ABSTRACT
Impurities arising from synthesis and formulation processes, or from degradation of the active pharmaceutical ingredient (API) and excipients, may be present in finished drug products (DPs) The safety of impurities is a concern for both pharmaceutical companies and regulatory authorities, and the control of impurities is generally addressed by International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) guidelines from a regulatory point of view However, these documents do not provide any specific guidance related to the determination of acceptable levels for genotoxic impurities (GTIs)
GTIs are a subset of impurities that may induce genetic mutations, chromosomal breaks, and/or chromosomal rearrangements and have the potential to cause cancer in humans In 2002, the European Medicines Evaluation Agency (EMEA) published a draft position paper that addressed the issue of GTIs from both toxicological and quality perspectives1 The original paper became a draft guideline in 2004, and the concept of threshold of toxicological concern (TTC) was introduced first in this
131 Introduction 365 132 Industrial Practice 366 133 Case Studies 367
1331 Case Study 1 368 1332 Case Study 2 372 1333 Case Study 3 376
13331 Experimental Section 377 13332 Results and Discussion 377 13333 Conclusions 382
References 383
guideline2 The TTC concept is part of a risk assessment approach that aims to establish a conservative while virtually acceptable maximum daily intake below which a GTI poses negligible safety concerns to human health The utilization of TTC establishes a 15 µg/day limit based on a lifetime exposure resulting in a cancer risk of 1 in 100,000, a risk level that the EMEA guideline considers justified because of the benefits derived from pharmaceuticals The EMEA guideline was finalized in 20063 and became effective in 2007 Later, durational adjustments to the TTC limit were also accepted by the EMEA for medicinal products in development phases4 As shown in Table 131, a set of staged TTC limits were published dependent on the duration of exposure in clinical trials Following the EMEA, the US Food and Drug Administration (FDA) also published its own draft guidance in 2008 providing recommendations on how to evaluate the safety of genotoxic and carcinogenic impurities during clinical development and for marketing applications5 The recommended exposure thresholds in this guidance are generally aligned with the staged TTC limits recognized by the EMEA
The pharmaceutical industry recognizes its obligation and strives to make substantial efforts during drug development to control all impurities, including GTIs, at safe levels
Early in the discovery stage, structural alert assessment is performed to assist lead optimization and final candidate selection In silico structure-activity relationship (SAR) tools are usually used to predict genotoxicity of the drug candidates Recently, it has also been recognized that early SAR screening of potential hydrolysis products and known metabolites may facilitate the elimination of high-risk candidates from the pipeline and save time and resources in the development phase
Prior to the first in-human clinical investigation, synthetic schemes to manufacture the API and key raw materials must be evaluated Potential impurity structures, including starting materials, isolated intermediates, reagents and solvents, by-products, catalysts, and counterions, are submitted for in silico genotoxicity analysis For structural alert hits, the impurities that have proved challenging
TABLE 13.1 Staged TTC Limits for GTIs
to control are isolated and purified and are investigated by in vitro genotoxicity assays, such as the Ames test If GTIs are found, the optimum option is to develop an alternative synthetic route that can remove these impurities If this is not technically feasible, impurity specifications should be set based on maximum administered doses and maximum duration of study using the staged TTC concept Accordingly, analytical methods with adequate selectivity and sensitivity are commonly required In cases where testing is not performed, scientific justification must be given, which usually includes the evaluation of compound reactivity, purification steps, point of introduction, and so on However, analytical data may still be needed later in development to support the indicated rationale (For some very highly reactive impurities, purge is obvious; so no data would be required Additionally, for families of GTIs data are often generated for the impurity with the highest probability of presence and the others do not have to be tested) For formulation development, a similar practice may also apply While, the impurities of interest are often degradation products which may be introduced in the manufacturing process of the final dosage form or during storage To identify potential degradation impurities for genotoxicity assessment, stress studies are commonly needed
