Analytical Testing and Control for Genotoxic Impurities in Drug Substances

The heightened awareness and need for control of genotoxic impurities (GTIs) present increasing challenges and concerns during drug development Their toxicity at trace levels requires accurate analysis of these impurities at the parts-per-million level or lower, posing significant challenges to analytical methodologies For both the pharmaceutical industry and the health authorities alike, the characterization and control of GTIs to levels considered acceptable for patient safety is an evolving and complicated process Drug companies strive to find ways to avoid the introduction of GTIs in drug substances However, the intrinsic nature of synthetic chemistry renders this avoidance an insurmountable task due to the necessary use of electrophilic agents for carbon-carbon and carbon-nitrogen bond formation and due to formation from side reactions Therefore, as a common practice, known and potential GTIs are strategically controlled to meet safety requirements by regulatory authorities One essential part of this strategic effort is the development of sensitive, specific, and robust analytical methodologies to enable the understanding of these impurities, including their origins of generation and fate through the process leading to the final drug substances Chromatography combined with conventional detection has generally been the method of choice to achieve specificity and robustness with acceptable sensitivity The advancement in new analytical instrumentation and technologies has pushed sensitivity for GTIs to levels unattainable by conventional methods This chapter intends to provide an overview of GTIs from regulatory, toxicology, and analytical testing and control perspectives by combining the authors’ own experience with approaches shared by the pharmaceutical industry

In an ideal world, drugs would contain no impurities since the impurities are generally of no therapeutic value To minimize impurities, steps should be taken during drug development to avoid, control, or remove potential impurities Impurities in drugs may originate from many sources, including raw materials, reagents, intermediates, synthetic by-products, leachables, and extractables due to incompatibilities with excipients and packaging, and degradation products The control and toxicological qualification of impurities has been well understood for many years and harmonized internationally (International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use [ICH] Q3 guidelines)1 In these guidelines, the concept that some impurities “are unusually potent” or “produce toxic or unexpected pharmacological effects” is highlighted; however, the concept of genotoxicity is not specifically addressed

Genotoxic compounds are generally reactive with or intercalate into DNA, leading to cancer or heritable changes in patients, with carcinogenicity being more easily detected Genotoxicity includes induced genetic damage and fixation, gene mutation, larger scale chromosomal damage, and recombination and numerical chromosome changes Genotoxicity is characterized by a standard test battery (ICH S2)2 Guidelines to specifically address GTIs were first introduced in 2002 by the European Medicines Agency (EMA)3 This initial guideline acknowledged the gaps

in ICH Q31 and proposed a default threshold of toxicological concern (TTC) for GTIs to minimize patient risk as follows: “A TTC value of 15 μg/day intake of a GTI is considered to be associated with an acceptable risk (excess cancer risk of <1 in 100,000 over a lifetime) for most pharmaceuticals From this threshold value, a permitted level in the active substance can be calculated based on the expected daily dose Higher limits may be justified under certain conditions such as shortterm exposure periods” The TTC was based on a lifetime exposure; to define a strategy for shorter term exposures, the concept of a “staged TTC” was developed by a Pharmaceutical Research and Manufacturers of America working group and published by Mueller and others4 The staged TTC approach has been recognized and incorporated into both Food and Drug Administration (FDA) (draft)5 and EMA6 guidances An internationally harmonized guidance is in preparation7 The current guidances were compared and summarized8 Although there are some differences between the guidances, the staged TTC concepts are similar, as shown in Table 111

The staged TTC concept allows analytical method development to evolve along with a clinical drug candidate as the drug advances from early phase development to commercialization Ultimately, the analytical impact of staged TCC means that dosedependent analysis potentially to low parts-per-million levels is typically needed for the drug substance (eg, for long-term dosing of a 500 mg/day drug product, limits for GTIs would be 30 μg/g) These limits are approximately 1000 times lower than those for a nongenotoxic impurity, as described in ICH Q31 Many GTIs are intrinsically reactive, and process chemists often do not expect them to persist through a multistep synthesis Therefore, regulatory specifications may not always be needed for these products; however, methods capable of demonstrating the purging power of the chemical process to these low levels are needed to ensure patient safety Several recent articles have provided overviews of control strategies in active pharmaceutical ingredient (API) development and analytical considerations in developing methods for GTIs9-12 ICH guideline M7, Assessment and Control of DNA Reactive (Mutagenic) Impurities in Pharmaceuticals to Limit Potential Carcinogenic Risk, is under development as one way to recognize the existence of mechanisms leading to a dose response that is nonlinear or has a practical threshold This applies not only to compounds that interact with non-DNA targets but also to DNA-reactive compounds, whose effects may be modulated by, for example, rapid detoxification before coming into contact with DNA, or effective repair of induced damage The

