapproximately 5% moisture content when the liquidwas dried at a rate of 15 mL/min (i.e., �1 L/h). Toincrease production scale (output), a pilot-scale spray

dryer, MS-150, which enhances drying capacity 4-foldbased on drying air flow rate, was assessed in the studypresented here. The comparison of MS-35 and MS-150 in drying chamber dimensions and calculated airresidence time is summarized in Table II. The dataindicate that it takes almost 4 times longer for thedrying air to pass through the MS-150 than it does topass through the MS-35.

All formulations (Table I) were spray-dried usingMS-35 and MS-150, and the drying conditions andpowder/process characterization results (yield, particlesize, and water content) are summarized in Table III.Particle size (median value, D50) was slightly largerwhen powders were spray-dried using MS-150 (�15�m) than when using MS-35 (�12 �m), but size was

TABLE IIISpray-Drying Conditions of MS-150 and MS-35 and Summary of Results

Spray Dryera FormulationTi

(°C)To

(°C)Liquid Feed

Rate (mL/min)Yield(%)

Particle Size(�m, D50)

MoistureContent (%)

MS-150 A-1 190 90 53 82 14.7 4.4

A-2 190 90 49 93 16.0 3.8

A-2* 190 90 50 �50 NA NA

A-2 190 100 20 90 14.1 3.6

A-2 190 84 102 31 15.5 7.7

A-2 225 95 84 45 14.9 5.6

A-3 190 89 66 54 15.5 4.5

A-4 187 88 58 99 14.8 6.9

A-5 187 88 39 100 15.7 5.8

A-6 187 87 46 99 17.7 4.9

MS-35 A-4 189 87 15 95 12.9 5.8

A-5 189 88 14 97 11.6 5.0

A-6 189 97 16 100 12.6 4.3

MS-150 B-1 190 90 52 84 15.1 5.0

B-2 190 90 51 98 14.7 4.2

B-2* 190 90 50 �50 NA NA

B-3 190 90 58 41 16.1 4.7

B-4 187 88 55 92 14.6 6.3

B-5 187 88 48 100 15.1 6.1

A-6 187 87 55 100 16.6 5.8

MS-35 B-4 189 87 16 100 12.3 5.3

B-5 189 87 16 100 11.2 5.5

A-6 189 87 17 100 12.7 5.0

Ti, drying air inlet temperature; To, drying air outlet temperature; NA, not available.a MS-150 drying air rate � 154 kg/h and nozzle gas pressure � 10 psig; MS-35 drying air rate � 35 kg/h and nozzlegas pressure � 10 psig.* Spray-dried on MS-150 using the cyclone in Figure 1c.


not affected by drying conditions or formulation. Thewater content of all spray-dried powders was mostlyaround 5% and showed no obvious correlations withformulation, dryer size, or drying conditions. Powderyield was almost fully recovered (�95% yield) for allarginine-containing formulations (A-4, A-5, A-6, B-4,B-5, and B-6) and was very good (�90% yield) forformulations with low levels of trehalose (�2:1 mAb:trehalose w/w, i.e., A-2 and B-2) for both MS-35 andMS-150. However, yield appears to be influenced byseveral factors:

1. Formulation. It is consistent with the previousfinding (1) that formulations with high trehaloseconcentrations (for example, 1:2 mAb:trehalosew/w for A-3 and B-3) are difficult to recoverbecause a significant amount of the powder sticksto the drying chamber.

2. Cyclone design. The dual-cyclone system (Fig-ure 1d) outperforms the original single cyclonesystem (Figure 1c). Each cyclone of the dualsystem has a design concept similar to that ofthe MS-35 as described previously (1). The dualcyclone was designed for the collection of finebiopharmaceutical powders and has a cut-offsize of 1.5 �m, as calculated using a developedmathematical model (27). The original single-cyclone design failed to effectively collectsmall particles because it is designed for gen-eral cyclone application and has a cut-off sizeof 6 �m. This resulted in �50% collectionefficiency for A-2* and B-2* (Table III) com-pared with �90% yield for A-2 and B-2, whichwere collected using the dual cyclone system.

