Pharmaceutics,PreformulationandDrugDelivery
`Characterization of Protein Rheology and Delivery Forces for
`Combination Products
`
`NITIN RATHORE,1 PRATIK PRANAY,2 JOSEPH BERNACKI,1 BRUCE EU,1 WENCHANG JI,1 ED WALLS1
`
`1Drug Product Engineering, Amgen, Thousand Oaks, California 91320
`
`2Department of Chemical Engineering, University of Wisconsin-Madison, Wisconsin 53706
`
`Received 17 March 2012; revised 3 July 2012; accepted 31 July 2012
`
`Published online 31 August 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.23297
`
`ABSTRACT: Characterization of a protein–device combination product over a wide range
`of operating parameters defined by end-user requirements is critical for developing a product
`presentation that is convenient for patient use. In addition to the device components, several
`product attributes, such as product rheology and product–container interactions, govern the
`functionality of a delivery system. This article presents results from a characterization study
`conducted for a high-concentration antibody product in a prefilled syringe. Analytical models
`are used to study the rheological behavior and to estimate delivery forces over a broad design
`space comprising temperature, concentration, and shear stress. Data suggest that high-viscosity
`products may exhibit significant shear thinning under the shear rates encountered under de-
`sired injection times. © 2012 Wiley Periodicals, Inc. and the American Pharmacists Association
`J Pharm Sci 101:4472–4480, 2012
`Keywords: extrusion; rheology; viscosity; drug delivery systems; injectables; mathematical
`model; protein delivery
`
`INTRODUCTION
`The need for higher dosage amounts coupled with pa-
`tients’ preference for fewer injections has resulted in
`the need for high-concentration parenterals. The com-
`mercialization of high-concentration antibody formu-
`lations poses several challenges related to formula-
`tion stability, analytics, and manufacturability.1 In
`addition, the product presentation needs to mitigate
`any potential challenges related to the administra-
`tion of viscous drug products. For many target ther-
`apeutic indications, the drug is expected to be self-
`administered by the patient via a delivery device such
`as an autoinjector. Such autoinjectors can be very ef-
`fective in enhancing user convenience, driving a com-
`petitive advantage.2 Market surveys and historical
`experience related to the therapeutic area in ques-
`tion can be used to develop end-user requirements
`and drive the final product presentation.3 When de-
`veloping high-concentration formulations, it is impor-
`tant to assess their injectability and functionality ear-
`lier during process development. This article presents
`
`Correspondence to: Nitin Rathore (Telephone: +805-313-6393;
`Fax: +805-375-8251; E-mail: nrathore@amgen.com)
`Journal of Pharmaceutical Sciences, Vol. 101, 4472–4480 (2012)
`© 2012 Wiley Periodicals, Inc. and the American Pharmacists Association
`
`theoretical and experimental framework on how such
`evaluations can be applied, with the goal of develop-
`ing a robust formulation that is stable and optimal
`for delivery.
`Three specific aspects of the protein–device com-
`bination product are explored in this study: (a) the
`characterization of product viscosity over a broad con-
`centration and temperature range, (b) the characteri-
`zation of product–syringe interactions and its impact
`on friction forces associated with delivery, and (c) the
`characterization of product rheological behavior un-
`der the high shear rates (∼100,000 s−1) associated
`with syringe injection. The effect of component vari-
`ability on the delivery forces and injection time has
`been studied earlier and the results have been pre-
`sented in a separate article.4
`In addition to product rheology and product–
`syringe interactions, the relationship between the
`measured force and viscosity has been examined us-
`ing theoretical predictions for laminar flow in a tube
`(the Hagen–Poiseuille equation).5 Mechanistic mod-
`els have been verified with experimental data, al-
`lowing for characterization over a wide design space
`beyond the operating set points. Such broad design
`space characterization is essential to understand the
`functionality of the device system, set appropriate
`
`4472
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 101, NO. 12, DECEMBER 2012
`
`Novartis Exhibit 2020.001
`Regeneron v. Novartis, IPR2020-01318
`
`

