Effect of Excipients on the Stability and Structure of Lyophilized
`Recombinant Human Growth Hormone
`
`HENRY R. COSTANTINO,*,† KAREN G. CARRASQUILLO,‡ ROCIO A. CORDERO,‡ MARCO MUMENTHALER,†
`CHUNG C. HSU,† AND KAI GRIEBENOW‡
`
`Contribution from Pharmaceutical Research and Development, Genentech, Inc., 1 DNA Way,
`South San Francisco, California 94080 and Department of Chemistry, University of Puerto Rico, Rio Piedras Campus,
`San Juan, Puerto Rico 00931-3346.
`
`Received February 27, 1998. Accepted for publication April 9, 1998.
`
`Abstract 0 We have investigated the effect of mannitol, sorbitol,
`methyl R-D-mannopyranoside, lactose, trehalose, and cellobiose on
`the stability and structure of the pharmaceutical protein recombinant
`human growth hormone (rhGH) in the lyophilized state. All excipients
`afforded significant protection of
`the protein against aggregation,
`particularly at levels to potentially satisfy water-binding sites on the
`protein in the dried state (i.e., 131:1 excipient-to-protein molar ratio).
`At higher excipient-to-protein ratios, X-ray diffraction studies showed
`that mannitol and sorbitol were prone to crystallization and afforded
`somewhat less stabilization than at lower ratios where the excipient
`remained in the amorphous, protein-containing phase. The secondary
`structure of rhGH was determined using Fourier transform infrared
`rhGH exhibited a decrease in R-helix and
`(FTIR) spectroscopy.
`increase in (cid:226)-sheet structures upon drying. Addition of excipient
`stabilized the secondary structure upon lyophilization to a varying extent
`depending on the formulation. Samples with a significant degree of
`structural conservation, as indicated by the R-helix content, generally
`In addition, prevention of protein-
`exhibited reduced aggregation.
`protein interactions (indicated by reduced (cid:226)-sheet formation) also
`tended to result in lower rates of aggregation. Therefore, in addition
`to preserving the protein structure, bulk additives that do not crystallize
`easily and remain amorphous in the solid state can be used to increase
`protein-protein distance and thus prevent aggregation.
`
`Introduction
`In the development of a lyophilized pharmaceutical
`protein, sugars (saccharides and polyols) are often added
`to the formulation in order to improve stability and
`increase the shelf life.1 Some specific examples of FDA-
`approved lyophilized protein formulations include sucrose
`in various human immunoglobulins and toxoid vaccines,
`lactose in glucagon and haemophilus b conjugate vaccine,
`and mannitol in urokinase and recombinant human growth
`hormone.2 Despite the success shown by such examples,
`the mechanism regarding how sugars protect lyophilized
`proteins is still not fully understood.
`Several factors are likely to play a role. For example, it
`has been shown that the secondary structure of some
`proteins may be altered upon lyophilization, and that
`sugars can help preserve the native conformation.3,4 It is
`postulated that the hydroxyl groups in sugars form hydro-
`gen-bonds with polar groups in proteins in the solid state,5
`substituting for water molecules which play a role in the
`structure. In this fashion, sugar molecules may “replace”
`
`* Corresponding author: Phone: 650-225-4710; Fax 650-225-3191.
`† Genentech, Inc.
`‡ University of Puerto Rico.
`
`water molecules in the solid state.5 Others have proposed
`alternative views, for example in the case of the lyopro-
`tectant trehalose.6 Recent moisture sorption studies have
`revealed that sugars can indeed interact with proteins in
`such a fashion as to satisfy protein water-binding sites.7,8
`Several mechanisms can cause a protein to undergo
`aggregation in the solid state. For various proteins it has
`been established that dehydration-induced structural al-
`terations exposing reactive groups (in particular disulfide
`bonds)9,10 is the initial step leading to covalent protein
`aggregation.11 Such structural alterations are also ex-
`pected to promote noncovalent aggregation of proteins. It
`follows that structural conservation will improve solid
`protein stability if the mechanism of deterioration is
`conformation-dependent. But even when the degradation
`occurs without conformational change, a sugar excipient
`may still provide stability for lyophilized proteins. For
`example, the therapeutic protein recombinant humanized
`monoclonal immunogloulin G, which largely retains its
`native secondary structure upon spray drying, is stabilized
`against solid-state aggregation by the addition of lactose.12
`An alternate view to explain this is that sugar excipients
`can serve to “dilute” proteins in the solid state, decreasing
`protein-protein contacts and preventing intermolecular
`degradation reactions such as aggregation.9,13 Yet another
`conception is that sugar excipients provide a glassy matrix
`wherein protein mobility and hence reactivity are mini-
`mized.14,15 In all of these views of the mechanism of solid-
`state stabilization, it is critical that the sugar remains in
`the amorphous, protein-containing phase. Various envi-
`ronmental factors, such as increased temperature and
`moisture, can induce sugar crystallization.8,16,17
`In the present study we have investigated the effect of
`various sugar excipients on the solid-state stability of a
`model pharmaceutical protein, recombinant human growth
`hormone (rhGH). Growth hormone, or somatotropin, is
`susceptible to various deterioration pathways in the solid
`state, predominantly aggregation.18,19 This protein is an
`FDA-approved drug for the long-term treatment of children
`with growth failure, currently available in both liquid and
`lyophilized forms containing mannitol.2
`In the present
`investigation, we have examined the protective effect of
`various excipients on aggregate formation during incuba-
`tion at an elevated storage temperature.