TABLE 11.1 Staged TTC Limits

regulatory approach to such compounds can be based on the identification of a critical no-observed-effect level and use of uncertainty factors (ICH Q3C) when data are available rather than the more general TTC concept

GTIs have been widely studied and modeled, and classified into five categories,13 which are summarized in Table 112

The alerting structures shown in Figure 111 are known to be involved in reactions with DNA The figure does not contain all potential alerting structures, and a more thorough analysis of a particular structure can be performed using various software programs whose results correlate well with the results from ICH S22 and are often used by regulatory agencies for predictive purposes14-16

To build rational and sufficient control strategies for GTIs, it is crucial to have analytical methods that are sensitive, selective, and robust for analyzing the impurities The methodology for analyzing GTIs depends on target specifications and expected values for these impurities In some cases, the level of a GTI is expected to be well below the targeted specification and supported by robust analytical data and the selected method may therefore be described as a limit test to alleviate routine testing burdens

For most of the potential and confirmed GTIs, analytical methods of superior sensitivity are required to support the development of control strategies because of the low TTC limit of 15 μg/day set by the regulatory authorities for commercial products To achieve the necessary sensitivity for low parts per million-level determinations, techniques with both selective separation and highly sensitive detection are generally required Thus, hyphenated techniques such as gas chromatographymass spectrometry (GC-MS), liquid chromatography-mass spectrometry (LC-MS), or LC-MS/MS are often applied to resolve and selectively quantify GTIs

Selection of the appropriate analytical technique is directly related to the properties of the GTI The great variety of structures, reactivities, responsiveness to selected detection methods, and potential matrix effects has posed many unique challenges to analytical method development for these impurities The genotoxic properties and the chemical properties of GTIs can be correlated in general terms and paired with appropriate analytical techniques in an attempt to minimize the development challenges This logic is illustrated in Table 113

Sections 1132 and 1133 are organized following this consideration and logic to describe the development of appropriate analytical methods and provide practical examples for each approach

Direct analysis of GTIs without sample pretreatment has been successful, but it is not without its challenges In many cases, determination of GTIs down to low partsper-million levels in the presence of high concentrations of drug substances makes sample preparation very critical in method development Excessive presence of the matrix can detrimentally affect analytical method development, making it challenging or impossible to achieve the desired sensitivity Headspace sampling of liquid or solid samples provides a simple sampling method for gas chromatography (GC) analysis of volatile analytes including GTIs For complicated samples, the matrix needs to be removed by applying an appropriate sample pretreatment or preparation technique This usually involves extraction of the analytes from the sample matrix Well-established extraction techniques include liquid-liquid extraction (LLE) and solid-phase extraction or solid-phase microextraction (SPME) These extraction methods are routinely applied for robust and accurate analysis by GC or liquid chromatography (LC) For highly volatile or reactive GTIs that may be labile to GC or LC conditions, sample preparations generally involve derivatization to improve the

TABLE 11.3 Properties of GTIs and Analytical Considerations

stability of the GTIs and to render them suitable for accurate analysis Derivatization has been widely and successfully employed to enhance the method capabilities and detector sensitivities for both high-performance liquid chromatography (HPLC) and GC

11.3.2.1 Direct Genotoxic Impurity Analysis by Gas Chromatography Many methods have been reported for GTI analysis by GC coupled with flame ionization detection (FID), mass spectrometry (MS), or other specific detectors17-23 When possible, direct analysis without extensive sample preparation is preferred to streamline the method development and testing requirements This approach can be achieved by increasing sample concentrations and column loading and has shown to meet the testing needs for different GTIs Li18 reported a gas chromatography-flame ionization detection (GC-FID) method for the trace analysis of methyl methanesulfonate (MMS), ethyl methanesulfonate (EMS), and isopropyl methanesulfonate (IMS) in pharmaceutical drug substances Samples at 40 mg/mL were dissolved in mixtures of acetonitrile and water and directly injected into a gas chromatograph in the splitless mode The method achieved a detection limit of approximately 1 ppm and a quantitation limit of 5 ppm for the analysis of these esters in pharmaceuticals, although quantitation of these compounds at levels less than 5 ppm would require a more sensitive GC-MS method In another example for atenolol analysis, residual levels of the GTIs allyl chloride, 1,3-dichloro-2-propanol, and 2,3-dichloro-1-propanol were determined by direct injection GC-SIM-MS (SIM stands for single ion monitoring) analysis of standard and sample solutions24 The method was able to achieve a limit of quantitation (LOQ) of 3 ppm in relation to 50 mg/mL atenolol concentration Both methods suffered the major issue of matrix accumulation and contamination of injector liners, leading to poor method recovery Therefore, the liners needed replacement or cleaning after every 15-20 injections, making the methods less desirable for routine or high-throughput analysis

