Approaches to Assess, Analyze, and Control Genotoxic Impurities in Drug Substance Development

Impurities are introduced in the form of starting materials, reagents, by-products, intermediates, and degradation products to drug substances As they do not provide therapeutic benefits to patients, they need to be controlled for the safety and quality of pharmaceutical substances The International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) Q3A guideline Impurities in New Drug Substances [1] provides guidance

121 Introduction 343 122 Classification and Qualification of Impurities 344 123 Flow Diagram for Genotoxicity Assessment and Control 347 124 Approaches to Control Process-Related Impurities 349 125 Factors to Consider When Developing Analytical Methods to Control

Genotoxic Impurities 349 126 Case Study #1: Intermediate Preparation for Ames Testing 352 127 Case Study #2: Unexpected Impurity in an Outsourced Intermediate 353 128 Case Study #3: Control of 2-Chloropyridine Impurities by Gas

Chromatography/Mass Spectrometry 355 129 Case Study #4: Control of Genotoxic Intermediates by LC/MS/MS 357 1210 Case Study #5: Control of Methane Sulfonic Acid and Tosic Acid in

Lieu of Testing Corresponding Alkyl Esters 359 1211 Conclusion 360 Acknowledgments 361 References 361

on the content and qualification of impurities in new drug substances produced by chemical syntheses for registration applications The ICH Q3B guideline Impurities in New Drug Products [2] provides guidance on the content and qualification of impurities in new drug products produced from chemically synthesized new drug substances for registration applications Residual solvents are regulated by the ICH Q3C guideline Impurities: Guideline for Residual Solvents [3], which recommends acceptable amounts for residual solvents in pharmaceuticals for the safety of the patient Though these ICH Q3X guidelines are intended for registration applications and not for clinical research stages of development, impurity assessment, analysis, and control during drug development are carried out with these guidelines as guiding principles in the pharmaceutical industry ICH Q3A lists three thresholds (reporting, identification, and qualification) for impurities The guideline states that “identification of impurities present at an apparent level of not more than (≤) the identification threshold is generally not considered necessary However, analytical procedures should be developed for those potential impurities that are expected to be unusually potent, producing toxic or pharmacological effects at a level not more than (≤) the identification threshold” In addition, Footnote 3 of the qualification threshold states that “lower thresholds can be appropriate if the impurity is unusually toxic” It is generally accepted that genotoxic impurities are in the category of unusually potent and unusually toxic impurities

Genotoxicity is a broad term that refers to any deleterious change in genetic material regardless of the mechanism by which the change is induced [4] It encompasses effects from mutagenicity through DNA reactivity, DNA damage, and chromosomal damage, both structural chromosome breakage and aneuploidy [5] Genotoxic impurities with limited toxicological data should be controlled at the level of the (staged) threshold of toxicological concern (TTC) (sTTC stands for staged TTC) according to the European Medicines Agency (EMA) guideline [6,7] and the Food and Drug Administration (FDA) draft guideline [8] Readers are referred to Chapters 1 and 3 for details on the (s)TTC concept and calculations of acceptable exposures To distinguish genotoxic impurities from nongenotoxic (nonmutagenic) impurities, nongenotoxic impurities are sometimes referred to as “regular,” “typical,” or “ordinary” impurities

Toxicological overviews and regulations of genotoxic impurities are discussed in the literature [9-11] Some interesting examples of genotoxic impurity assessment and control in pharmaceutical products in the literature are listed in the references section of this chapter [12-31]

This chapter presents the approaches employed to assess genotoxic risks and analyze and control genotoxic impurities in drug substance development with a few case studies discussing the rationale and procedures

The ultimate risk concern for genotoxic impurities is carcinogenicity; however, carcinogenicity data are not available for most impurities that are encountered Therefore, available alternative data are used for risk assessment purposes These

alternative data include structure-activity relationships, bacterial reverse mutation (Ames) tests, and a generic determination of virtually safe exposure levels (TTC and sTTC) Mueller and others [5] proposed a classification of impurities with respect to their genotoxic potential This system is not a general classification of genotoxicity, rather this classification would be used solely for the purpose of deciding whether an impurity possesses a high level of risk and is therefore to be controlled at very low levels of daily intake

The proposed classification of impurities is as follows:

