Drug development costs are rising and pharmaceutical companies cannot afford tomake mistakes in late-stage development — especially when it comes to packaging.
The science behindsorbent selection
Drug products subjected to degradation because ofenvironmental stresses can be salvaged if proper
packaging and protection is provided by sorbents. Bythe time the drug product is handed from the formulationchemists to the packaging engineers, mechanisms ofdegradation, such as hydrolysis, oxidation, dehydration,isomerization, racemization, elimination andphotodegradation, are usually known. It is then upto the packaging engineers to plan a ‘defence’ againstthese degradation pathways.
In accordance with various regulatory guidelines,stability studies are required to prove that a drug willmaintain its physical and chemical characteristics duringa given time frame (expiration dating) to ensure the safety,identity, strength, quality and purity of the medicines.The International Conference on Harmonisation (ICH) haspublished a widely known guidance document regardingthe outline of such stability studies.1
Whether dealing with a new drug application forinnovators or an abbreviated new drug application
(ANDA) for generics, time is money and quickly bringinga quality product to market is key. This is particularlytrue for generic companies filing ANDAs for referencelisted drugs, as they are in competition for the 180-dayexclusivity provisions set forth by the Hatch-WaxmanAmendments to the Federal Food, Drug andCosmetics Act.2
Many modern drug substances and products aresusceptible to environmental humidity, and may physicallyor chemically degrade, or lose potency and efficacy whenexposed to atmospheric moisture. Although higher barrierpackaging can combat this issue, it is often cost-prohibitivecompared with less expensive packaging solutions,such as high-density polyethylene (HDPE) bottles andincorporating sorbents.
Historically, basic calculations were made for productsrequiring a sorbent regarding sorbent selection forregistration stability lots. A sorbent-ranging study wasperformed prior to this to determine whether or not thedesiccant recommended and its quantity were correct.
Thomas J. HurleyStanislav E. Solovyovfrom MultisorbTechnologies.
Imag
e So
urce
/Get
ty Im
ages
PEERREVIEWED
Sorbent selection
Performing these additionalcharacterization tests, such asdissolution, assay and degradantmonitoring, takes time and money.
Our goal was to eliminatesorbent-ranging studies or ‘guesswork’by creating a predictive model ofmoisture permeation and adsorption todetermine the appropriate sorbent andits amount to achieve shelf-life targetsfor given pharmaceutical formulationsand dosage forms.
The scientific modelA pseudo-empirical modellingprogramme has been developed byMultisorb Technologies to mathematicallypredict the stability outcomes of testing.No model can truly replace empiricaltesting, but the programme decreasesexcessive testing, and demonstrates
and predicts the effects of the selectedpackaging and its incorporated sorbents.The model predicts the internalconditions of a drug product packagebased on a given set of externalconditions and selected input criteria.This modelling is based on the integrationof internal and external equilibriumrelative humidity (ERH) conditionswith time, and the adsorptionprofiles/isotherms of the desiccantand drug product.
One of the assumptions necessaryregarding the dynamics of permeationand adsorption was that, at any giventime, the system inside a package is ina state of equilibrium:�in�ERH1�ERH2�ERH3�ERHA�ERHD
Each component, i (i�D, 1, 2, 3,…)inside the bottle has its own sorptionisotherm, Si(�) (Figure 1).
The water vapour transmission rate(WVTR) of the bottle is measured atsome known humidity difference (�RH)between internal and externalenvironments: �RH��in��out
A simplified picture of permeationand adsorption is considered:● The RH outside the bottle is constant,
as is the case with stability testing.● Moisture permeation through
container/package surface area isrelatively slow, allowing for fairlyrapid equilibration and an uniformERH of all components.
● Quasi steady-state pattern ofpermeation through the bottlewall is present.
● Package WVTR is linearly scalablewith �RH between separatedenvironments, which is the case withmany polymers including polyolefins.