`
`Materials and Methods
`ProteinsRecombinant human growth hormone (rhGH) was
`produced at Genentech, Inc. (South San Francisco). The protein
`bulk containing 2 mg/mL rhGH, 88 mM mannitol, and 5 mM
`sodium phosphate, pH 7.8, was buffer-exchanged into a 100 mM
`ammonium bicarbonate solution and lyophilized to form excipient-
`free protein.20 Samples were lyophilized in a Leybold (Germany)
`
`1412 / Journal of Pharmaceutical Sciences
`Vol. 87, No. 11, November 1998
`
`10.1021/js980069t CCC: $15.00
`Published on Web 08/13/1998
`
`© 1998, American Chemical Society and
`American Pharmaceutical Association
`
`MYLAN INST. EXHIBIT 1093 PAGE 1
`
`MYLAN INST. EXHIBIT 1093 PAGE 1
`
`

`

`model GT20 unit at a chamber pressure of 150 (cid:237)mHg and a shelf
`temperature of -35 °C for 48 h (primary drying) followed by a
`shelf temperature of 30 °C for about 10 h (secondary drying). All
`dried samples were stoppered under dry N2 when the vacuum
`pressure was <127 mmHg. The lyophilized material was sealed
`in glass vials and stored at 2-8 °C until use. The lyophilization
`and cold-temperature storage did not adversely affect protein
`quality, particularly with regard to aggregation and clipping.
`ExcipientssMannitol, sorbitol, methyl R-D-mannopyranoside,
`lactose, trehalose, and cellobiose, all analytical grade, were
`obtained from Sigma Chemical Co. (St. Louis, MO) and were used
`as supplied.
`Ratios of Excipients to rhGH in Lyophilized Sampless
`Lyophilized samples containing different excipient-to-protein ratios
`were prepared by adding an appropriate amount of concentrated
`excipient solution to the excipient-free protein solution prior to
`lyophilization. The ratios selected were based on the molar
`amount of various strongly and weakly water-binding sites in the
`rhGH molecule, as described elsewhere.21 The molar ratios chosen
`for our study were 31:1 (lyoprotectant-to-rhGH; representing 50%
`of strongly water-binding sites present in the rhGH molecule),
`131:1 (100% of strongly and weakly water-binding sites), 300:1
`and 1000:1 (excess beyond the total of all water-binding sites).
`Preparation of Lyophilized Excipient/rhGH Sampless
`Lyophilized excipient-free rhGH was reconstituted with deionized
`water to form a stock solution containing 20 mg/mL protein.
`Protein concentration was confirmed by UV absorption at 278 nm.
`Concentrated excipient stock solutions were also prepared and
`combined with rhGH stock solutions to obtain the precise excipi-
`ent-to-rhGH ratio desired. Solutions were filtered (0.22 (cid:237)m), and
`the protein concentration was again confirmed by UV absorption.
`Aliquots of 4 mL of the excipient-to-protein solution were filled
`into 10-cc glass vials and lyophilized as described above for
`excipient-free protein. The sample of 1000:1 methyl R-D-manno-
`pyranoside:rhGH exhibited “collapse” upon lyophilization and was
`not assayed for solid-state stability.
`Residual Moisture ContentsThe residual moisture content
`was assayed by the Karl Fischer titration method.21 All samples
`were found to contain a residual moisture of approximately 2-3%
`(w/w).
