Vous êtes sur la page 1sur 79

GENOTOXIC IMPURITY ANALYSIS IN PHARMACEUTICALS

by

Matthew Austin Janson

A Thesis Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree MASTER OF SCIENCE (PHARMACEUTICAL SCIENCE) December 2009

Copyright 2009

Matthew Austin Janson

Acknowledgements

The following individuals are being acknowledged for their contributions in this thesis: Dr. Wei-Chiang Shen for his role as committee head and for his invaluable guidance throughout the process of writing this thesis. Dr. Curtis Okamoto and Dr. Clay Wang for serving on the thesis committee. Dr. Peter Zhou and Dr. Sophie Wang for scientific discussion and review of this thesis.

ii

Table of Contents

Acknowledgements List of Figures List of Abbreviations Abstract Introduction: Genotoxicity Genotoxicity in Pharmaceutics Genotoxic Impurity Assessment and Control Regulatory Aspects Analytical Methods to Monitor Genotoxic Impurities Chapter 1: Analytical Challenges and Techniques Sample Handling Issues Selectivity from Matrix Other Separation Techniques Sensitivity Method Validation and QC Transfer Chapter 2: Analytical Method Details Hydrazines Aldehydes Nitro Compounds Arylamines Chapter 3: Conclusion References Appendix A: Analytical Method Details

ii iv vi viii 1 2 5 11 17 19 20 21 23 25 28 30 31 37 45 49 53 58 62

iii

List of Figures

Figure 1: Genotoxic Impurity Decision Tree Figure 2: Decision Tree for GTI (Genotoxic Impurity) Testing Strategy Figure 3: Scaled TTC: Concentration vs. Daily Dose Figure 4: Functional Groups Associated with Genotoxic Impurities and Their Possible Transformations/Reactions Figure 5: Decision Tree for Selecting an Appropriate Detector (Chrom = Chromophore and Sens = Sensitivity) Figure 6: Acetone Azine in the LOQ (0.1 ppm) and 1 ppm Levels. Black Trace is Blank Figure 7: Effect of Various Alcohols in HILIC Mobile Phase to Separate Hydrazines (1 = 1,2-dimethylhydrazine, 2 = 1,1-dimethylhydrazine, 3 = methyl hydrazine and 4 = hydrazine) Figure 8: Derivatization Reaction of Formic Acid and Formaldehyde with Acidified Alcohol Figure 9: HPLC/UV Chromatograms of Derivatized Aldehydes (1 = formaldehyde, 2 = acetaldehyde, 3 = propanal, 4 = butanal, 5 = pentanal and 6 = hexanal) in the Presence of Common Excipients (A = Aldehyde Standard with no Excipients, B = Hydroxymethyl Cellulose and C = Hypromellose) Figure 10: GC/MS Method 2 Chromatograms of Derivatized Aldehydes (1 = formaldehyde-PFBHA, 2 = PFBHA, 3 and 4 = acetaldehyde-PFBHA) in the Presence of Common Excipients (A = Hypromellose, B = Hydroxylpropyl Cellulose and C = PFBHA Blank)

6 9

14 18

26

33

35

39

41

43

iv

List of Figures: Continued

Figure 11: Comparison of HPLC (top) and UPLC (bottom) Stability-indicating Methods. Impurities I and II (8nitro-6-methoxyquinoline) are Well Resolved from the Drug Substance, Primaquine Figure 12: Schematic Showing Spiking Studies of Impurity III at Various Stages in the Synthesis of the API (GW786034)

47

51

List of Abbreviations

ADI: Acceptable Daily Intake ALARP: As Low as Reasonably Possible API: Active Pharmaceutical Ingredient CAD: Charged Aerosol Detector CE: Capillary Electrophoresis CLND: Chemiluminescent Nitrogen Detector DEREK: Deductive Estimation of Risk from Existing Knowledge ECD: Electron Capture Detector ELSD: Evaporative Light Scattering Detector FID: Flame Ionizing Detector GC: Gas Chromatography GTI: Genotoxic Impurity HILIC: Hydrophilic Interaction Liquid Chromatography IC: Ion Chromatography LOD: Limit of Detection LOQ: Limit of Quantitation MCASE: Multi-computer Automated Structure Evaluation MNC: 1-methyl-4-nitro-3-propylpyrazole-5-carboxamide MNP: 1-methyl-4-nitro-3-propylpyrazole-5-carboxylic acid MS: Mass Spectrometry

vi

List of Abbreviations: Continued

NMP: N-methylpyrrolidone PDE: Permitted Daily Exposure PFBHA: O-2,3,4,5,6-(pentaflurorbenzyl) hydroxylamine PPM: parts per million QbD: Quality by Design QC: Quality Control RP-HPLC: Reversed Phase High Performance Liquid Chromatography RSD: Relative Standard Deviation SFC: Supercritical Fluid Chromatography SIM: Single Ion Monitoring S/N: Signal to Noise SPE: Solid Phase Extraction SPME: Solid Phase Microextraction SRM: Selected Reaction Monitoring TTC: Threshold of Toxicological Concern UPLC: Ultra Performance Liquid Chromatography v/v: Volume/Volume

vii

Abstract

This review paper covers the analytical methods used to monitor genotoxic impurities. Specifically, the impurities discussed here contain hydrazine, nitro, arylamine or aldehyde functional groups as found in pharmaceuticals. These genotoxic impurities can come from many places including starting materials, reagents, intermediates, solvents or unwanted side reactions of the active pharmaceutical ingredient (API) synthetic process that get carried over into the final product. In addition, the API itself can decompose to form genotoxic impurities or they can form in the drug product by reaction between excipients or containers and the API. The control of these potent impurities is critical which are generally monitored at very low levels in the ppm and ppb range. This poses many challenges on the analytical methods to achieve high sensitivity as well as separation of the impurities from the sample matrix for selectivity.

viii

Introduction: Genotoxicity

Genotoxic substances are those which impact genetic material by means of mutations. Mutations can be chromosomal breaks, rearrangements, covalent binding or insertion into DNA during replication. Mutations may also occur indirectly by activating a cell to produce genotoxic substances. These changes to the genetic material, which can be caused by exposure to very low levels of a genotoxin, can lead to cancer [27]. Because of this, it is important to identify genotoxic substances followed by monitoring and control at very low levels to ensure safety to the public.

Genotoxins are ubiquitous and can be found in numerous sources such as food products, food supplements, air and water supplies. Genotoxins may also be found in pharmaceutical products, which will be the main focus of this paper. Specifically, the class of genotoxins covered will be those which are electrophilic compounds that can react with DNA. An example from this class is alkylating compounds that form covalent derivatives on the N7 nitrogen of guanine. One potential result of this can lead to a mutation such as DNA base mispairing. Another example is the formation of an adduct of a nitrogen-containing genotoxin with DNA, an event that could inhibit DNA synthesis.

Genotoxicity in Pharmaceutics

Genotoxic substances in pharmaceutics, known as genotoxic impurities, are gaining more attention. These substances add risk without any benefit to pharmaceuticals. Genotoxic impurities in relation to pharmaceuticals can come from many places including starting materials, reagents, intermediates, solvents or unwanted side reactions from the API synthetic process that get carried over into the final product. In addition, the API itself can decompose to form genotoxic impurities or they can form in the drug product by reaction between excipients or containers and the API. The employment of these compounds within the synthetic process is logical as these substances are reactive building blocks that come together to form complex drug substances. This reactivity, however, also means that they can readily react with and damage DNA if carried over into a product taken by patients.

