Navigating the Physicochemical Maze
The way a human body interacts with and processes a drug molecule is dependent upon its physical properties that are baked into the structure at the design stage. But of the many properties we consider during drug design, which ones actually matter? Clearly, across “drug discovery space”, one size does not fit all, but certain key descriptors consistently come to the fore.
Several of a molecule’s properties can give drug developers an insight into how it might behave in the body, particularly those pertaining to lipophilicity, described by the partition coefficient, (Log P) and the distribution coefficient (Log D). The latter is particularly informative, because it gives an insight into the dynamics of ADME behaviours, i.e. how a molecule is Absorbed, Distributed, Metabolised and Excreted to site of action. In recent times, insights into shortcomings in the measurements of Log P and Log D (at given pH) have led to new more reliable measurements and, in turn, better predictive data.
Log D, usually at measured at pH 7.4, is perhaps the most critical parameter in drug discovery. But this measurement on its own is not without challenge. While the shake-flask octanol–aqueous buffer method for measuring Log D is simple in principle, it can be difficult to get right, particularly for poorly soluble compounds. More lipophilic molecules are inherently less soluble, so estimating their behaviour is difficult with two phase systems, but the dynamic range can be extended or complemented by using chromatographic techniques, irrespective of solubility. [1,2]
Looking beyond Lipophilicity
So, lipophilicity remains a critical factor, though it’s important to recognise, not the only factor. A couple of papers published a few years back, one by Novartis  and the other from Pfizer , looked into these extra parameters in order to develop what they termed an “extended clearance classification”. Clearance remains a factor in many projects, but it’s worth considering that a simple hepatocyte-based screen is unlikely to be successful in predicting in vivo clearance for very polar non-permeable acids, as transporters are likely to be a key determinant of DMPK and distribution. Here, the authors classified a range of marketed drugs according to whether they were acids, bases, zwitterion or neutral, and considered attributes such as permeability and molecular weight. The aim was to understand the molecule’s DMPK profile and disposition based on these properties. Here, factors such as ionisation, molecular weight and permeability were found to be key determinants of transporter-mediated active efflux and thus clearance from the body – so limiting therapeutic exposure.
Whilst molecular weight is easily and accurately estimated, ionizability (or pKa) permeability can be more challenging. Predicting pKa is not easy and current computational methods are limited by the extent, quality and diversity in available data to train machine learning or ab initio models. Clearly, poor pKa estimations contribute to poor Log D prediction, even with well-curated and reliable models for Log P prediction.
Estimations of permeability remain contentious – cheaper, simpler PAMPA permeability assays have tended to fall from favour as they too often do not correlate with cell activity or oral bioavailability. The elephant in the room with permeability is poor understanding of the roles of influx and efflux transporters, yet experience shows that the likelihood of oral exposure can be estimated with plots of CMR (calculated molar refractivity – a physical factor related to molecular size) and Log D. 
So, while the medicinal chemistry community, overly influenced by the Rule of 5, retains a staunch focus on molecular weight, this number could seemingly just be a surrogate for other properties that it often happens to track. While the Rule of 5 puts an upper limit on molecular weight of 500, in reality, many larger than this have made successful drugs, particularly in the antibacterial field where big, complex natural products are commonplace. But how much does molecular weight actually matter?
Weighing up Molecular Weight
AbbVie took a look at the DMPK characteristics for “beyond rule of 5” chemical space in its drugs and compound collection in 2017,  and the parameters that contribute to the oral bioavailability of such molecules. The resulting multiparametric scoring function (MPS) correlates preclinical PK results with cLog D7.4 (actually the magnitude of calculated Log D – 3, the optimum value for permeation), the number of rotatable bonds, and the number of aromatic rings. This analysis delivered a useful set of principles to predict the chances of achieving oral bioavailability based on empirical data, and it suggested that molecular weight did not influence the outcome.
In many cases, new drug molecule often comply with Rule of 5 stipulations, notably within lipophilicity and Hydrogen Bond Donor guidelines, but for challenging targets such as protein–protein interactions there are a growing number of higher molecular weight drugs that necessarily require more hydrogen bond acceptors to modulate their lipophilicity.  But of all the physicochemical properties, one that repeatedly causes on-going headaches is solubility.
Sense and Solubility
Solubility – or the lack of it – is a growing issue. Even those compounds that initially appeared acceptable may hit problems as the compounds progress and new, more stable polymorphs are encountered. And if solubility and/or rate of dissolution is inadequate, increasingly larger doses will not increase the exposure. The average volume of water in the human stomach is about 250 mL, so if a delivered dose of an oral drug cannot dissolve in that volume, it’s likely that there will be little or no absorption. Given this, GSK noted that for clinical success, solubility > 100 µg/mL and/or a predicted human dose of < 100 mg were common characteristics of successful drugs. 
