ABSTRACT
The presence of genotoxic impurities (GIs) within pharmaceutical products has been a subject of considerable recent interest following the issuance of guidelines from European [1,2] and US [3] regulatory agencies These guidelines typically mandate routine control of both documented GIs and potential genotoxic impurities (PGIs) at the parts-per-million level until suitable scientific data are acquired to dismiss the
101 A Brief Introduction to Genotoxic Impurities 294 102 Quantitative Analysis of Genotoxic Impurities 296
1021 Method Development 296 1022 Sample Preparation and Derivatization 298 1023 Gas Chromatography 299 1024 High-Performance Liquid Chromatography 302
10241 Multiple Reaction Monitoring 303 10242 High-Resolution Mass Spectrometry 309
1025 Ion Chromatography 309 1026 Element-Specific Analysis by High-Resolution Inductively
Coupled Plasma-Mass Spectrometry 310 1027 Sulfur, Phosphorus, Chlorine, Bromine, and Iodine Detection 310 1028 Instrumental Limit of Detection 311
10281 Comparison of GC-FID, GC-MS, and HR-ICP-MS 311 1029 Acceptable Elemental Limits Calculation for Genotoxic
Impurity Alerts 311 10210 Validation 314
103 Future Outlook and Conclusions 315 References 315
concern, significantly adding to the number of compounds impacted by these guidelines [4-6] This chapter describes the challenges surrounding the trace analysis of GIs; however, the majority of the principles presented are similarly applicable to the analysis of PGIs
Responding to this new GI trace analysis mandate has required considerable innovation in the creation of new methods for pharmaceutical synthesis, purification, and analysis In particular, the accurate and reproducible trace analysis of GIs has been especially challenging, as many of the traditional methods used for routine pharmaceutical analysis are simply not suitable for GI analysis at the parts-per-million level This problem is further exacerbated by the fact that many GIs owe their potential genotoxicity to an ability to chemically react with DNA This propensity toward reactivity renders these GIs extremely difficult to analyze, as samples can easily degrade under standard sample preparation or analysis protocols Nevertheless, analytical chemists within the pharmaceutical industry and academia have risen to the challenge, creating new methods and approaches for accurately measuring GIs-a vital first step to inform the development of a suitable control strategy Excellent reviews of current GI methods based on alerting functional groups in the chemical structure of analytes have been published [7-12] In this chapter, we survey the state of the art of this rapidly evolving analysis field, pointing out a number of current challenges, existing solutions, and still unmet needs
All active pharmaceutical intermediates (APIs) will always contain impurities as a result of the imperfect nature of chemical reactivity Typically, these API impurities originate from incomplete consumption of starting materials and intermediates in chemical reactions, formation of by-products, or decomposition of reagents and intermediates Among this host of potential impurities, a few known as GIs are of special concern owing to their ability to react with DNA and cause genetic mutations Additionally, genotoxins may also produce their effects by interfering with DNA replication, potentially leading to tumor development or other major health problems When this damage can be transferred from cell to cell, or from generation to generation, the compounds are said to be mutagenic [5] At very low levels of exposure, protective mechanisms can sometimes “correct” DNA damage, resulting in threshold effects [13] However, demonstration of such a threshold of activity for genotoxic compounds is very difficult; hence, for most compounds a linear dosereactivity relationship is assumed
To ensure the safety of patients and healthy volunteers in clinical trials for developmental new drugs, the levels of impurities present in APIs must be controlled within appropriate limits International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) guidelines Q3A and Q3B were collaboratively created by industry and regulators to address control of impurities in APIs and drug products, respectively [14,15] These guidelines define the reporting, identification, and qualification thresholds of impurities in APIs and final drug products based on the projected clinical dose, route of administration, and additional mitigating factors These guidelines recognize that the ICH-defined
impurity thresholds are not appropriate guidance for impurities that are “unusually toxic” such as GIs