The Viracept (nelfinavir mesylate) contamination incident is an example of a case that contributed to the heightened awareness and potential dangers of genotoxic impurities in a pharmaceutical product. Various lots of this HIV drug distributed by Roche Pharmaceuticals were pulled off the market in 2007. Strange odors were noticed from different batches of the drug product and further analysis revealed abnormally high levels of ethyl mesylate (ethyl methanesulfonate). Ethyl mesylate is an alkylating class of genotoxic impurity that can covalently

bind to DNA and enhance cancer risk. The source of ethyl mesylate was determined to be from residual ethanol used in manufacturing equipment cleaning and methane sulfonic acid contained in the drug product. This incident prompted much discussion regarding specifications for the amounts of this genotoxic impurity as found in one particular pharmaceutical product. This highly publicized incident helped bring more attention to the issue of genotoxic impurities in pharmaceuticals.

A major challenge with genotoxic impurities in pharmaceuticals is the balance between appropriate control with minimal impact to the time and costs associated with developing life-improving drugs. Some may argue that setting such low limits for genotoxic impurities is not always practical and lacking solid scientific justification.

If the genotoxic impurity is formed during the API synthesis, elimination may be achieved by changing the synthetic route and reagents used. Or, introduction of a genotoxic reagent should be introduced in the synthetic scheme as early as possible to allow elimination of this substance during subsequent reaction and purification steps. However, this may not be feasible as API synthesis can be quite complicated and limited based on chemistry and available reagents making these approaches impractical. Another means of genotoxic impurity control in this situation can be implementation of additional purification steps to the

synthetic route to drive off the impurities. However, if this is not effective, or the genotoxic impurity forms after the API synthesis (e.g. API degradation or reactions with excipients/containers), the impurity levels still need to be measured and monitored.

The additional work put into controlling the genotoxic impurity levels and the analytical method development, validation and regular testing will significantly impact the drug development timeline. More time and resources will be required, further delaying the time for the drug to reach the market, as well as increase the drug development costs. Because of this, a strong risk assessment approach needs to be applied to the development process that balances patient safety with control of the genotoxic impurities.

Genotoxic Impurity Assessment and Control

Genotoxic impurities can be identified by various methods: 1) as already known genotoxins, 2) possessing similar functional groups with known genotoxins, 3) testing positive by genotoxicity assays or 4) marked as a potential genotoxin by one of many computer-based structure-activity software programs. Multicomputer automated structure evaluation, MCASE (http://www.multicase.com/products/prod01.htm), Topcat (http://www.accelrys.com/products/topkat/) and deductive estimation of risk from existing knowledge, DEREK (http://chem.leeds.ac.uk/luk/derek/) are common software programs used to evaluate and provide alerts for impurities which may be genotoxic [6]. In addition, published data can be used to evaluate a potential genotoxin. If the software program provides a structural alert, the impurity will go into AMES bacterial mutagenicity testing. Results from the AMES test will be considered more definitive than the in-silico data.

Muller classifies impurities into 5 groups [22] which can assist with setting safe specification levels. Class 1, being the most dangerous, are known genotoxic carcinogens that need to be avoided as much as possible. Class 2 compounds are genotoxic, but with unknown carcinogenicity and need to be handled with a threshold of therapeutic concern (TTC) approach, which will be discussed later. Class 3 impurities have alerting structures that are unrelated to the parent. The

assessment of this compound will place it into Class 2 or as an ordinary impurity (Class 5). Class 4 have parent-related alerting structure, in which the genotoxicity studies on the API have already been performed and can be applied to the related impurity. Class 5 impurities have no alerts and are considered ordinary impurities that falls within the scope of ICH guidelines (ICH Q3A, B and C) [12, 13, 14].

Muller [22] describes a genotoxic impurity assessment decision tree that includes the 5 impurity classes.

Figure 1: Genotoxic Impurity Decision Tree

In the above decision tree, Class 1 compounds are first assessed to see if they can be eliminated from the product. If all practical attempts for elimination are fruitless, a risk assessment is performed, and if data exists (i.e. animal toxicology data), an impurity limit is set. In most cases, data is not available to support a limit, and a default level, known as the TTC is applied. This concept will be described in greater detail later in this paper. Class 2 compounds are assessed to see if they have toxicity threshold mechanisms, or levels at which no genotoxicity is observed from in-vivo testing. If a threshold exists, this level will be applied to the compound based on permitted daily exposure (PDE) calculations taken from the ICH guidelines. If a threshold cannot be ascertained, the default TTC will be applied. Class 3 compounds need to be tested to determine if they are positive for genotoxicity. If genotoxic, the decisions for the impurity follow the Class 2 path, whereas if it tests negative the impurity is controlled as an ordinary impurity according to ICH guidelines. Class 4 compounds can be assessed and qualified based on parent (API) genotoxicity. If the parent compound is genotoxic, then the Class 2 path is followed, otherwise it is treated like an ordinary impurity. Class 5 has no alerting groups and is also treated like an ordinary impurity.

A scientifically-based, practical approach to testing a specific genotoxic impurity or impurities for a particular product and process is applied. Only those genotoxic impurities likely to form or carry through the API synthetic scheme may

be monitored and those that are not likely to form will not be considered. In addition to these genotoxic impurities, genotoxic impurities from other sources such as API degradation or reactions with drug product excipients/containers may also be monitored.

The genotoxic impurity testing point can vary and needs to be determined with scientific basis and supporting data [24]. For example, an impurity that forms during the API synthesis may only require testing in the intermediate (and not the final product) that immediately follows introduction of the impurity. The synthetic process may significantly purge the impurity over various reaction steps, processing and purification. With supporting data, this allows for higher impurity specifications, but tested at an earlier point in the synthetic process. Further testing will occur in later intermediates only if the amount of the impurity in the first intermediate tested exceeds a predetermined level. This predetermined level can be found by spiking studies that relate impurity survival throughout the synthesis. An excellent case study involving genotoxic impurity testing of intermediates and not the final product will be described later in this paper. Obviously, the testing point and impurity level specifications will vary on a caseby-case basis.

Figure 2: Decision Tree for GTI (Genotoxic Impurity) Testing Strategy

Figure 2 was presented by Pierson [24] and describes suggested points for testing of genotoxicity impurities present in the API synthetic scheme. If the genotoxic impurity is introduced in the final synthetic step, the API will need to be tested. If the impurity is introduced at the penultimate step, penultimate testing will assess if further testing is necessary. If the impurity level in the penultimate is lower than the API specification, testing of the API is not required. If the genotoxic impurity is introduced before to the penultimate, but within four steps prior to API formation, API testing may not be required if impurity levels in the intermediate does not surpass a pre-defined level. This is based on prior data

and spiking studies that correlated a limit in the intermediate with a safe or nonexistent level in the API. If levels are low enough, only the intermediate in which the impurity is introduced may require testing. If the impurity introduction is more than 4 steps before the API synthesis, a no-testing rational may be applied with support that the synthetic process purges the genotoxic impurity to insignificant levels.

Industry is more in favor of controlling genotoxic impurities in a phaseappropriate manner. Stringent control over genotoxic impurities in very early phase drug development is not appropriate because at this phase attrition is high and synthetic schemes are not completely set. Therefore, developing highly sensitive methods is less of a requirement than that for later phase and commercial products in which the chemical process is fully established and the drug is going into larger patient populations at more frequent dosing.