While formulation tricks can be applied to a compound to improve its solubility and dissolution rate, such as hot melt extrusion and amorphous spray dispersion, it would be preferable if compounds were more soluble in the first place. Inherently more soluble compounds take less time to develop and enable better predictions in PK/PD simulations, enabling better clinical outcomes due to less dosing variability.
Yet despite the recognition that insolubility is a problem, it persists. Broadly speaking, Big Pharma companies were – and still are – excellent at identifying compounds with good DMPK properties to advance into the clinic, but perhaps less good at de-risking efficacy and safety. Biotechs, while more risk-averse, are more focused on speed, and often measured solubility is one of the corners that may be cut in the drive to reach a milestone quickly.
Considerations and Context
Perhaps, at the heart of the issue, the ability to carry out so many experiments with tiny amounts of material contributes to this situation: if only a couple of milligrams of amorphous drug have been made for testing, there will be no clues about its crystalline form. If no one who has physically handled the compounds is involved in the testing and development decisions, it is easy for important parameters such as solubility to be overlooked.
Whilst these properties are important, the chosen route of administration will also have a bearing on the optimal properties of a molecule, and those for an orally available drug will differ from one that will be administered, for example, via infusion in a hospital setting, as might be the case for a last-line-of-defence antibacterial. In this critical care setting, solubility is the most important consideration, along with clearance and the target unbound concentration, and these parameters eclipse other considerations.
Of course, with determination, all these challenges can be overcome. But they do increase risk and cost of a project, and so there is a clear case to be more DMPK aware. And even if these issues are tolerable in the discovery phases, the inability to attain high dosing and exposure in, for example, non-GLP toxicity and Cyp induction studies, may cause unanticipated delay and challenge further down the development track.
Ultimately, the key requisite for any potential therapeutic compound is to achieve enough unbound drug concentration at the target for long enough to cause the required biological phenotype. Early human dose prediction is also often overlooked, but can be very helpful in terms of identifying key risks and future strategies for optimisation.
But to really understand which properties are important, we need to understand molecular risks and what issues need to be addressed moving forward. To fully assess and understand our lead molecules earlier in the drug discovery process. To acknowledge that each project is different and will require different solutions to the encountered challenges. And to engage early with our DMPK colleagues, to truly understand the appropriately measured properties of the molecules we elect to work with and the complexities those properties may cause us over the life of our projects.
This article summarises the vibrant discussions during a recent roundtable event hosted by Sygnature Discovery and chaired by Dr Rob Young, Principal of Blue Burgundy Ltd. We hope you found it thought provoking and stimulating. We love to connect and interact with scientists across the globe, to discuss all aspects of drug discovery science and to help drive forward the progression of experimental therapies toward the clinic and patient benefit. If you’d like to strike up a conversation, or if we can help you accelerate your own projects toward patients, we’d love to hear from you. Please reach out using any of the contact forms.
Young, R. J.; Green, D. V.; Luscombe, C. N.; Hill, A. P., Getting physical in drug discovery II: the impact of chromatographic hydrophobicity measurements and aromaticity. Drug Discov Today 2011, 16 (17-18), 822-30
Lombardo, F.; Shalaeva, M. Y.; Tupper, K. A.; Gao, F., ElogD(oct): a tool for lipophilicity determination in drug discovery. 2. Basic and neutral compounds. J. Med. Chem. 2001, 44 (15), 2490-7
“Application of the extended clearance concept classification system (ECCCS) to predict the victim drug-drug interaction potential of statins”, G. P. Camenisch et al., Drug Metabolism and Personalized Therapy, 2015, 30(3), 175
“Extended Clearance Classification System (ECCS) informed approach for evaluating investigational drugs as substrates of drug transporters”, M.V. Varma et al., Clinical Pharmacology and Therapeutics, 2017, 102(1), 33
Tinworth, C. P.; Young, R. J., Facts, Patterns, and Principles in Drug Discovery: Appraising the Rule of 5 with Measured Physicochemical Data. J. Med. Chem. 2020, 63 (18), 10091-10108.
“Beyond the Rule of 5: Lessons Learned from AbbVie’s Drugs and Compound Collection”, D. A. DeGoey et al., Journal of Medicinal Chemistry, 2018, 61(7), 2636
Shultz, M. D., Two Decades under the Influence of the Rule of Five and the Changing Properties of Approved Oral Drugs. J. Med. Chem. 2019, 62 (4), 1701-1714.
Bayliss, M. K.; Butler, J.; Feldman, P. L.; Green, D. V.; Leeson, P. D.; Palovich, M. R.; Taylor, A. J., Quality guidelines for oral drug candidates: dose, solubility and lipophilicity. Drug Discov Today 2016, 21 (10), 1719-1727.]