The identification and control of GTIs in APIs and formulation is a dynamic approach that needs to be reassessed throughout the drug development process For example, if the synthesis and/or formulation process has changed the new process needs to be assessed If the acceptable toxicology limit has changed due to the changes in a clinical program, the capabilities of the manufacturing process and analytical method for control at the new level need to be assessed as well
Three industrial case studies are discussed in this section As mentioned in Section 132, when the maximum daily dose and/or the duration of study are changed in a clinical program the limit of GTIs may also have to be updated based on the staged TTC concept In the first case study, the specification of a GTI in a drug substance was lowered 10-fold when the clinical program went from early phase (phase I to phase IIa) to late phase (phase IIb to phase III) Accordingly, a new API-manufacturing process was developed and risk assessment followed by design of experiments (DOE) was applied to assess the GTI removal capability of the new process An analytical method with improved sensitivity was also developed to fit its intended use in this case GTI determination at trace levels requires highly sensitive analytical methods, which may possess tremendous challenges in a quality control (QC) environment The most conventional analytical instrumentations in pharmaceutical analysis are high-performance liquid chromatography-ultraviolet (HPLC-UV) and gas chromatography (GC) with a flame ionization detector However, they are often inadequate for accurate analysis of GTIs at low parts-permillion levels, depending on properties of the analytes and sample matrices6 In the second and third case studies, attention is focused on specific analytical challenges related to GTI determination
Quality by design (QbD) is a systematical approach to understand and optimize product quality attributes to meet specific objectives The focus of this concept is that quality should be built rather than tested in a product QbD approaches have been described in a number of regulatory guidelines7 For example, ICH Q10, Pharmaceutical Quality System, introduced the concept of control strategy, defined as a planned set of controls derived from product and process understanding that ensures process performance and product quality8 Hence it is essential to fully understand the manufacturing process to ensure that the quality of both drug substance and DP is appropriate and consistent7 As part of pharmaceutical QbD approaches, DOE is efficient in process development for evaluating the effects and possible interactions of several factors This case study discusses the control strategy to regulate a GTI during the clinical phase transition period The control strategy included the development, risk assessment, and utilization of the specification settings, purification process, and analytical testing method DOE was applied in process development to facilitate the full understanding and hence optimization of the purification process for GTI removal
API #1 was under development to treat certain cancers By-product A can be formed by the hydrolysis of API #1 This impurity is mutagenic and therefore should be strictly controlled During early development (phase I to phase IIa), the maximum daily dose of API #1 was 500 mg and the maximum duration of treatment was 6 months Accordingly, the limit of GTI compound A in the drug substance was set to 20 ppm using the staged TTC limit (see Table 131):
1 µg
5 mg 2 ppm
00 0= (131)
A specific reverse-phase high-performance liquid chromatography (RP-HPLC) method with ultraviolet (UV) detection was used for its analysis This method has a linear range from 2 to 30 ppm, the limit of quantitation (LOQ) is 2 ppm, and the limit of detection (LOD) is 1 ppm The chromatograms for LOQ and LOD are shown in Figure 131
When the clinical program entered the late phase (phase IIa and beyond) and was prepared for registration, the drug substance specification of compound A was reassessed in accordance with the EMEA GTI guideline for marketed products A limit of 2 ppm was proposed based on the new maximum daily dose for API #1 of 800 mg/day:
= ≈ 15µg
800 mg 19 ppm 2 ppm (132)