To address this shortcoming, the Committee for Proprietary Medicinal Products of the European Medicines Evaluation Agency released a guidance document in 2007, which defined a new framework to address GIs in APIs [1] The EMEA guidelines recommended that the structures of impurities detected in APIs, as well as those highly likely to form in APIs via chemistry arguments, are assessed in silico for PGI structural alerts with respect to an established public access database The genotoxicity of these identified PGIs can be confirmed using a suitable bacterial reverse mutation test like the Ames test, but long-term in vivo studies in animals are required to determine whether a compound is a carcinogen [5,16,17]
For compounds with sufficient evidence for a threshold-related mechanism, limits can be calculated following the “permitted daily exposure” (PDE) approach as per ICH guideline Q3C for residual solvents [18] This approach calculates a PDE derived from the “no observed effect level” or the “lowest observed effect level”
For compounds without sufficient evidence for a threshold-related mechanism, the EMEA guideline recommends that all due diligence is performed to avoid the presence of genotoxic compounds in an API, either by eliminating the particular reagent of concern from the synthesis or by using alternative synthetic processes to avoid the formation of the GI This approach can often be difficult to implement, as the highly reactive nature of many synthetic intermediates and reagents (eg, benzyl bromides and epoxides) is essential to their function in the chemical synthesis, but is often also the underlying cause of their genotoxicity During early development of a new API, where understanding of chemical processes and impurity control are still relatively limited, such efforts are further complicated and become impractical In cases where the presence of the GI cannot be avoided, levels should be reduced using reaction workup steps and/or purification technologies [5,19]
To establish acceptable levels of GIs in APIs, the EMEA guidance based its approach on a previous approach used by the Food and Drug Administration (FDA) to establish acceptable levels of contaminants leaching from food packaging [20], renaming it the “threshold of toxicological concern” (TTC) The TTC approach was established so as to set acceptable limits for GIs in APIs, which will not expose patients to a significantly increased risk of developing cancer during a lifetime of taking a given medication Based on the carcinogenic potency in rodents of over 700 carcinogens, exposures of less than 015 µg/day were estimated to be a “virtually safe dose,” unlikely to increase a lifetime cancer risk by more than 1 in 106, for all but the most potent carcinogens Using this rationale, the EMEA recommended a limit of 15 µg/day for chronic exposure to a GI/PGI, representing an excess lifetime cancer risk of 1 in 105 This small level of added patient risk was justified by the significant positive health benefits received by the patient taking the medicine The acceptable concentration of the GI in the API is calculated based on the expected dose as follows:
( ) ( )( ) = µ
Concentration limit ppm TTC g day
Dose g day (101)
The EMEA guidance acknowledges that lower TTC values should be used for compounds of high potency and that higher TTC values can be justified for shortterm exposure The EMEA guidance does not provide specific guidance for the limits of PGIs in investigational APIs during shorter duration clinical trials, leaving its applicability to development somewhat open to interpretation
To address this issue, the Pharmaceutical Research and Manufacturers of America (PhRMA) proposed a staged TTC approach, which extrapolated the allowable daily intake (ADI) for GIs from lifetime levels [21] This approach targeted a 1 in 106 level of added patient risk for early clinical trials less than 1 year in duration, since healthy volunteers do not receive any benefits from exposure to the API For longer exposures, the risk is kept as 1 in 105 as only patients receiving a benefit from exposure to the API are likely to be used in such long-term clinical studies The ADIs were therefore set as shown in Table 101
The staged TTC approach was adopted by the EMEA in 2010 [2] The daily allowable levels were calculated in a way similar to what was proposed by the PhRMA group, but a factor of 2 was introduced to account for deviations from the linear model A similar TTC approach was also considered acceptable by the FDA in a draft guidance document in 2008 [3]