The methodology follows a massbalance of moisture inside the bottle,including the moisture in the headspaceair (typically minimal as can be calculatedusing psychrometric equations) andin each solid component such as drugproduct and sorbent. The model isisothermal with T(t)�constant. Thesorption isotherms, Si(�), and densityof saturated water vapour in air areobtained at a specific environmentaltemperature, T. The initial ERH of thepackage system is determined from theinitial RH of all packaged componentsand atmospheric air at the time ofpackage sealing. Through a seriesof mathematical operations andintegrations, the time, t, to reach a givenERH, �, within the package is found.Consequently, the rate of water vapouringress is controlled by the WVTR ofthe package material and the sorptioncapacity of the components within thepackaging system. As the internal ERHof the bottle increases, the �RHbetween the inside of the package andthe external conditions decreases, andthe rate of moisture permeation throughthe package surface area decreases.
Water content versuswater activityMoisture can play a critical role insolid-state degradation. There are two
Real-time simulations (25 �C/60% RH) 180 Days 365 Days 545 Days 730 Days
Table 3 Summary of simulations and customer data.
Simulation 350 160-cc 2 g 26.0% (0.93%) 28.0% (1.00%) 29.5% (1.06%) 31.5% (1.12%)
Actual data 350 160-cc 2 g �28.0% (1.00%) �35.5% (1.30%) �33.5% (1.20%) �33.5% (1.20%)
Simulation 30 75-cc 2 g 13.5% (0.51%) 18.5% (0.68%) 23.5% (0.84%) 27.5% (0.98%)
Actual data 30 75-cc 2 g �13.5% (0.51%) �31.0% (1.10%) �25.0% (0.90%) �28.0% (1.00%)
Simulation 30 75-cc 1.5 g 15.5% (0.58%) 21.5% (0.78%) 27.0% (0.96%) 32.0% (1.15%)
Simulation 30 75-cc 1 g 19.0% (0.70%) 26.5% (0.95%) 32.5% (1.17%) 38.0% (1.42%)
Simulation 350 160-cc 2 g 27.5% (0.94%) 28.5% (0.97%) 30.0% (1.03%) 33.0% (1.15%)
Simulation 30 75-cc 2 g 13.0% (0.45%) 17.0% (0.58%) 20.5% (0.70%) 30.0% (1.03%)
Simulation 30 75-cc 1.5 g 15.0% (0.52%) 20.0% (0.68%) 24.0% (0.81%) 34.0% (1.19%)
Simulation 30 75-cc 1 g 18.0% (0.62%) 24.0% (0.81%) 29.0% (0.99%) 40.0% (1.49%)
Note Product Bottle Desiccant Predicted package ERH (drug product moisture content)count size (silica gel)
Accelerated simulations (40 �C/75% RH) 30 Days 60 Days 90 Days 180 Days
Figure 1 A typical bottle containersystem.
In
Out
WVTR
Sorbent selection
kinds of water potentially present in agiven drug product: bound water and‘free’ or ‘available’ water.3 Bound water(i.e., water of crystallization) is anintegral part of a molecule and may notbe available for chemical activity. Theenergy of activation required to releasethis moisture is not usually achievedby solely lowering the RH or adding adesiccant. Unbound or ‘free’ water isloosely held by powder particles andavailable for chemical activity. Theenergy of activation required to releasethis moisture can be achieved bylowering the RH or adding a desiccant.
Water content is an extensive propertyof a given material. The quantitativemeasurement of water is typicallyperformed by Karl Fischer titration,near-infrared spectroscopy or loss ondrying (LOD). Karl Fischer titrations havedominated the measurement of waterin pharmaceutical products for manyyears.4 “The Karl Fischer methodmeasures surface moisture (unboundmoisture) and most types of boundmoisture. For some products, thismeans that the Karl Fischer methodwill measure more moisture than thegravimetric method.”5
Water activity (aw) or ERH (aw�100and expressed as a percentage) isderived from the fundamental principlesof thermodynamics and physicalchemistry. aw is an intensive property ofa material and a qualitative measurementof the energy status of water in a systemdetermined by equilibrating the liquidphase water in the sample with thevapour phase water in the headspaceand measuring the RH of the headspace.3
Water content and aw can becorrelated by analysing theadsorption/desorption isotherm of amaterial. Sorption isotherms delineatea substance’s affinity for moisture atvarying RH and temperature.