`Glass Transition TemperaturesThe glass transition (Tg) for
`the excipients was measured by differential scanning calorimetry
`(DSC) using a Perkin-Elmer Model 7 unit. Approximately 10 mg
`of each sample was loaded into an aluminum sample pan, sealed,
`and placed in the calorimeter. An empty pan was used as a
`reference. Following an equilibration of 10 min at 30 °C, samples
`with sorbitol were cooled at a rate of 20 °C/min to -20 °C, whereas
`all other samples were heated at a constant rate of 10 °C/min to
`120 °C. Tg was taken as the midpoint in the thermogram as
`measured from extensions of the pre- and posttransition baselines,
`using Perkin-Elmer software provided with the calorimeter.
`X-ray Powder Diffraction (XRD)sXRD of the lyophilized
`excipient:protein samples was conducted as described elsewhere.22
`Solid-State Stability StudiessLyophilized excipient:rhGH
`samples were stored for up to 28 days at 50 °C. At selected
`timepoints, two vials at each excipient:protein ratio were recon-
`stituted with sterile water-for-injection (WFI) to 1 mg/mL initial
`protein and assayed as follows. The amount of insoluble ag-
`gregates was determined by measuring the protein concentration
`(UV detection at 278 nm) of the reconstituted sample after
`centrifugation (3000 rpm for 30 min) and filtration (0.22 (cid:237)m).
`Insoluble aggregates smaller than 0.22 (cid:237)m were not necessarily
`removed by the filtration. The amount of soluble aggregates was
`determined by size-exclusion HPLC (UV detection at 214 nm) on
`a Tosoh TSK20000SWXL column (7.8 mm i.d. (cid:2) 30 cm length,
`particle size 5 (cid:237)m). Typically, 10 (cid:237)L of each filtered, reconstituted
`sample was loaded onto the column at a flow rate of 0.5 mL/min.
`The mobile phase consisted of 150 mM sodium chloride, 50 mM
`sodium phosphate, pH 7.2.
`FTIR SpectroscopysFTIR studies were conducted with a
`Nicolet Magna-IR System 560 optical bench as described previ-
`ously.4 A total of 256 scans at 2 cm-1 resolution using Happ-
`Ganzel apodization were averaged to obtain each spectrum. For
`all experiments involving aqueous solutions, a Spectra Tech liquid
`cell equipped with CaF2 windows and 15-(cid:237)m thick spacers was
`used. Lyophilized protein powders were measured as KBr pellets
`(1 mg of protein per 200 mg of KBr).4,23,24 Each protein sample
`was measured at least five times. When necessary, spectra were
`
`Table 1sInfrared Band Positions, Areas, and Assignments in the
`Amide I Region for Various Formulations of rhGHa
`band position (cm-1)
`
`sample
`
`SDb
`
`lyophilized (excipient-free)
`
`aqueous solution (excipient-free) 1686
`1682
`1670
`1655
`1639
`1633
`1692
`1682
`1672
`1655
`1639
`1631
`1690
`1683
`1675
`1654
`1637
`1629
`1690
`1681
`1671
`1656
`1640
`1631
`
`co-lyophilized with 131:1
`mannitol:protein
`
`co-lyophilized with 131:1
`lactose:protein
`
`area
`Gaussian
`assignment
`(%)
`curve-fittingc
`6 – 1 unorderedd
`1687 – 1
`10 – 1 unordered
`1678 – 1
`6 – 1 unordered
`1670 – 1
`57 – 3 R-helix
`1655 – 1
`14 – 2 unordered
`1640 – 1
`7 – 3 (cid:226)-sheet
`1634 – 1
`14 – 1 (cid:226)-sheet
`1696 – 2
`15 – 1 unordered
`1683 – 3
`20 – 1 unordered
`1670 – 1
`29 – 3 R-helix
`1655 – 1
`12 – 2 unordered
`1640 – 1
`10 – 2 (cid:226)-sheet
`1629 – 1
`7 – 3 (cid:226)-sheet
`1694 – 2
`18 – 3 unordered
`1682 – 2
`15 – 2 unordered
`1670 – 1
`39 – 3 R-helix
`1655 – 0
`12 – 1 unordered
`1640 – 1
`9 – 1 (cid:226)-sheet
`1630 – 0
`8 – 3 (cid:226)-sheet
`1691 – 3
`12 – 2 unordered
`1682 – 1
`16 – 1 unordered
`1671 – 1
`48 – 2 R-helix
`1656 – 0
`5 – 1 unordered
`1641 – 1
`11 – 2 (cid:226)-sheet
`1632 – 1
`a Data are the average and standard deviation of four to five independent
`determinations. b Second derivative. c Gaussian curve-fitting was performed
`of Fourier self-deconvoluted amide I spectra. d Unordered structures include
`random coil, extended chains and turns.