Trace level (parts-per-million) quantitation of GIs presents the pharmaceutical analytical chemist with many technical challenges First, an analytical technique appropriate for the properties of a GI (volatility, thermal stability, presence of a chromophore, hydrophobicity, etc) must be selected as the basis for the analysis method to be developed Second, the reactive nature and stability of the GI must be adequately addressed during method development to ensure that requisite reproducibility and accuracy are achieved Third, the clinical dose and duration of the study must be understood well to guide the development of an adequately sensitive analysis method meeting project needs The targeted allowable GI limit will directly impact the selection of critical parameters of the analysis method to be developed, including choice of detection technique (ultraviolet [UV], light scattering, electrochemical detection, mass spectrometry, etc) Additionally, certain components of the sample matrix may also present substantial method development challenges due to analytical interference from the API itself, process impurities, or degradation products The elimination of these interfering matrix components is typically achieved by (1) isolation of the analyte of interest by sample preparation, (2) chromatographic resolution, or (3) using a more selective detector
A wide range of analytical techniques can be used to analyze GIs, depending on the properties of the analyte Due to the high structural diversity of GIs, and the complexity of the sample matrix, no single approach is applicable to address all problems However, analytical laboratories within the pharmaceutical industry and academia have developed systematic strategies to guide GI method development, which have proved to be quite useful [5,6,11,22-25] The majority of these strategies
TA B
LE 1
0. 1
D ai
ly A
llo w
ab le
L im
it s
fo r
G Is
a s
Sp ec
ifi ed
b y
Ph R
M A
,E M
EA ,a
nd F
D A
consist of the following two steps: (1) evaluation of the volatility of the analyte, which typically informs the choice of chromatographic technique to be used, and (2) evaluation of the detection technique based on the properties of the analyte (eg, presence of a chromophore and presence of a halogen atom within the molecule)
Traditionally, API impurity analysis in pharmaceutical laboratories has been carried out using high-performance liquid chromatography-ultraviolet (HPLC-UV) for nonvolatile compounds and gas chromatography-flame ionization detection (GC-FID) for volatile compounds; therefore, method development for GIs also typically begins from a chromatographic analysis dictated in large part by analyte volatility For low parts-per-million detection of GIs, the use of nonspecific detectors such as flame ionization detection (FID) or UV may not be feasible owing to insufficient detector sensitivity and artifacts produced by minor interferences from the sample matrix Hyphenated mass spectrometry techniques such as gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) have gained popularity in GI analysis due to their superior sensitivity and selectivity [25] For GC-MS, both electron ionization (EI) and chemical ionization ionization modes have been employed, with EI being the more popular approach [24,26] Ionization modes for LC-MS include electrospray ionization (ESI) and atmospheric pressure chemical ionization (APCI) For some nonpolar and low-polarity compounds that are not efficiently ionized by either ESI or APCI, atmospheric pressure photoionization (APPI) may be used Selectivity is usually improved by conducting selected ion monitoring (SIM) on a single quadrupole mass spectrometry (MS) or multiple reaction monitoring (MRM) on an MS-MS instrument LC-MS-MS methods are typically used in the analysis of very complex samples where masses are likely to overlap despite the high resolution of the MS instrument MRM mode provides a better signal to noise (S/N) ratio and reduced baseline offset as compared to SIM mode, resulting in better sensitivity and selectivity [24]
For situations where the chromatographic resolution of the analyte from the sample matrix is not sufficient for the development of a suitably sensitive and selective method, sample preparation is critical Several extraction/preconcentration techniques to isolate and/or concentrate the analyte of interest from the sample matrix have been reported In particular, techniques relying on the partitioning of the analyte of interest between two phases to enrich one of the phases in the analyte have been investigated The simplest of these techniques, liquid-liquid extraction (LLE), where the analyte is portioned between two immiscible liquids, often requires an additional concentration step before analysis Liquid-phase microextraction (LPME) uses a capillary hollow membrane filled with microliters of an extracting solvent, which is introduced in a liquid sample prepared in an immiscible solvent The analyte is concentrated in the extracting solvent, which typically can be directly analyzed via chromatography Additionally, solid-phase extraction (SPE) can be employed to partition the analyte of interest and other sample components between a solid stationary phase and a liquid mobile phase Various stationary phases are