Headspace GC-MS analysis of trace-level benzene, a by-product in the synthesis of a Genentech (South San Francisco, California) API starting material, was performed at sample concentrations of 200 mg/mL Benzene is a known human carcinogen with an ICH concentration limit of 2 ppm The analytical separation was performed using an Agilent DB-624 column (30 m × 032 mm × 18 μm) With selected ion detection at m/z 78, benzene levels as low as 01 ppm were readily quantitated Figure 112 is a total ion current chromatogram of benzene at a concentration of 025 ppm and the corresponding SIM-MS spectrum The method demonstrated excellent linearity up to 100 ppm (R2 = 0999) and was successfully applied for the analysis of a number of development samples to ensure sufficient control of benzene when this starting material was carried on to API synthesis

Novel approaches without extensive extractions or derivatization have been explored and reported in the literature Sun and others25 developed a matrix deactivation method for the analysis of some GTIs by GC-MS This approach was based on the hypothesis that the instability of some analytes at trace levels was caused by

reactions between the analytes and some low-level reactive species in the sample matrix Adding chemical scavengers or neutralizing species to the sample matrix would chemically quench these hypothetical reactive species and stabilize the reactive and unstable analytes This approach was applied for the analysis of trace-level unstable alkylators such as O-1-chloroethyl S-methyl carbonothioate, resulting in significant improvement of sensitivity to less than 1 ppm

11.3.2.2 Genotoxic Impurity Analysis by Gas Chromatography with Sample Extraction

GTI analysis by direct injection GC methods suffers significant issues of column contamination and matrix effects that are caused by the injection of high concentrations of API or matrix onto the heated GC column along with low-level analytes Therefore, samples are often pretreated through various extractions to remove the matrix and/or to enrich the analytes for increased sensitivity Extraction was almost the default method of choice early on before instruments and detectors such as MS

detectors were routinely explored for GTI analysis, and it still holds significance and applicability in enabling many trace analyses

Numerous methods have been reported for the analysis of low-level alkyl sulfonates in pharmaceutical development due to their prevalence in API synthesis and significance as GTIs26-28 The development of one internal drug substance involved the use of methanesulfonic acid (MSA) as a reagent for salt formation and methanol as a solvent, leading to the possibility of MMS formation during the process A direct injection GC-MS method was developed to analyze the API for residual MMS using external standard for quantitation The API samples (~100 mg/mL) were prepared in a mixture of water and acetonitrile (95/5, v/v) and extracted with dichloromethane for GC analysis and MS detection in SIM mode The method was shown to be linear between 01 and 25 μg/mL (or 1 and 25 ppm relative to sample concentration) with R2 ≥ 099 The lower limit was verified as LOQ, although the achievable LOQ was lower than 01 μg/mL Acceptable recovery was demonstrated at multiple levels The method was used to test the API for both development and release purposes The same method conditions were demonstrated to be suitable for the trace analysis of EMS and IMS that may come from commercial MSA supplies Figure 113 shows the chromatograms for the working standard (1 μg/mL) and the API