`

`CHARACTERIZATION OF PROTEIN RHEOLOGY AND DELIVERY FORCES FOR COMBINATION PRODUCTS
`
`4473
`
`performance specifications, and ensure that the final
`product remains within the specifications.
`
`MATERIALS AND METHODS
`Siliconized glass and plastic syringes from different
`vendors were used in the study. Brookfield standards
`of different viscosities (Brookfield 5, 10, and 50 cP)
`were used as Newtonian liquid standards for cali-
`bration purposes. The rheological behavior of several
`high-concentration antibody products (>100 mg/mL)
`was evaluated under different temperature and shear
`conditions.
`Product viscosity measurements were performed
`using a double-gap rheometer (Malvern, Worcester-
`shire, UK) under low-shear-rate (10–150 s−1) condi-
`tions. The extrusion force measurements in syringes
`were performed using a mechanical testing system
`(Instron 5564; Instron, Norwood, Massachusetts),
`equipment designed to evaluate the mechanical prop-
`erties of materials and components. The force mea-
`surements from the Instron 5564 were further uti-
`lized for viscosity characterization over a wide range
`of shear rates; the details of this methodology are dis-
`cussed in the Theory section.
`Mathematical models have been applied to model
`the rheological behavior over a wide range of temper-
`atures and product concentrations. Analytical models
`discussed earlier in the context of estimating extru-
`sions forces4 have been applied to characterize vis-
`cosity over a range of shear stresses associated with
`syringe injection.
`
`THEORY
`The system under consideration is flow through a pre-
`filled syringe as shown in Figure 1. The syringe con-
`sists of a barrel of radius rb with an attached needle
`of length Ln and radius rn. An external force is ap-
`plied through the plunger rod to drive the fluid with
`an injection speed ¯v (plunger rod speed in length over
`time dimensions). The break-loose force refers to the
`maximum force required to set the plunger rod into
`motion. The extrusion force is the force required to
`sustain the plunger rod in motion while maintaining
`the desired flow rate of product through the needle.
`This study characterizes the total extrusion force as-
`sociated with delivery of a product through syringe
`injection.
`
`Friction and Hydrodynamic Forces
`The total extrusion force required to deliver an injec-
`tion can be described as:
`
`Ftotal = f friction + Fhydrodynamic,
`
`(1)
`
`Figure 1. A schematic of the various components associ-
`ated with the syringe delivery system.
`
`where Ftotal is the total force needed for driving the
`plunger, ffriction is the friction force between the stop-
`per and the syringe wall, and Fhydrodynamic is the hy-
`drodynamic force required to drive the fluid out of the
`needle. Hydrodynamic forces arising from the pres-
`sure drop associated with syringe barrel and entry
`losses into the needle are assumed to be negligible.
`The ffriction arises from the interaction between the
`stopper and the glass syringe barrel. For siliconized
`syringes, a silicone oil layer serves as lubricant be-
`tween the stopper and barrel surface. As discussed
`in a previous article,4 the ffriction is dependent on the
`syringe and stopper geometry, the level of siliconiza-
`tion, and the injection speed. In addition, the fric-
`tion can also be impacted by interactions between the
`filled product and glass barrel. As a result, in order
`to properly estimate the Ftotal using Eq. 1, it is impor-
`tant to estimate the ffriction for the specific product–
`syringe system under consideration. The ffriction for
`such a system (syringe barrel wetted with product)
`can be estimated by removing the needle from the
`syringe (making the hydrodynamic component negli-
`gible) and measuring the total extrusion force.
`The Fhydrodynamic results from the pressure drop
`required to drive the fluid out of the syringe. For
`Newtonian fluids, the relationship between the pres-
`sure drop P required to drive the fluid at flow rate
`Q (units: volume/time) through a cylindrical tube is
`given by the Hagen–Poiseuille equation
` P = 8:LQ
`Br4
`
`(2)
`
`,
`
`where : is the viscosity of fluid, r is the radius of the
`tube, and L is the length of the cylindrical tube.5 The
`equation assumes laminar flow of an incompressible
`liquid through a channel of constant cross-section di-
`ameter of 2r.
`The hydrodynamic component of extrusion force
`has previously been estimated4 as a function of so-
`lution properties, syringe dimensions, and injection
`
`DOI 10.1002/jps
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 101, NO. 12, DECEMBER 2012
`
`Novartis Exhibit 2020.002
`Regeneron v. Novartis, IPR2020-01318
`
`