The new specification represented a reduction in the level of compound A from not more than (NMT) 20 ppm to NMT 2 ppm in the drug substance, requiring an update in all regulatory dossiers and the corresponding analytical method for this specification An RP-HPLC method with mass spectrometry (MS) detection was developed to support the proposed specification This method has an LOQ of 06 ppm, as shown in Figure 132
Compound A ranged from less than 2 ppm to 20 ppm in the drug substance in early batches A new purification process was then developed using charcoal to reduce the level of this compound A risk analysis of batches that failed specification was conducted to investigate critical parameters for the new process The following parameters required further investigation based on the risk assessment: starting concentration of compound A, amount of charcoal, aging time, and temperature with charcoal A DOE was run to test for these parameters, and the results are shown in Table 132 The design is a standard half fraction of a 24 factorial of resolution 4 with two-way interactions confounded with other two-way interactions It has three center points added to the design, denoted using “*” in Table 132 Figure 133 shows the DOE normal plot of the standardized effects, assessing the purification process based on the data in Table 132 Basically, compound A starting concentration and charcoal amount had the most significant impact on this processing step The higher
the compound A starting concentration, the higher the final concentration of compound A in API #1 A larger amount of charcoal was more effective in reducing the compound A level in API #1 Additionally, aging time and the interaction of both compound A starting concentration and charcoal amount had some effect over the process but were not as significant as the other two factors Based on the DOE results, the following specification was set for the upstream process as part of the control strategy: compound A starting concentration = NMT 55 ppm Hence, the three replicates at center points with the compound A starting concentration of 55 ppm represented a worst-case scenario in real-life manufacturing
Using the replicates from the center points, a mean of 11 ppm and a standard deviation of 03 (026) ppm were obtained The standard deviation represented the combined process and analytical variability Using three standard deviations (3σ) from the mean to achieve a confidence interval, the process would support a specification of NMT 11 + (3 × 03) = 2 ppm of compound A after purification To assess analytical method variability alone, a standard deviation (σ analytical) of 010 ppm was estimated using the analytical method transfer data, where the same three batches were tested using the new high-performance liquid chromatography-mass spectrometry (HPLC-MS) method in different analytical laboratories at different sites (Table 133) Using the calculated analytical variability, the standard deviation of the process (σ process) was determined to be 024 ppm using Equation 133:
σ = σ − σprocess 2
(133)
TABLE 13.2 DOE Design and Results Assessing the Purification Process to Remove GTI Compound A from the Drug Substance
Therefore, the main component of variability for compound A level in the drug substance comes from process variability
To assess the variability in compound A that could be expected throughout the retest period of API #1, three laboratory-scale drug substance batches were placed on supportive stability and three full-scale drug substance batches were placed on commitment stability In both the stability data sets, no correlation was observed for compound A level with time Hence, the environment or stability of API #1 with respect to the GTI compound A does not add to the variability for determining the compliance of compound A level to the proposed specification limit
As discussed earlier, the risk of failure for the proposed specification of compound A of NMT 2 ppm was judged to be low Very little variation is observed in the analytical method, and essentially no variation is contributed by any instability of the drug substance The main component of variability for compound A level in the drug substance comes from the process variability This process has been studied in a DOE, and a worst-case manufacturing condition was used to estimate the level and variation of compound A that could be expected All risk assessment and DOE results supported the proposed new specification for compound A at release of the drug substance to ensure that the product meets the specification over the retest period