3. Drying conditions. The liquid feed rate is a criticalparameter. Powder yield decreased when the liquidfeed rate was increased from 50 mL/min for A-2(31% at 102 mL/min and 45% at 84 mL/min). Yieldloss occurred primarily in the drying chamber.

The scalability of the spray-drying process is obviousbetween MS-35 and MS-150 based on the rate ofdrying air and the rate of water evaporation (liquidfeed rate). Under the similar drying conditions, bothdryers produced powders of similar water content,particle size, and powder yield (if the similar high-efficiency cyclone was used). The 4-fold increase indrying rate, that is, the liquid feed rate of 50 mL/min(MS-150) versus 15 mL/min (MS-35), reflected the4-fold increase in the drying air rate, 154 kg/h versus

38 kg/h. The attempt to further enhance the productionrate by increasing the liquid feed rate failed, suggest-ing that the current drying conditions were optimizedand further scale-up can only rely on using an evenlarger spray dryer.

MAb Stability/Potency of Spray-Dried PowderFormulation

It was previously reported (1) that, despite a high inlettemperature (�190 °C), the physical stability (deter-mined by SEC-HPLC) of spray-dried mAbs wascomparable to or greater than that of freeze-driedcounterparts. In addition, a carbohydrate sugar (e.g.,trehalose) was essential in stabilizing mAbs duringspray drying and long-term storage, particularly underthe stressed condition (40 °C). The stabilization ben-efit increased with increasing amounts of trehalose,but the lowest amount of trehalose tested in that studywas a 1:220 molar ratio of mAb:trehalose. Stability ofthe mAb was only assessed based on the protein sizingassay (SEC-HPLC).

In the current study, the amount of trehalose wasfurther reduced to decrease osmolality in the arginine-containing formulations, which contained either notrehalose (A-4 and B-4) or one of two mAb:trehalosemolar ratios, 1:38 (A-5 and B-5) and 1:150 (A-6 andB-6). All three formulations were spray-dried usingthe MS-35 and MS-150 units, and the powder sampleswere stored in vials for up to 3 months at 40 °C and upto 6 months at 25 °C. Two liquid controls, liquidformulations prior to spray drying and reconstituted(into 100 mg/mL) from spray-dried powders, werestored under the same conditions to compare withpowder samples. In addition to SEC-HPLC analysis(quantifying size variants), the biological potency ofmAbs and turbidity of the solution were also deter-mined. Figure 3 summarizes the size variant (SEC-HPLC) data for mAb A formulations for 3 months at40 °C and 25 °C, while the size variant data for mAb Bare presented in Figure 4 for 3 months at 40 °C and 6months at 25 °C. The size variant data include high-molecular-weight species (HMWs) due to aggregation,low-molecular-weight species (LMWs) due to fragmen-tation, and total percent monomer (%monomer).

High-Molecular-Weight Species (HMWs): Both an-tibodies aggregated slightly, �1% increase in HMWs,right after spray drying at 190 °C, even for formula-tions containing no trehalose (A-4 and B-4; Figures 3aand 4a, respectively) although the starting level of


HMWs was different between mAb A (�5%) andmAb B (�0.5%). The differences between the twospray dryers were minimal. After long-term storage,mAb aggregation in powder formulations increasedwith enhanced storage temperature and decreasing tre-halose concentration. When powder formulations werecompared with the liquid formulations prior to spraydrying, mAb A and mAb B behaved differently. Theliquid formulations of mAb A aggregated faster thanthe powder counterparts, while the liquid formulationsof mAb B were more stable than the correspondingpowders. The stabilization effect of trehalose wasmore prominent in the solid state than in the liquidstate. When liquid formulations reconstituted from thespray-dried powders were compared with liquid for-mulations prior to spray drying, their tendency toaggregate or fragment was similar, suggesting that the

effect of high-temperature spray drying on long-termstorage is benign.

Low-Molecular-Weight Species (LMWs) (Frag-mentation in Figures 3b and 4b): Fragmentationoccurred minimally in the solid state, while it was amajor degradation pathway in the liquid state for bothmAb A and mAb B (Figures 3b and 4b, respectively).The ability of trehalose to protect mAb from fragment-ing played an insignificant role.

