Guidelines for the Quality of Compounds from Medicinal Chemistry
Final compound QC for registration
One fundamental question to be answered by Chemistry Department Management is who actually does this. For practical reasons and speed it is very likely to be the Medicinal Chemist making the compound using the same Open Access LC/MS system that is used for reaction monitoring (see my previous article for a discussion of the problems this can cause and their mitigation) and Open Access NMR. But is there any independent data checking and, if so, who by? The data may be checked by his or her supervisor, or team/project leader. At least they will be looking with a fresh pair of eyes but the acceptability of this is open to question. There is an argument that the final approval of the data should come from an experienced analytical chemist. This task need not be onerous or cause delays especially if managed within an ELN.
The goal here is high quality data for clear decision making. This is only realised through high quality compounds. Problems may not be seen in potency assays but you can be sure they will be picked up in other assays, maybe first in ADME leading to a significant waste of time for those scientists. It is far better for these issues to be picked up at source. As an aside, there are many benefits to having Compound Management QC, Medicinal Chemistry support, Phys Chem and ADME/bioanalysis scientists working side by side at least in the office environment. There is a lot of common ground in what they do. Admittedly this becomes more difficult as companies become larger.
QC of DMSO stock solutions by LC/MS
The main purpose of this activity is to assure the integrity of the Compound Management process, i.e. the right compounds have ended up in the right wells. It should be run in parallel with potency screening (in vitro Pharmacology) and hence assures the validity of that data, adding to the accuracy and security of the screening workflow.
It needs to be carried out to a high level by an experienced analytical chemist familiar with interpreting mass spectrometry data (single quad) for compound ID confirmation as well as LC purity issues – main peak within acceptable retention time window, peak shape, peak height, quality of integration etc – and who has the skills to generate acceptable data if not afforded by the first pass analysis.
Stability of DMSO stock solutions
Where these have been stored for a significant period of time before a re-test is requested, e.g. for a secondary (selectivity) assay, QC data from the fresh stock is effectively ‘out of date’ and may not reflect the current quality of the test sample. Hence, a repeat QC should be triggered by a request for DMSO stock that is, say, more than 4 weeks old. Four weeks generally means that we are beyond the initial screening cycle. All of this is amenable to automation within compound management software that is fit for purpose.
Certificates of Analysis
For certain studies, a higher level of compound quality assurance is required. These studies include key in vitro tox studies such as an AMES test and key pharmacological tests. The CoA tests will be run on the solid or a freshly prepared solution of the same batch to be used in the key biological studies. It may include the following sections.
Ideally, the sample should be analysed in two different systems, e.g. at low and high pH. The gradient times should be extended compared to “fast” LC (say 9 minutes compared to 1.5) with a broad gradient, typically 5-95% organic solvent, in most cases. Careful integration of the chromatograms is important as is the analysis of exactly-matched blanks to identify any peaks not derived from the sample itself. The peak height or area should be controlled within acceptable limits by adjustment of injection volume and I would recommend dissolution initially in acetonitrile at 1mg/ml with subsequent 2X dilution in water if possible. DMSO should be avoided unless necessary.
Elemental (CHN) Analysis
The primary purpose of CHN analysis is to uncover the presence of inorganics, e.g. salts, and solvents missed by LC. Data may be routinely fitted to account for the presence of water but other species should only be allowed if there is good reason to believe their presence is plausible. Hence, an HCl salt may reasonably contain excess HCl and residual solvents of crystallisation are also to be expected. The latter may also be seen in the NMR and good agreement of the calculated levels between the two techniques should be achievable.
Ideally, all batches for CoA should be re-crystallised from a Class II (non-AMES positive) solvent and the melting point range recorded.
A single quadrupole mass spectrometer is adequate and the spectra can be collected via LC/MS or direct injection. The expected M+H and/or M-H ion(s) should be found even if they are not the base peak. The identity of the base peak, either a fragment or adduct, should be understood and consistent with the structure. If both the M+H and M-H ions are particularly weak, effort should be made to adjust the MS source conditions to increase their abundance. If available, accurate mass (±5ppm) data is, of course, preferable.
The registration spectra should usually be adequate to unambiguously prove the structure. Where this is not the case, additional experiments should be performed. Where solvent is identified in the spectrum, an estimate of its weight% in the sample should be calculated and should be consistent with the calculated level from CHN if available. The spectrum should be checked for grease contamination around 0ppm, which should be absent.
What is Purity?
The following are typical purity criteria in use at different stages of the drug discovery process
In general, these are LC-UV “purities” i.e. they are the area% of the main peak on a particular LC system (column/mobile phase/gradient/temperature/flow rate) with given detection parameters (wavelength). This is obviously not the same thing as weight% which might be considered the ‘true’ purity of a batch of compound.
Clearly, LC is capable of missing numerous components that may either be transparent to it or are not eluted or are hidden beneath the solvent or the compound of interest itself. Hence the need for (a) multiple separation and detection systems and (b) unrelated techniques such as CHN, MS and NMR to corroborate LC “purity” data.
Also, generally, only a single sample is taken with no thought to homogeneity of a solid sample or sampling protocols. An exception to this is CHN analysis which is usually at least carried out in duplicate and variations in this data outside the specification can reflect sample inhomogeneity.
If any of the issues discussed in these articles are of interest to you, please feel free to contact me directly on firstname.lastname@example.org for further information.