10

Regulatory Aspects

ICH provides guidelines for impurities (Q3A, B and C), but does not specifically provide acceptable levels for those genotoxic in nature. It does address the need of lowered levels for unusually potent impurities, such as those that are genotoxic. The conventional approach to genotoxic impurities has been that any amount of these substances are harmful and should be eliminated. Ideally, the genotoxic impurity should be non-existent, however, its presence may not be completely avoided and its detection may also be limited by the current analytical detection capabilities. In these instances, a risk-benefit assessment would be applied as recommended in the CHMP position paper on genotoxic impurities in pharmaceutical excipients [10].

There is evidence for potential threshold mechanisms that would allow for low levels of impurities to be present without certain genotoxicity. The impurity threshold-related mechanisms may not be known at the time it is being addressed. If toxicology testing provides a threshold for a genotoxic impurity, then a PDE can be extrapolated and applied to humans. However, if there is no sufficient evidence for a threshold mechanism, the genotoxic impurity levels should be controlled to as low as reasonably practical (ALARP) levels balanced with appropriate risk assessments.

11

Many times data to support a reasonable ALARP is not available, so a default limit known as the threshold of toxicological concern (TTC) is applied. TTC represents a level at which a patient can be exposed to a genotoxic impurity in a pharmaceutical with minimal risk while balanced with the therapeutic benefits of the pharmaceutical. This TTC concept was originally applied to foods [18, 25]. The current TTC default for pharmaceuticals is recommended at 1.5 microgram per day, which at this level represents one additional cancer case out of 100,000 if dosed over a lifetime [10]. This limit was originally based on animal carcinogenicity studies covering more than 700 compounds that was extrapolated and scaled to humans along with additional safety factors [18].

However, some argue the TTC default value, which was originally based on acceptable food levels, is overly conservative and hinders the development of important pharmaceuticals [5]. PhRMA published a paper that suggests a staged TTC under certain circumstances in which patients can be exposed to more than 1.5 micrograms per day [22]. With this approach, genotoxic impurities may be present in pharmaceutics at levels that are scaled to the dosing regimen of the drug. For example, drugs with higher, more frequent doses will have lower TTC values than a drug that has a lower overall dose. The cancer risk for doses less than 12 months in duration was lowered 10-fold to offset the shorter timeframe. Other factors that may effect the TTC value is the drug indication, patient life expectancy, pediatric patient populations, highly potent structural groups (e.g.

12

aflatoxin-like and N-nitroso compounds) and the coexistence of other genotoxic impurities with similar functional groups.

Figure 3 depicts the scaled TTC approach suggested by Muller [22] in terms of the drug dose and acceptable daily intake (ADI) of the genotoxic impurity. Here the ADI is expressed both relative to the drug dose (% and ppm) and in absolute impurity amount (g/day)

13

Figure 3: Scaled TTC: Concentration vs. Daily Dose

14

Figure 3: Continued

15

In the upper shaded area of Figure 3 are impurity concentrations that are 100 ppm relative to the API dose. At these very low concentrations, developing analytical methods becomes extremely challenging. On the other hand, the lower shaded area represents impurity concentrations 0.5%. For most cases, the authors recommend capping the ADI at this level as the quality of the drug substance may be impacted at impurity concentrations higher than 0.5%.

Additional concerns arise with regard to multiple genotoxic impurities in the same pharmaceutical. Bercu [2] and colleagues suggest that up to three genotoxic impurities, structurally related or not, may be acceptable during pharmaceutical development. An approach in this case is to add up the levels of the similar genotoxic impurities and to treat those that are different separately.

16

Analytical Methods to Monitor Genotoxic Impurities

There are many published analytical methods that describe the analysis of genotoxic impurities in pharmaceutics. The analytical method should be phase appropriate and evolving as the drug moves toward commercialization. Having a method in place early in the drug development process can help guide process chemistry and may later be fine tuned with enhanced sensitivity and robustness. There have been two excellent reviews on the analysis of genotoxic alkylating agents that have been published [8, 9]. Genotoxic impurities encompass a broad range of chemicals and this paper will focus on those not covered by those reviews. In all cases, the goal is maximum analyte sensitivity and resolution from matrix interferences. Figure 4 shows alerting functional groups commonly associated with genotoxic impurities and their possible transformations [5].

17

Figure 4: Functional Groups Associated with Genotoxic Impurities and Their Possible Transformations/Reactions

This paper will provide a review of the analytical methods within the context of pharmaceuticals for genotoxic impurities possessing the following structurally alerting groups: hydrazine, nitro, arylamine and aldehyde.

18

Chapter 1: Analytical Challenges and Techniques

There are many analytical challenges associated with genotoxic impurity analysis. The highly reactive nature of these impurities makes sample handling tricky and many precautions need to be taken. In addition, these low-level impurities are commonly in the presence of a large relative amount of API, and possibly excipients if the product is in the formulated stage. This leads to significant matrix interference which can impact proper analysis. As mentioned previously, the TTC sets the very low levels these impurities can exist in pharmaceuticals. Because of that, sensitivity is a major component that needs to be addressed. Once a method has been developed and validated, the next step can be implementation in a more routine quality control (QC) environment. The method should be robust, simple and run using standard equipment to ensure a smooth transfer from R&D to a regulated laboratory.

19

Sample handling issues

The reactive nature that defines a genotoxic impurity also makes its analysis and handling problematic. The impurity under investigation can easily react during extraction, sample preparation and analysis, yielding artificially low and variable results as well as possible interferences from its degradation products. This necessitates careful measures to maintain the integrity of the impurity up to the point of analysis. In addition, many of genotoxic impurities are low molecular weight compounds that render them quite volatile and therefore able to escape prior to and during analysis. Some ways to combat this is to limit the number of preparation steps, keep temperatures low and stabilization in a particular solvent or by derivitization.

20

Selectivity from Matrix

Major selectivity problems arise from the fact that the analytes are at very low levels in the presence of relatively large amounts of API and excipients. As a result, detrimental matrix effects can be amplified as evidenced with interference and poor recovery. As a result, separating the analyte from the matrix can be a major hurdle that needs to be addressed early on.

Separation from the sample matrix may be handled during sample preparation by extraction procedures or filtration. The choice of extraction procedure [liquidliquid, solid phase extraction (SPE) or solid phase microextraction (SPME)] depends on the nature of the compounds being separated. Of course, additional development work needs to be completed if the matrix changes, which can readily occur during the drug development process.

On the other hand, if sample preparation itself cannot fully eliminate the sample matrix, analytical instrumentation may be used for this task. For example, a triple quadrapole mass spectrometer can help eliminate overwhelming background interference from the sample matrix. In what is referred to as selected reaction monitoring (SRM), the target ion(s) is selected and further fragmented, allowing for exquisite selectivity in which only the target analyte ions from the genotoxic impurity are detected. The signal from SRM is lowered somewhat given the

21

nature of this detection technique, however, the signal to noise (S/N) greatly improves with a significant decrease in background interference and can lead to much better sensitivity.

Headspace sampling can also significantly reduce sample matrix effects in which only volatile compounds are analyzed. This is an excellent technique because in many cases the genotoxic impurity is more volatile than the sample matrix. Headspace coupled to gas chromatography (GC) eliminates potential contamination of large matrix amounts into the chromatographic system. Challenges with headspace sampling include the need for analyte volatility and possible thermal degradation of the analyte during the sample heating phase.