Change (Decrease) in %Monomer: For total de-crease in %monomer (relative to liquid formulationprior to spray drying), similar trends were observed:powder formulations being more stable than liquidcounterparts; increasing trehalose concentration re-sulting in less degradation; stability profile of liquid

Figure 3

Effect of spray drying and effect of long-term storage of liquid and spray-dried formulations (A4 –A6 formAb A) on (a) HMWs before and after 3 months at 40 °C, (b) LMWs before and after 3 months at 40 °C,(c) total %monomer before and after 3 months at 40 °C, (d) total %monomer before and after 3 monthsat 25 °C.


formulations prior to spray drying and reconstitutedfrom spray-dried powders being identical; powdersspray dried from MS-35 and MS-150 showing nosignificant differences (Figures 3c, 3d, 4c, and 4d,respectively). Overall, trehalose effectively stabilizedmAbs at mAb:trehalose molar ratios as low as 1:38.

In Vitro Potency Stability

The in vitro potency assay for mAb A and mAb Bwas based on the ability of the two mAbs to inhibitthe proliferation of HUVEC cells (13) and BT-474cells (14), respectively. All liquid and dry powderformulations at t � 0 and t � 3 months at 40 °Cwere analyzed, and the results are presented for

mAb A and mAb B in Figure 5a and Figure 5b,respectively. The activity of all mAb formulationsafter spray drying was comparable to that of therespective liquid formulations prior to spray drying,suggesting the impact of high-temperature spraydrying is insignificant. Incubation of liquid formu-lation for 3 months at 40 °C substantially reducedtheir activity by 20 –30% for mAb A and 50 –70 %for mAb B. Incubation of powder formulations for 3months at 40 °C has little impact on the potency ofmAb A and approximately 30% potency decreasefor mAb B. Although aggregation is the main deg-radation pathway for dry powder formulations, it ispossible that aggregated mAb species remain active.The LMWs might be less potent or inactive, as

Figure 4

Effect of spray drying and effect of long-term storage of liquid and spray-dried formulations (B4 –B6 formAb B) on (a) HMWs before and after 3 months at 40 °C, (b) LMWs before and after 3 months at 40 °C,(c) total %monomer before and after 3 months at 40 °C, (d) total %monomer before and after 6 monthsat 25 °C.


fragmentation is the major degradation mechanismfor liquid formulations. Although trehalose protectsmAb from degradation (aggregation), it shows noobvious effect in retaining mAb potency, either inthe liquid state or the solid state.

Solution Turbidity

Solution turbidity may be an implication of proteininstability (i.e., the presence of insoluble aggre-gated protein particles). Turbidity is also importantfrom the process perspective because the spray-dried powder will be reconstituted into a high-concentration mAb liquid prior to bioburden reduc-tion and sterile filtration. A turbid solution may tendto clog the filter pores and foul the membrane toslow down or interrupt the manufacturing process.All liquid formulations, including those prior tospray drying and reconstituted solutions from spray-dried powders before and after 3 month incubation

at 40 °C, were UV-scanned before and after filtra-tion through a 0.22 �m filter. The turbidity resultsfor mAb A and mAb B are summarized in Figures 6aand 6b, respectively.

Solutions of reconstituted spray-dried powders aregenerally more turbid than the respective liquid for-mulations prior to spray drying, suggesting the pres-ence of insoluble protein aggregates. The amount ofthese aggregates, however, is difficult to quantify andmay be insignificant, as protein concentration remainsthe same after filtration. Solution turbidity appeared toincrease slightly after 3 month incubation at 40 °C forboth powder and liquid formulations, but there was noobvious trend among the different formulations. Theseparticles were removable upon sterile filtration (0.22�m); after filtration, solution turbidity was compara-ble to the unfiltered liquids even after 3 month storageunder a stressed condition. The impact of solutionturbidity on filterability (i.e., filter fouling) is out of

Figure 5

Effect of spray drying and effect of long-term storage of liquid and spray-dried formulations on in vitropotency of (a) mAb A formulation (A4 –A6) and (b) mAb B formulation (B4 –B6).


the scope of this study and will be evaluated sepa-rately.