EDC or EDCI, acronyms for 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide, a water-soluble carbodiimide usually obtained as the hydrochloride, is generally used as a carboxyl activating agent for the coupling of primary amines to yield amide bonds Additionally, EDC can be used to activate phosphate groups Common uses for this carbodiimide include peptide synthesis, protein cross-linking to nucleic acids, preparation of immunoconjugates, and many drug substance syntheses EDC has not been identified as a known carcinogen, but its toxicology properties have not been fully understood by thorough investigations Therefore, the limit of EDC in a drug substance needs to be assessed case by case and varied in different study phases In this case study, a 50 ppm initial limit was posed on EDC in API #2 based on good-laboratory-practices toxicity studies
TABLE 13.3 GTI Compound A Levels Determined Using HPLC-MS during Analytical Method Transfer
The analysis of EDC in the drug substance was a challenge Conventional HPLC-UV and GC-FID methods are not suitable as EDC lacks strong UV adsorption and volatility However, the EDC molecule has three nitrogen atoms (one saturated and two unsaturated) and hence can easily adopt H+ and be ionized and detected by MS Therefore, a liquid chromatography-mass spectrometry method was developed for the analysis of EDC in API #2 As mentioned earlier, EDC is usually in its hydrochloride form when used in organic syntheses During method development, it was found that the stability of EDC hydrochloride in the sample solution may become a severe issue When dissolved in water, EDC hydrochloride can convert to another form following the scheme in Figure 134 This alternative form of EDC has decreased H+ adoptability, and its retention behavior in high-performance liquid chromatography is also altered Figure 135 shows chromatograms of 25 ppm EDC hydrochloride (relative to drug substance) in an acetonitrile/water (80/20, v/v) solution that was repeatedly injected The intensity of EDC peak in the initial injection of freshly prepared solution was approximately 240,000, and the peak intensity decreased to 220,000 in the second injection After 24 hours, the EDC peak disappeared The kinetics of EDC conversion to its alternative form can be significantly accelerated if the drug substance is a hydrochloride salt In this case study, API #2 is a dihydrochloride salt Figure 136 shows the chromatogram of a freshly prepared drug substance solution spiked with 25 ppm EDC Apparently, the EDC peak is not observed in the chromatogram
To solve the solution stability issue, various attempts have been made to maintain the preferable form of EDC in the sample solution, and the most effective approach was to prepare the drug substance containing EDC in an alkaline solvent In this case, the sample solvent consisted of acetonitrile/water/ammonia at 80/20/02 (v/v/v); 20 mM of ammonia formate plus 005% of formic acid in water and acetonitrile were used as the mobile phases A and B, respectively, for gradient elution of EDC on a C18 column EDC was detected in selected ion monitoring (SIM) at +156 m/z Although the initial limit of EDC was set at 50 ppm, the LOQ of the analytical method was validated at 2 ppm, anticipating a decreased limit in later development phases The chromatograms in Figure 137 demonstrated the stability of EDC after 24 hours at 2 ppm level, whereas the chromatograms in Figure 138 show the stability of 2 ppm EDC spiked in the drug substance in an alkaline sample solution In both cases, the difference of EDC peak areas in fresh and aged solutions was less than 10%, which met the predefined acceptance criterion of NMT 30% for solution stability The accuracy/recovery result could be obtained by comparing the peak areas of EDC in the chromatograms in Figures 137a and 138a The recovery value was 95%, which met the predefined acceptance criterion (70%–130%) as well
Alkyl sulfonates can be formed from the esterization of alkyl and aryl sulfonic acids with low-molecular-weight alcohols Alkyl and aryl sulfonic acids, for example, methanesulfonic acid, benzenesulfonic acid, and p-toluenesulfonic acid, are common counterions for salt formation in drug substances Meanwhile, low-molecularweight alcohols, for example, methanol, ethanol, and isopropanol, are frequently used as organic solvents to synthesize and/or purify drug substances and hence may become residual solvents in APIs Alkyl esters of alkyl and aryl sulfonates are a known class of GTIs, and their presence in a drug substance or a DP has been closely scrutinized by regulatory agencies and the pharmaceutical industry9 Since the European Directorate for the Quality of Medicines and Healthcare expressed its concerns on the alkyl mesylate impurities in mesylate salts in 2000,10 a significant amount of research and development work has been focused on this group of compounds11