`

`4474
`
`RATHORE ET AL.
`
`Newtonian: Fhydrodynamic =
`
`speed for both Newtonian and non-Newtonian fluids:
`
`(cid:3)
`
`(cid:2)
`
`8B:Lnr4
`b
`r4
`n
`
`viscosity of the fluid as:
`
`¯v
`
`(3)
`
`: = Jw
`(w
`
`= (Ftotal − f friction)
`
`(cid:2)
`
`(cid:3)
`
`r4
`n
`8BLnr4
`b ¯v
`
`.
`
`(7)
`
`This equation allows the Instron 5564 to be used
`as a microcapillary rheometer to estimate viscosity
`for Newtonian fluids based on injection force and di-
`mensions of syringe components.
`
`Non-NewtonianFluids
`The shear stress for a non-Newtonian fluid can be
`calculated in an analogous fashion:
`Ftotal − f friction
`Br2
`b
`
`(cid:2)
`
`(cid:3)
`
`Jw = P rn
`2Ln
`
`=
`
`rn
`2Ln
`
`.
`
`(8)
`
`The expression for the shear rate is:
`
`,
`
`(9)
`
`(app = 4Q
`Br3
`n
`where (app is known as the apparent shear rate. The
`wall shear rate, (w, is related to the apparent shear
`rate as8:
`(w =
`
`(cid:4)
`
`(cid:5)
`
`3n+ 1
`4n
`
`=
`
`4Q
`Br3
`n
`
`(cid:4)
`
`(cid:5)
`
`3n+ 1
`4n
`
`(app.
`
`(10)
`
`The calculation of viscosity of non-Newtonian fluid
`involves estimation of two parameters: K and n. Equa-
`tions 8–10 can be used to calculate the apparent vis-
`cosity as
`
`(cid:6)
`
`(cid:4)
`
`: app = Jw/(app =
`
`K
`
`(cid:5)n(cid:7)(cid:8)
`
`3n+ 1
`4n
`
`(cid:9)n−1
`
`(app
`
`.
`
`(11)
`
`b
`
`¯vn
`
`Non − Newtonian: Fhydrodynamic
`(cid:4)
`(cid:5)n 2BKLnr2n+2
`3n+ 1
`=
`r3n+1
`n
`where n is the power-law index (n = 1 representing
`the Newtonian case), K is defined as the flow consis-
`tency index, : is the viscosity, rb is the radius of the
`syringe barrel, rn is the radius of the needle, Ln is
`the length of the needle, and ¯v is the linear speed of
`the stopper.
`
`(4)
`
`Estimation of Product Viscosity from Injection Force
`Data
`Equations 3 and 4 can be used to estimate the
`extrusion force associated with a syringe injection,
`given the knowledge of product rheological behavior.6
`Product viscosity measured at
`low shear rates
`(100–1000 s−1) using common laboratory rheometers
`may not be completely representative of the rheolog-
`ical behavior under the high shear rates associated
`with commonly used injection times (100,000 s−1).7
`The following sections present the theoretical frame-
`work used to characterize product viscosity under
`high shear rates. Standards of known viscosity have
`been used for calibration.
`
`NewtonianFluids
`If the pressure drop across the syringe is known, then
`Eq. 3 can be used to calculate the rheological prop-
`erties of a Newtonian fluid. The shear stress for the
`syringe system can be expressed as:
`
`(cid:2)
`
`Jw = Prn
`2Ln
`
`=
`
`Ftotal − f friction
`Br2
`b
`
`(cid:3)
`
`rn
`2Ln
`
`,
`
`(5)
`
`where Jw is the shear stress at the wall or barrel sur-
`face and P is the pressure drop primarily governed
`by the flow through the syringe needle.5 The shear
`rate is given by:
`
`(w = 4Q
`Br3
`n
`
`= 4¯vr2
`r3
`n
`
`b
`
`,
`
`(6)
`
`where (w is the shear rate at the wall.8 Equations 5
`and 6 can be combined to give the relation for the
`
`where : app is the apparent viscosity. The power-law
`index and K can then be calculated by fitting : app
`versus (app. The true viscosity and apparent viscosity
`are related as:
`
`(cid:4)
`
`(cid:5)
`
`: true =
`
`4n
`3n+ 1
`where : true is the true viscosity.
`
`: app,
`
`(12)
`
`RESULTS AND DISCUSSION
`Complete understanding of the injectability of a drug
`product requires knowledge of the product–syringe
`interactions (and their impact) as well as the prod-
`uct rheology. Small variations in the product concen-
`tration or operating conditions such as temperature
`could significantly impact the product viscosity and
`the resultant extrusion forces. The following sections
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 101, NO. 12, DECEMBER 2012
`
`DOI 10.1002/jps
`
`Novartis Exhibit 2020.003
`Regeneron v. Novartis, IPR2020-01318
`
`

