Residual Moisture Testing Methods for Lyophilized Drug Products

Many drug products that are unstable in the presence of water are lyophilized to increase the stability and shelf life. Lyophilization is a process of preservation by removing water from the product to sufficiently low levels for adequate stability. This stability is achieved by limiting the potential for degradation upon storage caused by hydrolysis reactions with any remaining residual water, referred to as residual moisture. Residual moisture-considered a critical quality attribute (CQA)-can have a significant impact on the stability of a lyophilized product when stored in the dried state. The remaining water at the end of the process can be involved in hydrolysis reactions, which typically lead to degradation. The interaction of residual moisture, or more specifically of available water, on the lyophilized material can result in changes including degradation, crystallization, and collapse. Reconstitution of the drug product can also be affected. Residual moisture testing, therefore, requires a method that provides accuracy and precision to low levels.

Different methods can be employed to measure residual moisture content in lyophilized materials. Understanding the advantages and suitability of each method and its respective limitations aids in selecting the most appropriate method for testing. This paper reviews the common methods for residual moisture testing: loss on drying, thermogravimetric analysis, Karl Fischer coulometric titration, and near infrared (NIR) spectroscopy.

The significance of residual water

Lyophilization is a way to preserve products that are unstable in the presence of water. Removing the water from the product decreases the potential for degradation by hydrolysis and, therefore, improves the product’s shelf life. Residual moisture content is a CQA of lyophilized finished products. According to the International Council for Harmonization (ICH) Q6A guidance, the water content of a finished product should have a test procedure and a specification when appropriate (1).

It is commonly assumed that removing as much water as possible from the finished product is crucial; however, over-drying a finished product, or removing too much water, can have detrimental effects, such as altering a crystalline hydrate or inducing protein unfolding (2). Therefore, understanding the role water plays and how the water interacts with the finished product is important. There are two ways water can be present and interact with finished products: adsorbed and absorbed water (3). Adsorbed water is on the surface of the product, and absorbed water resides in the bulk of the finished product (3).

Water interacts differently with amorphous and crystalline structures. Crystalline structures generally do not have as high an affinity to absorb water as amorphous structures because of the high degree of order, tight crystal packing, and rigid structure. Amorphous structures are not highly ordered or rigid, allowing the structure to more readily sorb water. Water also acts as a plasticizer and will affect the dried product glass transition temperature: As the residual moisture content increases, the glass transition temperature decreases. Lowering the glass transition temperature can cause long-term stability issues, sometimes allowing the amorphous material to crystallize. When material crystallizes, water will be released. This freed water can then affect the API and cause hydrolysis reactions to occur, further degrading the finished product. The effects of water on a finished product can lead to a number of other changes including chemical or biochemical degradation and can alter reconstitution behavior or promote collapse of an open, porous structure.

Because water can be present in a lyophilized product in a number of ways, measurement of the residual moisture and the method used to measure the water content must be accurate, precise, and, preferably, discriminating. The residual moisture content of a lyophilized product can be quantified by using a number of different methods. It is imperative the method used to quantify the residual moisture is appropriate for the finished product being tested, the quantity of water, and the type of water present (adsorbed or absorbed).

Sample preparation considerations

The majority of lyophilized products are amorphous, have a significantly large surface area, and a propensity to sorb moisture when exposed to the atmosphere. Therefore, the most important aspect in all residual moisture testing methods is preventing moisture uptake during sample preparation and handling. Care should be taken to minimize the exposure of the product to atmospheric moisture and the potential risk of skewing the test results.

Lyophilized products are sometimes stoppered at reduced headspace pressure. During sample preparation, the product and its container must be reverted to atmospheric pressure. If this is not performed prior to testing and the test method requires weighing of the sample pre- and post- testing, the initial pre-testing weight will be incorrect. The sample can be brought to atmospheric pressure by introducing dry air or nitrogen piercing the stopper with a needle; or, the stopper can be removed in a low humidity glove box. Piercing the stopper is preferred as removing the stopper without first equilibrating the pressure may disrupt the cake and lead to product blow-out.

Residual moisture testing methods

The methods listed are not the only methods available, but rather the most commonly employed, and consequently will be covered within the scope of this review. For each approach, it is essential to understand potential interactions and assess the benefits and risks.

Loss on drying (LOD) method. LOD is a gravimetric method that determines the weight loss of a sample after driving off volatile residuals. In this method, the temperature used needs to be warm enough and the time long enough to release all volatile components. Because this method determines the weight loss of any volatile constituent, not only water, weight loss of other volatile components, such as an organic solvent impurity, is reflected in the measurement.