A similar approach was described for the simultaneous determination of alkyl mesylates and alkyl besylates in finished drug products by LLE and direct injection GC-MS26 The sulfonates, including MMS, EMS, IMS, methyl besylate, and ethyl besylate, were extracted from finished drug products by n-hexane for direct injection into the gas chromatograph for analysis N-hexane as the extraction solvent brought the double benefits of (1) extracting only sparse amounts of API or other related impurities from the matrix and (2) ensuring a longer stability period for the analytes over solvents of higher polarity Using methyl tosylate as the internal standard for quantitation, the method afforded a quantitation limit of 001 μg/mL, corresponding to 004 μg/g (or parts per million) in tablet or capsule powders A direct injection LLE-GC-MS method was also developed for the alkylating agent dimethyl sulfate (DMS) and was applied to study the efficiency of DMS removal from the process29 DMS and the internal standard d6-DMS were extracted together with methyl tert-butyl ether to overcome the matrix interference A sensitive headspace SPME GC-MS method was reported for the determination of alkyl sulfonates at the parts-per-billion level in pharmaceutical APIs and drug products27 The standard addition method was employed to ensure uniform sample matrix effect, and the limit of detection (LOD) was achieved at 50 ppb for alkyl mesylates and 5 ppb for besylates and tosylates In another study, seven alkyl sulfonate esters were determined at the 5 ppm level by GC-MS, and SPME was demonstrated to be superior to liquid-phase microextraction or LLE for the detection of these sulfonate esters in the APIs30 A strong anion-exchange (SAX)-SPE GC-MS method was reported21 for the trace analysis of 2-chloroethanol in cloperastine fendizoate, an antitussive API After removing fendizoate by SAX-SPE, purified sample solutions were analyzed by GC-MS with selected ion detection at m/z 80, corresponding to the molecular ion of 2-chloroethanol The LOD was achieved at 17 ppm for 2-chloroethanol, and the method was used to test multiple batches of the API

11.3.2.3 Genotoxic Impurity Analysis by Gas Chromatography with Derivatization

Various extraction methods have proved to be effective in overcoming the sample matrix issues for GC analysis in a high-temperature environment However, many GTIs are highly reactive and labile to normal sample preparation or testing conditions Consequently, low recovery due to analyte decomposition or interaction with the matrix is a significant challenge and concern for the trace-level analysis Headspace sampling and injection avoids introduction of matrix components (mostly APIs) into the GC system, therefore improving method accuracy and robustness However, not all GTIs are suitable for the headspace approach, particularly if they

have limited volatility and/or thermal stability Fortunately, chemical derivatization has shown to be a viable alternative in filling the gaps and addressing issues

Hydrazine is a prime candidate for chemical derivatization for trace-level GC analysis Its analysis at trace levels has been reviewed and highlighted in a number of articles9,10,31-33 Hydrazine is a highly polar and reactive GTI; the reactivity makes it challenging to test hydrazine reliably under normal LC or GC conditions For GC analysis, hydrazine has very limited volatility (boiling point > 100°C) and does not respond well to FID Therefore, most hydrazine analyses are performed by derivatization, followed by GC or LC analysis coupled with MS detection Acetone is known to work well both as a solvent and as a derivatizing agent for the analysis of hydrazines by GC,34,35 and low parts per billion-level detection has been achieved with a nitrogen-specific detector34 Using acetone or acetone-d6 as a derivatizing agent, a general in situ derivatization-headspace-GC-MS method was developed and validated for the determination of hydrazine in drug substances at low parts-per-million levels36 The derivative acetone azine was detected by MS, and acetone-d6 was used as an alternative reagent in the attempt to differentiate the preexisting acetone azine in drug substances due to the presence of acetone from the manufacturing process The LOQ was demonstrated to be as low as 01 ppm with good recovery The method was used for hydrazine analysis in a number of API matrices to support API release and process development

Aldehydes are commonly analyzed by derivatization as well, especially the shortchain aldehydes with no chromophores that are not suitable for direct GC-FID or liquid chromatography-ultraviolet analysis37-43 Li and others42 reported a headspace GC method for the quantification of low-molecular-weight aldehydes in pharmaceutical excipients C1-C8 aliphatic aldehydes and benzaldehyde were derivatized by O-2,3,4,5,6-(pentafluorobenzyl)hydroxylamine hydrochloride (PFBHA), analyzed by static headspace GC for PFBHA aldehyde oximes, and detected by negative chemical ionization mass spectrometry (NCIMS) The method demonstrated detection limits less than or equal to 01 ppm (relative to a 10 mg/mL sample concentration) for all aldehydes Over 30 common excipients were screened for these aldehydes, and formaldehyde and acetaldehyde were found to be the most abundant This GC-NCIMS method was later expanded to the analysis of aldehydes in excipients used in liquid/semisolid formulations43 The method employed the derivatization of aldehydes in an aqueous medium to be compatible with the liquid and surfactant types of excipients followed by direct injection for GC analysis and MS detection