commercially available to support SPE, including a large variety of reversed phases, normal phases, and ion exchange phases An alternative to SPE, solid-phase microextraction (SPME) is a solvent-free extraction technique in which a polymer-coated fused silica fiber is exposed to a gas (headspace) or liquid sample (immersion) The analyte is partitioned between the sample matrix and the silica fiber stationary phase coating and subsequently desorbed at a high temperature in a gas chromatography (GC) injection port or redissolved in a solvent prior to analysis by GC or LC These techniques can be applied successfully to the analysis of GIs but typically require method optimization and validation to address matrix effects before routine application to samples of interest, often making these techniques difficult to transfer between samples [27]
Derivatization may be used to improve analyte properties by enhancing volatility or stability, reducing polarity, or improving the extraction profile However, development of derivatization procedures requires optimization of the reaction parameters (temperature, solvent, time, etc) to maximize recovery, stability of the derivatized species, and selectivity of the derivatization reaction [28,29] A generic derivatization procedure that produces the same product for a given class of GIs can be used, as in the method reported by Alzaga and others [28], where the same product is obtained from different alkylating agents With this procedure, the derivatization reaction, shown in Figure 101, gives the same product for the methyl ester of methane sulfonic acid or of benzene sulfonic acid
In this case, a good understanding of the impurities that may be present in the sample is required to ensure that the results obtained from the analysis can be correctly used to calculate the level of GIs present in the sample Several in situ derivatization procedures have been designed to occur during the sample heating required for GC headspace sampling, thus simplifying the sample preparation procedure Examples include the in situ derivatization of hydrazine [30] or of alkylating reagents [28,29] Optimization of the parameters can be very complex as they must be suitable for both the derivatization and the headspace sampling [28]
GC coupled with FID is commonly used in the analysis of volatile GIs [7,9,10,31] For example, several mesylate esters were successfully analyzed in a drug substance using a simple GC-FID method with a limit of quantitation (LOQ) of 5 ppm
in the drug substance [31] If the GI has sufficient vapor pressure, headspace GC or dynamic headspace is usually preferred to minimize the contamination of the injection port or column head by nonvolatiles such as APIs [25] As an example, a recently published report described the analysis of the oxidation catalyst 2,2,6,6-tetramethyl-1-piperidinyloxy in 13 batches of filibuvir drug substance using headspace-GC-MS [32] Development of a headspace GC method requires optimization of the headspace parameters (temperature, solvent, and duration of heating); but the resulting method can often be applied to a range of similar compounds, such as structurally related alkylating agents [28,29] In cases where the GI is not volatile enough for headspace GC analysis, direct injection techniques may be used [33] In such instances, frequent cleaning of the injector is recommended, as the buildup of API impurities and associated thermal degradation products can interfere with GI quantitation [26,31] In addition, the thermal stability of GIs must be assessed and appropriately accounted for to allow proper quantitation [33] For GIs bearing a halogen such as alkyl and benzyl halides, electron capture detectors (ECDs) have been investigated due to the unique halogen selectivity of these detectors [7] This technique is particularly useful due to its specificity if no other component of the sample matrix contains halogen atoms, but the overall sensitivity of the approach typically rivals that of FID for monohalogenated compounds [26] The sensitivity of ECD detection can vary significantly depending on the nature and number of halogen atoms present on the molecule; hence, most published methods opt for MS detection [7,24] The use of a nitrogen-phosphorus specific detector, also known as a thermionic specific detector, has also been reported for the analysis of trace formaldehyde in antibiotics after derivatization to the oxime with hydroxylamine [34]
