The FDA regulates the presence of impurities in APIs, formulated drug products, food ingredients and cosmetics, and sets thresholds at which these impurities must be identified or adequately tested in safety and clinical studies. Investigation of these impurities must be initiated during the early stages of drug or product development, such as the pre-investigational new drug (pre-IND) stage.

Modern mass spectrometry (MS) and nuclear magnetic resonance (NMR) techniques can have a tremendous impact on the speed and sample requirements for structure elucidation of trace impurities or metabolites. A rapid protocol used for the isolation and characterization of trace impurities in drug substances and formulated drug products has been developed. The approach has been successfully employed for the characterization of several hundred impurities from more than 130 different substances covering a wide variety of structural classes.

The sources of impurities vary greatly. These could include starting materials, intermediates, by-products of the synthesis of an API, or degradation products of the API or its impurities arising during manufacture or storage. Impurities may also be present that are not related to the API, arising from the synthetic or extractive process or as a result of contamination from unrelated chemicals. The stability of APIs is determined using short- and long-term stability studies with resulting degradants often requiring identification. In addition, new impurities can suddenly appear during drug or product development due to changes in the synthetic protocol, starting materials, source of starting materials, or even variability as processes are scaled up.

The International Conference on Harmonization (ICH) sets standards for the purity of drug substances¹ and drug products.² These guidelines set levels at which impurities must be reported, identified, or qualified and vary dependant on the dosage. For example, for drug substances to be given at levels of less than two grams per day, the guidance states that impurities between 0.10% and 0.15% should be identified and those that reach 0.05% must be reported. Impurities present at levels greater than or equal to 0.15% must be evaluated according to the ICH standards.

Similar regulations provide specifications for food and cosmetic ingredients, which set thresholds at which impurities need to be reported, identified, and qualified. The early isolation and identification of impurities often allows improvements or modifications in the synthetic pathway or purification process, which can prevent the formation of an impurity or reduce it to sub-threshold levels.

The Impurity ID Isolation and Characterization Process

Our rapid protocol used for the isolation and characterization of trace impurities in drug substances and formulated drug products uses a combination of spectrometric and spectroscopic techniques to analyze impurities during and after isolation, which minimizes the total analysis time. The application of capillary NMR facilitates this process by reducing the amount of an impurity that must be isolated to allow acquisition of NMR data, which is traditionally the technique requiring the most sample.

Capillary NMR is a powerful tool for structural elucidation due to the ease of use and the fact that only approximately 20 to 50 micrograms of an impurity need to be isolated depending on the set of NMR experiments required. This need to isolate only minute amounts for MS and NMR analysis accelerates the isolation and structure elucidation process. The reduced sample requirements of capillary NMR are also advantageous in cases where the amount of an impurity that can be isolated is limited due to limitations of the starting material. The following impurity identification protocol has been applied in the isolation and characterization of a large number of impurities in drug substances and products as well as impurities in food ingredients and of drug metabolites. This capability also logically extends to the analysis of Schedule I to Schedule V controlled substances and potent compounds.

An outline of the impurity characterization workflow is provided in Figure 1. API or drug product is provided containing the impurity or impurities of interest. Typically anywhere from one to five impurities are present in the sample at the 0.01% to 0.5% (w/w) level. An analytical liquid chromatography (LC) method is also commonly available and, if amenable, this analytical method is used to acquire MS data by LC/MS and to follow the scaled-up isolation of the impurities. The MS data can often provide the molecular weight of the impurity, which is typically the first piece of structural information obtained. The use of non-volatile mobile phase additives, such as ion-pairing agents or phosphate buffers, which was common before the routine availability of LC/MS, precludes the in-line acquisition of MS data.

If the analytical LC method is not compatible with MS, either an alternate LC/MS compatible analytical method may be developed or the acquisition of MS data will be undertaken after isolation of the impurity. Tracking the isolation by LC/MS allows confirmation that the correct compound has been isolated and that the impurity is stable. Many impurities are found to be unstable during isolation and require special handling during isolation and concentration, such as limiting exposure to light or moisture, or removal of acidic mobile phase modifiers. In extreme cases it may be necessary to isolate and characterize the degradant of an impurity, which can then allow the structure of the original impurity to be inferred.