For the GC analysis of sulfonate and sulfate esters, pentafluorothiophenol (PFTP) is a common derivatizing agent44,45 An in situ derivatization headspace GC-MS method was reported44 for the simultaneous determination of some trace-level alkyl sulfonates and dialkyl sulfates using PFTP An LOQ less than 1 ppm was achieved for each analyte by this method, and an internal standard was employed for accurate quantitation and robust analysis The use of deuterated internal standards reduced the interference from matrix, and the method was developed as a potential generic method for the determination of these alkylating agents in different API matrices The major drawback of the method is the lack of selectivity with respect to the alkyl groups For the same alkyl group, the same PFTP derivative is formed for mesylates, besylates, tosylates, or sulfates Therefore, knowledge of the synthetic process is

required for targeted determination and control of these GTIs The same derivatizing agent was used for a method to monitor the formation of EMS from reaction mixtures containing ethanol and MSA45 After its formation, EMS was derivatized by PFTP and then analyzed using a static headspace GC-MS method

Other examples of derivatization for GC analysis include the determination of volatile amines through acid chloride derivatization and detection by nitrogenphosphorus detection for aziridine (a potential carcinogen) and electron capture detection for 2-chloroethylamine (precursor of aziridine in API synthesis)46 The bulk of the drug substance was removed by extraction or precipitation, and the derivatization occurred in a water-organic solvent two-phase system The derivatives were extracted to the organic phase and analyzed by GC The method achieved detection limits of about 02 mg/kg (or parts per million) for aziridine and 05 mg/kg (or parts per million) for 2-chloroethylamine; however, the sample preparation procedures were complex due to the multiple steps of extractions that were performed in addition to the derivatization

11.3.3.1 Genotoxic Impurity Analysis by High-Performance Liquid Chromatography-Ultraviolet (UV) Spectroscopic Detection

Many pharmaceutical compounds, including API starting materials and intermediates, have alerting structures such as the ones shown in Figure 111 and need to be tested as potential GTIs in APIs for process development and control purposes These compounds are generally nonvolatile and have aromatic or other structural features that render them suitable for analysis by HPLC with ultraviolet (UV) detection For a particular API, structural similarity between these impurities and the API demands excellent selectivity of the HPLC methods in ensuring accurate quantitation Reversed-phase HPLC is the method of choice for API impurity analysis and is often optimized to meet sensitivity requirements for genotoxic analysis47-50 Severin48 described an HPLC-UV method for the determination of N-nitrosohexamethyleneimine, a potential carcinogen, in tolazamide bulk drug and pharmaceutical dosage forms N-nitrosohexamethyleneimine was extracted with diethyl ether, enriched by evaporation of the solvent, and determined by HPLC-UV at 228 nm The method afforded a LOD as low as 1 ppb; however, it involved multiple steps of sample treatment The procedure was labor intensive and prone to mistakes An HPLC-UV-DAD method was reported for the low-level analysis of 4-amino-2-ethoxy-cinnamic acid and its ethyl ester in a drug substance and in prototype formulations47 With sample extraction or derivatization, detection limits of 2-15 ppm in the drug substance were achieved using this method for the potential GTIs 4-amino-2-ethoxy-cinnamic acid, the HCl salt of 4-amino-2-ethoxy-ethyl cinnamate, and 4-bromo-3-ethoxy-nitrobenzene Conversion of the crystalline free base drug to its partially crystalline potassium salt helped to improve the solubility of the drug substance and enable the preparation of sample solution at 15 mg/mL for increased sensitivity

For GTIs labile to typical LC conditions or with a low response to UV detection, chemical derivatization is widely applied to enable the analysis by HPLC-UV As an example of this approach, residual methanesufonyl hydrazide (MSH) in equipment cleaning was determined by HPLC-UV for an API production51 MSH was used as the source of a reducing agent in the first step of the API production; therefore, the cleaning samples from the production needed testing for this toxic material For the analysis, MSH was derivatized with benzaldehyde to form benzalazine, which was then analyzed by HPLC and detected by UV for which LOQ was achieved at 03 ppm An internal method was established for the determination of acid chlorides by HPLC-UV52 The analytes were derivatized with methanol to form methyl esters for the analysis and quantitation For HPLC analysis of aldehydes, the analytes are typically derivatized with 2,4-dinitrophenylhydrazine The resulting hydrazones are analyzed by HPLC-UV for selectivity and sensitivity37,39,40 Nageswari and others39 applied this derivatization approach for the low-level quantitation of formaldehyde in a drug substance by HPLC-UV The LOD and LOQ were achieved at 05 and 15 ppm, respectively, and the method was used to monitor the formaldehyde level in scale-up batches of bulk API