• Class 1: impurities known to be both genotoxic (mutagenic) and carcinogenic – This group includes known animal carcinogens with reliable data for a

genotoxic mechanism and human carcinogens Published data on their chemical structures exist, demonstrating the genotoxic nature of the impurities

• Class 2: impurities known to be genotoxic (mutagenic), but with unknown carcinogenic potential

– This group includes impurities with demonstrated mutagenicity based on testing of the impurities in conventional genotoxicity tests, but with unknown carcinogenic potential

• Class 3: alerting structure, unrelated to the structure of the active pharmaceutical ingredient (API) and of unknown genotoxic (mutagenic) potential

– This group includes impurities with functional moieties that can be linked to genotoxicity based on structure but have not been tested as isolated compounds They are identified based on chemistry and using knowledge-based expert systems for structure-activity relationships The alerting functional moiety is not present in the structure of the parent API Some generic rule-based alerts may be quite unspecific, and further consideration must be given to chemical structural constraints, chemical environment, or experimental data in the assessment of potential genotoxicity Due to the uncertain relevance of structural alerts, regulatory action should not be based solely on the presence of a particular functional group; rather, the accuracy for predicted genotoxicity should be evaluated case by case based on the available scientific literature, additional data on the chemical class, and further available (genotoxicity) test results on closely related structures

• Class 4: alerting structure, related to the API – This group includes impurities that contain an alerting functional moi-

ety that is shared with the parent structure The genotoxicity of the isolated impurity is unknown, but the genotoxicity of the active principle is characterized through conventional genotoxicity testing Similar chemical constraints and chemical environment exist for the alerting substructure in the impurity and the API

• Class 5: no alerting structure or sufficient evidence for absence of genotoxicity

– This group is adequately covered by the existing ICH Q3A, Q3B, and Q3C guidelines

A qualification strategy for each class of impurities proposed [5] is as follows:

• Class 1: carcinogenic impurities should be avoided by modification of formulation options, synthetic route, starting materials, reactants, or purification steps When it is not practical or realistic to avoid these impurities, specifications should be determined Compound-specific risk assessment should be conducted if sufficient 2-year rodent data are available to determine the carcinogenic potency of an impurity The appropriate data could be cancer slope factors, TD50 (the average daily dose estimated to halve the probability of remaining tumor free throughout a 2-year study) values, or maximum tolerated dose information If insufficient information is available to calculate a compound-specific risk, then general risk assessment based on the (s)TTC concept is performed

• Class 2: conventional genotoxicity tests are designed to identify genetic hazards, and the data generated are not suitable for quantitative characterization of risk Impurities for which there is strong evidence of mutagenicity should be assessed on the basis of whether or not there is evidence for a threshold-mediated mechanism

– For impurities with sufficient evidence of threshold-based genotoxicity (ie, compounds that do not react with DNA, such as spindle poisons, topoisomerase inhibitors, and DNA synthesis inhibitors), calculate permitted daily exposures (PDEs) from no-observed-effect levels or lowest-observed-effect levels as described in ICH Q3C for class 2 solvents

– For impurities without sufficient evidence for a threshold-related mechanism, including compounds with direct DNA reactivity such as DNA binding or DNA adduct formation, levels should be controlled according to the (s)TTC approach These compounds are usually bacterial mutagens, unequivocal in vitro clastogens, and frequently positive in in vitro genotoxicity assays

• Class 3: when an impurity contains a structural alert without experimental data, the impurity may be controlled according to the (s)TTC levels Alternatively, the impurity may be tested in a genetic toxicology assay (typically a bacterial reverse mutation assay, the Ames test) to gain toxicological understanding of structural predictions If the assay results are positive (Ames positive), determining the impurity to be a bacterial genotoxin, then evaluate the compound as a class 2 impurity If the compound is not mutagenic (Ames negative), the experimental data override the structural alert and the impurity is controlled as an ordinary impurity

• Class 4: when an impurity has a structural alert that is shared with the API, thorough genotoxicity testing of the API is usually sufficient to qualify the structurally similar impurity When genotoxicity tests on the API are negative, the impurity is controlled as an ordinary impurity

• Class 5: these impurities are controlled as ordinary impurities

A summary of the impurity classification and the aforementioned qualification strategy is shown in Table 121