Table IV summarizes the time required to reconstitutethe spray-dried powders (A4 –A6 and B4 –B6) into 25mg/mL concentration. All powders were completelydissolved within 3 min. It took slightly longer time todissolve powders produced from the pilot-scale dryer,2–3 min, than the lab-scale dryer, 1–2 min, probablybecause of the larger particle size of the powders fromMS-150. Reconstitution into higher concentrationswill certainly take much longer time. However, recon-stitution will take place at the DP manufacturing siteusing qualified mixing equipment and pre-character-ized mixing conditions, which is beyond the scope ofthis study.

Figure 6

Effect of spray drying and effect of long-term storage of liquid and spray-dried formulations on solutionturbidity before and after 40 °C storage, and before and after 0.22 �m sterile filtration for (a) mAb A (A4 –A6)and (b) mAb B (B4 –B6).

TABLE IVTime Required for Reconstituting Spray-DriedPowders into 25 mg/mL

FormulationReconstitution Time (s)MS-35 MS-150

A-4 89 143

A-5 92 134

A-6 85 144

B-4 134 133

B-5 102 169

B-6 141 173


Reconstitution into High-Concentration/High-ViscosityFormulations

Spray-dried powders were reconstituted into liquidswith mAb concentrations of 100, 150, and 250 mg/mL.Their viscosity and osmolality were measured and aresummarized in Table V. Compared with liquid formu-lations prior to spray drying, reconstituted liquids atthe same concentration (100 mg/mL) had comparableviscosity (�3– 4 cP) and osmolality (200 –300mOsmol/kg). To further increase mAb concentrationvia UF/DF, the associated increase in viscosity maylimit the capability of the standard UF/DF process.Internal manufacturing data suggested that UF re-mains a feasible method to concentrate mAbs to theviscosity range of 15 to 30 cP but with reduced pro-cess yield. UF is not capable of producing mAb for-mulations beyond 30 cP, which requires substantiallylonger processing time and results in lower yields.Although performing UF/DF at a higher temperature(e.g., 30 or 40 °C) can decrease viscosity to enable theprocess, it may risk compromising mAb quality. Thereconstitution approach has no such limitations; A-6formulation viscosity reached beyond 200 cP and be-yond 300 cP for B-6 at 250 mg/mL mAb concentra-tion. Comparing formulations containing trehalose(A-4, B-4, A-5, and B-5) with other formulations, itwas apparent that enhanced viscosity is attributed tothe increased trehalose concentration. In addition, thepresence of arginine salt and trehalose in spray-driedpowders greatly increased solution osmolality. Thus,an even more flexible and effective approach in pro-ducing high-concentration mAb formulations is to in-clude only a carbohydrate sugar at a molar ratio(sugar:mAb) of �150:1, or preferably �50:1 in caseswhere cold storage (2– 8 °C) is allowed. The arginine-containing diluent can be used to reduce liquid formu-lation viscosity during spray-dried powder reconstitu-tion.

Conclusions

This study successfully demonstrated the scale-up ofthe spray-drying process and the supreme advantagesof producing high-concentration/high-viscosity mAbformulations based on spray-dried powder reconstitu-tion. The spray-dried formulations maintained goodmAb stability/potency during long-term storage evenwith mAb:trehalose molar ratios significantly greaterthan 1:150. This low-sugar concentration benefitedspray-drying process yield and enabled lower solutionviscosity upon reconstitution. There is no process lim-itation in solution viscosity when high-concentrationmAb formulations are prepared from spray-dried pow-der reconstitution compared with concentration via theconventional UF process.

Conflict of Interest Declaration

The authors declare that they have no competing in-terests.

References

1. Bowen, M.; Turok, B.; Maa, Y.-F. Spray drying ofmonoclonal antibodies: investigating powder-based biologic drug substance bulk storage. Dry-ing Technol. 2013, 31 (13–14), 1441–1450.

2. Broadhead, J.; Rouan, S. K.; Hau, I.; Rhodes,C. T. The effect of process and formulation vari-ables on the properties of spray-dried beta-galac-tosidase. J. Pharm. Pharmacol. 1994, 46 (6),458 – 467.

