The stories of impurities in drugs usually reach the public when a disaster happens – an enantiomeric impurity present in thalidomide (used to treat nausea in pregnant individuals) gained worldwide notoriety in 1962 when its teratogenic effect was discovered. As recently as 2009, at least 25 children died in Bangladesh after taking paracetamol syrup that contained ethylene glycol instead of glycerol as the sweetening agent – ethylene glycol was much cheaper because it is used by the automotive and textile industries. Another recent example is platinum-containing anti-cancer drugs (such as cisplatin and oxaliplatin) that were found to contain traces of the toxic Pt(IV) ions that were present as salts rather than complexes. Such horror cases form a very long list with multiple new items added every year. The situation goes all the way back to Paracelsus, who famously declared that even “the dose makes the poison”. Here we shall focus on the impurities.
An impurity is defined as any component that is not the chemical entity described in the formulation. Pharmaceutical impurities are tightly regulated. For every impurity, there exists a chemistry and a safety aspect. The chemical aspect is purely technological: any impurity can cause secondary reactions, both directly and catalytically. Chemically significant impurities come in a variety of types: degradation compounds resulting from decomposition of the active substance; synthetic side products resulting from branches in the reaction path; solvent residues, intermediate compounds, catalyst residues and other artefacts of the synthesis procedure; constituents of the mechanical parts of the synthetic apparatus (filter threads, tubing materials) as well as secondary leachables from those parts, e.g. plasticisation agents, excipients, preservatives, and adjuvants. A good example of a secondary chemical process is catalytic decomposition of hydrogen peroxide in the presence of manganese salts: even a tiny manganese-containing impurity would result in rapid degradation. This dictates particular attention to the material of the container: in the case of hydrogen peroxide, it cannot be brown glass.
The safety aspect refers to the toxic, teratogenic, carcinogenic, hallucinogenic, and other undesired effects on the target organism. This aspect tends to be both more important and harder to avoid because fundamental laws of physics, chemistry and biology tend to enter the picture. Impurities with a safety aspect tend to be more significant in practice.
This document provides an overview of the types of impurities in pharmaceutical ingredients and comments on the practical significance of each type, with recent examples where available.
This section analyses the impact of various types of chemical impurities on the function, regulatory compliance, marketability, storage stability and other attributes of a pharmaceutical formulation.
The starting materials are the initial point of the chemical reaction. Reactions are rarely quantitative, and the initial substances are sometimes close enough in properties (or high enough in concentration) that they can survive the purification and preparation process, and make their way into the final product. A good practical example is the synthesis of tipranavir – an antiviral medication, where aminobenzene is reagent used in one of the stages of the synthesis. Because both the reactant and the final pharmaceutical substance are aromatic amines, they are hard to separate. At the economically sustainable level of effort, the residual concentration of the (fairly toxic) aminobenzene in the final product is about 0.1%. However, because it is a known substance with a well understood toxicity profile, the admixture is tolerated by the regulatory agencies because the toxic effect at the level of exposure actually reached during therapeutic use of tipranavir is negligible.
Because each stage of the reaction chain introduces new materials, the network of possible substances that can present themselves at the final stage is fairly extensive. It is unlikely that very chemically dissimilar molecules would make it through the multiple purification stages, and so it is usually expected that the impurities in the final product are substances that are very closely chemically related. In practice this means that they are hard to separate. An example often cited in the literature is the synthesis of flumecinol (an itch suppressor), where the desired substance has a trifluoromethyl group on one position, and the undesired admixture has it next door. Another example is the methylation site in tolperisone (a muscle relaxant), where the principal admixture differs in the position of the methyl group.
In both cases, this positional uncertainty is inherited from the impurity found on one of the reagents. Such impurities are exceedingly hard to eliminate, but if their toxicological properties are known, even significant concentrations can be tolerated by the regulatory agencies, so long as the therapeutic dose of the main substance does not co-administer a toxic dose of the impurity.