22

Other Separation Techniques

Reversed phase high performance liquid chromatography (RP-HPLC) is most commonly used to separate pharmaceutical compounds. This technique relies on hydrophobic interactions of analytes dissolved in liquid mobile phases with non-polar stationary phases. However, if the analytes are ionized or very polar, either ion chromatography (IC) or hydrophilic interaction liquid chromatography (HILIC) may provide better retention and separation. IC is based on ion-ion interactions between charged analytes and oppositely charged groups embedded in the stationary phase. HILIC separation is becoming more common and is based on retaining very polar compounds on a polar stationary phase. The mobile phase contains water that is believed to form a layer upon the stationary phase. Separation is driven by the analytes partitioning between this layer and the mobile phase. One aspect to keep in mind with HILIC is balancing the analyte solubility in the sample diluent with mobile phase compatibility.

GC is a widely used technique that is commonly used to analyze low molecular weight and low boiling point compounds in the gas phase. Compounds are introduced to the system either as a gas by headspace or injecting a small amount of liquid sample that gets vaporized. The compounds are retained on a wax-based stationary phase and eluted with a carrier gas prior to detection.

23

Though not utilized in the methods reviewed in this paper, supercritical fluid chromatography (SFC) is a technique that is also becoming more common and can be applied to genotoxic impurity analysis. The system is very similar to conventional GC and HPLC with the exception of the mobile phase. The mobile phase, which is at its supercritical phase by means of pressure and temperature variations, behaves like both a gas and a liquid. SFC is essentially a hybrid of GC and HPLC separation mechanisms, resulting in ideal solvating and viscosity properties. In addition, the low temperatures used in SFC may be especially fitting for thermally labile substances.

24

Sensitivity

The extremely low impurity levels from the TTC and staged TTC approaches inevitably indicates the need for high sensitivity. It is critical to maximize the analyte solubility, allowing larger amounts to be exposed to the detection system to increase sensitivity. Analytes can also be pre-concentrated by the previously mentioned matrix-separating techniques of SPE, SPME and liquid-liquid extraction. Analysts need to be vigilant against cross contamination and interference that may yield false positives given analysis these extremely low levels.

In many cases, the requirements for such low levels of detection dictates the mode of detection [31]. For example electron capture detectors (ECD) are highly sensitive to compounds containing halogens. This provides an advantage over the more commonly used flame ionizing detectors (FID). ECD can be an appropriate detector because genotoxic impurities commonly contain halogen groups. Mass spectrometry (MS) is another more sensitive detection mode, especially when used in single ion monitoring mode (SIM). MS can also provide qualitative information to identify a compound when in scan mode. This is particularly advantageous during method development in which this would allow for peak identity assignments in a chromatogram.

25

In addition to these detectors, others such as fluorescence, electrochemical, charged aerosol detection (CAD), chemiluminescene and evaporative light scattering (ELSD) can be used for trace genotoxic impurity analysis.

Figure 5: Decision Tree for Selecting an Appropriate Detector (Chrom = Chromophore and Sens = Sensitivity)

Figure 5 [31] presents a general scheme for selecting an appropriate detector for low-level analysis. Here, the detector is initially chosen based on if the analyte has a chromophore. If it does, a traditional UV method should be evaluated, but if sensitivity is poor, MS detection should then be evaluated. Without a

26

chromophore, ELSD is evaluated and if the sensitivity is not low enough, CAD is assessed. If sensitivity once again is not adequate, MS detection is evaluated. It is important to keep in mind Figure 5 is a generic decision tree and each genotoxic impurity analysis case should also consider the chemical properties of the impurity. These properties may allow use of other detectors or incorporation of sampling and sample processing techniques that can aid in the targeted sensitivity.

27

Method Validation and QC Transfer

Transferability of the analytical method to a quality control environment for routine testing is another concern with regard to genotoxic impurity analysis. QC labs may not have the same, elaborate instrumentation that may have been used during the research and development stages. In light of this, it may be prudent to utilize more standard detectors such as FID coupled to GC and UV for HPLC during the early stages of method development. Also, sample preparation can be problematic and should be kept fairly simple, if possible. A method with complicated, multi-step sample preparation can be challenging to validate and run routinely.

Validation of the instrumentation and the methods are additional requirements to ensure their intended purposes. The extent of the method validation will vary with the stages of drug development, becoming more rigorous as the drug progresses to marketing. Validation should cover sensitivity, specificity, accuracy, linearity and precision in order to prove the method is capable of its intended use.

This paper will cover the analytical methods used to monitor the genotoxic impurities with hydrazine, nitro, arylamine and aldehyde functional groups. This will serve to provide background and possibly a starting point for analysts who

28

need to monitor for the previously listed impurities. However in most cases, the sample matrix will be different in each case and will need to be accounted for.

29

Chapter 2: Analytical Method Details

The following chapter will cover aspects, including some validation results, of the analytical methods for hydrazines, aldehydes, arylamines and nitro compounds. These classes of genotoxic impurities have varying chemical properties, which in many cases, dictates the nature of the sample preparation, sampling, separation and detection. This is evident from the assortment of methods that are described here. The method operating parameters can be found in Appendix A.

30

Hydrazines

Hydrazines are very polar, basic molecules that pose many analytical challenges. These compounds lack chromaphores and therefore cannot be detected by UV. Non-substituted hydrazines do not contain carbon and are not suitable for FID detection. Hydrazines are low molecular weight and with MS detection, interference from other species can be a problem, especially at trace level analysis. However, their polarity and basicity make them good candidates for IC and HILIC separation techniques. In addition, their reactivity makes them suitable for derivatization. Many reported methods derivative hydrazines with carbonyl-containing compounds [30] which can greatly improve sensitivity and selectivity.

Khuhawar [17] developed a GC method to determine hydrazine in isoniazid formulations (syrup and tablets) using a FID and derivatization with ethyl chloroformate. Hydrazine is a degradation product of isoniazid. Interestingly, this method was applied to formulated products, including tablets. For the tablet sample preparation, there was a requirement for pre- and post-derivatization chloroform extractions which separated hydrazine from other ingredients in the tablet formulation. Without the additional extraction step, a particular tablet ingredient would be injected onto the GC column and would degrade, creating an unstable baseline. The limit of detection (LOD) for this method was found as

31

1.75 g/mL and linear coefficient of correlation >0.9958 for 3.5-35 g/mL hydrazine. Rather low relative percent error values for test solutions were found as 1-2%.

Although FID is a more common detector, sensitivity may be lacking. A highly sensitive, direct-injection GC/ECD method to determine hydrazine in commercial isoniazid formulations was developed by Carlin [3]. Hydrazine is derivatized with two molecules of benzaldehyde to form benzalazine prior to GC analysis. The limit of quantitation (LOQ) for hydrazine is 10 ppm relative to 10 mg/mL isoniazid and with a linearity correlation coefficient of 0.998 for a 10-100 ppm concentration range. Assay variability was 9.5% for 25 ppm and 11.3% for 100 ppm hydrazine. ECD was chosen over FID for selectivity in case of formulation changes and not for increased sensitivity.

A generic headspace-GC/MS approach for hydrazine determination in drug substances was developed by Sun [28] with similar sensitivity to that of the prior GC/ECD method. Samples were derivatized using acetone to form acetone azine. However, acetone will compete with the higher-boiling acetone azine in the headspace, leading to sensitivity and reproducibility issues. The authors investigated this issue and optimized the derivatization solution, which was also the sample solvent, to contain 5% acetone (v/v) in NMP. NMP is a solvent that adequately dissolved various hydrophobic APIs to the required concentrations.

32

An LOQ of 0.1 ppm relative to 100 mg/mL drug substance was achieved and the authors point out that a lower quantitation limit can possibly be achieved by increasing the sample concentration. The method described has very good sensitivity given MS detection in SIM mode as seen in the Figure 6. Figure 6: Acetone Azine at the LOQ (0.1 ppm) and 1 ppm Levels. Black Trace is Blank.