To facilitate isolation of the target impurity, the analytical method may be scaled up. Method development or modification is often necessary at this stage if the analytical method is not easily applicable at the larger scale. This typically involves the substitution of non-volatile mobile phase additives, an optimization of the separation between the impurity of interest and other components in the sample, and/or a reduction of the total separation time.

Formulated drug products also often require additional workup at initial stages to separate the API and related substances from excipients present in the formulation. This typically involves dissolving or suspending solid formulations in a suitable solvent, followed by filtration or centrifugation to remove insoluble excipients, which may be followed by one or more solvent or solid phase extraction steps.

Once a suitable LC method has been determined, the next step is to enrich the impurities by depleting the sample of the major component followed by additional purification of the individual impurities as needed. One process used is reverse or normal phase high-pressure liquid chromatography (HPLC) at the semi-preparative or preparative scale for isolation. Once an isolation method has been optimized, the purification of the impurity can be automated using a Waters Autopurification system, which allows peak collection triggered by mass, ultraviolet, or evaporative light scattering (ELS) detection.

The typical structural characterization methods include LC/MS, LC with tandem MS detection (LC/MS/MS), high resolution mass spectrometry (HRMS), and 1D and 2D NMR experiments using either a capillary or standard 5 millimeter gradient NMR probe. MS data can often be acquired during the isolation procedure on the starting material or partially purified fractions, which speeds up the overall process. Preliminary NMR data is used to monitor purity with a
full NMR dataset being acquired once a sample of sufficient purity and quantity has been obtained. In cases of unstable or volatile compounds LC/NMR can be the method of choice. It is an important tool to generate NMR data in-line after separation of mixtures, but its drawbacks include limited LC methods available due to the need to use deuterated solvents, as well as the potential overlap or masking of critical signals in the NMR spectrum with the mobile phase components.

Initial characterization by LC/MS facilitates the isolation of the selected impurities, as well as provides preliminary structural information. Ionization can be achieved for most substances using electrospray (ESI) or atmospheric pressure chemical ionization (APCI). An example is provided by the recently completed characterization of five impurities (See Figure 2, Compounds 2-6) of AMRI’s ALB109654(a), Compound 1, a novel tubulin inhibitor under development.^3,4 LC and MS analysis provided the molecular weight for each component. Based on the MS data, Compound 2 was tentatively assigned as vinblastine, which is a starting material in the synthesis, and Compound 3 was determined to be an isomer of the parent Compound 1.

Acquisition of HRMS data can allow a determination of the molecular formula in most cases or at a minimum narrow the list of possibilities. In Figure 2, for example, the molecular weight and isotope pattern observed for Compound 4 by LC/MS suggested substitution with bromine, which was confirmed by accurate mass determination. Similarly, Compound 5 was found to result from the substitution with iodine, and Compound 6 was found to result from loss of a methyl group. Fragmentation data obtained from LC/MS/MS experiments often yields information on the region of the molecule that has been changed and is crucial for structure identification. A comparison of the fragmentation pattern of the parent compound with that of Compounds 4-6 indicated that each of these impurities was modified in the lower indoline moiety.

In less complex structures, MS and MS/MS data may allow a structure to be proposed without the need to isolate material for NMR analysis. In more complex structures, such as the example described here, NMR data is also required to complete or confirm the structural assignment. The structure of Compound 2 was confirmed by a comparison of the 1H NMR spectrum with that of an authentic sample of vinblastine.

The use of a capillary NMR probe required isolation of less than 50 µg of pure compound that was obtained by semi-preparative HPLC in a single run from 8 mg of substrate. Similarly, capillary NMR analysis of Compound 3 allowed its structure to be determined as the positional isomer of the tubulin inhibitor, Compound 1, which has the SCH3 substituent at C-13’. NMR data was also used to determine that both Compounds 4 and 5 had the halogen substituent at
C-17. Depending on the impurity, NMR datasets can include 1D 1H NMR, 13C NMR, and DEPT spectra as well as 2D 1H-1H COSY, TOCSY, HSQC, HMBC, NOESY or ROESY, and HSQC-TOCSY spectra. MS and NMR data of the drug substance, starting materials, and intermediates are also acquired during the process to provide a set of comparison data for analysis of the structure of the impurity.