GC-MS applications have been widely used in the low-level quantitation of volatile GIs when other nonselective detectors do not offer the required sensitivity Sulfonic acids (methane sulfonic acid, benzene sulfonic acid, and p-toluene sulfonic acid) are commonly used in the synthesis of active pharmaceutical ingredients as acid catalysts, or as counterions to form salts to upgrade purity or control pharmaceutical properties Crystallization in solvent systems containing alcohols can potentially lead to the formation of corresponding sulfonate esters, which are well-publicized GIs [21] Similarly, the presence of other strong inorganic acids (eg, HCl) in solvent systems containing alcohols can lead to the formation of alkyl halides, which are also considered GIs [21] These volatile sulfonate esters and small polar alkyl halides have been analyzed using very sensitive GC-MS methods such as the one shown in Figure 102 [35] This method utilizes an Rtx®-200 column and if used in the SIM mode can achieve detection limits as low as 1 ppm Excellent linearity, precision, and spike and recovery values have all been reported
GC-MS has also been used for the quantitation of hydrazine at levels as low as 01 ppm In one approach developed by Sun and others [30], hydrazine was first converted to an acetone azine by in situ derivatization with acetone, which was then analyzed by headspace GC-MS Residual parts-per-million levels of alkylating agents, epoxides, halogenated sulfides, and halogenated amines have all been successfully analyzed by GC-MS as well [7,24,25,28]
HPLC is widely used as a separation technique for the analysis of nonvolatile GIs The most widely used separation mode is reverse-phase (RP) HPLC [7,9,10] For extremely polar analytes, hydrophilic interaction liquid chromatography (HILIC) can be used to achieve sufficient retention [25,36,37] HILIC employs eluents rich in acetonitrile with low aqueous content, a mobile phase mixture that is believed to form a water-rich layer at the surface of the polar stationary phase Separation is then achieved by partitioning the polar analytes between the mobile phase and the aqueous layer coating the stationary phase The use of HILIC has been reported for the analysis of quaternary ammonium ions, which were either compounds of interest (quaternary hydrazine derivatives [37]) or derivatization products (of alkyl sulfonates and dialkyl sulfates [38])
For the detection of analytes bearing a chromophore, the UV detector is the conventional first choice for RP-HPLC and HILIC [24,39] In particular, liquid chromatography-ultraviolet techniques can be very useful tools to control GIs during the early stages of developing an API manufacturing process, where clinical studies are shorter in duration and corresponding GI specifications may be looser and not require quantitation at low parts-per-million limits [39] However, UV detection is often neither sensitive nor selective enough to achieve low parts-per-million levels of detection As a result, alternate detectors such as evaporative light scattering detectors (ELSDs) or charged aerosol detectors (CADs) may be used, as applied to the analysis of carcinogenic pyrrolizidine alkaloids in plant products [40] However, the examples of using ELSDs or CADs for the analysis of GIs in APIs are quite limited owing to the lack of robustness and/or selectivity needed for GI analysis One major disadvantage of ELSDs is that the response can be highly dependent on analyte volatility and eluent composition CAD response is also dependent on analyte volatility [41] CAD is becoming more popular due to its more universal response, wide dynamic range, precision, and simplicity The use of a chemiluminescent nitrogen detector has also been reported for the analysis of two nitrogen-bearing GIs, hydrazine and 1-1-dimethylhydrazine, as shown in Figure 103 [36]
LC-MS and LC-MS-MS applications have been widely used in low-level quantitations of nonvolatile GIs when UV detectors, ELSDs, or CADs do not afford the required sensitivity or specificity Analytes containing basic nitrogens can often be easily detected by atmospheric pressure ionization LC-MS in the positive ion mode due to their high proton affinity Similarly, acidic analytes can often be detected in the negative ion mode via deprotonation [39] Analytes with weak acidic groups are better ionized by negative mode APCI than ESI [24] For neutral compounds, coordination ion spray-MS can be used to detect adducts (Li+, Na+, K+, and NH4+) in the positive ion mode [24] APPI can also be efficiently used to ionize neutral and lowpolarity analytes For very reactive intermediates, the compound of interest is often converted to a stable, easily ionized derivative prior to analysis Direct analysis of GIs such as aziridines, arylamines, and aminopyridines by LC ESI-MS in SIM mode has been reported [25] Aldehydes and ketones can be derivatized with 2,4-dinitrophenyl hydrazine, and the corresponding derivatives can be analyzed by LC-MS using ESI in the negative ion mode [25] Figure 104 shows the analysis of 1 ppm solutions of derivatized formaldehyde, propionaldehyde, hexanal, and nonanal [25]