Due to the different threshold requirements for “ordinary” (“typical” or “regular,” denoting “nongenotoxic” in this context) impurities versus “genotoxic” impurities, determining the genotoxicity (or nongenotoxicity) of an impurity is an important step in establishing control strategies Figure 121 shows a typical flowchart for impurity assessment and control Prior to the manufacture of the drug substance batch used for human clinical trials, the structures in the synthetic scheme, including starting materials, intermediates, reagents, and known by-products, of the drug substances are submitted for in silico evaluation If the in silico evaluation of a structure is negative, the impurity is considered nongenotoxic and controlled according to the ICH Q3A/B limits If the in silico assessment gives a structural alert, then the impurity is assigned to one of the five categories [5] Depending on which class the impurity is categorized into, different strategies could be applied, as described in Section 122 If the compound is not expected to be present above TTC in the drug substance due to the chemical properties of the compound, a justification is written and the compound is controlled “as low as reasonably practicable” (ALARP principle) according to the pharmaceutical assessment described in EMA and FDA guidelines [6,8] When dealing with an impurity with insufficient toxicological data and the impurity level is expected to be above (s)TTC, one can proceed to control the impurity by

TABLE 12.1 Classification, Qualification, and Risk Assessment of Impurities

chemical and analytical efforts, or to test the impurity in a genetic toxicology assay (typically the Ames test) If the Ames test result is negative, then the impurity is controlled as an ordinary impurity according to ICH Q3A and Q3B If the Ames test result is positive, then the impurity is controlled at the level of (s)TTC

Ames testing and chemical/analytical control could happen sequentially or simultaneously depending on the resource availability and the project time lines The two possible outcomes at this step are as follows: (1) the impurity is controlled according to (s)TTC regardless of the Ames test result, and (2) the Ames test result is negative and the impurity is controlled as an ordinary impurity according to ICH Q3A/B Thus, a rational practice at this stage is as follows: if the anticipated Ames test result is positive based on prior knowledge of compounds with similar structures, then one would proceed with chemical/analytical control effort without spending precious resources on Ames testing If the compound structure is frequently encountered in drug development and it has a reasonable chance to be Ames negative based on prior knowledge, then the compound is submitted for Ames testing If the Ames test result of the compound is negative, then it saves all future chemical/analytical efforts to control the impurity at (s)TTC levels If prior chemical/analytical efforts show that controlling the impurity at (s)TTC levels is very challenging, then one would proceed with Ames testing to ensure that the demanding efforts are justified

The compound may not be expected to be present in the drug substance at the TTC level due to the properties of the compound For example, a compound with low boiling point is not expected to be present in a drug substance that was manufactured by high-temperature procedures such as distillation or hot filtration Another example

is compounds with high reactivity that can be reasonably expected to be present at a level below TTC in the drug substance An example of such a compound is acyl chloride Acyl chloride is a strong alkylating reagent commonly used in chemical synthesis for alkylation purposes Therefore, acyl chloride could alkylate DNA and RNA and cause mutation and genotoxicity However, due to its high reactivity, acyl chloride readily reacts with water (or other protic solvents) under process conditions or ambient conditions and becomes carboxylic acid (esters when reacting with alcohols) In this case, acyl chloride is not expected to be present in the drug substance above the TTC level This is an example of the “process control” described in the option 4 approach for process-related impurity control (see Section 124)

Control of process-related impurities could be achieved by one of the following four approaches [4] The central concepts of each approach are set in italics:

• Option 1 includes a test for the impurity in the drug substance specification with an acceptance criterion at or below the acceptable limit using an appropriate analytical procedure

• Option 2 includes a test for the impurity in the specification for a raw material, starting material, or intermediate, or as an in-process control (IPC), with an acceptance criterion at or below the acceptable limit using an appropriate analytical procedure

• Option 3 includes a test for the impurity in the specification for a raw material, starting material, or intermediate, or as an IPC, with an acceptance criterion above the acceptable limit using an appropriate analytical procedure coupled with demonstrated understanding of fate and purge and associated process controls that ensure that the level in the drug substance is below the acceptable limit without the need for any additional testing