`

`CHARACTERIZATION OF PROTEIN RHEOLOGY AND DELIVERY FORCES FOR COMBINATION PRODUCTS
`
`4475
`
`Figure 2. Concentration dependence of the viscosity for
`model antibody formulation. Note that the viscosity in-
`creases exponentially with protein concentration.
`
`Figure 3. Temperature dependence of the viscosity of two
`different formulations of Product A. Note that the viscosity
`increases exponentially as temperature decreases.
`
`injection time specifications should take into account
`the allowed operating temperature range and the as-
`sociated impact on product viscosity.
`A statistical analysis was performed to explore any
`cross-interaction between temperature and concen-
`tration of an antibody product. A leverage plot anal-
`ysis using JMP statistical software (SAS Institute,
`Cary, North Carolina) showed that a cross-interaction
`term is indeed required to adequately capture the vis-
`cosity dependence on product concentration and tem-
`perature. The following equation was obtained after
`performing linear regression between ln(:), 1/T, C,
`and a C/T interaction term:
`
`(13)
`
`ln : = −12.103 + 0.0189C + 3297
`(cid:4)
`(cid:5)
`T
`+ 26.7(C − 132.3)
`− 0.00352
`
`1 T
`
`where C is the concentration of product in mg/mL, T
`is the temperature in Kelvin, and : is the predicted
`viscosity in centipoise. This model can be used to rep-
`resent product viscosity over a wide temperature and
`concentration space, as shown in Figure 4.
`
`Product–Syringe Interaction and Its Impact on ffriction
`As described earlier, the total delivery force is a sum
`of the ffriction and the extrusion force needed to over-
`come the pressure drop across the syringe. Friction
`force in a syringe depends on the nature of the in-
`ner surface of barrel (e.g., lubrication with silicone
`oil) and the stopper properties. Friction force would
`be independent of the properties of filled product, pro-
`vided the formulation does not interfere with the lu-
`bricating medium. However, in most practical cases,
`
`present the results from characterization studies con-
`ducted to understand the friction forces and the prod-
`uct rheological behavior under high-shear conditions.
`
`Product Viscosity Characterization Over A Temperature
`and Concentration Range
`Viscosity is an important factor in the development of
`high-concentration therapeutics because of its strong
`dependence on protein concentration and tempera-
`ture. This dependency is the result of macromolecular
`crowding (non-product specific) as well as product-
`specific intermolecular interactions.1,7 It is common
`for biotechnology products to have an allowed tol-
`erance of about 10% around the target protein con-
`centration. Such tolerances are usually set based on
`the consistency of the manufacturing process as well
`as the variability associated with protein concentra-
`tion measurement. Small variations (≤10%) in pro-
`tein concentrations may not contribute to significant
`changes in product viscosity if the target concentra-
`tion is low. However, for high-concentration products,
`such variations could result in a significant increase
`in product viscosity. Figure 2 shows that the solu-
`tion viscosity of a model high-concentration antibody
`(Product A) increases exponentially as the protein
`concentration is increased beyond 100 mg/mL, high-
`lighting the importance of accurately characterizing
`the viscosity over a wide concentration range.
`Viscosity dependence on product temperature is
`equally critical, and as such has been studied ex-
`tensively in the literature.9–11 A robust device design
`should ensure functionality not only at a specific tem-
`perature, but also over a broad temperature range as
`defined by the end-user requirements. Figure 3 shows
`the exponential temperature dependence of the vis-
`cosity of two different formulations of Product A. Any
`
`DOI 10.1002/jps
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 101, NO. 12, DECEMBER 2012
`
`Novartis Exhibit 2020.004
`Regeneron v. Novartis, IPR2020-01318
`
`