For this gravimetric method, a sample is accurately weighed, then placed in an oven and dried at a high enough temperature for an appropriate amount of time to drive off any volatile residues. Testing can be performed in the primary packaging container if the container can withstand the test conditions. For example, if the sample is contained within a glass vial, then testing could be performed without removing the sample from its container. If the sample was lyophilized in bulk, then a sample of the material should be weighed into a glass vessel or other suitable container for testing. For any sample that would sorb moisture when exposed, samples should be prepared in a controlled environment to prevent moisture uptake from humidity in the atmosphere.

LOD is a relatively simple and straight-forward method to perform, but method development is required to determine the appropriate test conditions of temperature, time, and sample weight. When using this method, it is important to consider that the sample is exposed to elevated temperature conditions that will drive off all volatile matter. In addition,  if the material is sensitive to elevated temperatures, thermal degradation may influence the results.

The sample weight needs to be sufficient for an adequate level of detection and resolving weight differences to accurately quantify residuals. For example, if the sample is 1 g, then the sample should be weighed to at least 0.0001 g to be able to detect weight changes and calculate the weight loss to suitable significant figures; in this example, to resolve to 0.1% moisture.

A common temperature and condition to perform this method are 105 °C and atmospheric pressure for one hour. Other temperatures and pressures may be appropriate, depending on the product being tested. Examples of test conditions used for determining residual constituents in pharmaceutical drug substances and products are summarized in Table I.

Oven temperature control should be within ± 2° and the chamber pressure to within 2 mmHg of the intended conditions. In addition, the oven chamber should be qualified for temperature uniformity; all locations in the chamber should be within ± 3° of the target setpoint.

A vacuum oven may be required to dry the sample under a reduced pressure. In this case, the oven should be fitted with multiple ports for a vacuum source, pressure gauge or instrument, and dry air or inert gas inlet to revert to atmospheric pressure at the conclusion of the test.

Adequate control of the test parameters is essential. Control of the temperature should be within a reasonable variation for reproducibility of the conditions and consistency of the results. If a reduced pressure is part of the test conditions, then the proper measurement is also important. To assure the test conditions are reproduced, it would be beneficial if these test parameters were recorded on an instrument as well. The oven instruments should be calibrated on a regular basis for confidence in the results.

Points to consider for loss on drying. LOD is an effective method, particularly when there are potential chemical reactions involving the product and solvents or reagents. This method presents challenges when the dry weight of the finished product and water content are low. The weight of the container relative to the weight of the cake can make determining the small weight loss changes difficult to discern as small weight changes can be lost due to the weight of the container or vial. For example, a cake with a solid content of 100 mg and a residual moisture content of 1% would contain 1 mg of water. The difference in weight obtained on the balance prior to and after the LOD test would only be 0.001 g. Therefore, the balance needs to be able to accurately weigh beyond 0.001 g, to at least 0.0001 g.

Sample handling should be minimized; gloves should be worn to avoid material contamination. When testing is complete, the sample container should be cleaned and dried in an oven to ensure an accurate weight. When drying at a higher temperature, the sample container should be allowed to cool in a desiccator. Prior to the obtaining the final weight, the container should be allowed to equilibrate outside of the desiccator; this allows the inherent sorbed moisture to be present on the outside of the container when the sample was weighed initially.

When testing the final product container at higher temperatures in an oven, volatile components in the elastomeric stopper may vaporize, leading to an erroneous post-test weight. Therefore, any primary packaging component that may contain volatile constituents should not be included with the sample during the test.

Thermogravimetric analysis (TGA) method. TGA is an instrumental method that records the mass of a sample as a function of temperature or time. During testing, the mass of the sample is continuously monitored by a microbalance. The changes in weight are recorded by the instrument. During the analysis, the sample is protected from atmospheric moisture as nitrogen, or another inert dry gas, is continually purged through the instrument sample chamber. Figure 1 is an example of a TGA instrument.

Figure 1: Example of thermogravimetric analysis (TGA) instrument. (Figure courtesy of authors)

The sample is weighed into a sample pan. Depending on the instrument, there may be more than one type and size of sample pan available. The sample pan, for example, may be made of platinum metal or ceramic, which can be reused, or made of aluminum and be disposable. The sample pan is then hung from the hang down wire, which is connected to a microbalance to record the mass of the sample as the sample is heated. A thermocouple is positioned close to the sample to record the temperature during testing. When the testing is initiated, the furnace moves up into position surrounding the sample pan. In Figure 1, the furnace is in the down position.