• Option 4 involves understanding of process parameters and impact on residual impurity levels (including fate and purge knowledge) with sufficient confidence that the level of the impurity in the drug substance will be below the acceptable limit such that no analytical testing is needed for this impurity This option relies on process control in lieu of analytical testing

The options and approaches employed to control genotoxic impurities could evolve and change over the course of drug development as we gain process understanding and confidence in the process capability and control

Frequently used analytical techniques to measure genotoxic impurities are chromatographic methods including high-performance liquid chromatography (HPLC) and gas chromatography (GC) methods The most typical detector is an ultraviolet

(UV) absorption detector for HPLC and a flame ionization detection (FID) detector for GC These analytical techniques are well established and provide numerous desirable characteristics including linearity and sensitivity Furthermore, these instruments are ubiquitous in almost all analytical laboratories, making method transfer straightforward When a genotoxic impurity is volatile enough for GC analysis, GC offers advantages as most pharmaceutical intermediates and products (and their associated impurities) are not volatile, effectively reducing matrix interferences and possibly improving specificity and sensitivity of the methods

HPLC-UV and GC-FID methods provide retention time-based specificity; therefore, when overlapping interferences are present in the sample they suffer loss of specificity This problem is demonstrated in Figure 122 Each chromatogram is overlaid with a chromatogram of a hypothetical genotoxic impurity standard at 30 ppm (against a product to be monitored) The genotoxic impurity elutes at 13  minutes of retention time In Figure 122a, the genotoxic impurity elutes at 13 minutes and the product elutes at 24 minutes There is no interference, and this method is capable of detecting the genotoxic impurity at a 30 ppm level In Figure 122b, the genotoxic impurity elutes at 13 minutes and the product elutes at 27 minutes Even though the product elutes at a retention time that is well separated from the retention time of the genotoxic impurity, there is interference from a minor component of the product, making this method not suitable Figure 122c shows a product that elutes very close to the genotoxic impurity; there is also interference from a minor component, and the method is not suitable The interference problem could be a factor in determining where in the synthetic scheme to control the genotoxic impurity Obviously, it is desirable to avoid the interferences shown in Figure 122b and c It should be noted that even when the product is chromatographically well separated from the genotoxic impurity, as in Figure 122b, any minor components (impurities) in the product could present an interference issue Impurity profile of the product may differ considerably from batch to batch causing an analytical method not reliable, when dealing with trace level (typically parts per million) impurity control The challenges stemming from varying impurity profiles are severe during the early stages of development, where manufacturing changes are made frequently and process understanding and control is still being established To manage the variability, analytical methods that provide more specificity than HPLC-UV or GC-FID might be desired Mass spectral detection provides such specificity [25-27,31] The EMA guideline [6] states that “detection and/or quantification of these (genotoxic) residues should be done by state-of-the-art analytical techniques,” requiring pharmaceutical companies to employ cutting-edge analytical technologies for genotoxic impurity control

Another important factor to consider is solubility of the product (starting materials, intermediates, or API) in which the genotoxic impurity needs to be controlled In general, a high solubility of samples is required to detect a very low (typically parts per million) level of genotoxic impurity Obviously, judicious selection of a suitable dissolving solvent is essential to achieve high solubility

Consider the following hypothetical synthetic scheme, in which the genotoxic compound A is a starting material of the synthesis:

Genotoxic A (GTA) → intermediate B → intermediate C → intermediate D → API

The ultimate goal is to control the genotoxic compound A (GTA) in the API at a level at or below (s)TTC As discussed in Section 124, control points could be any compound after GTA Generally speaking, the level of GTA becomes lower as more synthetic steps are performed In other words, the level of GTA in intermediate C is expected to be lower than that in intermediate B, and the level of GTA in intermediate D is expected to be lower than that in intermediate C, and so forth It would be prudent to choose a control point where GTA is reasonably expected to be controlled, such as after a reactive condition that can destroy GTA or after a crystallization step that can reject GTA It is preferable to control a genotoxic impurity as early as possible in the synthesis One reason is to allow the possibility of cleaning up the impurity further in cases where the analytical results are higher than the acceptable limit at the planned control point If the planned control point is at the API, and the measured level is higher than the acceptable limit, then one has very limited options to clean up the API with an unacceptable impurity profile Another factor to consider is that a genotoxic impurity may undergo reactive conditions to generate other (and potentially more than one) genotoxic impurities, making impurity control more challenging downstream