`
`corrected for the solvent background in an interactive manner
`using the Nicolet OMNIC 3.1 software4,23,24 to obtain the protein
`vibrational spectra. We have confirmed that this procedure is
`reliable for water background subtraction when using 15 (cid:237)m thick
`spacers.23 Prior to further analysis, small water vapor bands
`present were eliminated from the spectra.
`FTIR Data AnalysissSecond DerivatizationsAll spectra were
`analyzed by second derivatization in the amide I and amide III
`regions for their component composition.4,23-26 Second derivative
`spectra were smoothed with an 11-point smoothing function (10.6
`cm-1).4
`Fourier Self-Deconvolution (FSD)sFSD27-29 was applied to the
`unsmoothed spectra to enable quantification of the secondary
`structure in the amide I region by Gaussian curve-fitting24,30 using
`the program OMNIC 3.1. The parameters chosen, a value of 24
`for the full width at half-maximum (fwhm) and k ) 2.4 for the
`enhancement factor, are in the range of those published.30-33 Note
`that FSD alters the band shapes, but preserves the integrated
`band intensities when over-deconvolution is avoided.27,31 The
`values chosen for FSD in our analyses were checked for the risk
`of such over-deconvolution (which could result in distorted band
`areas)28,30 by employing the strategy outlined by Griebenow and
`Klibanov.23
`Gaussian Curve-FittingsThe frequencies of the band centers
`found in the second derivative spectra in the amide I region were
`used as starting parameters for the Gaussian curve-fitting (per-
`formed using the program GRAMS/386 from Galactic Industries,
`Inc.). The secondary structure contents were calculated from the
`areas of the individual assigned bands and their fraction of the
`total area in the amide I region.23,24,30 Gaussian curve-fitting was
`performed in the amide I region after band-narrowing of the
`protein vibrational spectra by FSD.30,33
`In all cases, a linear
`baseline was fitted in addition to the Gaussian bands. In most
`cases, the discrepancies between component frequencies obtained
`by second derivatization and the Gaussian curve-fitting were below
`4 cm-1 (Table 1). The secondary structure content is reported as
`the average and standard deviation of the value calculated for at
`least four independently obtained spectra.
`Band AssignmentssThe band assignment in the amide I region
`followed those in the literature and is summarized for some typical
`
`Journal of Pharmaceutical Sciences / 1413
`Vol. 87, No. 11, November 1998
`
`MYLAN INST. EXHIBIT 1093 PAGE 2
`
`MYLAN INST. EXHIBIT 1093 PAGE 2
`
`

`

`Figure 1sFTIR spectra of rhGH (A) in aqueous solution at pH 7.8 and (B)
`the lyophilized powder. The solid lines represent the superimposed FSD and
`the curve-fit, and the dashed curves represent the individual Gaussian bands.
`
`samples in Table 1.23,24,33,34 Shown are the results for the
`excipient-free aqueous and lyophilized protein, in addition to
`lyophilized formulations containing a molar ratio of excipient-to-
`protein of 131:1 lactose and mannitol, representing samples with
`the greatest and least degree of structural conservation, respec-
`tively. For the aqueous solution, the main band at 1655 cm-1 was
`assigned to R-helices (Figure 1A) and a band at 1634 cm-1 to
`(cid:226)-sheets. All other bands were assigned to unordered structural
`elements ((cid:226)-turns, random coil, extended chains). The secondary
`structure content determined by Gaussian curve-fitting in the
`amide I region using these assignments (57 ( 3% R-helix and 7 (
`2% (cid:226)-sheet) were the same, within the error limits, as those
`determined by others utilizing the amide III spectral region35 and
`also agrees well with the X-ray crystal structure (60% R-helix).36
`When analyzing the spectra of lyophilized rhGH (Figure 1B), a
`new band at ca. 1696 cm-1 was assigned to (cid:226)-sheets. Bands at
`such frequencies are often assigned to intermolecular (cid:226)-sheets.33,37
`Otherwise, the frequencies of the Gaussian bands found for
`lyophilized rhGH were similar as for the aqueous solution and
`assigned the same.
`
`Results and Discussion
`Solid-State Aggregation of rhGH and the Effect of
`ExcipientssIn the dried state, intermolecular pathways
`predominate as the main degradation mode for the growth
`hormone molecule.18 Thus, to assess the stability of
`lyophilized rhGH, we monitored the solid-state formation
`of soluble and insoluble aggregates. Samples of rhGH were
`co-lyophilized with various excipients at excipient:protein
`(mol:mol) ratios of 31:1, 131:1, 300:1, and 1000:1. Protein
`aggregation in the various samples was monitored follow-
`ing incubation at the accelerated storage condition of 50
`°C.