The change in weight can be correlated to temperature. The thermogram in Figure 2 reflects a material having three distinct weight loss events, noted by three distinct downward slopes in the percent weight loss line (green), as well as the three sharp peaks in the derivative line (blue) (4). When analyzing thermograms, the start of the weight loss event is when the derivative begins to stray from the baseline; the end of the event is when the derivative then returns to the baseline. The temperature at which the weight loss is complete can be correlated to the peak of the derivative. As seen in the example, the derivative value begins to increase, reaches a maximum, then abruptly decreases, indicating first weight loss is complete at 150 °C, the second at 500 °C, and the third at 750 °C.

Figure 2: Example of a thermogram. (Figure courtesy of authors)

The TGA method requires only small quantities, as little as 2 mg of material (2). When evaluating a TGA scan, adsorbed water can be distinguished from absorbed water, which can be distinguished from the water of hydration. In conjunction with high-temperature differential scanning calorimetry (HT-DSC), TGA can distinguish water evolved during degradation (5). This method, however, will measure both surface and bound water in addition to all volatiles. Therefore, just like LOD, the type of water present, absorbed or adsorbed, must be distinguished.

Most often, the surface adsorbed water will be vaporized at a lower temperature and earlier time. Absorbed water, required to first diffuse to the sample surface then vaporize, is reflected in a more gradual weight loss at higher temperatures and over a longer time. If a hydrate is present, a sudden weight loss at a distinct temperature can be observed as the crystal hydrate melts.

Figure 3 illustrates a gradual weight loss from surface adsorbed water over a temperature range of 40 °C to 65 °C. Note the change in the rate of weight loss beginning at 67 °C reflected in the weight percent and derivative weight (%/°C).

Figure 3: Example of a gradual weight loss thermogram. (Figure courtesy of authors)

Two approaches to testing are commonly used for TGA: dynamic and isothermal. In dynamic testing, the temperature is increased at a constant, linear rate. A typical heating rate is 10 °C per minute. In the isothermal method, the temperature is kept constant for the duration of the test and the endpoint is established when the sample reaches a constant weight.

The main instrument parts are a microbalance and a programmable controller for controlled heating. A weighing platform attached to a microbalance is suspended in a heated chamber. The sample is dispensed into open pan, and the weighed pan is placed on the platform. The chamber is then heated, and the weight is continuously recorded.

The instrument needs to be calibrated on a regular basis for both weight and temperature (5,6). Some instruments can be connected to a gas chromatograph,  mass spectrometer, or other type of analytical instrument, which can aid in identifying each volatilized substance.

Points to consider for TGA. TGA measures mass changes as the sample is heated in a controlled environment. Samples should be prepared in a low-humidity glove box. If the TGA is located outside of a low-humidity glove box, the sample should be protected to prevent moisture uptake from the atmosphere during transfer to the TGA instrument by placing the sample pan in a small glass dish with desiccant and a lid.

The common practice of targeting a sample size of 10 mg of lyophilized material in the sample pan can be difficult depending on the sample bulk density. Problems can arise when working with samples that are light and fluffy, or electrostatic. Any variation in the packing of the material and dispensing into the sample pan can have an influence on the desorption rate, and therefore, weight loss results.

Valuable data can be obtained from the TGA method; adsorbed water can be distinguished from absorbed water, which can be distinguished from the water of hydration or water due to degradation (5). Because a small sample size is appropriate, this method is particularly useful with very small quantities of product.

However, analyzing the thermograms can be more difficult. With larger sample sizes, the weight-loss events on the resulting thermograms are of greater magnitude and more defined, making for easier detection and analysis. The transitions will still be there in a smaller sample size; however, may lack definition and be more difficult to detect and analyze. Analyzing thermograms can be subjective due to challenges defining the start and end point of an event.

Karl Fischer titration method. Karl Fischer titration is one of the most widely used analytical methods for residual moisture determination for lyophilized products (7,8). The method is suitable for a variety of dry products and, depending upon the amount of moisture to be measured, has multiple options for the best applicable method. Direct volumetric dispensing of the reagent is suitable for materials that have a relatively high level of residual moisture content in the range of 25–250 mg of water present in the sample. Lyophilized products typically have a low level of residual moisture content; a more sensitive method is warranted.

Generating the reagent electrochemically, known as the coulometric method, is considered a micro-method and is best suited when the residual moisture content is in the range of 0.5–5 mg of water. The ability to generate the iodine electrochemically allows the method to be better suited for the small quantities of water, such as the levels typically found in lyophilized finished products. Direct Karl Fischer titration is the most commonly used method; however, it may not be the most appropriate method for all products. Both the volumetric and coulometric titration methods use the same principle with respect to the chemical reaction to determine the water content.

The method is based on the Bunsen Reaction between iodine and water in the presence of excess sulfur dioxide. A chemist, Karl Fischer, discovered the reaction could be employed for moisture determination. To perform the titration, an alcohol, commonly methanol, is used as the solvent; pyridine acts as a base. The reaction is stoichiometric, where iodine and water are consumed at a 1:1 ratio.

