Test protocols for trace metals in pharmaceuticals are currently being updated. Nikki Schopp, Team Leader, Analytical Laboratory Services, SGS Life Science Services, looks at some of the issues that can arise when seeking the appropriate test method
There is always a possibility that trace quantities of heavy metal elements may be present in pharmaceutical products. Historically, the biggest risks came from those that might arise from natural sources – arsenic, cadmium, lead and mercury. In the past few decades, however, transition metal catalysts have increasingly been used in the manufacture of APIs, leaving the possibility that traces of more unusual metals might also be present. Metals commonly found in catalysts include palladium, platinum, nickel, rhodium, ruthenium, chromium and copper, among others.
To say that the US Pharmacopeia’s (USP’s) testing protocol for heavy metals is outdated is something of an understatement. The method, enshrined in USP <231> Heavy Metals, prescribes the qualitative method of sulfite precipitation followed by a visual comparison, and requires checks for 10 metals – antimony, arsenic, bismuth, cadmium, copper, lead, mercury, molybdenum, silver and tin – missing out many of those that might be present as catalyst residues. Almost 20 years since the issue was first raised, replacement chapters USP <232> Elemental Impurities – Limits and <233> Elemental Impurities – Procedures are now close to being adopted. USP <232> lays down permitted limits for a range of metals, or elemental impurities, as they are now denoted. USP <233> sets out permitted testing protocols.
USP <231> relies on a visual colour comparison, rendering very low levels undetectable, and provides no quantitative measure of the amount of heavy metals. In contrast, the new, state-of-the-art protocols rely on inductively coupled plasma (ICP) techniques. They allow for the use of ICP atomic or optical emission spectroscopy, or ICP mass spectrometry, depending on how a metal can be detected but do not preclude the use of other techniques, as long as all the stringent validation requirements are met. This allows concentrations down to low-ppb levels to be detected, and more quantitative measurements can be made.
ICP exploits the fact that excited electrons emit energy at certain wavelengths, and the excited ions they produce emit a wavelength characteristic to that specific element. This is another important improvement – it is possible to identify precisely which metal is present, which was not the case with the century-old USP <231> test.
Metals commonly found in catalysts include palladium, platinum, nickel, rhodium, ruthenium, chromium and copper
Although these tests are still under discussion, a good deal of work is going on ahead of implementation in terms of method development. High on the list is developing and validating methods for sample preparation. There are four main ways in which a sample might be prepared. The simplest, on paper at least, is to use a neat sample where the substance being analysed is a liquid. If it is soluble in water, a direct aqueous solution can be made; if the drug product has been formulated for injection or infusion, then the drug product itself would be an appropriate sample.
If it dissolves in an organic solvent, then a direct organic solution can be used, although these samples are a little more harsh on the instrument, and identifying the correct analytical parameters tends to prove a little more difficult. If none of these is appropriate, then an indirect sample must be made, usually via microwave digestion in a closed vessel, to break down the sample into soluble components. This is the last resort, as it takes by far the longest time. It is also the most costly as much more method development is necessary, as the best acid and heat profile must be determined.
When developing a microwave digestion protocol for a specific sample, the ideal is to find the lowest possible temperature necessary to break bonds in the shortest timespan. However, sometimes what may initially be considered a ‘more simple’ sample fails in validation studies, and microwave digestion must be used after all.
In theory, if a direct aqueous solution of the drug product is available, preparing a suitable sample should be straightforward. However, this is not always the case. In this example, the sample was supplied as a formulation for infusion, so all that was required was the spiking studies for validation. Three different sample preparations were needed for these studies. The first used a 2% nitric acid diluent to create a sample that would work for silver, mercury, ruthenium, cadmium, lead, chromium, molybdenum, palladium and vanadium. The second used a stronger concentration of a combination of acids – 12% nitric acid plus 8% hydrochloric – for nickel, copper, iridium, platinum and rhodium. Finally, the third solution used 2% hydrochloric acid for osmium.
These samples are then spiked with metals at three different levels before testing – at half the limit, the limit, and 1.5 times the limit. These spiked samples were run on the instrument, and while the accuracy requirements laid down in the USP chapter were met for many of the elements, they were not for nickel, copper, rhodium, osmium or platinum – all fell below the necessary 70% average recovery.