`First, we tested the stability of rhGH over time in the
`absence of any excipients. The formation of soluble ag-
`gregates was slight, increasing from about 2% upon lyo-
`philization to about 4% following four-week storage (Figure
`2A). The formation of insoluble aggregates was much more
`dramatic; whereas virtually no insoluble aggregates were
`formed upon lyophilization, more than half of the protein
`had formed insoluble aggregates after the four-week incu-
`bation (Figure 2B). Aggregate formation in human growth
`hormone is detrimental since it may lead to reduced
`bioactivity38 and increased immunogenicity.39
`Next, we tested various excipients for their potency in
`stabilizing rhGH against solid-state aggregation. Figure
`
`1414 / Journal of Pharmaceutical Sciences
`Vol. 87, No. 11, November 1998
`
`Figure 2sSolid-state stability of excipient-free rhGH. (A) Formation of soluble
`aggregates. (B) Formation of insoluble aggregates. (C) Loss of monomeric
`rhGH modeled as a pseudo first-order deterioration (calculated rate constant
`of 4.5 – 0.1 day-1).
`
`Figure 3sSoluble aggregate formation of rhGH co-lyophilized with (A) mannitol,
`(B) sorbitol, and (C) methyl R-D-mannopyranoside.
`Insoluble aggregate
`formation of rhGH co-lyophilized with (D) mannitol, (E) sorbitol, and (F) methyl
`R-D-mannopyranoside. Ratios of excipient-to-protein (mol:mol) were 31:1 (b),
`131:1 (9), 300:1 (2) and 1000:1 (1).
`3 shows both soluble and insoluble aggregation data for
`rhGH co-lyophilized with varying amounts of the straight-
`chain polyols mannitol and sorbitol and the more hydro-
`phobic sugar, methyl R-D-mannopyranoside. No clear
`conclusions can be drawn regarding these excipients’ ability
`
`MYLAN INST. EXHIBIT 1093 PAGE 3
`
`MYLAN INST. EXHIBIT 1093 PAGE 3
`
`

`

`Table 2sFormation of Non-Native Monomeric rhGH in Various
`Lyophilized Formulationsa
`
`non-native monomer (%)b
`
`1.2 – 0.1
`
`4 weeks
`
`1.9 – 0.2
`1.5 – 0.2
`
`14.7 – 0.5
`16.7 – 0.3
`22.3 – 0.3
`
`excipient:rhGH
`(mol:mol)
`
`31:1
`131:1
`300:1
`
`31:1
`131:1
`300:1
`1000:1d
`
`31:1
`131:1
`300:1
`1000:1
`
`3 weeks
`2 weeks
`1 week
`methyl R-D-mannopyranoside:rhGH
`- - -
`c
`1.0 – 0.1
`1.4 – 0.1
`1.5 – 0.1
`1.2 – 0.1
`lactose:rhGH
`- - - -
`11.8 – 0.1
`13.5 – 0.6
`13.6 – 0.7
`15.7 – 0.3
`15.8 – 0.3
`19.3 – 0.5
`12.4 – 0.1
`cellobiose:rhGH
`26.2 – 0.6
`25.0 – 0.8
`19.8 – 1.3
`17.8 – 0.7
`16.5 – 0.3
`12.9 – 1.0
`19.8 – 1.3
`15.4 – 0.4
`12.9 – 0.3
`18.6 – 0.4
`16.3 – 0.3
`14.4 – 0.4
`5.6 – 0.3
`a Species evidenced by HPLC with a retention near the native monomer,
`e.g., shoulder in the main peak. This species was not seen in rhGH co-
`lyophilized with the nonreducing sugar trehalose, mannitol, or sorbitol, and is
`probably a result of reaction with excipient, i.e., glycosylation. b Incubation at
`c Not detected. d Formation of non-native monomer was 8.3 – 0.3%
`50 (cid:176)C.
`upon lyophilization.
`
`Figure 4sSoluble aggregate formation of rhGH co-lyophilized with (A) lactose,
`(B) trehalose, and (C) cellobiose. Insoluble aggregate formation of rhGH co-
`lyophilized with (D) lactose, (E) trehalose and (F) cellobiose. Ratios of excipient-
`to-protein (mol:mol) were 31:1 (b), 131:1 (9), 300:1 (2) and 1000:1 (1).