By determining the water contentand water activity of a product andgenerating temperature-dependentsorption isotherms for all componentsin a package system, it is possible tomodel the internal conditions of thepackage system with time given constantexternal temperature and RH conditions.The entire package, including drugproduct and desiccant, acts as acomplete system and will reachequilibrium with time. For packagingconfigurations that exclude desiccants,the moisture that ingresses through thebottle with time is adsorbed only by the
drug product and evidenced by a higherERH within the bottle during the stabilitytesting time. For packaging configurationsthat include a desiccant, the desiccantwill first adsorb moisture within theheadspace (minimal). Because of theadsorption characteristics of the drugproduct versus the desiccant, thedesiccant will adsorb moisture fromthe drug product. The drug product inproximity to the desiccant will desorbor transfer moisture in a shorter periodof time than the drug product locatedfurther away from the desiccant. Themoisture that ingresses through thebottle with time will be adsorbed by thedrug product and desiccant accordingto their respective sorption isotherms.
ExampleA pharmaceutical company asked forrecommendations for a specificdrug product that was experiencingdegradation with time. In total, 30 tabletswere examined. The drug productsamples were representative of the
physical and chemical characteristicsas would be at the time of commercialpackaging.
The company achieved acceptablestability results in one packageconfiguration, but an undesired negativetrend in another. Therefore, the goal ofthis evaluation was to provide supportingdata for a desiccant recommendationfor the one bottle that would mostclosely mirror that of the other bottle.
The tablets were weighed on aModel AG245 analytical balance(Mettler-Toledo, OH, USA) and anaverage weight of 260 mg was obtainedwith a relative standard deviation of0.86%. Samples of the product werecrushed in a mortar and pestle to obtainaw of the product at 25 °C and 40 °Cusing a Series 3TE Water Activity Meter(Decagon Devices Inc., WA, USA). Thesamples were allowed to equilibratein the instrument for 15–17 min whilecontinuous readings were taken and aconstant value was obtained. Prior totesting, the instrument was calibrated
Table 1 Water activity measurements.
Sample aw @ 25 �C aw @ 40 �C
Calibration 0.252 (target 0.250 0.003) 0.268 (target 0.266 0.003)
Drug product 0.337 0.298
50
45
40
35
30
25
20
0 10
Ads (25C)Des (25C)Ads (40C)Des (40C)
20 30 40 50
Sample RH (%)
Wei
ght
(% c
hang
e)
60 70 80 90 100
15
10
5
Figure 2 Drug product sorption isotherms at 25 and 40 �C.
Sorbent selection
using a LiCl 13.41 molal solution inwater. The resultant values are listedin Table 1.
Accelerated stability testing andreal time stability testing are generallyperformed per ICH-prescribed conditionsof 40 °C /75% RH for 6 months and25 °C/60% RH for the duration of theexpiration date label claim (typically2 years),1 respectively. Recognizing this,water vapour sorption testing of thedrug product was performed at25 °C and 40 °C. Using a symmetricalgravimetric analyser, the water vapouradsorption of the product wascharacterized from 10% to 90% RHin 10% increments. The water vapourdesorption of the product wascharacterized from 90% to 10% RH in10% increments. The equilibrium criteriawere established such that the weightpercent change of the tablet was0.005% in 10 min or after a period of300 min as maximum. These parametersdefined the conditions necessary toprogress from one RH step to the next.The data logging interval was every2 min or a 0.200% change in weight.
For each experiment, one tablet wascrushed with a mortar and pestle andthe entire contents placed onto theweighing pan in the instrument. Thesamples were preconditioned withinthe instrument at their respective testingtemperatures (40 °C or 25 °C) and 2%RH for a period of time to removeresidual moisture and obtain a baselineof water vapour adsorption. Thebaseline is used to compare the drugproduct ERH with respect to residualmoisture content. The moistureremoved from the tablet during thepreconditioning process is usually agood indication of how much moisturea desiccant could potentially remove.