GC-FID and GC-MS are highly sensitive techniques for the analysis of sulfonic acid esters, but many of the aromatic sulfonates have very high boiling points and poor GC sensitivity LC-MS has been demonstrated as a highly sensitive analytical technique, with limits of detection (LODs) less than 01 ppm being easily achievable [42]
Figure 105 shows the methyl, ethyl, and isopropyl esters of toluenesulfonic acid at 1 ppm concentration with respect to the API The MS utilized SIM in the positive ion electrospray mode, monitoring m/z 204, 218, and 232 (the ammonium adducts of each compound) The instrument’s scan switches between SIM ions at appropriate times in the chromatographic run to capture the relevant peaks Although all three ions could be monitored simultaneously, selecting a specific ion by time increases the duty cycle of the instrument and thus improves the sensitivity Good linearity and recovery are observed [42]
10.2.4.1 Multiple Reaction Monitoring Once the mainstay of low-level metabolite and peptide quantitation, triple quadrupole mass spectrometers are now finding uses in the determination of GIs MRM experiments allow the differentiation of isobaric compounds based on their unique fragment ions generated after a collisionally induced dissociation, known as a transition This highly selective detection technique can eliminate background
interference from chromatography systems and significantly improve signal to noise, and therefore improve detection limits Triple quadrupole mass analyzers can be coupled to both HPLC and GC systems, and in both cases they can enable the analysis of low-level GIs in complex matrices In addition, the fragment ions and neutral loses observed can be used as indications of similar molecular functionality
In our laboratories, we recently demonstrated the selectivity and sensitivity of a triple quadrupole mass spectrometer utilizing an MRM scan A number of sulfonate esters were detected to sub-parts per million levels with a very fast chromatographic separation In each case, the monitored transition was that associated with the loss of the alkylsulfonate substituent; an example of such a transition is shown in Figure 106
A further application of the MRM experiment includes the determination of a substituted hydrozinopyridine The available HPLC-UV method was only capable of a GI detection limit of 100 ppm, with a target API concentration of 2 mg/mL (Figure 107) Monitoring the transition, using an MRM scan, from m/z 144 to m/z 117 (Figure 108) using a Sciex API3000 triple quadruple mass spectrometer enabled
the detection limit to be reduced to 1 ppm with very little method development (Figure 109) The huge improvement in selectivity enabled significant reductions in chromatographic run time, provided there was no ion suppression associated with coeluting peaks In this example, a recovery greater than 80% was achieved in a 6-minute isocratic separation
Generic screening methods for alkyl halides have been demonstrated [43], using 4-dimethylaminopyridine (4-DMAP) as a derivatizing agent The reaction scheme is shown in Figure 1010
The nature of the derivatizing agent is such that any potential reactive alkylating agents will be derivatized This facet of the sample preparation ensures that all potential alkylating agents in the sample matrix will react and therefore be detected, not just those species initially targeted from a theoretical scrutiny of the synthetic route The resulting quaternary nitrogen species provided excellent sensitivity in ESI+MS and can be easily resolved from the sample matrix using HILIC The derivatives of a number of GIs are shown in Table 102 along with the fragment ions and DMAP fragments
Initial screening can be conducted using a precursor ion scan, or a neutral loss scan of m/z 122, which is indicative of the 4-DMAP fragment Peaks displaying the fragments of m/z 121 and 123 or a neutral loss of m/z 122 can be assumed to have reacted with the derivatizing agent and thus are a potential alkylating species The main drawback to this approach is that alkyl halides with the same alkyl group molecular weight, but different halogens, cannot be differentiated since the resulting derivatives are isobaric However, knowledge of the synthetic process would enable a structure to be proposed
TA
B LE
1 0.