`
`to stabilize rhGH against soluble aggregate formation
`(Figures 3A-C); indeed, it seems that the addition of
`sorbitol may destabilize the protein (Figure 3B). However,
`the insoluble aggregate data (Figures 3D-F) reveal some
`interesting trends. For instance, at a mannitol-to-protein
`ratio of 31:1, rhGH is significantly more stable toward
`insoluble aggregate formation, with less than 10% insoluble
`aggregates formed after four-week storage, compared to the
`case of excipient-free protein (stability of rhGH:mannitol
`and excipient-free protein shown in Figure 3D and Figure
`2B, respectively). As the excipient content is increased to
`131:1 mannitol:rhGH, the stability is further improved.
`However, at the higher ratios of 300:1 and 1000:1, rhGH
`shows increased insoluble aggregation. Similarly to the
`case of mannitol, the data for sorbitol (Figure 3E) and
`methyl R-D-mannopyranoside (Figure 3F) also suggest that
`a 131:1 level of excipient-to-protein is optimal in stabilizing
`the protein against solid-state insoluble aggregate forma-
`tion. The significance of this ratio is further discussed
`below.
`Figure 4 depicts soluble and insoluble aggregate forma-
`tion for rhGH co-lyophilized with the disaccharides lactose,
`trehalose, and cellobiose. All three of these excipients show
`a dramatic protective effect against both soluble and
`insoluble aggregate formation. For example, lactose and
`trehalose afforded essentially complete stabilization to the
`protein with regards to insoluble aggregation when present
`at 131:1 excipient:rhGH and above (Figures 4D and 4E).
`Cellobiose was perhaps even more potent, with essentially
`no insoluble aggregates formed even at the lowest level
`tested, 31:1 excipient:rhGH (Figure 4F).
`It is important to note that lactose, a disaccharide of
`glucose and galactose, and cellobiose, which is comprised
`of two glucose units, contain (cid:226)1f4 linkages and are thus
`both reducing sugars. The two glucosyl units of trehalose
`are bonded via an R1f1 linkage; trehalose is not a reducing
`
`Table 3sPsuedo First-Order Rate Constants for Deterioration of
`Monomeric rhGH When Lyophilized in the Presence of Various
`Excipientsa
`
`excipient
`
`k (·10 -3 days-1) at an excipient:rhGH ratio of:
`31:1
`131:1
`300:1
`1000:1
`3.9 – 0.5
`3.8 – 0.4
`5.2 – 0.7
`9.9 – 2.0
`mannitol
`3.9 – 0.2
`3.5 – 0.3
`6.7 – 0.7
`9.2 – 0.7
`sorbitol
`methyl R-D-mannopyranoside 3.1 – 0.1
`0.9 – 0.5
`4.6 – 1.0
`4.1 – 0.6
`1.3 – 0.2
`1.1 – 0.3
`1.7 – 0.5
`lactose
`2.1 – 0.5
`0.4 – 0.2
`0 – 0.2b
`0.2 – 0.2
`trehalose
`0.3 – 0.1
`0 – 0.2
`0 – 0.1
`0 – 0.1
`cellobiose
`a The negative of the slope of the plot of ln[monomer] vs time. b In cases
`where a slight positive slope was observed, the rate constant was taken as
`zero (within the error of the fit).
`
`sugar. Reducing sugars have the potential to react with
`amino groups in proteins via the Maillard reaction.40,41
`In the case of lactose- and cellobiose-containing samples,
`we found that some 10-20% of soluble rhGH was in the
`form of an altered monomer over the four-week incubation
`at the accelerated stability condition, as distinguished by
`size-exclusion HPLC (Table 2). The presence of this species
`did not interfere with the quantitative analysis of soluble
`and insoluble aggregation. It is probable that this species
`is a glycosylated form the rhGH monomer. A very small
`amount (1-2%) of a similar species was also seen in rhGH
`co-lyophilized with methyl R-D-mannopyranoside. Al-
`though the latter is not a reducing sugar, it is possible that
`it contained a reducing sugar impurity, e.g., mannose,
`which reacted with rhGH upon solid-state storage. As
`discussed elsewhere,41 reducing sugars should be avoided
`in biopharmaceutical formulations, even though they may
`be potent stabilizing excipients.
`To more quantitatively compare the difference in stabil-
`ity of the various rhGH formulations, we analyzed the
`aggregation as a pseudo first-order process with respect
`to monomer (an example plot is shown in Figure 2C for
`excipient-free rhGH). A summary of the rate constants for
`all excipient:rhGH samples is listed in Table 3.