Adsorption/desorptionisotherms/profiles were generated at40 °C and 25 °C. The sorption isothermsfor the drug product are illustrated inFigure 2.
The company provided drawingsfor the bottles utilized for commercialpackaging: a 75-cc Blake bottle and a160-cc wide-mouth Blake bottle. TheHDPE resin used to produce the bottlesis Marlex HHM 5502-BN, as manufacturedby Chevron Phillips Chemical Company(TX, USA).
The quantity of tablets intended forcommercial packaging was 350 in the160-cc bottle and 30 in the 75-cc bottle.For the stability simulation models
of the drug product, the followingassumptions and principles applied:● The water vapour permeability for
the Marlex resin is0.381 g�mL/(100 in2�day; nominalvalue at test conditions of 38 °C/90%RH), and 0.085 g�mL/(10 in2�day;nominal value at test conditions of23 °C/60% RH).6
● The bottle dimensions and thicknessobtained from the drawings wereutilized to calculate the approximatesurface areas of the bottle andcorresponding moisture ingression,excluding the foil induction seal. Theinside diameter of the mouth of thebottle and the base-to-shoulder heightwas utilized for ergonomics of thedesiccant (i.e., ability to fit within thebottle).
● The reported bottle wall thicknessis the minimum wall thickness atmidpoint of the bottle. As there wasno tolerance for thickness providedon the 160-cc bottle drawing, theworst case minimum value plus0.25 mm was used in all calculationsand simulations. For the 75-cc bottle,
it was previously calculated andreported by the company thatthe average wall thickness was�0.84 mm. It is noted that variationsin the bottle wall thickness areinherent because of the moldingprocess. Thicker portions of thebottle wall will have a lower WVTRthan thinner portions and vice versa.WVTR of the whole sealed bottlecan also be measured directly usingIllinois Instruments’ (IL, USA) series7000 water vapour permeationanalyser.
● The actual volume of space withinthe container that the drug productconsumed was unknown. Becausethe moisture contribution to thesystem from the headspace of thecontainer is minimal, the total volumein the container was estimated andutilized as void volume.
● The bottles maintain a foil-inductionseal, and the simulations providedherein account for a steady-statemoisture transmission through thebottle and do not account for poorseals, pinholes or thickness variations
Bottle t�0 1 3 6 9 12 18 24
Table 2 Actual drug product moisture content (real time stability testing).
75-cc 0.8 1.0 0.9 0.5 0.6 1.1 0.9 1.0
160-cc 0.8 1.4 1.2 1.0 1.0 1.3 1.2 1.2
45
40
35
30
25
0 73 146 219 292 365Days
Inte
rnal
pac
kagi
ng R
H (%
)
438 511 584 657
Sim (160-cc)Actual (160-cc)Sim (75-cc)Actual (75-cc)Trend (160-cc)Trend (75-cc)
730
20
15
10
5
Figure 3 160-cc versus 75-cc comparison, simulated versus actual data real time (25 °C/60% RH).
Sorbent selection
becomes more stable and changeis more gradual, the predictionsbecome more accurate.
The moisture removal that occurredduring preconditioning/drying of thedrug products within the instrumentduring the testing was �1.15 and�1.05% at 40 °C and 25 °C,respectively. This corresponded fairlywell with the moisture contentreported by the company (1.21% and1.18%). Analysing the respectiveisotherms in Figure 2, moisturecontents of 1.15% and 1.05%approximately equate to an ERH of29–33%. The ERH is determined by firstlocating the initial moisture content ofthe product on the Y-axis ofa graphical isotherm and thenfollowing the line perpendicular to theY-axis until the said point crosses theisotherm line. The X-coordinate of thispoint is determined to be the ERH.Referring to the aw test resultsreported for crushed tablets in Table 1of this report (0.298–0.337), thecorrelation is reasonable and withinacceptable variation.
The company forwarded actualmoisture content results from real-timestability testing of the drug productin 75-cc and 160-cc bottles. The watercontent measurements determined byLOD at given time points are presentedin Table 2 (time in months).