Karl Fischer titration is based on a two-step reaction represented by Equation 1:

Equation 1

The sulfur dioxide and the base react with the alcohol to form an intermediate alkylsulfite salt. The alkylsulfite salt is oxidized by the iodine and forms an alkylsulfate salt. During the oxidation, product residual water is consumed. The presence of ions during the reaction is sensed by a dual conductivity probe and indicates the endpoint of the reagent titration. To determine the amount of water present in the product, the instrument performs a calculation based on the amount of iodine consumed during the titration (7).

In volumetric titration, the iodine is added from the instrument’s burette to the solvent in the reaction cell containing the sample. The amount of the water present is quantified and calculated based on the volume of reagent consumed in the reaction with water and the reagent concentration. In coulometric titration, the iodine is produced in-situ electrochemically during the titration. The reaction cell containing the reagents houses an anode and cathode as shown in Figure 4. Iodine ions are generated at the anode by electrochemical oxidation. These iodine ions are formed when the negative iodide ions release electrons and subsequently react with water (Equation 2). The cathode allows for the completion of the electrochemical reaction, by a reduction reaction (Equation 3). Positive hydrogen ions are reduced to hydrogen.

Figure 4: Example of a reaction cell. (Figures courtesy of the authors)

Equation 2

Equation 3

The amount of iodine ions generated for the reaction with the water present is quantified based on the total charge required in generating the ions, expressed in the equation as Q, where the charge generated is measured by current (amperes) and time (seconds), as per Equation 4:

Equation 4

Two main types of titration systems are used to perform Karl Fischer coulometric method: a fritted cell or fritless cell. In a fritted-cell system, the anode and cathode that form the electrolytic cell are separated by a semi-permeable membrane called a diaphragm or frit (Figure 5). The frit prevents the generated iodine at the anode from being reduced back to iodide at the cathode. Two different reagents, an anolyte and catholyte, are required.

Figure 5: Fritted cell. (Figures courtesy of authors)

In a fritless cell, only one reagent containing all the necessary components is used, and through the design of the cell, the iodine will not react at the cathode and be reduced back to iodide (Figure 6). In both systems, the cathode and anode are platinum electrodes, which induce a current through the cell.

Figure 6: Fritless cell. (Figures courtesy of the authors)

Both types of reaction cells are appropriate for testing lyophilized product. The fritted cell is preferred when the sample being tested is known to interact with constituents within the reagents, and therefore, two different Karl Fischer reagents are required.  Reactions with the Karl Fischer reagents are discussed in more detail in the “Points to Consider” section. The fritless cell requires only one reagent and is easier to maintain.

Over the years, improvements have been made to the Karl Fisher reagents to remove pyridine, a noxious carcinogen. Currently, the reagents generally contain imidazole or other primary amines.

Sample preparation and introduction into the reaction cell may entail using an anhydrous solvent, such as methanol, to dissolve the dried product or extract the water when a suspension forms. An alternate method employs evaporation: The sample is heated and a dried gas such as nitrogen is used as a carrier, passing the gas over the heated sample then percolating the carrier gas through the reagent in the reaction cell.

The instrument should be routinely inspected for effective seals of the reaction cell and cleanliness of the sensing electrode, anode, and cathode. The instrument does not require calibration; however, it is appropriate to test a known water standard to verify the instrument is functioning properly prior to use. The instrument should be qualified, proving the instrument will perform as expected.

Routine preventative maintenance is needed to maintain the instrument in prime operating condition. At a minimum, the reagents should be drained every three months and the reaction cell and electrodes cleaned as per the manufacturer’s recommendations. All parts of the instrument including the electrodes, tubing, and ports should be inspected and replaced as needed.

A suitable method should be developed to yield confidence in the accuracy and precision across the range of possible moisture levels for a material. A significant amount of time and effort is often required to develop a specific water extraction and titration technique including selection and quantity of solvent, the soak time for water extraction, and sample injection volume. When performing an oven method, heating temperature and carrier gas flow rate variables must be explored and established. Each variable to consider in developing the method, as described later in this presentation, can significantly affect the results and needs to be justified in detail.

Points to consider for Karl Fischer method. Because of the sensitivity and resolution that can be obtained, Karl Fischer coulometric titration is the preferred method for lyophilized product moisture determination. There are many advantages, the major being specificity for water and low level of detection. However, this method is time-consuming, sample limited, and destructive. Much effort is required to establish and validate the method for variables such as the appropriate extraction solvent, extraction time, sample size, and injection volume.