`

`4476
`
`RATHORE ET AL.
`
`Figure 4. Product A viscosity as a function of temperature
`and concentration. The highlighted box indicates the range
`of interest for this product (c0 ± 0.1c0 mg/mL, 5–25◦C) based
`on intended usage.
`
`the product formulation can interact with the lubri-
`cated glass barrel and modify the ffriction. Figure 5
`provides a qualitative assessment of the impact of dif-
`ferent formulations on the ffriction for different syringe
`types (glass and plastic). The friction forces in glass
`syringes for water and buffer without Polysorbate-20
`(PS-20, a commonly used surfactant) are similar to
`that of an empty syringe, indicating that these so-
`lutions do not interact significantly with the lubricat-
`ing silicone oil layer on the glass barrel. The extrusion
`force profiles for the PS-20 free buffer samples showed
`reasonable consistency with a coefficient of variation
`(CV) of about 15%. However, buffer samples contain-
`ing PS-20 showed significant variability in extrusion
`force profiles (CV of 40%). Measured friction forces in
`glass syringes for all buffer samples with PS-20 were
`higher than corresponding PS-20 free buffer samples,
`suggesting potential interactions between the surfac-
`
`tant formulation and the lubricated glass barrels. The
`behavior is different for plastic syringes, wherein the
`ffriction for drug product without PS-20 is significantly
`higher than that for buffer and product with polysor-
`bate.
`Surfactants such as PS-20 can leach silicone oil and
`also serve as potential lubricant themselves, although
`silicone oil is a better lubricating agent than PS-20.
`Silicone oil leaching likely explains the observation
`that the presence of PS-20 increases the ffriction in
`glass syringes. Conversely, plastic prefilled syringes
`do not have any silicone-oil-driven lubrication, and
`thus PS-20 may serve as a lubricating agent to re-
`duce the ffriction. Protein present in the active formu-
`lation can also increase the surface activity of the
`solution, resulting in the leaching of silicone oil from
`glass syringes. Depending on product type, the pro-
`tein molecules could also bind to the surface of the
`syringe barrel, particularly in poorly siliconized ar-
`eas, and impact the necessary injection force. Com-
`plete understanding of protein–surface interactions
`is essential to estimate their impact on the friction
`forces over the duration of the shelf life of injection
`devices. Given the complexity and variability in these
`measurements, a quantitative assessment relying on
`a larger number of samples is needed to more accu-
`rately estimate the impact on friction forces.
`
`Rheological Characterization of Products at High Shear
`Rates
`As described earlier, the shear rates associated with a
`syringe delivery are typically much higher than shear
`rates employed by commonly used rheometers. For
`example, delivering a 1 mL volume through a 27 G
`needle over an injection time of 6 s would result in
`a shear rate of about 180,000 s−1, whereas viscosity
`measurements using a double-gap rheometer are typ-
`ically performed at shear rates under 1000 s−1.
`
`Figure 5. The effect of syringe material on the friction force for different injected liquids.
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 101, NO. 12, DECEMBER 2012
`
`DOI 10.1002/jps
`
`Novartis Exhibit 2020.005
`Regeneron v. Novartis, IPR2020-01318
`
`