The determination of alkyl sulfonates at the parts-per-million level may possess multiple analytical challenges, which have been reviewed previously in the literature11 Briefly, three types of approaches have been developed: (1) direct analysis using GC-FID12 or GC-MS-SIM13, (2) direct analysis using HPLC-MS-SIM14, and (3) indirect analysis with derivatization9,15-18 Chemical instability of alkyl sulfonates
is the major problem for direct analysis methods, which may raise concerns regarding sensitivity and recovery9 Derivatization may stabilize these highly reactive analytes; however, technical issues such as long reaction time,9 low recovery,16 and low specificity15 were still observed
In this case study, a generic headspace-GC method was developed for the determination of alkyl sulfonate with in situ derivatization As shown in Figure 139, this method used chloride ion as the derivatizing reagent and the derivatization product, alkyl chloride, was analyzed using GC-FID Factors that may affect the derivatization reaction, for example, choice of derivatizing reagent and its concentration, solvent pH, reaction time, and so on, were evaluated and optimized This method was successfully validated over the range of 05-5 ppm with an LOD of 02 ppm The specificity and sample matrix effect were also studied using a series of marketed APIs as model compounds
13.3.3.1 Experimental Section An appropriate amount of a drug substance containing alkyl sulfonate was dissolved in 1 mL of N-methyl-2-pyrrolidone (NMP) in a headspace vial One hundred microliters of the derivatizing reagent was then added, and the vial was sealed immediately After in situ incubation at 80°C for 15 minutes, the headspace sample was analyzed on a DB-624 column (Agilent J&W, Santa Clara, California) with the oven temperature held at 35°C for 10 minutes Two analytes, methyl methanesulfonate (MMS) and ethyl methanesulfonate (EMS), were studied in this case Their derivatization products, methyl chloride and ethyl chloride, elute at about 25 and 56 minutes, respectively
13.3.3.2 Results and Discussion 13.3.3.2.1 Using Hydrochloric Acid as the Derivatizing Reagent The capability of hydrochloric acid (HCl) as the derivatizing reagent was evaluated first HCl at various concentrations (1, 2, and 3 N, respectively) was added into headspace vials containing MMS and EMS, respectively, and analyte recovery values were obtained As shown in Figure 1310, the recovery of MMS was approximately 100% at all three HCl concentration levels, whereas for EMS the recovery increased slightly from 80% to 90% when the HCl concentration varied from 1 to 3 N Therefore, HCl is capable of derivatizing both MMS and EMS with satisfactory recoveries
Alcohols are frequently used as organic solvents to synthesize and/or purify drug substances, and hence they may become residual solvents in APIs As shown in Figure 1311, alcohols may also react with chloride ion to form alkyl chloride in the
acidic condition and hence interfere with the accurate determination of alkyl sulfonate by giving positively biased results
To evaluate if residual alcohols would interfere with the analysis of alkyl sulfonate, HCl at various concentrations (1, 2, and 3 N) was added into headspace vials containing methanol and ethanol The amount of methanol ranged from 60 to 3000 ppm, and the amount of ethanol ranged from 100 to 5000 ppm Note that 3000 ppm and 5000 ppm are the ICH limits for methanol and ethanol, respectively As shown in Tables 134 and 135, an acidic environment favors the derivatization of alcohols into alkyl chlorides, which might interfere with the analysis of alkyl sulfonate Therefore, it is not appropriate to use HCl as the derivatizing reagent if residual alcohols are suspected in the API
13.3.3.2.2 Using Tetramethylammonium Chloride as the Derivatizing Reagent Tetramethylammonium chloride (TMACl) was studied as an alternative derivatizing reagent In addition, the effect of solvent acidity was evaluated TMACl was prepared in water, 001 N HCl, and 01 N HCl at three concentration levels (1, 2, and 3 N, respectively) and was added into headspace vials containing 3000 ppm of methanol and 5000 ppm of ethanol Table 136 lists the relative amount of methyl chloride detected Basically, a higher TMACl concentration and a stronger acidic environment favored the derivatization of methanol into methyl chloride No ethyl chloride was detected in any of these experiments Therefore, to prevent the alcohol from interfering, the most suitable derivatizing reagent is 1 N or 2 N TMACl prepared in water Under such a derivatizing condition, the recovery of MMS in the presence or the absence of 3000 ppm methanol was about 100% and the recovery of EMS in the presence or the absence of 5000 ppm ethanol was above 80%