Five case studies of genotoxic impurity control are discussed in Sections 126 through 1210 These case studies were encountered during drug substance development and were selected to demonstrate diverse circumstances and different approaches employed for genotoxic impurity control

A synthetic intermediate (intermediate A) showed a structural alert in in silico assessment, and it was decided to submit the intermediate for Ames testing Hydrazine was one of the reagents used to make intermediate A (see the following reaction):

Starting material + H2N-NH2 (hydrazine) → intermediate A Hydrazine is a highly reactive and strong reducing agent, and it is a known muta-

gen (Ames positive) and carcinogen (see Chapter 3, Table 313) It was necessary to control the level of hydrazine in the intermediate A submitted for the Ames test, so that the Ames test results reflected the effects of intermediate A, not hydrazine A  gas chromatography/mass spectrometry (GC-MS) method was developed as a limit test on acetone-derivatized hydrazine Acetone was used as a dissolving solvent and a derivatizing agent [19]

The ICH Q2A guideline Text on Validation of Analytical Procedures [32] states that “testing for impurities can be either a quantitative test or a limit test for the impurity in a sample Either test is intended to accurately reflect the purity characteristics of the sample Different validation characteristics are required for a quantitative test than for a limit test” When it is necessary to develop an analytical method rapidly to determine the content of a particular impurity at a certain level, limit tests are often considered Limit tests require less validation than quantitative tests as the scope of limit tests is narrower Table 122 shows the validation requirements of quantitative and limit tests in the ICH Q2A guideline A limit test needs to be validated to show specificity and detection limit only Typically, recovery is measured

to demonstrate that the analytical method is capable of detecting a spiked analyte at the level of interest When an analyte is reactive, recovery is an especially important characteristic of the analytical method, because appropriate recovery data indicate that the analytical method is able to measure the analyte present in samples

One of the identification methods specified for an outsourced compound was infrared (IR) spectral comparison to a reference standard This procedure is typically known as “ID by IR” The acquired IR spectrum of the outsourced batch and that of the reference standard are shown in Figure 123 The two spectra match well except for an absorbance peak at 1740 cm-1, which is observed only in the IR spectrum of the outsourced batch Therefore, this batch failed the ID by IR test An IR absorbance peak at 1740 cm-1 suggests a carbonyl functional group After careful examination of the batch production record from the vendor, diethyl carbamyl chloride (DECC) was proposed as the identity of this impurity The identity of the impurity was confirmed by GC-MS experiments against a purchased authentic sample of DECC DECC is a known genotoxin and carcinogen

A GC-MS method was developed to follow and control the impurity DECC is a reactive compound similar to acyl chloride The appropriate selection of sample diluent and preparation is important as DECC can react with the diluent to hinder

TABLE 12.2 Validation Required for Quantitative versus Limit Tests

adequate analytical measurement For example, DECC can react with methanol to form methyl carbamate (similar to the formation of ester when acyl chloride reacts with alcohol) The mass spectra from GC-MS of DECC prepared in methanol are shown in Figure 124 Intact DECC shows a molecular ion at m/z 135 with a chlorine isotope pattern (bottom mass spectrum), whereas methyl carbamate shows a molecular ion at m/z 131 without a chlorine pattern (top mass spectrum), illustrating the importance of proper choice of diluent as an important part of analytical method development

Modification of the batch production procedure was requested to the vendor to avoid the formation of DECC for future batches

The ICH Q10 guideline Pharmaceutical Quality System [33] has a section titled “Management of Outsourced Activities and Purchased Materials” with the following statements:

The ultimate responsibility of ensuring quality drug substance lies with the pharmaceutical company even when a genotoxic impurity originates from an outsourced batch This case study demonstrates that setting well-informed and thoughtful specifications is a critical aspect of proper control of impurities in outsourcing activities, as well as good process understanding and effective communications with vendors

The FDA draft guideline on genotoxic impurities [8] states that “in cases where a class or family of structurally similar impurities is identified and is expected to have similar mechanisms resulting in their genotoxic or carcinogenic potential, the total

daily exposure to the related compound should be evaluated relative to the recommended threshold exposure”

