Biologicals cover a wide range of materials which are often labile in solution and require lyophilization to stabilize them. For instance Factor VIII activity begins to fall as soon as a sample of plasma is taken and whether a purified concentrate from plasma or a recombinant product of DNA technology is considered, both show loss of activity during processing and on storage and so one of the challenges of formulation is to stabilize the biological material for the shelf life of the product [1]. In order to do this water is removed by lyophilization and this usually reduces the degradation rate significantly. Towns [2] reviewed the interaction between water and proteins; stripping the water to below a monolayer covering of the surface of the protein generally engenders greater dry-state stability. Higher water content in lyophilized material plasticises the protein and allows for greater mobility as well as lowering the glass transition temperature, which can impact stability however formulation must be optimized so as to protect the dehydrated biological from damage [3].
One crucial area of quality control for such biological products then is the measurement of the residual water content of the lyophilized material [4,5,6]. As higher moisture content often correlates with poorer stability, this has resulted in recommendations for low residual moisture content of 0.5-3% w/w [4] and moisture induced instability is particularly an issue when preparing primary biological reference materials as these must maintain their potency for the life of the standard [7] although over-drying may also be deleterious to stability.
Freeze drying comprises three stages, freezing, primary drying (usually at sub ambient temperatures) and secondary drying (usually at ambient temperatures or above). Cycles are designed so as to achieve removal of most of the loosely-bound water and to achieve a pharmaceutically elegant cake with a high retention of product bioactivity. In the extreme example of a collapsed freeze dried cake water content is often high (Figure 1) and although high moisture in itself need not necessarily result in impaired stability [8], degradation is often increased in such materials. Even when an acceptable freeze dried cake is achieved residual moisture content may vary depending on cycle and formulation. Very dense freeze dried cakes, such as when there is a high protein content of say 10-20%, may dry slowly with formation of a resistant skin which will resist water vapor escape until it ruptures. It has been shown that to reduce water content sufficiently drying at above-ambient temperature may be required and it is the temperature rather than the length of time that governs the effectiveness of this drying step, even though in our experience the product temperature as judged by in-vial probes may never quite achieve these shelf temperatures.
To determine the residual moisture content then is critical and a number of methods have been used including gravimetric measurement, coulometric Karl Fischer titration, thermogravimetric analysis and others [6]. We share here our insights into the use of four such methods during the preparation of lyophilized biological reference materials at NIBSC.
Coulometric Karl Fischer
Coulometric Karl Fischer is widely used [9] and gives reliable results with a wide range of materials although some compounds (such as aldehydes, ketones and thiols) make the titration problematic. At NIBSC samples are analyzed within a dry box with humidity below 300ppmV water vapor (1.2% RH) and reconstituted with an aliquot of the anolyte solution and re-injected into the cell via the sample port. Ideally, it is important that the sample is fully soluble in the methanolic reagent but in practice this is not always possible with proteins of high concentration (such as sera) and at best a slurry of material is achieved, requiring the port stopper to be removed and a wider diameter disposable pipette used to transfer the product rather than direct injection into the cell via the port septum. Alternatively others have suggested that more polar solvents such as formamide can be added to aid dissolution, though this may require careful control to avoid artefacts [10] . This disruptive addition of partly soluble material results in a higher CV of determination than would be expected were the cell not to be so disturbed and the coulometric Karl Fischer operating system incorporates a delay of up to 1.5 minutes to cover the necessary addition time.
Most biological materials lyophilized at NIBSC have given good moisture determination by coulometric Karl Fischer and this is our method of first choice. A number of ampoules/vials (typically 12 for a production scale fill of several thousand ampoules) are tested across the fill and the mean and coefficient of variation recorded. Although low CVs are often seen (below 20%) if the variation rises above 35% this usually indicates a need for further investigation for potential heterogeneity in the product or problems with the assay. When freeze drying cycles have been appropriately designed the residual moisture across a batch should not differ significantly, but validation of the moisture content distribution should be performed when validating a new freeze drying process (Figure 2).
With the increasing number of preparations which have a low dry weight of non-volatile materials (of 10mg/ml filled or less) the impact of dry weight on moisture content determination has been evaluated. A series of two model freeze dried compounds (trehalose and human serum albumin) were prepared and residual moisture determined for dry weights from 50mg/ml to 1mg/ml. The apparent residual moisture rose exponentially below 10 mg and this may be an indication of the limit of detection of the method, although low moisture content determinations with narrow CVs are achievable in some reference material batches. If problems occur then a number of ampoules can be reconstituted and pooled in a single titration (as recommended in [5]). On occasions some formulations have proven problematic, with titration times becoming long (>4 min) - often this happens with materials with high protein content such as sera or blood product concentrates which can be difficult to suspend/ dissolve and in such cases the cell solutions will need to be more regularly changed.