Solvents and reagents. Selecting the appropriate solvent for extraction of the water from the sample is crucial. When using a solvent extraction method, the product should be either soluble in the solvent or the solvent should be able to adequately extract the water. The most common solvent used is anhydrous methanol (9). If the product is not soluble in methanol, or there is insufficient water extraction, other solvents-including anhydrous dimethyl sulfoxide (DMSO), formamide, or combinations of methanol and formamide may be used.

As an evaporative technique, an oven attached to the Karl Fischer instrument may be suitable for products that are not soluble in the solvent or interact negatively with the reagents. In this technique, the product is heated to release the water, and then water vapor is carried to the reaction cell by a dry inert gas, such as nitrogen. The temperature required to ensure all the moisture is released should be determined during method development and should not be so high as to cause the sample to decompose. The flow rate of the dry inert gas must be evaluated and be adequate to ensure all released water is transferred to the reaction cell. If the flow rate is too high, detecting the endpoint can be difficult.

A Karl Fischer instrument equipped with an oven typically contains a well to hold the sample for testing. The well size must be appropriate for the size of the vial; if the well is too large for the vial, then heating may not be even or complete. The sample can be transferred to an appropriately sized container if the correct well size is not available. However, the transfer may interfere with the accuracy and precision of the results if the material is exposed to the atmosphere during sample handling, causing the dried material to absorb moisture.

Interactions between the Karl Fischer reagents, components, and the product can occur; aldehydes, ketones, and thiols (mercaptan) are known to be problematic. Side reactions can lead to false results that are higher or lower than the actual residual moisture. Aldehydes and ketones will react with methanol in the presence of an acid, undergoing condensation reactions, yielding water as a byproduct. Aldehydes may also react with ingredients in the Karl Fischer reagents and cause a reaction that consumes water. It is also important that the product does not interact with iodine (9). Thiols (mercaptan) will react with the free iodine in the Karl Fischer reagent and consume the iodine. It is important, therefore, to know and understand the sample chemistry to ensure side reactions do not lead to inaccurate results.

Some products will not be readily soluble in methanol and will require evaluation steps to determine how much solvent to add, how long to let the sample soak to ensure all of the water has been extracted, and if sample agitation is required. To approximate the amount of methanol or other solvent to add, use the original fill volume and then adjust the volume based upon measured water and sufficient extraction quantity.

An alternate method is to inject solid particles instead of the solvent solution as the supernatant. A best practice, however, is to add only a solution to the reaction cell to minimize the amount of un-dissolved material injected; such material may coat the electrodes and interfere with the electrode function.

Preferably, the pH of the solvent should be between 5 and 8, as the rate of the reaction depends on the pH (7). At a pH below 5, the reaction will occur too slowly; at a pH above 8, the reaction will occur too fast (7). In both circumstances extreme reaction rates make endpoint detection difficult. Highly acidic or highly basic samples may, therefore, need to be buffered (7).

Blank samples of methanol or the solvent used for extraction should be tested during sample preparation and handled in the same manner as the test samples. The blanks allow for measuring the amount of water inherent in the solvent used for extraction and any contribution to the results by the sample preparation and handling technique. This water is then subtracted from the water measured during testing so that the results accurately reflect the water in the sample.

Care should be taken to prevent moisture from contaminating the extraction solvent. As well, the reaction cell should be protected and fitted appropriately to prevent atmospheric moisture ingress. In particular, a large number of penetrations of the septum at the injection port may allow air and the accompanying moisture to infiltrate the reaction cell. To ensure the solvent used for testing is sufficiently dry, a desiccant may be added to the bottle to remove any moisture uptake. Any extraneous addition of moisture into the reaction cell may manifest in erroneously high moisture results or difficulty in detecting a clear titration endpoint. In addition, an injection of the solvent “neat” should be titrated to ensure the moisture in the solvent is not high before any samples are prepared.

Sample injection. Due to variations in the reaction cell configurations of Karl Fischer titrators from different manufacturers, the injection port can either be above or below the level of the Karl Fischer reagents. If the injection port is above the level of the reagents, the needle must be long enough to reach below the level of the reagents, once inside the cell to ensure all of the sample will be added to the Karl Fischer reagents. This will ensure the sample will not adhere to the side of Karl Fischer reaction cell. However, this may require the use of an extremely long needle that can then pose safety challenges. It is not uncommon to use needles as long as 3.5 inches to ensure the sample can be eluted below the level of the Karl Fischer reagents. Extreme care must be taken when working with such needles to avoid bending or breaking.

The injection volume is also a variable that should be considered during method development. A typical injection volume is 1 mL; however, if the product has a fill volume of 1 mL or lower, then a smaller injection volume, or pooling the extraction solution from multiple samples may be warranted. Pooling samples entails combining sample from two or three containers to inject into the instrument. For example, a product with an original fill volume of 0.5 mL and poor solubility in methanol may require up to three vials to obtain a sufficient test sample size to equal or exceed the level of quantification.