Many organic reactions yield more than one product. The products can differ in substituent location, in cis-trans or optical configuration, in the deprotection of side groups, and in many other ways. Reactions can proceed incompletely or too far, the products may undergo isomerisation or rearrangement once they had formed (synthesis of cisplatin is a good example). Some products may oligomerise and polymerise. Secondary undesired reactions can occur between the products and the reagents or catalysts used to obtain them. A good example of an insufficiently selective reaction that results in esterification of multiple sites (when only one is needed) is the esterification of nandrolone with decanoic acid. It is the last step in the synthesis of the nandrolone deaconates, which is the depot form of nandrolone (an anabolic substance). Nandrolone has multiple hydroxyl groups, and the reaction proceeds to some extent on all of them. The result is not only nandrolone decanoate, but also the esterification product where the decanoic acid ester moiety is attached to a double bond.
The practical impact of the by-products is chiefly the expense of their elimination and the additional consumption of (possibly expensive) reagents used up in their generation. A trained organic chemist would normally be able to predict the by-products of each individual reaction and design ways of isolating the desired substance, but predicting by-products of a chain of multiple reactions where purification at intermediate stages is not done (so-called “one pot” synthesis) is a much harder task because side reactions can branch in effectively unpredictable directions. Help chiefly arrives from nuclear magnetic resonance, where chemical shifts and J-couplings are sufficiently sensitive to tell even very structurally similar substances apart. However, NMR is a very expensive and specialised technique that requires upfront investment in the millions of pounds to even establish the facility.
A degradation product is an impurity that appears at some point during manufacture or storage. The causes can vary as far as wide as light, temperature, moisture, pH, contact with atmosphere or, for that matter, any secondary reactions with what is present around the substance. Some degradation pathways are encluntered often enough to be classified:
The impact of degradation reactions on pharmaceutical industry is severe. It is the primary reason why most drugs have a fixed shelf life and must be discarded when that period is over. Stabilising agents are a possibility, but for many drugs some form of degradation is virtually inevitable. This is a particular problem for vitamins because they tend to be easily oxidised and chemically rather tender molecules. Severely toxic outcomes are rare, but many drugs would degrade below permissible concentration in a matter of days if storage requirements are not fulfilled.
A surprising number of drugs contain a bound heavy metal: lanthanide containing MRI contrast agents are one example, and platinum-containing anti-cancer drugs are another. The worst thing that can happen to an MRI contrast agent (which is administered in huge doses during MRI scans) is that the lanthanide would find its way out of the very tight ligand cage – in that case the death of the host organism is a near certainty. Similar considerations apply to substance like oxaliplatin – a complex of platinum used to cross-link DNA and make cells die on division – the four-coordinate form of Pt(II) is relatively non-toxic while encaged by the ligand, but free Pt(II) ion and, worse still, Pt(IV) ion from partial oxidation are very toxic indeed, particularly to kidneys.
Heavy metals also occur in catalysts that are increasingly often used in stereoselective synthesis. Heterogeneous catalysts are not normally a problem – the surface stays in one piece, but incompletely eliminated homogeneous catalysts can carry their heavy metal all the way into the final formulation of the drug. Grubb’s catalyst contains Ru, Ziegler-Natta catalyst contains Ti, Adams catalyst contains Pt – most of the d– and f– elements are involved some way or another into some catalytic process. The problem is bad enough and frequent enough that a formal ICH Q3C classification exists (Table 1).
Table 1. Examples of metals belonging to different toxicity classes within ICH Q3C classification.
|Class||Examples of metals||Notes|
|Class 1||arsenic, lead, cadmium, mercury||Extremely toxic|
|Class 2A||vanadium, molybdenum, cobalt||Very toxic|
|Class 2B||silver, thallium, platinum and other platinoids||Toxic|
|Class 3||barium, lithium, chromium, copper||Moderately toxic|
|Class 4||sodium, calcium, iron, zinc, magnesium, aluminium||Low toxicity|
Most organic reactions used in drug synthesis are carried out in liquid phase, and the residual solvent, left either as a result of the substance being incompletely dried, or as crystallographic solvent, can therefore present a problem. ICH 3QC recommends using the least toxic solvent possible at each synthetic step. The classification is given in Table 2.