Degradants arising from the manufacture or storage of a drug are characterized in the same manner. Compound 6 was first observed during an accelerated stability study of the parent compound. Subsequent isolation and characterization indicated that this degradant arises by hydrolysis of the ester at C-3. Once the structures of the impurities have been determined it is often possible to suggest plausible routes of formation relating the impurities to each other and
the substance under study. This information can then be used to alter the synthetic (or storage) conditions to reduce their accumulation.

Most of the impurities we have identified were present in drug substances or drug products prepared by synthetic processes or in some cases by fermentation. However, our protocol is also applicable within the food and beverage industry. For example, it is well known that the major constituents of the leaves of S. rebaudiana are the remarkably high-potency, low-calorie sweeteners stevioside, rebaudiosides A and C, and dulcoside A; glycosides of the diterpene steviol, ent-13-hydroxykaur-16-en-19-oic acid.⁶ Extracts of the leaves of S. rebaudiana have been used for centuries to sweeten food and beverages in Japan, South America, and China.⁷ Purification of the crude extract obtained from the leaves of S. rebaudiana resulted in the isolation of two new minor diterpene glycosides, along with the known steviol glycosides stevioside, rebaudiosides A-F, and dulcoside A. During another study,⁸ the viability of the sweetener monatin⁹ was tested in a model lemon-lime beverage system. A total of seven compounds were identified as major degradation products of monatin, which included two novel structures.

In addition to the isolation and characterization of impurities, our protocol is applicable to characterization of drug metabolites. A number of metabolites have been isolated and identified from mixtures generated by either human or rat liver microsome preparations, or isolated from plasma samples or urine.¹⁰ Unlike other sources, the amount of a metabolite preparation or extractable biological fluid is often limited, requiring isolation and identification of metabolites, which can only be isolated in µg quantities.

In conclusion, modern MS, NMR, and separation techniques can have a tremendous impact on the identification and structure elucidation of trace impurities or metabolites. We have developed and employed an impurity ID protocol to characterize several hundred impurities from more than 130 different substances covering a wide variety of structural classes. Our impurity ID protocol has proven useful in the isolation and characterization of a large number of impurities in drug substances and drug products as well as impurities in food ingredients and drug metabolites.

With continuing advances in separation and spectroscopic techniques, modern technology has allowed a formerly arduous and difficult process to become easier to achieve in a relatively straightforward and rapid series of steps.


References

1. International Conference on Harmonization (ICH), Guidelines for Industry, Q3A, Impurities in New Drug Substances (Revision 2), June 2008.
2. International Conference on Harmonization (ICH), Guidelines for Industry, Q3B, Impurities in New Drug Product (Revision 2), July 2006.
3. Voss ME, Ralph JM, Xie D, Manning DD, Chen X, Frank AJ, Leyhane .J, Liu L, Stevens JM, Budde C, Surman MD, Friedrich T, Peace D, Scott .L, Wolf M, Johnson R, Bioorg. Med. Chem. Lett. 2009; 19: 1245-1249.
4. Milanowski DJ, Keilman, J, Guo, C, Mocek, U, J. Pharm. Biomed. Anal., 2011; 55: 366-372.
5. Chaturvedula SP, Rhea J, Milanowski D, Mocek U, Prakash I, Nat. Prod. Comm., 2011; 6: 175-178.
6. Brandle JE, Starrratt AN, Gijen M, Can. J. Plant Sciences, 1998; 78: 527-536.
7. Genus JM. Phytochemistry, 2003; 64: 913-921.
8. Upreti M, Somayajula KV, Milanowski DJ, Kowalenko P, Mocek U, San Miguel R, Prakash I, Food Chemistry, 2012; 131: 413-421.
9. Vleggaar R, Ackerman, LG J, Steyn PS, J. Chem. Soc., Perkin Transactions 1: Organic and Bio-Organic Chemistry (1972-1999), 1992; 22: 3095-3098.
10. Unpublished results.

Dennis J. Milanowski is senior scientist II, Structural Analysis at Albany Molecular Research, Inc., Bothell Research Center. He can be reached at dennis.milanowski@amriglobal.com. Ulla Mocek is section head, Structural Analysis at the same AMRI center. She can be reached at ulla.mocek@amriglobal.com.