06 October 2014

Physical properties in drug design

This is the title of a magisterial review by Rob Young of GlaxoSmithKline in Top. Med. Chem. At 68 pages it is not a quick read, but it does provide ample evidence that physical properties are ignored at one’s peril. It also offers a robust defense of metrics such as ligand efficiency.

The monograph begins with a restatement of the problem of molecular obesity: the tendency of drug leads to be too lipophilic. I think everyone – even Pete Kenny – agrees that lipophilicity is a quality best served in moderation. After this introduction Young provides a thorough review of physical properties including lipophilicity/hydrophobicity, pKa, and solubility. This is a great resource for people new to the field or those looking for a refresher.

In particular, Young notes the challenges of actually measuring qualities such as lipophilicity. Most people use log P, the partition coefficient of a molecule between water and 1-octanol. However, it turns out that it is difficult to experimentally measure log P for highly lipophilic and/or insoluble compounds. Also, as Kenny has pointed out, the choice of octanol is somewhat arbitrary. Young argues that chromatographic methods for determining lipophilicity are operationally easier, more accurate, and more relevant. The idea is to measure the retention times of a series of compounds on a C-18 column eluted with buffer/acetonitrile at various pH conditions to generate “Chrom log D” values. Although a stickler could argue this relies on arbitrary choices (why acetonitrile? Why a C-18 column?) it seems like a reasonable approach for rapidly assessing lipophilicity.

Next, Young discusses the influence of aromatic ring count on various properties. Although the strength of the correlation between Fsp3 and solubility has been questioned, what’s not up for debate is the fact that the majority of approved oral drugs have 3 or fewer aromatic rings.

Given that 1) lipophilicity should be minimized and 2) most drugs contain at most just a few aromatic rings, researchers at GlaxoSmithKline came up with what they call the Property Forecast Index, or PFI:

PFI = (Chrom log D7.4) + (# of aromatic rings)

An examination of internal programs suggested that molecules with PFI > 7 were much more likely to be problematic in terms of solubility, promiscuity, and overall development. PFI looks particularly predictive of solubility, whereas there is no correlation between molecular weight and solubility. In fact, a study of 240 oral drugs (all with bioavailability > 30%) revealed that 89% of them have PFI < 7.

Young summarizes: the simple mantra should be to “escape from flatlands” in addition to minimising lipophilicity.

The next two sections discuss how the pharmacokinetic (PK) profile of a drug is affected by its physical properties. There is a nice summary of how various types of molecules are treated by relevant organs, plus a handy diagram of the human digestive track, complete with volumes, transit times, and pH values. There is also an extensive discussion of the correlation between physical properties and permeability, metabolism, hERG binding, promiscuity, serum albumin binding, and intrinsic clearance. The literature is sometimes contradictory (see for example the recent discussion here), but in general higher lipophilicity and more aromatic rings are deleterious. Overall, PFI seems to be a good predictor.

The work concludes with a discussion of various metrics, arguing that drugs tend to have better ligand efficiency (LE) and LLE values than other inhibitors for a given target. For example, in an analysis of 46 oral drugs against 25 targets, only 2.7% of non-kinase inhibitors have better LE and LLE values than the drugs (the value is 22% for kinases). Similarly, the three approved Factor Xa inhibitors have among the highest LLEAT values of any compounds reported.

Some of the criticism of metrics has focused on their arbitrary nature; for example, the choice of standard state. However, if metrics correlate with a molecule's potential to become a drug, it doesn’t really matter precisely how they are defined.

The first word in the name of this blog is Practical. The statistician George Box once wrote, “essentially, all models are wrong, but some are useful.” Young provides compelling arguments that accounting for physical properties – even with imperfect models and metrics – is both practical and useful.

Young says essentially this as one sentence in a caveat-filled paragraph:

The complex requirements for the discovery of an efficacious drug molecule mean that it is necessary to maintain activity during the optimisation of pharmacokinetics, pharmacodynamics and toxicology; these are all multi-factorial processes. It is thus perhaps unlikely that a simple correlation between properties might be established; good properties alone are not a guarantee of success and some effective drugs have what might be described as sub-optimal properties. However, it is clear that the chances of success are much greater with better physical properties (solubility, shape and lower lipophilicity). These principles are evident in both the broader analyses with attrition/progression as a marker and also in the particular risk/activity values in various developability screens.

In other words, metrics and rules should not be viewed as laws of nature, but they can be useful guidelines to control physical properties.


Pete said...