Something within the sample was clearly inhibiting the instrument from detecting these elements
Clearly, the simple weigh and dilute method was not going to work for this sample after all. As it was already in solution, the only other possibility was to use microwave digestion to break down the drug product into smaller fragments, as something within the sample was clearly inhibiting the instrument from detecting these elements.
After microwave digestion was applied, all the elements passed the requirements with the exception of osmium, and from our experience osmium appears to be a common problem across samples. The answer lay in returning to the standard weigh-and-dilute method to carry out a separate test for osmium, using hydrochloric acid as the diluent instead of nitric acid, as osmium is known to behave better in hydrochloric acid. This time, osmium passed.
This case study highlights that just because a drug product is already in solution, it does not mean that sample preparation is unnecessary. Rather, validation studies are essential to prove whether the sample is suitable for analysis or not, and to guide sample preparation studies if the simple diluted sample does not meet all the necessary requirements.
When developing and validating sample preparation methods, it is important to use the exact same material as the drug product. In this example, after the sample preparation method had been developed and all recovery criteria met, the customer supplied a different sample of the material for validation. The exact same techniques and tests were run, but this time the material being used in the test had been purified, in place of cheaper, unpurified material that had previously been provided as a cost-saving exercise.
While in some cases this might not be a problem, here it definitely was, as all of the results for palladium ran low when the samples were prepared and analysed using the purified material, but had not for the unpurified product.
The problem arose because the sample had been subjected to an acid-based purification process
The problem arose because the sample had been subjected to an acid-based purification process, and thus the acidity of the sample was different from that of the unpurified version. This affected the amount of palladium that could be recovered from the sample. While this was an issue only for palladium, and the other elements were not affected at all, it highlights a real problem as palladium is one of the metals most commonly used in catalysts. It is also one of those that the existing colour comparison tests cannot identify.
This is still a work in progress, with the method having to be re-developed in the light of its impact on palladium. But it does highlight the importance of developing a specific preparation protocol for each individual sample, regardless of how similar they are, and not substituting purified for unpurified material after the protocol has been developed.
This third case study has similarities to the previous one, in that the palladium recovery levels fell below the required threshold during validation. While only the half-level spiked sample failed, if it fails at half the limit, it has failed regardless of whether the other results are acceptable.
The client had already developed a sample preparation method for this sample, using nitric acid and hydrogen peroxide and warming it to 105°C in a normal oven. We started out using this too, as it is simpler and quicker than going through a full microwave digestion process, and it had worked for all samples throughout development. Yet once more, when the purified product was sent through for validation testing, palladium recovery levels came out too low.
Sometimes no matter what is tried, nothing appears to work for all metals
This led us to return to the drawing board, with a desire to redevelop the whole method so that only one sample preparation was required, rather than create a separate one just for palladium. While this is the ideal, it is not always possible, as sometimes no matter what is tried, nothing appears to work for all metals. In those cases, two (or more) methods will be necessary even though the testing process will take longer.
Our starting point was the acid/peroxide combination that had been used previously in the oven-based digestion, as we felt it was unlikely to adversely affect the results for any of the other metals. This time, the digestion was carried out via microwave heating. It was a success – palladium recovery was well above the limits for all three spiked samples. The test has now been validated, and is in routine use for elemental impurities testing in this drug product.
Even when USP <232> and USP <233> have been adopted, the monographs will not provide specific sample preparation methods. Rather, the appropriate method will have to be determined for each drug product on a case-by-case basis. As can be seen from the examples above, while solubility plays a role in determining what is appropriate, just because a solution can be made does not mean it makes a suitable sample for elemental impurities testing.
Ultimately, the path forward will depend on the exact wording of the monographs when finalised. As USP delayed its implementation to harmonise requirements with ICH (International Conference on Harmonisation) and the European Pharmacopoeia ahead of adoption, this may still be some time off. However, a good deal of work is being carried out ahead of time to ensure that clients are ready to meet the new demands as quickly as possible once they are implemented.