The linearity coefficient of correlation was 0.999 for hydrazine concentrations of 0.1-10 ppm. Precision at 1 ppm ranged from 2.7-5.6% RSD. Recovery at 1 ppm in various drug substances was reported as 79-117%. Some low recovery values were explained by hydrazine potentially interacting with the API, preventing its derivatization and subsequent detection. The derivatization procedure coupled with headspace sampling eliminates drug substance from

33

contaminating the GC. The authors mention the possibility for automating the derivatization step which could significantly increase sample throughput. This generic method was applied using various drug substances for the trace-level determination of hydrazine.

An alternative to GC derivatization methods is direct analysis by IC. Jagota [15] developed a direct-injection IC with electrochemical detection method to determine hydrazine as found in the penultimate intermediate of an anti-infective agent. The method involved minimal sample preparation without the need for a time-consuming derivatization step. The authors emphasize the simplicity of the method in which the sample is dissolved in 50/50: methanol/water (v/v) and immediately injected into the IC, which has a short run time of 10 minutes. The LOQ of hydrazine is 100 ppm relative to 0.5 mg/mL of intermediate. Linearity correlation coefficient is >0.99 for hydrazine levels from 100 to 400 ppm. Assay precision is 1.2% for 100 ppm hydrazine. Recovery values were 90.2-104.9% for 100-300 ppm hydrazine.

In addition to the prior IC method, HILIC analysis of hydrazines does not require derivatization. A novel HILIC/chemiluminescent nitrogen detector (CLND) method to simultaneously determine hydrazine and 1,1-dimethylhydrazine in a pharmaceutical intermediate was reported by Liu [21]. An interesting aspect of this method is the simultaneous determination of substituted hydrazines, which is

34

not common in the published literature. HILIC separation is ideal for separation of polar molecules such as hydrazines. Tedious derivatization is not necessary in this case and the hydrazines are analyzed in their native state. To accommodate the HILIC separation mechanism, less polar components are included in the aqueous mobile phase to control polar compounds elution from the stationary phase. An alcohol-based mobile phase was investigated and ultimately used to accommodate sample solubility and CLND compatibility (lack of nitrogen-containing mobile components). Various alcohols were screened for optimal retention and separation of hydrazines as seen in the Figure 7, with ethanol selected for optimal resolution and run time. Figure 7: Effect of Various Alcohols in HILIC Mobile Phase to Separate Hydrazines (1 = 1,2-dimethylhydrazine, 2 = 1,1-dimethylhydrazine, 3 = methyl hydrazine and 4 = hydrazine).

35

The authors comment that the increasing alcohol chain length correlation with hydrazine retention times relate to partitioning effects and hydrogen bonding competition of the alcohol and analytes at the stationary phase aqueous layer.

The LOQ for both hydrazine and 1,1-dimethylhydrazine were found as 0.02% relative to 10 mg/mL. Linear coefficients of correlation were >0.999 for both analytes with concentrations of 0.01-0.1%. Precision values at the 0.02% level were 5.0 and 3.9% for hydrazine and 1,1-dimethylhydrazine, respectively. Hydrazine recovery at LOQ level was 80.0-87.8% while that for 1,1dimethylhydrazine was 99.4-102.5%. This method does not require derivatization and may be applied to other substituted hydrazines such as methylhydrazine and 1,2-dimethylhydrazine.

36

Aldehydes

Aldehydes, like hydrazines, pose many analytical challenges as they are quite volatile and many do not possess a chromophore. Low molecular weight aldehydes can be challenging to separate from one another as well as other lowmolecular weight species arising from the sample matrix or background. Because of this, derivatization is commonly used to assist with aldehyde detection and separation.

A GC/FID method to determine benzaldehyde in injectable formulations for diclofenac, piroxicam and Vitamin B-complex was developed by Kazemifard [16]. Benzaldehyde is an oxidation product of the excipient, benzyl alcohol, used as a preservative in these formulations. This method utilized a simple chloroform based liquid-liquid extraction step followed by direct injection into the GC. The LOQ for benzaldehyde is 0.4 g/mL and linearity demonstrated over the range of 0.5-100 g/mL with a correlation coefficient of > 0.995. Method precision was < 2.5% RSD at concentrations from 15 to 45 g/mL. Benzaldehyde recovery was excellent and ranged from 95.2-98.1% for samples at concentrations of 20.5-39.6 g/mL. The author suggests this method is quite suitable for routine analysis in a QC environment.

37

A very sensitive GC/MS method with derivatization to determine formaldehyde in common pharmaceutical excipients was developed by del Barrio [4]. Formaldehyde is challenging to analyze because it of its low response using FID as well as poor peak shape and separation issues. The authors utilized acidified ethanol to derivative formaldehyde to diethoxymethane to combat these issues. In this case, the sample diluent was also the derivatizing agent, which significantly simplified the sample preparation procedure. Ethanol is capable of either dissolving or dispersing excipients to enable extraction. In addition, ethanol is in excess during the derivatization which assists with completing the reaction. The LOQ value was found to be 0.02 g/mL. The linear coefficient of correlation was 0.9960 for a wide range of 0.02-1000 g/mL formaldehyde and with precision values for the same concentrations ranging from 1.3-10.1% RSD. Recovery at 5 g/mL in many different excipients varied from 87.5-105.5%. This method was also shown to detect and quantitate formic acid, an oxidation product of formaldehyde that can also be present in samples. In this case,

formic acid is derivatized and detected as ethyl formate. The derivatization reactions for formaldehyde and formic acid are listed in Figure 8.

38

Figure 8: Derivatization Reaction of Formic Acid and Formaldehyde with Acidified Ethanol

The authors also performed method validation for formic acid which had similar results to that obtained for formaldehyde.

Li [19] developed two head space GC methods with derivatization to determine aldehydes in many common pharmaceutical excipients. The derivatization is carried out with O-2,3,4,5,6-(pentafluorobenzyl) hydroxylamine hydrochloride (PFBHA), which can conveniently be carried out in aqueous conditions at room temperature. The derivatization products are quite volatile and suitable for head space analysis. The first method described is a GC/MS method with negative chemical ionization detection that was shown to be very sensitive. This method was used to quantitate formaldehyde, acetaldehyde, propanal, butanal, pentanal, hexanal, heptanal, octanal and benzaldehyde in the presence of more than 30 excipients. MS detection allowed for identification confirmation of the derivatized 39

aldehydes. Very low LOD values of 0.2-1 g/mL were achieved for all analytes with this method. Linear coefficients of correlation were all > 0.99 for concentrations from 0.5-20 g/mL. Precision was shown to be 2.0-5.6% at the same concentrations as the linearity samples. Though PFBHA bears a chromophore and can be analyzed by HPLC/UV, the author reported significant interference from the matrix that would preclude accurate analysis. Several chromatograms that show the derivatized aldehydes in the presence of various excipients are presented in Figure 9.

40

Figure 9: HPLC/UV Chromatograms of Derivatized Aldehydes (1 = formaldehyde, 2 = acetaldehyde, 3 = propanal, 4 = butanal, 5 = pentanal and 6 = hexanal) in the Presence of Common Excipients (A = Aldehyde Standard with no Excipient, B = Hydroxymethyl Cellulose and C = Hypromellose)

It is obvious from Figure 9 that the excipients will interfere with the aldehyde analysis by co-elution. However, the figure is shown only at one particular UV wavelength and other wavelengths could possibly allow for acceptable UV analysis. The use of headspace GC/MS allows for cleaner chromatograms and excellent resolution from the excipients as seen in Figure 10.