`The data show that all of the excipients employed in our
`study imparted significant stabilization to rhGH, particu-
`larly when present at a level of 131:1 excipient:protein and
`higher (Figure 5). The straight-chain polyols mannitol and
`
`Journal of Pharmaceutical Sciences / 1415
`Vol. 87, No. 11, November 1998
`
`MYLAN INST. EXHIBIT 1093 PAGE 4
`
`MYLAN INST. EXHIBIT 1093 PAGE 4
`
`-
`-
`-
`-
`-
`-
`-
`

`

`Table 4sGlass Transition Temperatures of Various Compoundsa
`Tg ((cid:176)C)
`4b
`
`excipient
`
`source
`
`mannitol
`sorbitol
`methyl R-D-mannopyranoside
`lactose
`trehalose
`cellobiose
`water
`
`Franks et al.45
`Franks et al. 45
`this study
`Saleki-Gerhardt and Zografi16
`Saleki-Gerhardt and Zografi16
`this study
`Slade and Levine 46
`
`-2
`70
`108
`115
`77
`-133
`a The data for the Tg of sugar/polyol excipients from the literature are for
`rigorously dried samples; data for methyl R-D-mannopyranoside and cellobiose
`contain about 2-3% residual moisture.
`b Mannitol is difficult to prepare in the
`amorphous state following lyophilization. In the present study, mannitol tended
`to crystallize upon lyophilization and therefore no Tg was observed.
`
`the range of 150-200 °C.40 The Tg of a lyophilized
`excipient:protein mixture may then be estimated as a
`contribution of individual Tgs.15
`Therefore, the addition of a component with a lower Tg
`than rhGH will serve to lower the Tg of the system relative
`to pure protein. The excipients used in our study all exhibit
`Tgs lower than that expected for pure rhGH (Table 4). In
`addition, there was some residual water present in our
`lyophilized samples (in the range of 2-3% for all samples
`in our study) which also decreases the Tg of the system.40
`Therefore, it is expected that addition of the relatively low-
`molecular-weight polyol and saccharide excipients used in
`our study would lower the Tg relative to rhGH alone.42 Even
`so, it was observed that the protein stability was improved
`in the presence of such excipients. Consequently, the
`stabilization afforded against solid-state aggregation effect
`cannot be explained solely in terms of the excipient’s ability
`to impact mobility through changes in the Tg of the system.
`A similar conclusion was made for the effect of various
`excipients on stabilizing bovine growth hormone toward
`thermal unfolding in the lyophilized state.42
`i.e.,
`Nonetheless, the physical state of the system,
`amorphous or crystalline, is likely to impact stability and
`Tg is an important parameter.15 Certainly it is required
`(but not necessarily sufficient) that the excipient remains
`in the amorphous, glassy phase with the protein at the
`storage condition. Above their Tg, amorphous polyols and
`sugars may be susceptible to crystallization at the crystal-
`lization temperature, Tc. If this occurs to an excipient in
`a pharmaceutical protein formulation, the stabilization
`effect may be lost as the excipient and protein phase
`separate. Also, in a closed system, crystallization may
`result in the release of water (e.g., formation of anhydrous
`crystals) which is then available for the protein, and may
`cause further destabilization, since water plays a key role
`in solid-state protein aggregation.11,40
`We also examined the physical state of lyophilized
`excipient:rhGH samples by X-ray diffraction (XRD). All
`samples, except for 300:1 and 1000:1 mannitol:rhGH, were
`amorphous upon lyophilization, as evidenced by the lack
`of any distinguishing features in their diffractograms (data
`not shown). However, there was the possibility that the
`excipient in the formulation may crystallize during high-
`temperature storage, particularly in the case where the
`protein content is relatively low, since proteins have been
`shown to inhibit sugar crystallization in the solid state.8,17,22
`To test this, we performed XRD on samples containing
`the highest levels of excipient, following their incubation
`for four weeks at 50 °C. The XRD data, depicted in Figure
`6, show that 1000:1 lactose, trehalose, and cellobiose all
`remained amorphous following the high-temperature in-
`cubation. However, the data for 1000:1 sorbitol:rhGH
`(Figure 6B) and 300:1 methyl R-D-mannopyranoside:rhGH
`(Figure 6C) revealed that the excipient had undergone
`
`Figure 5sPsuedo first-order constants for the deterioration of rhGH co-
`lyophilized with various amounts of (A) mannitol, (B) sorbitol, (C) methyl R-D-
`mannopyranoside, (D) lactose, (E) trehalose, and (F) cellobiose.