11.3.3.2 Genotoxic Impurity Analysis by High-Performance Liquid Chromatography-Mass Spectrometry

As is the case for GTI analysis by GC-MS, MS has become the detection of choice for superior sensitivity and selectivity in HPLC analysis of GTIs28,53-56 The advancement in both MS instrumentation and methodology for LC applications has routinely pushed GTI method sensitivity to levels unachievable by HPLC-UV

Both SIM and multiple reaction monitoring (MRM) have demonstrated great utility and success in GTI analysis for pharmaceuticals Various organohalides were analyzed at trace levels by HPLC via derivatization using 4-dimethylaminopyridine (DMAP)57,58 or dimethylamine (DMA)25,59 as the derivatizing agent The derivatizing agents react with the halides to form quaternary amine salt derivatives The DMA or DMAP derivatives were resolved from matrix and other components in the samples by hydrophilic interaction liquid chromatography (HILIC) using Waters (Milford, Massachusetts) Atlantis HILIC columns prior to quantitation by either ESI-SIM-MS (ESI stands for electrospray ionization) or tandem MS using MRM For the analysis of 4-fluorobenzyl chloride (4FBCl),57 the precursor to product ion transition of the DMAP-derivatized 4FBCl, m/z 231 to 109, was used for the selected reaction monitoring for high selectivity and sensitivity DMAP derivatization was also used as a generic approach for the screening and quantitation of potential genotoxic alkyl halides containing Cl, Br, or iodine in various positions58 In addition to monitoring the characteristic precursor to product transitions, the loss of DMAP (observed as a neutral loss of 122 Da) by the derivatives would suggest the presence of any molecule that reacted with DMAP, therefore allowing the screening of unknown potential GTIs at relevant levels The structure of originating compounds can be further elucidated using other fragmentation information of the same parent molecules, followed by the development of a more specific method for quantitative target analysis when needed This method could be applied for screening GTIs in different samples including APIs, drug products, chemical process samples, and samples from stability studies, although

further studies are needed to demonstrate the applicability For the DMA derivatizations,25,59 the derivatives were quantitated by ESI-MS due to their high proton affinity Sensitivity (signal to noise ratio [S/N] > 10) was achieved at less than or equal to 2 ppm for (S)-(+)-3-chloro-1,2-propanediol, bis(2-chloroethyl) ether59 and bis(2chloroethylamine)25 with typical sample concentrations of 5 mg/mL DMA was also used to derivatize (R)-2-(5-bromo-2,3-difluoro-phenoxymethyl)-oxirane, for which similar sensitivity was achieved59 As the authors60 pointed out, the spike recovery for all three compounds had to be carefully studied or concurrently determined at the time of sample analysis, and the method S/N and analytical results were corrected for recovery of each analyte29 A group of commonly encountered alkyl esters of sulfonates or sulfates in drug substances were determined at low parts-per-million levels using a general derivatization LC-MS method60 Trialkylamines (methyl and ethyl) were used as derivatizing agents for these esters to form stable quaternary ammonium ions HILIC was again selected to retain these highly polar derivatives and readily separate them from the API peak The quantitation was performed by ESI-MS in positive ion mode The method demonstrated excellent sensitivity for all alkyl esters at a target analyte level of 1 to 2 ppm (API at 5 mg/mL); however, no individual LOD/ LOQ was reported The method was tested for multiple APIs and successfully applied to the determination of various alkyl sulfonates or dialkyl sulfates in APIs and drug products Wu and others63 reported the analysis of some epoxides by derivatization with LC-MS The derivatization of the epoxides was achieved in the gas phase via Meerwein reaction with the ethylnitrilium ion, generated by the protonation of acetonitrile in the mobile phase under typical conditions of atmospheric pressure ionizations As a result, the gas-phase derivatization allowed direct LC-MS analysis of the epoxides, which would otherwise be challenging due to poor ionization Commonly used atmospheric pressure ionization techniques including ESI, APCI, and APPI were evaluated for optimal formation of the Meerwein reaction products, and APCI was found to offer the best sensitivity and the most robust detection compared to the other two techniques Quantitation of the epoxides was performed by both SIM and MRM Both methods afforded an excellent sensitivity less than or equal to 1 ppm detection limit (relative to 10 mg/mL API); however, the MRM method is preferred for selectivity reason when dealing with complex sample matrices