41

The second method (Method 2) was designed for more rapid and routine analysis. Here, GC/FID was implemented analyze formaldehyde and acetaldehyde. Only these two compounds were included in this method as the first method did not find any of the other higher molecular weight aldehydes at significant levels. For both aldehydes, LOQ values were 5 g/L and linear correlation coefficient values >0.99 for the concentration range of 5-200 g/L. Precision was 3% RSD for both analytes at the linearity concentrations and recoveries found as 85-97% in the presence of excipients.

The use of headspace GC/MS allows for cleaner chromatograms with excellent resolution from excipients as compared to the previously mentioned HPLC/UV method as seen Figure 10.

42

Figure 10: GC/MS Method 2 Chromatograms of Derivatized Aldehydes (1 = formaldehyde-PFBHA, 2 = PFBHA, 3 and 4 = acetaldehyde-PFBHA) in the Presence of Common Excipients (A = Hypromellose, B = Hydroxylpropyl Cellulose and C = PFBHA Blank)

This GC/MS method can distinguish between the syn and anti isomers that are expected to form from the derivatization reaction, as can be seen from the two acetaldehyde-PFBHA peaks in the chromatogram above. Li suggests sensitivity for this method can be increased by ECD given the high fluorine content of the derivatization agent. In addition, this method can be routinely applied in the drug product arena to screen excipients for aldehyde content.

43

Whereas all the previous aldehyde analysis methods used GC, Han [11] reported a capillary electrophoresis (CE) method coupled with chemiluminescence detection. This method determines 3,4-dihydroxybenzaldehyde in injectable and tablet formulations of the Chinese herb Salivia miltorrhrza. CE was selected for a desired high degree of resolution. CL was chosen because of its enhanced sensitivity of 3,4-dihydroxybenzaldehyde in the presence of the CL reagents, luminol and ferricyanide. The LOD for 3,4-dihydroxybenzaldehyde was 7.0 x 108

M and a linear coefficient of correlation of 0.999 for a concentration range of

0.60-30.0 M. Precision was found as 1.32% RSD for 0.4 M and recovery values were 95% for tablets (4.69 M) and 102% for an injectable formulation (2.25 M). This method was shown to efficiently resolve 3,4dihydroxybenzaldehyde from other components in the formulations within 6 minutes.

44

Nitro Compounds

In addition to hydrazines and aldehydes, nitro-containing impurities are considered genotoxic structural alerts.

An HPLC/UV method to determine the nitro compound 4-nitro-p-cresidine within a color additive was developed by Richfield-Kratz [26]. This compound was first observed as an unknown impurity that was initially isolated and identified in the same paper. Sample preparation of the nitro compound from the matrix involved a chloroform extraction procedure followed by acidification and injection onto the HPLC. A very low limit of detection was reported as 71 ppb. The linear coefficient of correlation was found as 0.9986 for the concentration range of 1025100 ppb and precision was 6% RSD. Though an older method, sensitivity is very low and in the sub-ppm range, rivaling that of more current methods.

Two nitro-containing compounds, 1-methyl-4-nitro-3-propylpyrazole-5-carboxylic acid (MNP) and 1-methyl-4-nitro-3-propylpyrazole-5-carboxamide (MNC), were monitored in an HPLC/UV method reported by Nagaraju [23]. These two compounds are process related impurities in the synthesis of Viagra. The LOD values for MNP and MNC were 0.02 ng and 0.03 ng, respectively. Linearity coefficients of correlation values were 0.992 for 0.10-0.50 ng MNP and 0.986 for 0.10-2.50 ng MNC. Precision determined at 5 g/mL concentration was 4.8 and

45

3.6% for MNP and MNC, respectively. Recovery at this same concentration for both compounds were 103.8% for MNP and 98.2% for MNC. The authors state that this method is applicable for both in-process and final product testing.

The previous two methods utilized less sensitive UV detection, but here Liu [20] describes an LC/MS method to determine two nitro-containing genotoxic impurities. These impurities, Impurity II and VIII, were seen during the production of the compound pazopanib hydrochloride. This same method was applied to the determination of an arylamine compound, Impurity III, as described elsewhere in this paper. The method has a LOD of 0.6 and 0.2 ppm, respectively for Impurities II and VIII. The linearity coefficient of correlation was >0.9995 for both compounds, at 2.4-60 ng/mL for Impurity II and 3-100 ng/mL for Impurity VIII. Precision values at the 1.7 ppm level were 5.8 and 2.1% RSD for Impurities II and VIII, respectively. Recovery at the 1.7 ppm level for Impurities II and VIII were 91 and 98%, respectively.

Dongre [7] exploited the relatively new technique of ultra performance liquid chromatography (UPLC) coupled to a UV detector to determine the nitro compound 8-nitro-6-methoxyquinoline. The LOQ was reported at 0.1 g/mL and linearity coefficient of correlation >0.999 for 0.25-0.75 g/mL concentration range. Precision at 0.5 g/mL was 3.78% and recovery values of 105.79107.48% for sample concentrations of 0.2-0.3 g/mL were obtained. The method

46

developed is a quick, 5-minute, stability indicating method that was also shown to have comparable results with conventional HPLC. UPLC uses scaled down aspects of HPLC with higher pressures. This exemplifies movement towards more time efficient and cost saving methods. Figure 11 shows a comparison of the UPLC and HPLC methods.

Figure 11: Comparison of HPLC (top) and UPLC (bottom) Stabilityindicating Methods. Impurities I and Impurity II (8-nitro-6methoxyquinoline) are Well Resolved from the Drug Substance, Primaquine.

As seen from the figure above, the 5-minute UPLC method provided comparable results to the 35-minute, more traditional HPLC method.

47

Though not in the context of pharmaceuticals, Zhao [32] describes a method to determine four nitrobenzene and nitrotoluene-based compounds in water samples. Interestingly, this is a GC/ECD method as opposed to the prior methods, which were LC methods. The author reports very low LOD values ranging from 0.02-0.4 g/L and acceptable linearity, precision and recovery values. This method can potentially be modified and adapted for pharmaceutical analyses.

48

Arylamines

The final group of compounds to be discussed here are arylamines. These compounds contain a chromophore and are be suitable for UV analysis, however, sensitivity may need to optimized or another detection technique selected.

Vanhoenacker [29] developed an HPLC/MS method to determine 13 arylamines in the presence of 4 different compounds. The 4 different compounds were chosen because of their varying polarity and pKas to represent a range of drug substances. Two sample preparations were evaluated (one in which the arylamines are derivatized with hexylchloroformate and the other without derivatization) and validated using the HPLC/MS method. Direct sample injection was used because of the range of polarities of the compounds, some of which are non-volatile and therefore not suitable for head space sampling. MS detection provided the required sensitivity. The LOD for all analytes was below 0.05 g/mL. Linear coefficients of correlation were >0.990 for analytes at 0.1-5 g/mL and precision was <2.1% RSD at 0.5 g/mL. The recovery studies were quite extensive, using 8 HPLC methods to determine all 13 arylamines as both derivatized and underivatized in the presence of the 4 previously mentioned compounds. Recovery values ranged from 70-120% at a 0.1 g/mL spiking concentration. Low recovery values were attributed to coelution with the API,

49

causing analyte signal suppression. The authors consider this a generic method that can be applied to other APIs and potentially cover other arylamines. They also suggest increased sensitivity by using tandem mass techniques, but caution that QC labs may not be equipped with the required instrumentation.

