Strategies for challenging compounds and separations encountered by the Drug Discovery Analyst

Not so many years ago, the biggest problem encountered in HPLC was, arguably, peak shape most notably of basic compounds interacting strongly with active sites present in the silicas in use at the time. However, the great improvements made in silica manufacturing processes leading to far greater purity and control of the surface chemistry has totally transformed the landscape such that now high efficiency is almost taken for granted. Another historical headache all but forgotten is the old bane of batch-to-batch and column-to-column reproducibility. Such are the improvements in synthetic and packing technology that, combined with similar ultra-precision in modern instruments, retention time precision from column to column and even instrument to instrument is typically around 0.1% (for new columns).

So, what problems have we got left? Let’s look at two areas.

Poor (UV) detectability

There is still no universal detector with adequate characteristics of sensitivity, linearity etc to replace UV absorbance as the best choice quantitative detector for many applications such as purity assessment (QC) and many others with analyte concentrations over, say, 1uM. However, when the analyte chromophore is so poor as to render the molecule effectively undetectable by UV or at least compromised such that, for example, it gives mis-leading results for purity applications (because the impurities absorb more strongly on a molar basis – of course no one complains when it is the other way around!) we have to think again.

The first consideration is wavelength. Generic applications frequently employ a diode array detector with a broad signal wavelength range, typically from 220nm up to somewhere between 300 and 400nm. This gives a kind of average chromatogram over the typical spectrum of absorbance of many drug-like compounds and is deliberately non-selective compared to a single wavelength (narrow bandwidth) which may be desirable (as in purity assessment). It is well known that “purities” determined at, for example, 254nm and 270nm can be very different depending on the spectral profiles of product and impurity molecules. However, detection of compounds with poor chromophores can often be greatly improved by switching to a single wavelength between 210 and 220nm (with a bandwidth of 4nm). This may result in a rather large solvent front, especially if the sample contains much DMSO, but can still be useable provided the compound of interest has good retention. Expect to see a sloping baseline in one direction or the other which will require attention to the integration parameters but, depending on the application, low wavelength UV detection should not be overlooked.

The next line of attack for compounds not suited to UV absorbance detection is an evaporative detector, either light scattering (ELSD) or charged aerosol (Corona CAD). While these inevitably add some complexity to an LC or LC-MS system, it is certainly worth having access to at least one system with such a detector. The same mobile phase restrictions that apply to MS detection apply to ELSD or CAD, i.e. no involatile additives are compatible. On the other hand, they can be very good at detecting inorganics in samples such as chloride and other (counter) ions.

ELSD can be rather forgiving in terms of purity measurement thanks to the sigmoidal shape of its response with analyte concentration which can result in the missing of low level impurities. As seen in the figure below, which shows the same separation on 2 different LC-ELSD systems, they are very reproducible.


The variation in response with gradient (% acetonitrile) due to a smooth change in the nebulisation efficiency can be (partly) negated by gas programming or by careful calibration procedures and software.

Evaporative-type detectors unsurprisingly lose sensitivity for molecules with appreciable volatility which tends to be for organic molecules below a molecular weight of around 250. However, these molecules may be well set up for GC separation and detection and, if you can do GC, there is certainly a good argument that you should. It will almost certainly give you the most efficient chromatography (sharpest peaks) and, from the detection point of view, opens up different possibilities including FID and EI mass spectrometry, both of which are very non-selective in nature and hence well-suited to purity applications. Obviously, any decision to make the (modest) investment in GC in a drug discovery setting will depend very much on how many amenable compounds you expect to encounter.

Polar Compounds

The other common problem, in my experience, is lack of (reverse phase) retention of polar compounds. Depending on the severity of the “problem” a number of strategies are available. Note I’m assuming that simple pH manipulation is either not relevant (no analyte ionisable groups) or has already proved ineffective (neutral compound still poorly retained) or the compound has permanent charge at all accessible pH values.

For a modest increase in retention, I have had particular success with Waters T3 column technology and this would be my first option in a lot of cases as it retains the convenience of reverse phase chromatography and large scale columns are available for preparative separations.

Second choice would be to switch to normal phase as, by definition, this will lead to higher retention of polar compounds. Preferably this would be by SFC to take advantage of all the benefits of this technique that result in much faster method development compared to normal phase LC. Again, SFC is a very favourable choice for large scale preparative separations.

My third choice would be to investigate Mixed Mode (MM) columns which combine both reverse phase and ion exchange ligands and hence allow multiple retention mechanisms. Obviously, a knowledge of analyte pKa values is necessary for appropriate choice of mobile phase pH. The buffers and solvents used are the same as one would use in reverse phase chromatography and are compatible with MS detection. The column I have enjoyed most success with is the Scherzo SS-C18 which combines C18 ligands with both strong anion and strong cation exchangers.

In my experience, HILIC has been less useful and much more difficult to work with than MM chromatography due to the long equilibration times (and ionic strength effects) encountered. Nevertheless, I have had good results with Waters Amide column in particular. Note that, as a rule of thumb, for HILIC to work it requires molecules with logD less than zero.

Finally, if you have access to Capillary Electrophoresis (CE), perhaps a system being used primarily for biomolecular separations, then certainly take advantage of this for charged molecules. Simple free solution capillary zone electrophoresis (CZE) is made for separating small, highly polar, charged molecules and results can be achieved very quickly and are highly predictable. CE also allows detection down to extremely low wavelengths, as low as 185nm, as well as indirect UV (and fluorescence) detection which has been used for detection of inorganic ions and other non-absorbing species. CE also generates negligible amounts of waste.

Further Information

If any of the issues discussed in these articles are of interest to you, please feel free to contact me directly on for further information and/or sign up at the bottom of this page to receive notifications of future articles and events.

Mark PortsmouthComment