`

`CHARACTERIZATION OF PROTEIN RHEOLOGY AND DELIVERY FORCES FOR COMBINATION PRODUCTS
`
`4477
`
`Table 1. Comparison of Viscosities Measured by Rheometer and Instron 5564 for Brookfield Standards
`
`Viscosity (cP)
`
`Standard Solution
`
`Brookfield 5 cP
`Brookfield 10 cP
`Brookfield 50 cP
`
`Temperature (◦C)
`20
`20
`20
`
`Double-Gap Rheometer
`(100 s−1)
`5.4
`9.6
`51.5
`
`Based on Instron 5564 Force
`Measurement (1500–300,000 s−1)
`5.7
`9.7
`50.0
`
`rate range of 1500–450,000 s−1. Figure 6 shows the
`shear dependence of viscosity as measured using the
`Instron 5564. The plot shows that all Brookfield stan-
`dards exhibit Newtonian behavior, as the viscosities
`are independent of the shear rate.
`several high-
`of
`The
`rheological properties
`concentration products were then measured using the
`Instron 5564 as a rheometer. Measurements were
`conducted over the above listed range of shear at
`both 20◦C and 5◦C. Results are provided in Figure 7;
`the symbols represent the viscosity as calculated from
`the extrusion forces measured on the Instron 5564,
`and the lines depict the corresponding exponential
`
`Product A 1.0c
`0
`Product A 1.6c
`0
`Product B 0.9c
`0
`Product C 0.9c
`0
`Product D 1.1c
`0
`
`Product rheology at 20 °C
`
`
`
`μ ∝ γ−0.16
`
`μ ∝ γ−0.15
`
`μ ∝ γ−0.20
`
`(a)
`
`102
`
`101
`
`Viscosity (cP)
`
`
`
`104
`
`105
`Shear rate (s−1)
`
`Product rheology at 5 °C
`
`106
`
`
`
`(b)
`
`μ ∝ γ−0.34
`
`Product A 1.0c
`0
`Product A 1.6c
`0
`Product B 0.9c
`0
`Product C 0.9c
`0
`
`μ ∝ γ−0.39
`
`μ ∝ γ−0.08
`
`μ ∝ γ−0.07
`
`102
`
`101
`
`Viscosity (cP)
`
`In light of this limitation, a new method utiliz-
`ing the Instron 5564 as a microcapillary rheometer
`was developed to measure the rheological properties
`of products at high shear rates. Three Brookfield
`standards of different viscosities (5, 10, and 50 cP)
`were used as standard Newtonian solutions. Mea-
`surements were performed using glass syringes from
`a single lot and vendor to minimize variability. An
`estimate of the needle internal radius was obtained
`by measuring the delivery forces for these standard
`solutions at a fixed plunger speed (205 mm/min) and
`using Eq. 7 to calculate rn. The optimum radius was
`estimated in an iterative manner by minimizing the
`sum of errors in the calculated viscosities for the three
`standard solutions. The optimum needle diameter for
`the 27 G syringes used in this study was found to be
`0.2217 mm, which lies within the ISO range for 27 G
`needles (0.191–0.229 mm). Table 1 lists the results
`comparing the reference viscosities (as measured us-
`ing a double-gap rheometer at a 100 s−1 shear rate)
`to the estimates from Instron 5564 using a needle
`diameter of 0.2217 mm.
`Once the method had been developed and the sy-
`ringe dimensions were calculated, the Instron 5564
`could be used for characterizing sample viscosity over
`a wide range of shear rates. Injection velocities from
`3 to 900 mm/min were used, corresponding to a shear
`
`106
`
`Figure 6. Viscosity as a function of shear rate for Newto-
`nian (Brookfield) solutions. Measurements were carried out
`at 20◦C.
`
`DOI 10.1002/jps
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 101, NO. 12, DECEMBER 2012
`
`
`
`104
`
`105
`Shear rate (s −1)
`Figure 7. Product shear thinning at (a) 20◦C and (b) 5◦C.
`As per Eq. 11, the power-law index (n) of each product is
`the listed viscosity versus shear rate exponent plus one; for
`example, the n for Product B is 0.80 at 20◦C and 0.61 at
`5◦C. The extrusion forces for these samples were collected
`using Instron 5564 and had a relative error in the 5%–10%
`range.
`
`Novartis Exhibit 2020.006
`Regeneron v. Novartis, IPR2020-01318
`
`