Five potential genotoxic impurities (named PGT-1, PGT-2, PGT-3, PGT-4, and PGT-5) needed to be controlled in an intermediate PGT-2 through PGT-5 shared the same 2-chloropyridine functionality as shown in Figure 125 The acceptable limit for genotoxic impurity control was 100 ppm All four potential genotoxic impurities displayed structural alerts in in silico assessment because of the 2-chloropyridine moiety Due to the structural similarity it was assumed that they would have the same mode of action for genotoxicity, and it was decided to control the four impurities as a sum [16] PGT-1 did not share the 2-chloropyridine functionality; therefore, the control limit for PGT-1 was set at 100 ppm A limit test method using GC-MS was developed for the four PGTs that had the 2-chloropyridine moiety The GC-MS chromatogram acquired in a timed selected ion monitoring (SIM) mode is shown in Figure 125 The chromatogram was obtained from a standard mixture made of 20 ppm each of the four potential genotoxic impurities If all four 2-chloropyridine PGTs showed not more than (NMT) 20 ppm individually, then the sum of the four 2-chloropyridine PGTs is NMT 80 ppm, satisfying the 100 ppm control limit A limit test HPLC-UV method was developed separately for PGT-1 (data not shown) The summary of three batches is shown in Table 123 Lot A and lot B exhibited a level of potential genotoxic impurities below the control limit for all five PGTs However, lot C showed more than 20 ppm for PGT-3 To meet the NMT 100 ppm control limit with a comfortable safety margin, the 20 ppm limit GC-MS method was revalidated with a 10 ppm limit for PGT-2, PGT-4, and PGT-5 The analytical results with the 10 ppm limit GC-MS method are shown for lot C in Table 123 This effort cleared all three lots to be used in the manufacture of the API

A nitro aromatic starting material was used to make an API via hydrogenation The synthetic scheme is shown in Figure 126 If the nitro aromatic starting material was not completely consumed, then the residual nitro aromatic starting material would undergo hydrogenation to the corresponding aniline impurity Both the nitro aromatic starting material and the potential aniline impurity showed structural alerts in in silico assessment as expected (see Chapter 2 for structural alerts)

The 10 ppm limit test LC/MS/MS methods were developed for the nitro aromatic starting material and the corresponding aniline impurity separately Figure 127 shows the specificity advantage of an LC/MS/MS method using a triple quadrupole mass spectrometer [26], compared to UV detection and single quadrupole mass spectrometric detection All four chromatograms were obtained from a single sample, which was the product dissolved in dimethyl acetamide (used as the diluent) at 100 mg/mL, spiked with the aniline impurity at 10 ppm The first trace (from the top) shows single reaction monitoring (SRM) of the aniline impurity, clearly showing a good signal to noise (S/N) ratio at 10 ppm at a retention time of 53 minutes The second trace is a UV

TABLE 12.3 Control of Potential Genotoxic Impurities with 2-Chloropyridine Functionality (PGT2, PGT3, PGT4, and PGT-5): Analytical Results Summary of Three Batches

chromatogram, and apparently this detection did not have suitable specificity for the aniline impurity The third trace was acquired in a SIM mode for the aniline impurity The peak at 53 minutes is evident, although not as clear as in the first trace, and other interference also showed up on this channel The fourth trace was acquired in a full mass spectrometry scan mode, and the molecular ion for the aniline impurity was plotted as an extracted ion chromatogram The peak at 53 minutes is noticeable, but the specificity and S/N ratio were not as good as in the SRM or SIM modes A better specificity, sensitivity, and detection limit could be achieved by using state-of-the-art analytical technologies such as triple quadrupole mass spectrometers

In this case study, the starting material was a genotoxic compound that needed to be monitored and controlled The chemical steps used to manufacture the API could potentially convert the genotoxic starting material to different genotoxic side products This case study illustrates the need to consider and control not just genotoxic intermediates but also potential side products that are genotoxic to ensure the quality and safety of drug substances