Vaporizer Karl Fischer Coulometric Titration
This solubility problem has led some to recommend the use of a vaporizer to drive the moisture off the sample and carrying this through on a stream of dry nitrogen into the KF cell [9,10]. However, in our experience this has proven problematic with biological materials due to thermal decomposition at comparatively low temperatures although the effect can be addressed by careful experiments to determine the decomposition temperature and this may be satisfactory where the same material is routinely assayed but would be too time consuming for our situation where many different materials/formulations have to be assayed. However, the technique has been useful in our experience to test the halobutyl rubber closures used in freeze drying, to determine the effectiveness of drying protocols used in their preparation.
Thermogravimetric analysis (TGA)
Thermogravimetric analysis is widely used [4-6,11] to determine the weight loss associated with the heating event to drive off the residual moisture. Unlike Karl Fischer which relies on a chemical reaction to detect water and therefore can be influenced by specific chemical contaminants (for instance iodine- binding components), thermogravimetric analysis accurately measures the weight loss as the sample is heated. This method measures not only the water but also any other volatiles which can be evolved by heating, so the composition of the lyophilized material must be well understood for it to be used to accurately measure the residual water content. The method is however susceptible to spurious water content due to thermal decomposition and this is particularly important for biological materials which may decompose at modest temperatures, for instance 100-150ºC. In order to address both of these problems evolved water is driven off the sample and into an evolved gas analyzer (low molecular weight range quadrapole mass spectrometer) using a stream of dry inert gas such as helium) (Figure 3).
In our experience the method has required some optimization in order to deliver results consistent with the data delivered by coulometric Karl Fischer for biological materials. One of the main difficulties with handling lyophilized materials is the artefacts introduced by the hygroscopic nature of the material when a container is opened to load the sample pan. Analysis of samples on open pans give moisture contents far higher than materials contained within inert environments. May [4] recommend the containment of the entire unit within a dry box with low relative humidity (0-20% RH) but we have investigated the effects of panning the materials then hermetically sealing the pans within a dry box with inert environment (1.2%RH or lower) and analyzing them on an uncontained thermogravimetric analyzer, using an autosampler and pan piercing device. Even use of a portable inflatable dry box device inflated with dry nitrogen can deliver a low moisture environment of 5-10% RH. Table 1 shows the moisture content of some biological materials panned in either a full dry box or panned on the open crucible.
Dense freeze dried cakes (such as plasma) adsorbed moisture from the atmosphere more slowly and so were more amenable to panning in an uncontrolled environment whereas those materials with more friable cakes (such as heparin) readily adsorbed moisture and so required panning within a controlled atmosphere of low RH.
Our initial studies with TGA resulted in higher moisture detection by TGA than by KF determination but this has been addressed following several modifications in our TGA protocol. The impact of a pre-equilibration step at ambient temperature before heating was investigated but not found to be beneficial as weight loss began almost immediately, different heating rates between 5 and 20ºC/min were also evaluated but did not result in consistency of the results obtained to those by Karl Fischer titration. However, a reduction in helium flow rate was implemented, and this resulted in more consistent results between TGA and KF (table 2) . The advantages of the TGA method are that it is more amenable to low dry weight samples (operating on typically 5mg of solid or less) whereas coulometric Karl Fischer although giving reliable results with freeze dried materials of greater than 10mg non volatile dry weight, gave increasingly higher relative moistures for materials dried with lower mass. As biological reference materials with low dry weight are increasingly being encountered in our Laboratory, TGA offers a valuable control measurement if the residual moisture determinations by Karl Fischer are unexpectedly high.
Near Infra Red Moisture Measurement
It has long been known that residual water can be detected by near infra red spectrometry and this method has been applied for the moisture determination in both small molecules and biological products [12]. Several authors have shown that the moisture detected correlated well with traditional coulometric Karl Fischer analysis [13,14]. At NIBSC non-invasive reflectance infra red spectrometry has been used to measure the moisture content in infectious biological reference materials that are not amenable to traditional Karl Fischer analysis due to the heightened risks of sharps injury with such materials. The method requires a standard curve to be generated using an identical lyophilized formulation and container diameter but which is amenable to Karl Fischer analysis. Figure 4 shows the calibration curve obtained for a 5ml screw capped vial format of lyophilized excipients, equivalent to those used for the stabilization of a live virus reference material. Lyophilized excipient in vials were opened to atmospheric air and allowed to equilibrate for different time lengths to increase the residual moisture content. The restoppered vials were then held and assayed by NIR and when stable over time moisture was confirmed in typical vials by coulometric Karl Fischer. This approach has been used recently at NIBSC to determine the moisture content in live virus reference materials under development as working standards for clinical diagnostictesting assurance. The method could lend itself well to incorporation into filling/labeling lines and because NIR measurement takes only seconds whole batches can be analyzed. However, each container format and formulation/fill depth will need its own calibration curve and there is an absolute requirement for integrity of the freeze dried cake. This makes it less appealing in our own situation where the formulation and presentation varies regularly depending upon the standard being lyophilized. This method has also proved useful in determining variations in the residual moisture content across batches of lyophilized materials. This has been used to test where processing problems have been suspected but could also be used to establish drying equivalence across the shelves of a freeze dryer during validation.