As confirmed by the company,the time-point, t�0, was the ‘initial’
moisture content of the product at thestart of stability testing and is not trulythe ‘release’ moisture content of thedrug product, as there was a delaybetween packaging and production oftablets. Also of importance is the factthat the t�0 data was that of the 75-ccbottle for both package configurations;a t�0 was not measured for the productin the 160-cc bottle.
Considering the aforementioned, theinitial moisture content (i.e., the moisturecontent measured at initiation ofstability testing) for the simulations ofthe 75-cc bottle was considered to be0.80%. Analysing the actual moistureresults on the previous page andconsidering the moisture content ofthe product, it follows reason that the‘initial’ moisture content of the drugproduct packaged in the 160-cc bottlewould be �0.80%. An average valueof 1.13% was utilized for simulations
that could possibly occur in themolding or induction sealing processof the bottle.
● There are time constraints andlimitations when performingisotherms on drug products in thefinished tablet form. To compensatefor the relatively short period of timea tablet will be subjected to givenconditions within the gravimetricsymmetrical analyser, as comparedwith extended periods of time in acontainer, the tablet was crushedprior to testing. This method increasessurface area and accelerates theadsorption/desorption of the tablet.The reasoning behind this test isthat, with extended periods of time(days/months), the moisture availableto the tablet will eventually migrateand equilibrate throughout the entiretablet. The isotherms for the crushedtablets were used in all simulations.
● Fundamentally, as the adsorptionisotherms did not display significantdesorption moving from 0% to 10%RH, it was assumed that the drugproduct was sufficiently dried anda good baseline was established.Therefore, variations in moisturecontent of the drug product,assuming no chemical changesoccurred, would only vary the initialequilibrium RH of the drug product,which was determined by evaluatingthe isotherm’s X and Y axes. Slightvariations in moisture content duringformulation and/or processing of thedrug product will alter the initial ERHslightly, but should not significantlyaffect the simulations as presented.
● The simulations account forsteady-state conditions andequilibrium capacity of the drugproduct and desiccant. The ‘initial RHof packaging system’ is consideredinstantaneous. In actuality and notshown in the simulations, the ERHwill gradually decrease and thenincrease with time to the reported‘initial RH of packaging system’,similar to a cosine curve. For thisreason, the simulations are lessaccurate during early stages of thetesting, and do not account for thelag time in desorption/adsorption ofthe drug product and adsorption bythe desiccant. This is especially truewith real-time simulations as thekinetics of the system are slowerthan those at accelerated conditions.As time progresses and the system
50
45
40
35
30
25
0 73 146 219 292 365Days
438 511 584 657
Sim (160-cc)
Sim (75-cc, 2 g)
Sim (75-cc, 1.5g)
Sim (75-cc, 1g)
730
20
15
10
5
Inte
rnal
pac
kagi
ng R
H (%
)
Figure 4 Real time stability conditions (25 °C/60% RH). 160-cc versus 75-cc comparison(simulated with various desiccant quantity).
● Drug products subjected to degradation as a resultof environmental stresses can be salvaged by usingsorbents.
● Using a modelling programme to mathematicallypredict the stability outcomes of testing eliminatessorbent-ranging studies.
● Formulation stability can be determined with basicknowledge of the formulations and pathways ofdegradation.
On the go…
Sorbent selection
pertaining to the 160-cc bottleconfiguration.
The 25 °C isotherm was used forreal-time simulations and the 40 °Cisotherm was utilized for acceleratedsimulations. Figure 3 displays thepredicted ERH and correspondingdrug product moisture content ofthe 350-count/160-cc bottle and30-count/75-cc bottle, respectively,packaged with 2 g of silica gel (730 daysreal-time stability at 25 °C/60% RH). Theactual moisture content data relayedby the company (Table 3) was convertedto drug product ERH per the 25 °Cisotherm and plotted on the graphs. Apolynomial trend line was fitted to thedata. As can be seen, the simulation’sendpoint is very similar to the endpoint ofthe polynomial trend of the actual data.
Figure 4 displays the predicted ERHof the 30-count/75-cc bottle if packagedwith 2 g, 1.5 g or 1 g of silica gel andtested at real-time conditions of25 °C/60% RH for 730 days. The160-cc bottle simulation (2-g silica gel)is also presented for comparison.