13.3.3.2.3 Active Pharmaceutical Ingredient Matrix Effect The sample matrix effect on derivatization was studied using a series of model compounds These compounds are marketed APIs with various physiochemical properties, and their chemical structures are shown in Figure 1312 In this experiment, 200 mg of the model compound was dissolved in NMP in headspace vials
TABLE 13.4 Methanol Interference with the Analysis of MMS Using Hydrochloric Acid as the Derivatizing Reagent
TABLE 13.5 Ethanol Interference with the Analysis of EMS Using Hydrochloric Acid as the Derivatizing Reagent
TABLE 13.6 Methanol (3000 ppm) Interference with the Analysis of MMS Using TMACl as the Derivatizing Reagent
and spiked with 2 ppm MMS and 2 ppm EMS, and the derivatizing reagent was 1 N TMACl in water As shown in Figure 1313, the recoveries of MMS and EMS in most model compounds were within 80%–120%, which was satisfactory for the determination of impurities at low-parts per million levels The only exceptions were acetaminophen and diclofenac sodium salt, in which low recoveries were observed In addition, as the ethyl ester group in clofibrate may react with chloride ion to form ethyl chloride, positive false signal was detected in clofibrate without spiking EMS EMS showed low recovery in acetaminophen, which may be due to the slow kinetics in this sample matrix Increasing chloride concentration or incubation time was found to improve the recovery For instance, in the case of acetaminophen the recovery of EMS increased to 86% using 2 N TMACl as the derivatizing reagent Neither methyl chloride nor ethyl chloride was detected in the diclofenac sodium salt, indicating that the recovery values of both MMS and EMS were close
to zero This phenomenon might have been caused by the high pH in the sample matrix of sodium salt To test this hypothesis, 1 N TMACl prepared in water, 3 N ammonium hydroxide, and 15 N ammonium hydroxide, respectively, were added to headspace vials containing 2 ppm MMS and 2 ppm EMS The peak areas of MMS and EMS are shown in Figure 1314 Obviously, a higher basicity in the solution decreased the recovery of both analytes, especially MMS To increase the acidity of the solution and hence improve the recovery, an appropriate amount of HCl was added into the sample matrix to neutralize the diclofenac sodium salt When the diclofenac sodium salt was half neutralized, the recovery of EMS was about 30%, whereas MMS was not yet detected When the diclofenac sodium salt was fully neutralized, the recovery of EMS improved to 88% However, the recovery of MMS was only about 48% After intensive investigations, it was found that the uncommonly low recovery of MMS was due to the competitive reaction of MMS with the carboxylic ion of the diclofenac sodium salt (see Figure 1315)
Theoretically, the product ratio of methyl chloride versus methyl ester can be calculated using Equation 134 This ratio is a constant independent of MMS concentration, if the amounts of chloride ion and carboxylic ion are in great excess:
[ ] [ ]
[ ] [ ]
[ ] [ ]
[ ] [ ]= = =
CH Cl
RCOOCH
Cl MMS
RCOO MMS
Cl
RCOO 3
k
k
k
k k
(134)
Therefore, the recovery of MMS can be normalized to 100% using a correction factor This correction factor is calculated as 100% divided by the observed recovery of MMS in the presence of the diclofenac sodium salt Table 137 shows the recovery values after correction using a factor of 164 The use of a correction factor can also be applied to other APIs if recovery is an issue due to a similar mechanism However, it should be noted that this approach has its limitation in situations where more than one component can interfere and the levels of interfering components are varied in different samples (eg, degradation products in stability samples)
13.3.3.2.4 Effect of Incubation Time A kinetic study was performed to evaluate the effect of incubation time on the recoveries of MMS and EMS As shown in Figure 1316, a period of 15 minutes is a robust incubation time for the derivatization of MMS and EMS with chloride ion in the presence of most model compounds
13.3.3.3 Conclusions A generic headspace GC-FID method was developed for the analysis of alkyl sulfonate using chloride ion to derivatize the analytes The advantages of this method include a short derivatization time, high recovery, and good QC friendliness This
TABLE 13.7 Normalized Recovery of MMS
method was validated over a range of 05-5 ppm with an LOD of 02 ppm In addition, the sample matrix effect was studied in a series of model compounds and the limitations of the method were studied Options for recovery improvement were also provided