Simultaneous analysis of several potential GTIs for one Genentech drug substance was achieved by HPLC-MS-MS in MRM mode These process impurities were generated at various steps during the API synthesis Their fate through the synthetic process was thoroughly studied to support the development of late-stage and commercial manufacturing process of the API Results from spiking studies and analysis of multiple batches of starting materials, intermediates, and the API allowed the establishment of appropriate control strategies for these impurities during API manufacture and supported the TTC limit in the API for confirmed GTIs For the analysis of these impurities in the API, chromatographic separation was achieved using an ACE 3 C18 column (150 mm × 46 mm), mobile phases of acetonitrile and 01% acetic acid in water, and a linear gradient of 10%–90% acetonitrile over 25 minutes The standard addition method was used for the quantitation of each impurity An internal standard was used to improve the reproducibility of the analysis Excellent linearity was achieved within the range of 1-50 ppm for each analyte

spiked into the 4 mg/mL API solutions, and an R2 of 099 or 100 was obtained for all impurities except two, for which R2 values were shown to be 098 and 097 The method demonstrated excellent reproducibility of relative standard deviation (RSD) less than 20% based on duplicate standard injections at five concentration levels between 1 and 50 ppm for each analyte Except one impurity, the average recovery based on duplicate injections of the API solutions at three spiking levels for each analyte was 83% or better with RSD of all individual recovery results below 10% The “last” impurity had an average recovery of 78% with an RSD of 12% for all the individual recovery results These results demonstrated the excellent recovery of the method for these impurities The sensitivity of the method was estimated to be in the low-to mid-parts per billion range for the impurities based on the reproducible responses obtained at the 1 ppm level The method was used to support the release of the API batches for late-phase development use Figure 114 shows the chromatograms of an unspiked API sample solution at approximately 4 mg/mL and an API solution spiked with 1 ppm of each GTI

Other examples include analysis of dimethyl sulfate via derivatization with 2-mercaptopyridine and quantification by HPLC-UV or HPLC-MS for improved sensitivity,53 and determination of arylamines using derivatization with LC-MS61

As shown in Figure 111, GTIs carry a wide range of structural features The unique characteristics of these impurities have led to some creative approaches from sample preparation to detection in advancing GTI analysis and control for drug development The matrix deactivation method for GC-MS and LC-MS46 offered a different approach to handle the analyte stability issue caused by the matrix David and others62 reported a two-dimensional GC method with Deans switching for the analysis of some potential GTIs in pharmaceuticals Sample solutions were injected onto an apolar column for first-dimension separation The GTI fractions were then heart cut and transferred to a polar second-dimension column for further separation prior to MS detection The second-dimension column was installed in a low-thermal-mass oven module that allowed independent temperature program and control This setup avoided the introduction of major components in the matrix (APIs, solvents, derivatizing agents, etc) into the second column or MS detector, greatly reducing contamination or column degradation The method demonstrated a reproducibility of RSD less than 15% and a sensitivity less than 1 ppm LOD for the testing of some Michael reactive acceptors and haloalcohols, indicating its potential to be an effective tool in GTI testing for pharmaceuticals

Trace-level analysis of neutral GTIs can be a challenging task due to their poor ionization efficiency in MS The gas-phase derivatization via Meerwein reaction took advantage of the acetonitrile used for HPLC and resulted in improved sensitivity for some epoxides when detected by MS63 Improved sensitivity for some trace-level epoxides was also demonstrated using coordination ion spray-mass spectrometry (CIS-MS) In CIS-MS, neutral molecules are converted to positively or negatively charged complexes in solution or gas phase, resulting in significant improvement in their detection sensitivity59,64,65 This ionization method is applicable for both positive

and negative ions For the analysis of trace-level epoxides,46 metal ions such as Na+, K+, or NH4+ were doped into the HPLC mobile phase and formed coordination complexes with the epoxides The adduct ions were detected by MS for improved sensitivity at a low parts-per-million level

Applications of special detection techniques have also been reported for GTI analyses Yuabova and coauthors50 explored the applicability of the charged aerosol detector (CAD) for GTI analysis Further studies are needed to establish CAD as a viable option for GTIs Liu and others66 described a HILIC method development using alcohol as a weak eluent for the simultaneous analysis of hydrazines in pharmaceutical intermediates The use of alcohol instead of acetonitrile as the weak eluent provided different selectivity and enabled the use of the chemiluminescent nitrogen detector The optimized HILIC-CLND (CLND stands for chemiluminescent nitrogen detector) method was applied for the analysis of trace-level hydrazine and 1,1-dimethylhydrazine in a pharmaceutical intermediate, with a reporting limit of 002% w/w for both analytes This reporting limit may be sufficient for the analysis and control of GTIs in intermediates; however, it is short of the typical low parts per million-level sensitivity required for the quantitation of GTIs in APIs