`

`4478
`
`RATHORE ET AL.
`
`fits (as per Eq. 11). An important observation from
`Figure 7 is that products having high viscosity
`(typically above ∼10 cP) exhibit
`strong shear-
`thinning behavior, whereas less viscous products do
`not. This is most evident from the profiles of Prod-
`uct A: the 1.6c0 mg/mL formulation shows strong
`shear thinning, whereas the 1.0c0 mg/mL concentra-
`tion product remains Newtonian at room tempera-
`ture. Figure 7b shows that lowering the temperature
`increases both the viscosity itself and the degree
`of shear thinning. Again examining Product A,
`Figure 7b indicates that the 1.6c0 mg/mL formulation
`becomes increasingly non-Newtonian with decreasing
`temperature (n = 0.84 at 20◦C, n = 0.66 at 5◦C), and,
`notably, the 1.0c0 mg/mL formulation exhibits shear-
`thinning behavior under low-temperature conditions
`(−0.08 slope), despite being Newtonian at room tem-
`perature. These data demonstrate the importance of
`taking into account the actual viscosity of the prod-
`uct at the representative shear rates before setting
`design space parameters for the delivery system.
`It is also deduced that the strength of shear thin-
`ning, as assessed by the power-law index n (recall n =
`1 for a Newtonian fluid), is dependent on viscosity of
`the product: the higher the viscosity, the greater the
`degree of shear thinning. This observation is shown
`in Figure 8, wherein the power-law index n is plot-
`ted against the product viscosity. The viscosity here,
`: 0, is the measured value of the viscosity from the
`double-gap rheometer at a low shear rate (100 s−1).
`Figure 8 demonstrates the strong correlation between
`product viscosity and non-Newtonian behavior, al-
`though available data were limited above 40 cP. As the
`viscosity of a given product decreases (at high tem-
`perature, low concentration, or varying buffer formu-
`lation, as per Figs. 2–4), we can anticipate that the
`
`Figure 8. Power-law index versus viscosity for all prod-
`ucts. The power-law index was obtained as explained in
`the text [results as ((n−1) in Fig. 7], and : 0 is the viscosity
`obtained using a standard rheometer.
`
`solution will behave in a more Newtonian manner.
`More importantly, this phenomenon allows for the
`anticipation of shear-thinning effects at high shear
`rate from readily available laboratory rheometer
`data. For high-viscosity products, the impact on de-
`vice design parameters would be significant and it
`would be prudent to characterize the shear-thinning
`behavior in detail.
`A detailed design space characterization of prod-
`uct rheology and injection forces is presented in the
`following section.
`
`Design Space Characterization for Syringe Delivery
`In the previous sections, we have discussed the im-
`pact of independently varying the product concentra-
`tion, temperature, and shear rate on product viscosity.
`However, the product viscosity is not a performance
`parameter for the final device design; rather, injec-
`tion time and the force required to provide it are
`the set points of interest. Design space parameters
`include temperature, product concentration, device
`attributes, and the plunger injection speed.
`The plunger speed and syringe barrel dimensions
`will dictate the injection time t0 directly (i.e., the vol-
`ume injected over the constant volumetric flow rate).
`The plunger speed and device dimensions also dictate
`the resultant Fhydrodynamic, both directly (the ¯v1 term
`in Eqs. 3 and 4, as well as via the ¯v dependence of
`4) and indirectly for non-Newtonian fluids via
`Ffriction
`the dependence of viscosity on the shear rate, which
`is based on ¯v (Eq. 6). The following discussion only
`considers the hydrodynamic portion of the force re-
`quired for injection, and assumes that this force can
`be consistently provided during injection.
`Figure 9 displays the force profile as a function of
`injection time and needle radius for a hypothetical
`1 mL injection; Figure 9a displays the profile for a
`20 cP Newtonian fluid, whereas Figure 9b displays
`the profile for a 20 cP power-law fluid that shear
`thins with a power-law fluid coefficient of 0.95. As per
`tinj = Br2
`material balance considerations (Q = Vinj
`b ¯v)
`and Eq. 9, longer injection time corresponds directly
`to lower velocity and shear rate, thereby explaining
`the greatly reduced force requirement at long injec-
`tion time. Larger needle radius also reduces the re-
`quired Fhydrodynamic at a given plunger speed, due to
`the lowered shear of the product through the nee-
`dle (Eqs. 3 and 4). As demonstrated by comparing
`panels a and b in Figure 9, shear thinning of pro-
`tein solution during injection also results in a reduc-
`tion of the Fhydrodynamic, due to the decreased viscosity
`at the high shear rate through the needle. Such de-
`pendence of injection force or time on shear rate will
`be further exaggerated for higher viscosity products
`with power-law coefficients significantly lower than
`one. This highlights the importance of characteriz-
`ing the shear-thinning properties of the product, so
`
`(cid:10)
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 101, NO. 12, DECEMBER 2012
`
`DOI 10.1002/jps
`
`Novartis Exhibit 2020.007
`Regeneron v. Novartis, IPR2020-01318
`
`