Table 2. Examples of solvents belonging to different toxicity classes within ICH Q3C classification.
|Class||Examples of solvents||Notes|
|Class 1 (to be avoided)||benzene, chlorinated hydrocarbons||carcinogenic, teratogenic, toxic, environmentally hazardous|
|Class 2 (to be limited)||hexane and other hydrocarbons||toxic or carcinogenic|
|Class 3 (low toxicity)||acetone, DMSO, ethanol and other polar solvents||not toxic in reasonable quantities|
|Class 4 (non-toxic)||water, edible oils||not toxic in any quantities|
All residual solvents must be reported, and the regulatory framework is based on the calculated daily intake of the solvent based on the limits placed on its concentration in the final substance.
Ever since the thalidomide story, enantiomers have been tightly controlled in pharmacology. Although racemic mixtures may be sold, the enantiomers must be characterised individually for their therapeutic profile, pharmacokinetics and toxicity. In some cases (notably ibuprofen), the pharmacological efficiency of different enantiomers is markedly different. In other cases (levofloxacin, albuterol, omeprazole) it is not. In some cases, only one enantiomer is permitted (e.g. timolol, clopidogrel) because the other is markedly toxic. The literature contains a rather memorable table of how different the properties of enantiomers can be:
|Substance||Enantiomer 1||Enantiomer 2|
|thalidomide||(S) – teratogen||(R) – anti-nausea drug|
|barbituric acid derivatives||(S) – sedative||(R) – convulsant|
|morphine derivatives||(R,S) – narcotic analgesic||(S,R) – cough suppressor|
|labetalol||(S,R) – alpha blocker||(R,R) – beta blocker|
|penicillamine||(D) – anti-arthritic||(L) – toxic|
The most significant matter in impurity classification in the pharmaceutical industry context is whether the impurity is identified or not. Much stricter standards apply to the impurities whose identity or toxicological properties are unknown. The following separation and identification methods exist.
The most straightforward way is to synthesize the pure substance that is suspected to be the admixture and to compare, for example, the NMR spectrum of that pure substance with the NMR spectrum of the admixture.
Perhaps the most informative structure elucidation methods are – in the order of increasing information content: optical spectroscopy, infrared and Raman spectroscopy, mass spectroscopy, nuclear magnetic resonance, X-ray diffraction. Having a sufficiently high quality dataset from the latter two methods essentially guarantees structure determination. However, laborious crystallisation may be required for X-ray, and expensive and fiddly isotope enrichment for NMR.
As analytical separation methods go, the different versions of chromatography with spectrophotometric or mass-spectrometric detection are likely the best. The most popular types of chromatography are high-performance liquid chromatography (HPLC) and gas chromatography (GC). The latter is only applicable to volatile substances. Industrial separation methods gravitate more towards the solvent extraction, which is cheaper and less laborious than industrial grade HPLC. Extraction can proceed from a solid to a liquid, or from a liquid into another liquid. The residual solvent problem discussed above presents itself fully at this stage.
Once the substance is fully characterised, the best method for detecting its presence in the industrial context is HPLC-MS or GC-MS. Mass spectra are highly characteristic, and mass-spectrometric equipment is cheaper and more sensitive than NMR equipment.
The following general principles apply to controlling and reporting pharmaceutical impurities in Europe and North America. Any manufacturer is required to produce the “impurity profile”, which is a list of:
In addition to the above, the following information, insofar as applicable, must be provided for a pharmaceutical formulation to be certified as safe:
In real pharmaceutical practice, when an impurity is detected, its structural identification is attempted. If structural identification fails, the concentration must simply be reduced to the NMT level (an expensive and laborious process). If structural identification succeeds, the concentration must be reduced to a level that is considered safe for humans.
Impurities have an existential importance in the design, synthesis, and manufacture of pharmaceutical ingredients. Given the current regulatory climate, it is either impossible or uneconomic to produce pharmaceutical formulations with unknown impurities. Large resources are therefore expended by pharmaceutical companies in order to design processes that minimise the production of undesired substances and identify those that do occur.