Liu [20] describes another LC/MS method that determined an arylaminecontaining impurity (Impurity III). This was performed in the context of an excellent case study that exemplifies quality by design (QbD) in relation to genotoxic impurity control during API synthesis. The trace impurity method has a LOD of 0.6 ppm. The linearity coefficient of correlation was >0.9995 for 3-100 ng/mL. Precision and accuracy values at the 1.7 ppm level were 2.6% RSD and 88%, respectively.

In addition to the trace analysis method, the author developed a less sensitive, but more robust, HPLC/UV method. This second method was used to monitor several genotoxic impurities, including Impurity III, during the API synthesis. Critical points in the process were identified in which the genotoxic impurity is introduced as well as proceeding steps in which the impurity may be purged. Studies were conducted such that impurities were spiked into the process at various steps and measured in subsequent intermediates. A schematic that shows the purging process for Impurity III spiked at various amounts at different process steps can be seen in Figure 12.

50

Figure 12: Schematic Showing Spiking Studies of Impurity III at Various Stages in the Synthesis of the API (GW786034).

This figure shows that for Impurity III spiked at 1% in Stage 2 and 5% spiked at Stage 3, the final product yields a concentration less than that of the final product specification (1.7 ppm). The significant decrease is attributed to several processing steps that assist with consuming the impurity. This information allowed higher specification settings for the impurities investigated at earlier phases in the synthetic process, without the need for final product testing. The HPLC/UV method was used to determine genotoxic impurity levels which were set at much higher specification levels. Another benefit to HPLC/UV is ease of use and availability of instrumentation.

Yet another highly sensitive, MS-based method was reported by Aznar [1], which utilized a UPLC method to detect 22 different primary aromatic amines. Cationic SPE was used to increase the method sensitivity by pre-concentrating the analytes as well as eliminating matrix effects that could interfere with the MS analysis. The UPLC method was quite similar to an earlier HPLC method as

51

mentioned by the author, however the UPLC method was chosen to quickly separate all 22 aromatic amines. SIM and SRM detection modes were evaluated and SIM mode was ultimately selected for the best overall sensitivity. LOD values ranged from 0.01-2.4 g/L for all analytes. Linear coefficients of correlation were > 0.992 for all analytes at 0.03-75 g/L. Precision and recovery ranges for all aromatic amines at 30 g/L were 4.5-13.4% RSD and 81-109%, respectively. Although the matrix involved in this method was polymer-based laminates, this method could be applied to aromatic amines in a pharmaceutical context, such as tablets. In this case, the SPE step can separate the analytes from excipients.

52

Chapter 3: Conclusion

Genotoxic substances as found in active pharmaceutical ingredients or its formulated product are known as genotoxic impurities. These impurities are generally derived from reactive reagents that are used to synthesize the pharmacologically active molecule. In addition, they can also arise from API degradation or from interactions with excipients or containers. The mere fact that these substances are utilized as reactive building blocks during synthesis also indicates they can be reactive with biological substances, which includes DNA. These interactions with DNA are considered mutations that can potentially lead to cancer. Given this, genotoxic impurities are of high concern. Assessment, control and monitoring these impurities are of paramount importance to ensure patient safety.

There are various approaches to assess if an impurity is genotoxic, ranging from in-silico determinations to in-vivo testing. Muller [22] presented a genotoxic impurity assessment tree that divides these impurities into 5 classes along with suggested impurity control levels. If significant similarities exist, the genotoxicity of an impurity may also be extrapolated from that of the parent compound.

Ideally, genotoxic impurities in a product administered to people should be controlled to a level of non-existence. However, many constraints prevent this

53

from being a practical goal. Synthetic schemes of the many novel pharmaceutical compounds in research may have no current alternative to the genotoxic reagent. Also, analytical methods and instrumentation have lower limits of detection, some which cannot extend below the ppm range. Currently, pharmaceutical companies have been targeting a TTC level of 1.5 g/day which ensures a safer product while satisfying regulatory agency guidelines. Recently, efforts by regulatory agencies and the pharmaceutical industry are moving towards a more practical, scaled TTC approach. This approach recognizes the challenges associated with the extremely low-level at which genotoxic impurities must be controlled and monitored, given risk-based allowable impurity levels. Here patients may be exposed to still low levels of genotoxic impurities, but scaled and balanced against various factors such as dosage duration, patient population and the benefits of the pharmaceutical.

Given this, there is a critical need for analytical methods to monitor genotoxic impurities. Most importantly, the analytical method needs to possess specificity and appropriate sensitivity. Genotoxic impurities have a broad range of chemical properties, and therefore, the handling and analysis of them will also vary. Genotoxic impurity reviews on analysis and control of alkylating agents such as alkyl halides, benzyl halides and organohalides [8] as well as alkyl esters of alkyl and aryl sulfonic acids [9] have been previously published. This paper focused

54

the analysis on four other groups of genotoxic impurities with hydrazine, aldehyde, arylamine and nitro functional groups.

Hydrazines are genotoxic impurities that can be quite challenging to analyze. They lack chromophores and cannot be directly analyzed with UV detection. In addition, they have very low molecular weights in which MS interferences can easily hinder analysis given the potential for similar molecular weight species found in the background. Given this and the fact that hydrazines are quite reactive, they are excellent candidates for derivatization [3, 17, 28]. Derivatization not only produces a more inert compound, but it adds molecular weight that can add sensitivity to methods using FID [17]. Despite the challenges with direct hydrazine measurement, two methods described here eliminate the need for additional derivatization steps by using IC and HILIC techniques [15, 21].

Aldehydes are similar to hydrazines in that they are reactive and can have low molecular weights. They too are good candidates for derivatization, which is commonly used to analyze aldehydes [4, 19]. Despite this, one method described here uses CE analysis [11], with a relatively simple sample preparation procedure and low detection limit.

55

Nitro and arylamine-containing compounds are the final two classes of genotoxic impurities discussed in this paper. Arylamines contain benzene and may be suitable for UV detection. Almost all nitro and arylamine analysis were carried out using liquid chromatography [1, 7, 20, 23, 26, 29], most likely because these compounds have higher molecular weights and are less suitable for GC analysis. Interestingly, a GC method was reported for a nitro compound [32] with very low LOD values in the sub-microgram per liter range.

In all the methods described in this paper, sensitivity is of utmost importance. Several approaches can be taken to decrease the limit of detection for a given method. Liquid-liquid extraction [26] and solid phase extraction [1] are techniques that can be used to achieve higher sensitivity. In addition, highly sensitive detectors such as MS [1, 4, 19, 20, 28, 29] and ECD [3, 32] are commonly used to analyze very low level impurities. These detectors are also quite selective, as is CLND, which was reported in several methods [11, 21].

Along with sensitivity, selectivity is a critical parameter necessary for impurity analysis. Matrix effects can create significant interference with the analytes of interest and may be circumvented with the previously mentioned extraction techniques as well as headspace sampling [19, 28]. Many times the instrumentation selected affords the best separation of the analytes from other components. Several methods reported here use techniques such as IC [15], CE

56

[11] and UPLC [1, 7], which are not quite as common as LC and GC methods. GC/FID [16, 17, 19] and HPLC/UV [20, 23, 26] tend to be more common given their robustness. They are also more suitable for a QC environment given their ease of use and availability in pharmaceutical laboratories.

An excellent QbD study [20] was performed in conjunction with trace analysis that assessed genotoxic impurity purging during an API synthesis. That study justified the use of less sensitive methods to determine genotoxic impurities upstream in the process. Also described here are methods that are fairly generic and can possibly be applied to other analytes or matrix environments [21, 28, 29].