The well-known Viracept case [12-15] in 2007 was caused by ethyl methane sulfonate (EMS) contamination Viracept is a protease inhibitor used for the treatment of HIV patients, and the contamination was first detected by a bad smell in blisterpacked tablets The final manufacturing step of the Viracept drug substance, nelfinavir mesylate, is the formation of the mesylate salt by adding methane sulfonic acid (MSA) to a suspension of nelfinavir in ethanol As MSA reacts with nelfinavir, the product is dissolved Subsequently, nelfinavir mesylate is isolated from the ethanolic solution by spray-drying However, it was neither MSA nor the manufacturing process itself that caused the EMS contamination found in the drug substance The root cause of the contamination was that tank drying was not performed after cleaning the MSA hold tank with ethanol Upon refilling the tank with neat MSA, reaction with residual ethanol over several months led to the formation of significant concentrations of EMS in the MSA that was used to produce the drug substance causing the contamination

Interactions of strong acids and alcohols during the process of salt formation may produce various alkylating agents such as alkyl halides and esters of alkyl sulfonic acids The alkylating agents reported in the literature include methyl methane sulfonate (MMS), EMS, besylates, and tosylates [12-16, 20, 22-24, 28] Most of these alkylating agents are known genotoxins

One of the intermediates used to make a drug substance was a p-toluenesulfonic acid (tosic acid) salt The process map is shown in Figure 128 Crude API in tetrahydrofuran (THF) was synthesized using intermediate A and intermediate B (tosic acid salt) The crude API was precipitated by adding water and isolated The final crystallization in n-propanol and water was performed for purification and form control Because intermediate B was a tosic acid salt and the final crystallization solvents included n-propanol, the formation of n-propyl tosylate was a concern as a potential contaminant in the drug substance

The n-propyl tosylate could be tested in the final API; however, if the detected level was above the acceptable limit then the options to remove n-propyl tosylate would have been very limited Considering the choices for controlling n-propyl tosylate in the API it was decided to test tosic acid in the crude API stage, and a limit test HPLC-UV method was developed for this purpose The limit test was performed as an in-process control (IPC) so that if the level of tosic acid was above the acceptable limit then the process step was repeated until the acceptable limit was reached This ensured that the level of tosic acid was below the acceptable limit at the crude API stage; thus, the testing of API for n-propyl tosylate was not necessary However, n-propyl tosylate was still tested (by GC-MS) in the API in early batches to ensure compliance and to understand the purging capability of the processes Purge reflects the ability of a process to reduce the level of an impurity, and the purge factor is defined as the level of an impurity at an upstream point in a process divided by the level of the impurity at a downstream point in the process [4] As more analytical testing data and process understanding accumulated, it was possible to set a target value of tosic acid in the crude API that would provide the API with an acceptable level of n-propyl tosylate with confidence

Another project had an intermediate that was an MSA salt The subsequent steps involved methanol, ethanol, and n-propyl alcohol Therefore, MMS, EMS, and n-propyl methane sulfonate (n-PMS) were potential genotoxic impurities in the drug substance The strategy of testing MSA (by LC/MS) at an appropriate synthetic step was used successfully to control MMS, EMS, and n-PMS in the drug substance, simplifying and alleviating analytical workload

The rationale, procedures, and practices employed to assess, analyze, and control genotoxic impurities are discussed with a few case studies Classification of impurities assesses the level of risk that a genotoxic impurity poses and provides a basis to set priorities for toxicological testing, analysis, and control The qualification strategy and flowchart for assessment and control present representative

procedures and practices The actual case studies presented demonstrate the different circumstances in which genotoxic impurities are encountered in drug substance development and highlight the various approaches employed to control them properly

The successful management of genotoxic impurity issues in a pharmaceutical environment involves many different disciplines including toxicology, chemical development, pharmaceutical development, analytical development, drug metabolism and pharmacokinetics, and regulatory affairs Effective communication and collaboration between disciplines is absolutely essential Genotoxic impurities present demanding challenges for pharmaceutical development Proper management of genotoxic impurities is a critical element in achieving the worthwhile goal of ensuring the quality and safety of drug substances and products for patients

The author acknowledges the following colleagues for their valuable discussions and contributions to the case studies presented in this chapter: Nelu Grinberg, Sherry Shen, Shengli Ma, Earl Spinelli, Xudong Wei, Bing-Shiou Yang, Dan Fandrick, Jason Mulder, Rogelio Frutos, Thomas Tampone, Yibo Xu, Steve Han, Xiao-Jun Wang, Li Zhang, Yongda Zhang, and Maurice Marsini