Near-infrared spectroscopy (NIR) method. NIR is a particularly attractive method as it is rapid, noninvasive, and nondestructive. This method is also safer than other methods as it does not require any reagents or exposure to the product (10). The typical scan time for a sample is less than one minute, which is much faster than the LOD, TGA, or Karl Fischer methods (11).

In spectroscopy, light is sent into or through the sample; the light that is not absorbed can be separated spatially into the different wavelengths; a sensor then determines the amount of light absorbed at each wavelength (12). NIR is considered a vibrational spectroscopy technique meaning the light interacts with molecular vibrations (12). The absorbance peaks are related to the different types of bonds present in overtones and combinations. The types of bonds that have a significant absorption in the NIR region are O-H, N-H, C-H, S-H. Focusing on the absorbance region specifically for water, the amount of light energy absorbed at the selective wavelength correlates to the quantity of water present; the relative water present will correlate to the intensity of absorption peaks (12).

NIR instrumentation can be run in transmittance or diffuse reflectance modes (13). The diffuse reflectance mode is used for solids such as lyophilized products. NIR can measure trace amounts of moisture due to the strong overtone absorption band for water at 1940 nm and 1450 nm (11). The measurement may be taken through a vial when the diffuse reflectance mode is used.  Little to no sample preparation or reagents are needed to perform the moisture measurement.

A correlation between the peak intensity and actual moisture content needs to be established for each material to be tested. Calibration samples with a known residual moisture content of the lyophilized product are required for the diffuse reflectance mode obtained. The calibration samples need to incorporate samples with higher and lower moisture content, in addition to the normal range for the lyophilized product. Moisture content for these samples needs to be determined by a primary method such as Karl Fischer or LOD. Ensuring the calibration samples are representative of the variation seen in typical production lots will ensure quantitative results are obtained. From the calibration samples, a calibration model is developed using a chemometric algorithm, such as multiple linear least-squares (MLR), partial least-squares (PLS), or principal component regression (PCR) (13). Choosing the type of chemometric algorithm to use will depend on the NIR spectra (10). The residual moisture can be easily obtained once the calibration samples and model are established.

An advantage of the NIR method is that many samples can be analyzed non-destructively. An entire batch of processed material can be tested in their primary packaging with no sample preparation or manipulation required. NIR can be used as a benchtop test where only representative samples of a batch are tested or integrated into manufacturing for measuring the moisture in each container within the batch.

Points to consider for NIR. NIR is rapid, non-invasive, and non-destructive. A major disadvantage, however, is it requires calibration samples with a known residual moisture content of the lyophilized product obtained by another validated method such as Karl Fischer titration, adding time and costs to the process.

The calibration samples must incorporate samples with higher and lower moisture content in addition to the normal range expected for the lyophilized product. A large number of calibration samples are required to develop a calibration curve. Generally, a calibration set must contain between 20 to 50 samples at different residual moisture levels to be statistically significant.

Preparing samples with higher moisture content can be challenging. One method is to inject water into the sample headspace and allow the water to equilibrate throughout the sample (11). The additional water will be absorbed by the lyophilized material, therefore increasing the moisture content. To create samples with lower moisture content, the material is exposed to extended secondary drying conditions during lyophilization (11). Once all the calibration samples have been created and tested, statistical analysis is used to correlate the results of the NIR method and the validated residual moisture test.

Because the NIR method is non-destructive, the samples can be tested a second time using another method (11). For example, the sample could be tested by NIR for residual moisture and then be tested by another residual moisture technique to verify the residual moisture results or for potency or assay. The ability to test the same sample by different analytical tests is very attractive when there are a limited number of samples or when the material being testing is extremely expensive.

Compendia Overview

The United States Pharmacopeia (USP), European Pharmacopoeia (Ph. Eur.), and Japanese Pharmacopoeia (JP) contain varying guidance on LOD, TGA, and Karl Fischer testing. This section compares the three different compendia, highlighting similarities and differences in methodology and guidance.

LOD. LOD is addressed in the following compendia:

• USP <731> Loss on Drying (14)

• Ph. Eur. Section 2.2.32. Loss on Drying (15)

• JP Section 2.41 Loss on Drying Test (16).

All three compendia briefly define “Loss on Drying.” According to the USP, LOD “determines the amount of all volatile matter of any kind driven off under the conditions specified” (14). In the Ph. Eur., LOD is stated as the loss of mass (15). In the JP, LOD is a method to “measure the loss in mass when dried under the conditions specified” (16). The JP also states this method can determine the “amount of water, all or part of the water of crystallization, or volatile matter in the sample, which is removed during the drying” (16). The important note for this method is “volatile matter;” not only water can be driven off by using this method, but other volatile components present in the product.