Genotoxic impurity analysis is a challenging, complex aspect of the drug development process. An appropriate balance needs to be found that takes into account patient safety against the amount of time and resources to quickly get a pharmaceutical to market. As seen here, there is a wide chemical variety of genotoxic impurities as well as the methods to analyze them. With such lowlevel analyses, more elaborate instrumentation and sample preparations become more common. Along with this is the need for skilled analytical chemists to operate these complicated methods and properly analyze the data. The need and evolution of genotoxic impurity analysis will continue so long as pharmaceuticals are in production.

57

References

1.

M. Aznar, E. Canellas, and C. Nerin, Quantitative determination of 22 primary aromatic amines by cation-exchange solid-phase extraction and liquid chromatography-mass spectrometry. Journal of Chromatography A, 2009. 27: p. 5176-5181. J. P. Bercu, et al., Quantitative assessment of cumulative carcinogenic risk for multiple genotoxic impurities in a new drug substance. Regulatory Toxicology & Pharmacology, 2008. 51(3): p. 270-7. A. Carlin, et al., Stability of isoniazid in isoniazid syrup: formation of hydrazine. Journal of Pharmaceutical & Biomedical Analysis, 1998. 17(45): p. 885-90. M. A. del Barrio, et al., Simultaneous determination of formic acid and formaldehyde in pharmaceutical excipients using headspace GC/MS. Journal of Pharmaceutical & Biomedical Analysis, 2006. 41(3): p. 738-43. E. J. Delaney, An impact analysis of the application of the threshold of toxicological concern concept to pharmaceuticals. Regulatory Toxicology & Pharmacology, 2007. 49(2): p. 107-24. K. L. Dobo, et al., The application of structure-based assessment to support safety and chemistry diligence to manage genotoxic impurities in active pharmaceutical ingredients during drug development. Regulatory Toxicology & Pharmacology, 2006. 44(3): p. 282-93. V. G. Dongre, et al., Development and validation of UPLC method for determination of primaquine phosphate and its impurities. Journal of Pharmaceutical & Biomedical Analysis, 2008. 46(2): p. 236-42. D. P. Elder, A. M. Lipczynski, and A. Teasdale, Control and analysis of alkyl and benzyl halides and other related reactive organohalides as potential genotoxic impurities in active pharmaceutical ingredients (APIs). Journal of Pharmaceutical & Biomedical Analysis, 2008. 48(3): p. 497-507. D. P. Elder, A. Teasdale, and A. M. Lipczynski, Control and analysis of alkyl esters of alkyl and aryl sulfonic acids in novel active pharmaceutical ingredients (APIs). Journal of Pharmaceutical & Biomedical Analysis, 2008. 46(1): p. 1-8.

2.

3.

4.

5.

6.

7.

8.

9.

58

10.

European Medicines Agency Committee for Medicinal Products (CHMP), London, Guidance on the Limits of Genotoxic Impurities. 2006. S. Han, et al., On-line chemiluminescence determination protocatechuic aldehyde and protocatechuic acid in pharmaceutical preparations by capillary electrophoresis. Journal of Pharmaceutical & Biomedical Analysis, 2005. 37(4): p. 733-8. ICH Q3A (R2), Impurities in New Drug Substances. 2006 [Available from:
http://www.ICH.org/.

11.

12.

13. 14. 15.

ICH Q3B (R2), Impurities in New Drug Products. 2006. ICH Q3C (R4), Impurities: Guidelines for Residual Solvents. 2009. N. K. Jagota, et al., Determination of trace levels of hydrazine in the penultimate intermediate of a novel anti-infective agent. Journal of Pharmaceutical & Biomedical Analysis, 1998. 16(6): p. 1083-7. A. G. Kazemifard, et al., Capillary gas chromatography determination of benzaldehyde arising from benzyl alcohol used as preservative in injectable formulations. Journal of Pharmaceutical & Biomedical Analysis, 2003. 31(4): p. 685-91. M. Y. Khuhawar, et al., Ethyl chloroformate as a derivatizing reagent for the gas chromatographic determination of isoniazid and hydrazine in pharmaceutical preparations. Analytical Sciences, 2008. 24(11): p. 14936. R. Kroes, et al., Structure-based thresholds of toxicological concern (TTC): guidance for application to substances present at low levels in the diet. Food & Chemical Toxicology, 2004. 42(1): p. 65-83. Z. Li, et al., Detection and quantification of low-molecular-weight aldehydes in pharmaceutical excipients by headspace gas chromatography. Journal of Chromatography A, 2006. 1104(1-2): p. 1-10. D. Q. Liu, et al., Analytical control of genotoxic impurities in the pazopanib hydrochloride manufacturing process. Journal of Pharmaceutical & Biomedical Analysis, 2009. 50(2): p. 144-150. M. Liu, et al., Hydrophilic interaction liquid chromatography with alcohol as a weak eluent. Journal of Chromatography A, 2009. 12: p. 2362-2370.

16.

17.

18.

19.

20.

21.

59

22.

L. Mueller, et al., A rationale for determining, testing, and controlling specific impurities in pharmaceuticals that possess potential for genotoxicity. Regulatory Toxicology & Pharmacology, 2006. 44(3): p. 198211. V. Nagaraju, et al., Separation and determination of synthetic impurities of sildenafil (Viagra) by reversed-phase high-performance liquid chromatography. Analytical Sciences, 2003. 19(7): p. 1007-11. D. A. Pierson, et al., Approaches to Assessment, Testing Decisions, and Analytical Determination of Genotoxic Impurities in Drug Substances. Organic Process Research & Development, 2008. 13(2): p. 285-291. A. G. Renwick, Structure-based thresholds of toxicological concern-guidance for application to substances present at low levels in the diet. Toxicology & Applied Pharmacology, 2005. 207(2 Suppl): p. 585-91. N. Richfield-Fratz, et al., Isolation, characterization and determination of trace organic impurities in FD&C red no. 40. Journal of Chromatography A, 1989. 467(1): p. 167-76. P. Skett, Low-level Measurement of Potent Toxins, in Analysis of Drug Impurities, R.S.a.M. Webb, Editor. 2007, Blackwell Publishing. M. Sun, L. Bai, and D. Q. Liu, A generic approach for the determination of trace hydrazine in drug substances using in situ derivatization-headspace GC-MS. Journal of Pharmaceutical & Biomedical Analysis, 2009. 49(2): p. 529-533. G. Vanhoenacker, et al., Determination of arylamines and aminopyridines in pharmaceutical products using in-situ derivatization and liquid chromatography-mass spectrometry. Journal of Chromatography A, 2009. 16: p. 3563-3570. M. Vogel, et al., Hydrazine reagents as derivatizing agents in environmental analysis--a critical review. Fresenius Journal of Analytical Chemistry, 2000. 366(8): p. 781-91.

23.

24.

25.

26.

27.

28.

29.

30.

60

31.

Z. Y. Yuabova, et al., Genotoxic impurities: A quantitative approach. Journal of Liquid Chromatography & Related Technologies, 2008. 31(15): p. 2318-2

32.

R. Zhao, S. Chu and X. Xu: Optimization of nonequilibrium liquid-phase microextraction for the determination of nitrobenzenes in aqueous samples by gas chromatography-electron capture detection. Analytical Sciences, 2004. 20: p. 663-666.

61

Appendix A: Analytical Method Details

62

Appendix A: Continued

63

Appendix A: Continued

64

Appendix A: Continued

65

Appendix A: Continued

66

Appendix A: Continued

67

Appendix A: Continued

68

Appendix A: Continued

69

Appendix A: Continued

70

Appendix A: Continued

71

Vous aimerez peut-être aussi