`

`CHARACTERIZATION OF PROTEIN RHEOLOGY AND DELIVERY FORCES FOR COMBINATION PRODUCTS
`
`4479
`
`Figure 9. The force (Fhydrodynamic) required to obtain var-
`ious injection times for a 1 mL solution through various
`needle radii for (a) Newtonian and (b) non-Newtonian so-
`lutions. A 20 cP solution viscosity with a power-law coeffi-
`cient of 0.95 was assumed for the non-Newtonian example
`(as per the shear-thinning trend with viscosity displayed
`in Figure 8). Note that the non-Newtonian fluid requires
`less force than the Newtonian because of shear thinning,
`particularly under high shear.
`
`that the force design considerations are accurate. As-
`suming that the product is Newtonian (i.e., that the
`low-shear viscosity measured in a viscometer accu-
`rately estimates the viscosity during needle injection)
`might result in a significant overestimation of the
`force required to achieve a desired injection time. In
`addition, the use of a mimic solution that does not
`follow the same rheological behavior as the drug of
`interest may also result in suboptimal design of the
`delivery system.
`Figure 10a demonstrates the effect of variations in
`temperature and product concentration on the force
`required to inject 1 mL of Product A in 6 s, whereas
`Figure 10b demonstrates their effects on the injection
`time when using a constant delivery force of 10 N.
`
`Figure 10. The effect of product concentration and tem-
`perature on the (a) required hydrodynamic force or (b) injec-
`tion time of a 1 mL injection. The force panel (a) assumes a
`6 s injection time, whereas the time panel (b) assumes 10 N
`provided hydrodynamic force. Product was assumed to be a
`Newtonian fluid for the purposes of this figure.
`
`Results show that combined effect of small variations
`in product concentration and temperature may re-
`sult in significant variability in the resulting injec-
`tion time or force, for example, the range of injection
`force within the relevant concentration and temper-
`ature bounds for a 6 s injection time (Fig. 10a) was
`3.2–12.3 N, whereas the injection time range with a
`constant 10 N force (Fig. 10b) was 1.9–7.4 s.
`These analyses act as useful tools to directly as-
`sess the impact of design parameters on tangible
`patient-focused outputs, such as the injection time,
`and device restrictions or requirements, such as the
`required injection force. The delivery device should be
`robust enough to function not only at target but also at
`the allowed extreme product concentrations. Also, for
`drug products recommended for storage under refrig-
`erated temperatures, it is important to either ensure
`device functionality at cold temperatures or make rec-
`ommendations to the user for adequate equilibration
`time to allow for room temperature delivery.
`
`DOI 10.1002/jps
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 101, NO. 12, DECEMBER 2012
`
`Novartis Exhibit 2020.008
`Regeneron v. Novartis, IPR2020-01318
`
`

`

`4480
`
`RATHORE ET AL.
`
`CONCLUSIONS
`This article describes the characterization work con-
`ducted to evaluate the effect of drug product rheol-
`ogy and drug product–syringe barrel interactions on
`the injection time and extrusion force for a protein–
`device combination injection product. The impacts of
`temperature, protein concentration, buffer formula-
`tion, and shear rate on viscosity and ffriction were mea-
`sured. Results show that interactions between prod-
`uct and container surface can impact the ffriction. By
`modeling the fluid dynamics of the device, the com-
`bined effect of these parameters on the injection time
`and extrusion force was predicted over a wide range
`of potential conditions. Such design space charac-
`terization provides a direct link between design pa-
`rameters (as concentration, temperature, and device
`dimensions) and patient-relevant outputs or limita-
`tions (injection time and force requirements). The re-
`sults demonstrate that high-viscosity drug products
`exhibit strong shear-thinning behavior, and this be-
`havior has a significant impact on the observed injec-
`tion times and hydrodynamic forces. Notably, if such
`products are assumed to be Newtonian, the force re-
`quired to inject them through a syringe at high ve-
`locity (low injection time) might be significantly over-
`estimated because of shear thinning. This problem
`may be exacerbated if mimic solutions are use