The method presented in all three compendia is similar, but there are some key differences. All three compendia agree the sample to be tested should be accurately weighed into a sample container that has been previously dried under the same test conditions. The USP and JP specify the sample container is to be dried for 30 minutes before testing, while the Ph. Eur. does not state a time nor directs a proper condition.

General test conditions described in the USP are highlighted in Table II.

The Ph. Eur. does not give as much guidance on sample preparation as the USP and JP for sample size, particle size, or sample depth. The USP states the test should be conducted on 1–2 g of material unless otherwise directed in the monograph. The JP further states to weigh within a range of ± 10% of the recommended amount in the monograph. The USP and JP specify the particle size of the sample should be not more than 2 mm and to reduce the particle size by quickly crushing if the particle size is greater than 2 mm. The USP and JP give guidance on the depth of the sample in the sample container: USP recommends a depth of approximately 5 mm and no more than 10 mm, and the JP recommends a depth of no thicker than 5 mm. According to the USP and JP, if the sample melts below the temperature specified in the general method, the sample should be exposed to a temperature 5–10 °C below the melting temperature for 1­–2 hours and then dried according to the conditions specified in the monograph.

The Ph. Eur. lists five methods for performing the test, which are as follows: “Where the drying temperature is indicated by a single value rather than a range, drying is carried out at the prescribed temperature ± 2 °C”:

• “in a desiccator: the drying is carried out over diphosphorus pentoxide R, at atmospheric pressure and at room temperature;

• “in vacuo: the drying is carried out over diphosphorus pentoxide R, at a pressure of 1.5 kPa at room temperature;

• “in vacuo within a specified temperature: the drying is carried out over diphosphorus pentoxide R, at a pressure of 1.5 kPa to 2.5 kPa within the temperature range prescribed in the monograph;

• “in an oven within a specified temperature range: the drying is carried out in an oven within the temperature range prescribed in the monograph;

• “under high vacuum: the drying is carried out over diphosphorus pentoxide R at a pressure not exceeding 0.1 kPa, at the temperature prescribed in the monograph” (15).

The USP provides some guidance on different procedures to follow, but not to the extent of the Ph. Eur. The JP provides limited general guidance on procedures to follow.

Thermogravimetric analysis. The compendia address thermogravimetric analysis in the following:

• USP <891> Thermal Analysis (5)

• Ph. Eur. Section 2.2.34. Thermal Analysis (17)

• JP Section 2.52 Thermal Analysis (6).

All three compendia agree that thermogravimetric analysis is a method in which the mass of the sample is recorded as a function of temperature, or time of heating, or both according to a controlled temperature program. The JP states this method can be used as an alternative method to a LOD or Karl Fischer method if it has been established that the material being tested contains only water and no other volatile substances.

The instrument is described as being comprised of a balance, a recorder, and a programmable controller for controlled heating. Calibration of the scale and balance are required per USP and JP. The Ph. Eur. requires verification of the balance by the mass loss of a certified reference material and temperature. The USP states the balance is to be calibrated using standard weights. Calibration of the thermocouples can be against a temperature reference or by the use of standard reference materials. When compared with standard reference materials, it is assumed the temperature of the reference material is the furnace temperature. The JP requires the temperature to be calibrated by the Curie temperature of a ferromagnetic substance. In the JP, there are two types of balance calibration, primary balance, and secondary balance calibration. The primary balance calibration is performed by using standard weights. The secondary balance calibration or confirmation is performed by using a calcium oxalate monohydrate reference standard. This type of calibration takes into account the effects of buoyancy and convection due to atmospheric gas flow in the real measurement. If the reference standard has a certified water content, and the results are within 0.3% of the certified reference standard, then normal operation is confirmed. If greater than 0.3%, then calibration must be performed with the reference standard using the procedure described in the guidance.

The Ph. Eur. gives guidance on verification of temperature and balance. The verification of the temperature is performed by using suitable standard reference material according to the manufacturer’s instructions. Verification of the balance is performed by using a reference standard with a known mass loss. All three compendia agree that the test atmosphere is critical and needs to be recorded. However, use of a specific inert carrier gas and flow rate are not specified in any of the compendia.

Coulometric Karl Fischer Titration. Coulometric Karl Fischer titration is discussed in the following sections of the compendia:

• USP <921> Water Determination, Methods Ia and Ic (Coulometric Titration) (18)

• Ph. Eur. Section 2.5.32. Water: Micro Determination (19)

• JP Section 2.48 Water Determination (Karl Fischer Method) (20).