Figure 5 displays the predicted ERHof the 30-count/75-cc bottle if packagedwith 2 g, 1.5 g or 1 g of silica gel andtested at accelerated conditions of40 °C/75% RH for 180 days. The 160-ccbottle simulation (2-g silica gel) isalso presented for comparison.
The ERH and corresponding drugproduct water content for allsimulations and actual customer datahave been summarized in Table 3.
Based upon the data provided by thecompany, and the testing and analyses,1.5 g of silica gel was recommended forthe 75-cc bottle configuration, such thatthe stability results would most closelyreflect those established with the160-cc bottle configuration with 2 gof silica gel.
ConclusionsIt can be seen from the exampleprovided previously that our modellingprogramme is valid and effective inpredicting the moisture content of drugproducts with time. With basic knowledgeof the formulations and pathways ofdegradation, the stability outcomeof formulations can be determined.
The example provided details a HDPEbottle package configuration, but themodel can be used with other packagedesigns, such as blisters, pouches,intermediate bulk containers andinhalation devices. The base
requirements for use of the model aresorption isotherms, a known WVTR atconstant internal and external RH,and some basic knowledge of theformulations and pathways ofdegradation.
Thomas J. Hurleyis a Senior Product Leader for HealthcarePackaging at Multisorb Technologies (NY,USA). He has worked at Multisorb for8 years providing technical support andconsultations to the company’s globalhealthcare customers. His involvementwithin the R&D team has allowed him toprovide new and innovative solutions forcustomers’ packaging needs. Prior to joiningMultisorb, he was a Production Managerat The Mentholatum Company (NY, USA)and Operations Supervisor at Kraft Foods(IL, USA). He holds a BSc in Chemistry fromClarkson University (NY, USA).
Stanislav E. Solovyovis a Principal Scientist at MultisorbTechnologies (NY, USA). His industrialexperience lies in polymer manufacturing,processing, and process and performanceproperty modelling in packaging applications.His research interests include activepackaging technology, nonlinear dynamicsof dissipative chemical systems, physicalchemistry of polymers, rheology andprocessing of polymer melts and composites,
powder processing, design and performancemodelling of reactive barrier packaging. Heis the author of 20 peer-reviewed articlesin scientific journals and six US patentapplications. He holds BSc and MS degreesin Applied Mathematics and PhD degreein Polymer Chemistry with postdoctoralexperience in Chemical Reaction Engineeringand Materials Engineering and Science.
References1. CDER/FDA Guidance for Industry — Q1A
(R2) Stability Testing of New DrugSubstances and Products, November2003.www.fda.gov/cder
2. CDER/FDA Guidance For Industry —180 day Generic Drug Exclusivity Underthe Hatch-Waxman Amendments to theFederal Food, Drug, and Cosmetics Act,June 1998.www.fda.gov/cder
3. A.J. Fontana, Fundamentals of WaterActivity Part I: Water Activity, DecagonDevices, Inc., 16 April 2008.www.on24.com
4. B. Snider, L. Peihong and N. Pearson,Pharm. Technol. 31(2) 56–71 (2007).
5. Centre for Biologics Evaluation andResearch — Guideline for theDetermination of Residual Moisture inDried Biological Products, January 1990.www.biopharm.com.tw
6. Actual testing provided by ChevronPhillips Chemical Co. (TX, USA),September 2005.
55
50
45
40
35
30
25
0 10 20 30 40 50Days
Inte
rnal
pac
kagi
ng r
elat
ive
hum
idity
(%)
60 70 80 90 100 110 120 130 140
Sim (160-cc)
Sim (75-cc, 2g)
Sim (75-cc, 1.5g)
Sim (75-cc, 1g)
150 160 170 180
20
15
10
5
Figure 5 Accelerated stability conditions (40 °C/75% RH). 160-cc versus 75-cccomparison (simulated with various desiccant quantity).
Article Reprinted from the ©June 2008 issue of
Art
icle
Rep
rint N
umbe
r:07
49