`
`sorbitol and methyl R-D-mannopyranoside were not quite
`as potent as the disaccharides tested, and exhibited an
`optimum stabilization at 1:131 excipient:rhGH. Lactose,
`trehalose, and cellobiose imparted a maximum stabilization
`effect when present at 131:1 excipient:protein and above.
`It is interesting that the 131:1 ratio represents the level
`of all potential strongly and weakly water-binding sites in
`the rhGH molecule.21
`It is possible that the excipient
`molecules afford stability to the protein by replacing water
`in the solid state, which would be consistent with the
`observed stability data (Figure 5). This conclusion is also
`supported by recent moisture sorption data showing that
`addition of sugars to rhGH decreases the accessibility of
`water-binding sites in a humidified atmosphere.7,8 These
`interactions may also have a relation to rhGH’s stability
`against lyophilization-induced structural alteration, as
`described below.
`Besides the potential of excipient-protein interactions
`to stabilize the protein, other factors may play a role, such
`as the excipient’s ability to dilute protein molecules in the
`solid-state and retard intermolecular reactions.12 Also, the
`excipient may provide a glassy matrix in which reactivity
`is retarded, and hence stability is improved.14,15 For
`instance, Hancock and Zografi15 have described how the
`amorphous state influences solid-state physical and chemi-
`cal properties in pharmaceutical formulations. It is im-
`portant for stability that the pharmaceuticals remain in
`the amorphous, glassy phase, below the glass transition
`temperature (Tg). The Tg is the temperature at which a
`material undergoes a change from a highly viscous glass
`to a viscoelastic rubber.
`For globular proteins such as rhGH, it is difficult to
`obtain Tg values.42 As discussed elsewhere, the difficulty
`in measuring Tg for proteins by standard techniques such
`as differential scanning calorimetry may be due to the large
`internal heterogeneity of domains and broad distribution
`of relaxation times.43,44 Most dry proteins exhibit Tgs in
`
`1416 / Journal of Pharmaceutical Sciences
`Vol. 87, No. 11, November 1998
`
`MYLAN INST. EXHIBIT 1093 PAGE 5
`
`MYLAN INST. EXHIBIT 1093 PAGE 5
`
`

`

`Table 5sFTIR Analyses of Various rhGH Formulations
`
`sample
`
`secondary structure content (%)a
`bandwidthb
`R-helix
`(cm-1)
`unorderedc
`(cid:226)-sheet
`10 – 1
`36 – 1
`7 – 3
`57 – 3
`aqueous solution
`17 – 1
`47 – 3
`24 – 3
`29 – 3
`lyophilized (excipient-free)
`co-lyophilized with mannitol
`10 – 1
`44 – 2
`21 – 4
`35 – 1
`31:1d
`11 – 1
`45 – 1
`16 – 1
`39 – 3
`131:1
`co-lyophilized with sorbitol
`9 – 1
`48 – 3
`16 – 2
`36 – 1
`31:1
`10 – 1
`38 – 2
`20 – 3
`42 – 2
`131:1
`co-lyophilized with methyl R- D-mannopyranoside
`14 – 1
`43 – 3
`41 – 1
`16 – 2
`31:1
`13 – 1
`47 – 2
`41 – 2
`12 – 2
`131:1
`co-lyophilized with lactose
`12 – 1
`43 – 3
`41 – 3
`16 – 2
`31:1
`11 – 1
`33 – 2
`48 – 2
`19 – 3
`131:1
`co-lyophilized with trehalose
`10 – 1
`41 – 2
`40 – 0
`19 – 1
`31:1
`9 – 1
`36 – 2
`46 – 2
`18 – 1
`131:1
`co-lyophilized with cellobiose
`12 – 1
`46 – 2
`43 – 2
`11 – 1
`31:1
`13 – 1
`45 – 3
`42 – 4
`13 – 2
`131:1
`a The secondary structure of rhGH was calculated by Gaussian curve-
`fitting of the Fourier self-deconvoluted amide I spectra. b Bandwidth corresponds
`to the full width at half-maximum of the (cid:24)1655 cm-1 band of all spectra in
`the amide I region. c Unordered secondary structure includes random coil,
`turns, and extended chains. d Values are the molar ratios of excipient-to-rhGH.
`
`spectrum of the lyophilized powder is significantly broad-
`ened due to the increase in intensity and bandwidth of
`other bands. While changes in the intensity of IR bands
`in this region demonstrate drastic structural alterations,
`broadening of the individual IR bands additionally dem-
`onstrates a loss of structural organization within the
`individual elements of the secondary structure.3 Quantita-
`tive analysis of the spectra by Gaussian curve-fitting
`revealed that lyophilization indeed caused a significant
`decrease in the R-he