3. Maury, M.; Murphy, K.; Kumar, S.; Mauerer, A.;Lee, G. Spray-drying of proteins: effects of sor-bitol and trehalose on aggregation and FT-IR

TABLE VViscosity and Osmolality of Liquid Formulations Reconstituted from Spray-Dried Powders

Formulation

Liquid Formulation Reconstituted Spray-Dried Powder

100 mg/mL 100 mg/mL 150 mg/mL 250 mg/mL

Viscosity(cP)

Osmolality(mOsmol/kg)

Viscosity(cP)


Viscosity(cP)


Viscosity(cP)


A-4 3.0 0.1 201 1 2.8 0.2 200 1 6.7 0.2 311 7 70.6 6.2 626 19

A-5 3.4 0.1 231 3 3.0 0.1 224 11 7.9 0.7 366 10 99.1 12.0 748 6

A-6 3.6 0.1 321 3 3.7 0.1 328 1 10.6 0.1 531 4 207.0 0.2 1190 42

B-4 3.1 0.0 202 1 2.6 0.0 185 6 6.1 0.3 288 1 49.9 1.2 563 8

B-5 3.6 0.0 255 4 2.8 0.2 204 3 7.1 0.3 336 2 59.3 1.9 635 6

B-6 4.0 0.1 314 4 3.3 0.2 303 2 10.1 0.2 405 3 328.7 5.7 953 40


amide I spectrum of an immunoglobulin G. Eur.J. Pharm. Biopharm. 2005, 59 (2), 251–261.

4. Maa, Y.-F.; Costantino, H. R.; Nguyen, P. A.;Hsu, C. C. The effect of operating and formula-tion variables on the morphology of spray-driedprotein particles. Pharm. Dev. Technol. 1997, 2(3), 213–223.

5. Andya, J. D.; Maa, Y.-F.; Costantino, H. R.;Nguyen, P.-A.; Dasovich, N.; Sweeney, T. D.;Hsu, C. C.; Shire, S. J. The effect of formulationexcipients on protein stability and aerosol perfor-mance of spray-dried powders of a recombinanthumanized anti-IgE monoclonal antibody. Pharm.Res. 1999, 16 (3), 350 –358.

6. Costantino, H. R.; Andya, J. D.; Nguyen, P.-A.;Dasovich, N.; Sweeney, T. D.; Shire, S. J.; Hsu,C. C.; Maa, Y.-F. The effect of mannitol crystal-lization on the stability and aerosol performanceof a spray-dried pharmaceutical protein, recombi-nant humanized anti-IgE monoclonal antibody.J. Pharm. Sci. 1998, 87 (11), 1406 –1411.

7. Schule, S.; Schultz-Fademrecht, T.; Garidel, P.;Bechtold-Peters, K.; Frieb, W. Stabilization ofIgG1 in spray-dried powders for inhalation. Eur.J. Pharm. Biopharm. 2008, 69 (3), 793– 807.

8. Maury, M.; Murphy, K.; Kumar, S.; Shi, L.; Lee,G. Effects of process variables on the powderyield of spray-dried trehalose on a laboratoryspray dryer. Eur. J. Pharm. Biopharm. 2005, 59(3), 565–573.

9. Andya, J. D.; Hsu, C. C.; Shire, S. J. Mechanismsof aggregate formation and carbohydrate excipi-ent stabilization of lyophilized humanized mono-clonal antibody formulations. AAPS PharmSci2003, 5 (2), 21–31.

10. Jovanovic, N.; Bouchard. A.; Hofland, G. W.;Witkamp, G. J.; Crommelin, D. J.; Jiskoot, W.Distinct effects of sucrose and trehalose on pro-tein stability during supercritical fluid drying andfreeze-drying. Eur. J. Pharm. Sci. 2006, 27 (4),336 –345.

11. Jain, N. K.; Roy, I. Effect of trehalose on proteinstructure. Protein Sci. 2009, 18 (1), 24 –36.

12. Chang, L.; Shepherd, D.; Sun, J.; Ouellette, D.;Grant, K. L.; Pikal, M. J. Mechanism of proteinstabilization by sugars during freeze-drying andstorage: native structure preservation, specific in-teraction, and/or immobilization in a glassy ma-trix? J. Pharm. Sci. 2005, 94 (7), 1427–1444.