Other Methods
There has been interest for some years in the use of non invasive methods to study the relationship between headspace gas moisture and moisture in the lyophilized product in contact with it [11,15] More recently headspace moisture analysis has become commercially deliverable from a number of manufacturers and these methods have been applied to assess residual moisture content [16]. The advantages of such methods are that there is no absolute requirement for the freeze dried cake appearance to be perfectly formed but there will inevitably still be a need as part of the validation to establish the impact of the particular formulation on the non-invasive near infra red response. However, if fully validated these methods would serve as a powerful quality control tool in assessing the moisture distribution across the entirety of a freeze dried batch to a greater degree than is possible with destructive test methods.
References
1. Hubbard A, Bevan S, Matejtschuk P. Impact of residual moisture and formulation on factor VIII and factor V recovery in lyophilized plasma reference materials. Anal Bioanal Chem 2007; 387(7):2503–2507.
2. Towns JK. Moisture content in proteins, its effect and measurement. J Chromatogr A 1995; 705:115-127.
3. Wang W. Lyophilization and development of solid protein pharmaceuticals. Int J Pharm 2000; 203:1–60.
4. May JC. Regulatory control of freeze dried products: Importance and evaluation of residual moisture in “Lyophilization/Freeze drying of Pharmaceutical & Biological Products” eds L Rey & JC May 2nd ed. Dekker, New York. 2004, pp 349-384.
5. May JC, Grim E, Wheeler RM, West J. Determination of residual moisture in freeze dried viral vaccines : Karl Fischer, gravimetric and thermogravimetric methodologies. J Biol Stand 1982;10:249-59.
6. May JC, Wheeler RM, Elz N, Del Grosso A Measurement of final container residual moisture in freeze dried biological products. Dev Biol Standard 1992; 74:153–164.
7. Matejtschuk P, Andersen M, Phillips P in “Lyophilization/Freeze drying of Pharmaceutical & Biological Products” eds L Rey & JC May 2nd ed. Dekker, New York. 2004, pp385-424.
8. Wang DQ, Hey JM, Nail SL Effect of collapse on the stability of freeze-dried recombinant factor VIII and alpha-amylase J Pharm Sci 2004;9395):1253-63.
9. Zhou L, Socha JM, Vogt FG, Chen S, Kord AS. A systematic method development strategy for water determinations in drug substances using Karl Fischer titrations. Amer Pharm Rev 2010; 13(1):74-84.
10. Stern M. Reagent selection in Karl Fischer analysis of pharmaceutical products. Amer Lab 2006;38(4):28-29.
11. May JC, Del Grosso A, Etz N, Wheeler R, Rey L. Thermogravimetry and vapor pressure moisture - Applications to determination of residual moisture in BCG vaccine. J Therm Anal Calor 2006;83(1): 31-3
12. Kamat MS, Lodder RA, DeLuca PP. Near infra red spectroscopic determination of residual moisture in lyophilized sucrose through intact glass vials. Pharm Res 1989; 6(11):961–965.
13. Savage M, Torres J, Franks L et al. Determination of adequate moisture content for efficient dry-heat viral inactivation in lyophilised factor VIII by loss on drying and by near infra red spectroscopy. Biologicals 1998; 26(2):119–124.
14. Lin TP, Hsu CC. Determination of residual moisture in lyophilized protein pharmaceuticals using a rapid and non-invasive method: near infrared spectroscopy. PDA J Pharm Sci Technol 2002; 56(4):196–205.
15. Rey L . Glimpses into the world of freeze drying in “Lyophilization/ Freeze drying of Pharmaceutical & Biological Products” eds L Rey & JC May 2nd ed. Dekker, New York. 2004, pp1-32
16. Mahajan R, Templeton AC , Reed RA ,Chern RT. Frequency Modulation Spectroscopy A Novel Nondestructive Approach for Measuring Moisture Activity in Pharmaceutical Samples. Pharm Technol 2005;29(10):88-104.
The authors work in the Standardization Science Section, of the National Institute for Biological Standards & Control, a centre of the Health Protection Agency in the United Kingdom whose mission is to safeguard the quality of biological medicines. Kiran Malik (BSc, MSc Medicinal Chemistry, University College London, UK) , Chinwe Duru ( BSc Pharmacology, MSc Biomedical Sciences, University of East London, UK) Mahammad Ahmed (BSc (Hons) Biology, Queen Mary College, University of London) are Scientists and the team is led by Paul Matejtschuk (BSc Biochemistry, University of York, UK, PhD Chemistry University of Warwick, UK) . The Standardization Science team have extensive experience of the formulation and lyophilization of a wide range of biological references materials with expertise in pilot scale freeze drying, thermal analysis, product characterisation and quality control. Website www.nibsc.ac.uk