A fluorescence detector has been used for the analysis of certain GTIs Hasei and coauthors67 developed a two-dimensional HPLC system coupled with online reduction and fluorescence detection for the analysis of 3-nitrobenzanthrone (3-NBA) The two-dimensional HPLC system consists of a reversed-phase HPLC dimension for purification and online reduction of 3-NBA to 3-aminobenzanthrone (3-ABA) and a normal-phase dimension for separation and analysis of 3-ABA, with a switch valve connecting the two dimensions Reduction of the nonfluorescent 3-NBA to fluorescent 3-ABA was achieved in ethanol with a catalyst column packed with alumina coated with platinum After reduction, 3-ABA was introduced and trapped onto a concentration column by switching the valve from the original position and was then introduced onto the normal-phase HPLC column by switching the valve back to the original position The normal-phase HPLC separated 3-ABA from other interfering components prior to the detection of 3-ABA by fluorescence at excitation and emission wavelengths of 490 and 560 nm, respectively The method had a reported detection limit of 0002 ng and reported quantitation limit of 0006 ng, which were calculated by S/N values of 3 and 10, respectively, based on the injections of 3-NBA from the linearity study at 2-2000 pg Even though the method was developed for the analysis of environmental samples, its excellent sensitivity makes it appealing for pharmaceutical applications also

A recent paper presented a capillary zone electrophoretic method for the determination of dimethyl sulfate and chloroacetyl chloride at trace levels in drug substances by indirect photometric detection68 Capillary electrophoresis offers highly efficient separations and consumes very little reagents69 However, it has the drawbacks of poor sensitivity and precision for trace analysis For the reported method, method optimization for sample injection techniques and the use of capillary with extended light path led to an enhanced sensitivity of LOQ = 10 ppm for both analytes, which was well below the limits required in the APIs based on the TTC approach This study showed that capillary electrophoresis could be a reasonable complement to the commonly used GC or HPLC method for determination of GTIs

GTI analysis poses many challenges technically and operationally Method development is often time consuming due to sensitivity and selectivity requirements When dealing with an extensive development portfolio with many compounds targeting a variety of therapeutic areas, it is not sustainable to develop specific GTI methods for each compound in the portfolio On the other hand, different pharmaceutical compounds could encounter the same GTIs or impurities of similar structural features Development of generic or general methods is therefore an effective way of simplifying method development and harmonizing GTI methods across various compounds Alzaga and others44 described an in situ derivatization-headspace-GC-MS method as a generic approach for the determination of residual sulfonates and sulfates in APIs A generic in situ derivatization LC-MS method was developed for the analysis of some arylamines and aminopyridines at the sub-parts per million level relative to APIs61 Another laboratory took a standardized approach to develop a “selective ion monitoring of analytes in real time” method for low-level detection of potential GTIs in pharmaceutical compounds70 The methods utilized complementary techniques such as GC-MS and LC-MS to enable the analysis of a wide range of GTIs and showed low parts-per-million sensitivity for each technique with sample concentrations less than or equal to 10 mg/mL

For any particular compound, general methods greatly reduce the effort required for individual method development; however, the applicability of these general methods still needs to be demonstrated or confirmed with respect to specificity, sensitivity, and accuracy

GTIs present unique challenges for the pharmaceutical industry from understanding of the toxicity to accurate analysis and control of the impurities in APIs and drug products71-76 Regulatory requirements have come a long way and continue to evolve to safeguard patient well-being

Analytical method development for GTIs has many factors to consider in order to achieve the required sensitivity To reliably quantitate the trace-level GTIs in the presence of high matrix concentrations, some sample treatment is required prior to chromatographic analysis For GTI detection, MS has been established as the detection of choice for superior sensitivity and selectivity that can be achieved by either the SIM or MRM approach

Over the last few years, quality by design (QbD) has become the new paradigm for pharmaceutical development The QbD principle has been applied to build the analysis and control strategies for GTIs through API synthetic processes53,77,78 Sound strategies demand extensive method development and validation followed by data collection for numerous samples This could prove overwhelming for many laboratories How to balance the drug development effort versus the regulatory requirements on GTIs remains a difficult question with few answers

The authors thank Dr Nik Chetwyn, Dr Peter Yehl, and Dr Christine Gu for their content and technical reviews Dr Kavita Mistry, Dr Christine Gu, and Edward Hoff of Genentech are acknowledged for granting permission to use their data Kyle Tse of Genentech is acknowledged for contributions to benzene analysis by GC-MS