13. Wang. Y.; Fei, D.; Vanderlaan, M.; Song, A.Biological activity of bevacizumab, a humanizedanti-VEGF antibody in vitro. Angiogenesis 2004,7 (4), 335–345.

14. Park, J. W.; Hongi, T. K.; Carter, P.; Asgarit, H.;Guot, L. Y.; Keller, G. A.; Wirthii, C.; Shalabyii,R.; Kotrsi1, C.; Woodt, W. I.; Papahadjopoulost,D.; Benz, C. C. Development of anti-p185IER2immunoliposomes for cancer therapy. Proc. Natl.Acad. Sci. USA 1995, 92 (5), 1327–1331.

15. Li, Y.; Kahn, D. W.; Galperina, O.; Blatter, E.;Luo, R.; Wu, Y.; Zhang, G. Development of aPlatform Process for the Purification of Therapeu-tic Monoclonal Antibodies. In Process Scale Pu-rification of Antibodies; Gottschalk, U., Ed.; JohnWiley & Sons: New York, 2011.

16. Shire, S. J.; Shahrokh, Z.; Liu, J. Challenges inthe development of high protein concentrationformulations. J. Pharm. Sci. 2004, 93 (6), 1390 –1402.

17. Liu, J.; Shire, S. J. Reduced-viscosity concen-trated protein formulations. U.S. Patent 8,142,776B2, 2012.

18. Liu, J.; Shire, S. J. High concentration antibodyand protein formulations. U.S. Patent ApplicationPublication US 2004/0197324 A1, 2004.

19. Jezek, J.; Darron, N. J.; Derham, B. K.; Royle, N.;Simpson, I. Biopharmaceutical formulations forpre-filled delivery devices. Expert Opin. DrugDeliv. 2013, 10 (6), 811– 828.

20. Bowen, M.; Armstrong, N.; Maa, Y.-F. Investi-gating high-concentration monoclonal antibodypowder suspension in nonaqueous suspension ve-hicles for subcutaneous injection. J. Pharm. Sci.2012, 101 (12), 4433– 4443.

21. Overcashier, D. E.; Chan, E. K.; Hsu, C. C. Tech-nical considerations in the development of pre-


filled syringes for protein products. Am. Pharm.Rev. 2006, 9 (7), 77– 83.

22. Sukumar, M.; Doyle, B. L.; Combs, J. L.; Pekar,A. H. Opalescent appearance of an IgG1 antibodyat high concentrations and its relationship to non-covalent association. Pharm. Res. 2004, 21 (7),1087–1093.

23. He, F.; Woods, C. E.; Trilisky, E.; Bower,K. M.; Litowski, J. R.; Kerwin, B. A.; Becker,G. W.; Narhi, L. O.; Razinkov, V. I. Screeningof monoclonal antibody formulation based onhigh-throughput thermostability and viscositymeasurements: design of experiment and statis-tical analysis. J. Pharm. Sci. 2011, 100 (4),1330 –1340.

24. Nony, P.; Girard, P.; Chabaud, S.; Hessel, L.;Thebault, C.; Boissel, J. P. Impact of osmolality

on burning and sensation during and immediatelyafter intramuscular injection of 0.5 ml of vaccinesuspensions in healthy adults. Vaccine 2001, 19(27), 3645–3651.

25. Zietkiewiz, W.; Kostrewska, E.; Gregor, A. Invivo studies on the action on the tissue of theosmolality of parenterally administered drugs.Grzyby Drozdzopodobne 1977, 23, 869 – 870;The Ralph McElroy Translation Company: Aus-tin, TX.

26. Keele, C. A.; Armsotrong, D. Production of Painby Hypertonic and Hypotonic Solutions. In Sub-stance Producing Pain and Itch; Edward Arnold(Publishers), Ltd.: London, 1964; pp 67–72.

27. Enliang, L.; Yingmin, W. A new collection theoryof cyclone separators. AIChE J. 1989, 35 (4),666 – 669.


manufacturing of high-concentration monoclonal …...research manufacturing of high-concentration...

Documents