All three compendia begin by briefly explaining the Karl Fischer reaction and general theory. Each compendia also describes the appropriate instrument in varying degrees of detail. Per the USP, calibration of the Karl Fischer instrument is not necessary. All agree, and state atmospheric moisture should be excluded from the system. The USP and Ph. Eur. emphasize the precision of the method is predominantly governed by how well atmospheric moisture is excluded from the system. The USP refers to coulometric titration as a micro-method for the determination of water. The Ph. Eur. states coulometric titration is restricted to the quantitative determination of water having a range of 10 µg to 10 mg of water.

Both the USP and Ph. Eur. highlight specific requirements for the method while the JP does not. These requirements are that each component of the test specimen is compatible with the other components, no side reactions or other reactions take place, and the volume and water capacity of the electrolyte reagent are sufficient (18, 19).

The methods described are generally similar for all three compendia, but there are some significant differences. The USP and Ph. Eur. both explain that adding solids directly to the reaction cell should be avoided and only performed if absolutely necessary. In the JP, the procedure is to add the sample specimen estimated to contain 0.2 mg to 5 mg of water directly to the anolyte to perform the titration.

According to the Ph. Eur., the accuracy of the test should be verified between every two successive sample titrations. This is performed by introducing an accurately weighed amount of water in the same order of magnitude as the amount of water in the sample being tested. Neither the USP nor JP requires such verification of the accuracy during testing.

Conclusion

Residual moisture is considered a CQA for a lyophilized preparation; sufficient accuracy and precision to measuring low residual moisture levels in a lyophilized product are essential.  Different methods are available to determine the residual moisture content of a lyophilized product; choosing and developing an appropriate method can be a challenging task. Once the most appropriate method has been selected, great effort and attention must be employed to develop and validate a specific method with the desired accuracy and precision.

The principles, advantages, and limitations of a method, as well as the quantity of water to be measured and the characteristics of the material to be tested should be considered when selecting a method. The guidance embodied within the applicable compendia should be reviewed. The influencing factors need to be explored, and the suitable variables determined when developing a method. When well understood and properly developed, the method for a lyophilized product will be able to provide the residual moisture content with suitable accuracy and precision for a high level of confidence in the measured results.

References

1. ICH, Q6A Specifications: Test Procedures and Acceptance Criteria for New Drug Substances and New Drug Products: Chemical Substances (ICH, October 1999).

2. FDA, Docket No. 89D-0140, Guideline for the Determination of Residual Moisture in Dried Biological Products (Center for Biologics Evaluation and Research, January 1990).

3. USP, <1241> Water-Solid Interactions in Pharmaceutical Systems, USP40–NF35, 2017.

4. TA Instrument, Universal Analysis 2000, Version 4.5A.

5. USP, <891> Thermal Analysis. USP40–NF35, 2017.

6. JP, 2.52 Thermal Analysis, Japanese Pharmacopoeia 16^th Edition, 2011.

7. EMD Chemicals, Inc., Aquastar, Karl Fischer Titration Basics.

8. L. Zhou, PhD, et al., American Pharmaceutical Review, 13(1):74-84, (January 2010).

9. Mitsubishi, Mitsubishi Chemical Corporation Karl Fischer Reagents Technical Manual (Mitsubishi, 2000). 

10. T.P. Lin and C. C. Hsu, PDA Journal of Pharmaceutical Science and Technology, 56 (4) (July/August 2002).

11. FOSS NIRSystems, Inc. “Analysis of Residual Moisture in a Lyophilized Pharmaceutical Product by Near-Infrared Spectroscopy” PH-AN-704 08/06.

12. Metrohm USA, “Moisture Analysis in Pharma: Expanding Near-Infrared Spectroscopy with Karl Fischer Titration,” Webinar, 2015, www.labroots.com/webinar/moisture-analysis-pharma-expanding-near-infrared-spectroscopy.

13. FOSS “A Guide to Near-Infrared Spectroscopic Analysis of Industrial Manufacturing Processes.”

14. USP <731> Loss on Drying, USP40–NF35, 2017.

15. EDQM, 2.2.32. Loss on Drying, European Pharmacopoeia 7.0, 2010.

16. JP, 2.41 Loss on Drying Test, Japanese Pharmacopoeia 16^th Edition, 2011.

17. EDQM, 2.2.34. Thermal Analysis, European Pharmacopoeia 7.0, 2010.

18. USP, <921> Water Determination, USP40–NF35, 2017.

19. EDQM, 2.5.32. Water: Micro Determination, European Pharmacopoeia 7.0, 2010.

20. JP, 2.48 Water Determination (Karl Fischer Method), Japanese Pharmacopoeia 16^th Edition, 2011.