I can’t currently see the article although I hope that ‘magisterial’ means that the author has moved on from:

“This graded bar graph (Figure 9) can be compared with that shown in Figure 6b to show an increase in resolution when considering binned SFI versus binned c log DpH7.4 alone.”

On a (slightly) more serious note, however, is the issue of which of logP of log D (let’s assume octanol/water) is the more appropriate (dare I say useful) measure of lipophilicity in drug design applications. So let’s suppose that I’ve got a compound with a carboxylic acid, a naphthalene ring and a benzene ring in its molecular structure. It has a PFI of 7.5 and I desperately want to transform this molecular turd into something less aromatic. So desperate am I that I’m even prepared to seek the blessings of the Sages of Stevenage. One thing I can do is get rid of one of the aromatic rings so let’s do that without further delay. Let’s turn that naphthalene into a naphthaquinone and being a molecular modeler that’s something I can do with a couple of mouse clicks. As an aside I should mention that I actually used to synthesize quinones when I was a post-doc in Minnesota. PFI is now 6.5 (actually a bit less because I introduced two polar atoms) and hopefully my modest efforts will meet the approval of the compound quality police. However, I really want to show how serious I am about being PFI-compliant so I’m going to do something about logD as well. There are two ways to reduce logD which are to reduce logP or to increase the extent of ionization. I’m going use the latter tactic and replace the carboxylic acid with a sulfonic acid and I’m confident that logD will decrease significantly. How much? Don’t know because sulfonic acids are strong acids but I’d expect at least 3 or 4 log units which should get PFI down below 3-ish.

Do I pass or fail Drug Design 101?

Anonymous said...

I have been involved in a great number of debates about PFI and one of the issues is that it treats all aromatic rings as having the same impact on solubility.

The other issue is how much of a REAL difference there is between PFI of 7.1 vs PFI of 6.9

Dr. Teddy Z said...

I think this comes down to, understand your metrics and let them GUIDE you. But don't be beholden to them.

Dan Erlanson said...

I agree with Teddy on guidance as opposed to tyranny, so Anonymous's comment about the difference between a PFI of 7.1 vs 6.9 is a bit of a straw man. However, the point about treating all aromatic rings the same does seem to be a weakness.

Pete, I'm afraid your example fails Drug Design 101 most PAINfully, by design I'm sure. However, you could replace the naphthalene (ClogP = 3) with a quinazolinone (ClogP = 1.2), which would take your PFI below 7. Getting rid of the naphthalene is likely to please not just the Sages of Stevenage but also the Authorities of Alderley Park and the Gurus of Groton.

Pete said...

Hi Dan, are you saying that the naphthaquinone is verboten? The Sages of Stevenage might even disagree with you on that point.

Dan Erlanson said...

I don't know that the word "verboten" belongs in drug discovery. That said, if you do depart from rules, guidelines, and conventional wisdom, you should have a compelling reason for doing so other than iconoclasm. Although naphthoquinones do appear in approved drugs, I'd wager that the vast majority of naphthoquinones that appear in the literature are spurious.

Anonymous said...

PFI does not treat all aromatics as having the same impact on solubility (or any other property), the differences will be reflected in the hydrophobic component of PFI not the ring count alone.

Pete said...

Anonymous (comment #7), you raise an interesting point and I need to think about this a bit more. My kneejerk reaction is that I disagree with you in that the aromatic ring risk associated with benzene is the same as that associated with pyridine even though replacement of benzene with pyridine will reduce the lipophilicity risk. One way to think about it is to ask whether PFI 'knows' whether the polarity is in the aliphatic or aromatic portions of the molecular structure. Like I said, you've raised an interesting point.

I'm not sure how effectively PFI separates the risks associated with lipophilicity and aromaticity. The analysis presented in support of SFI (a predecessor of PFI) appeared to consist of pointing at pictures and asserting that the trend shown in one was stronger than that shown in another. Hopefully those concerned have moved on from that.

Pete said...

Hi Dan/Teddy,

I have actually taken a naphthaquinone orally and I must admit that I was worried by its quninone-ness (and likelihood to cause a panel of self-appointed ‘experts’ to spit feathers) than I was of potential CNS effects of one of the alternatives (mefloquine). However, let’s not worry about this little diversion because there are more interesting things to discuss. Increasingly, I’m coming to the conclusion that we should simply stop talking about rules for drug discovery because any criticism of the rules leads to the kneejerk response that we should treat rules as guidelines. So why don’t we just talk about guidelines?

Guidelines (and metrics) for drug discovery need to be based on competent and honest analysis of data if you want people to use them for making decisions. Guidelines reflect trends in the data and the value (or usefulness, if you prefer) of a guideline depends on the strength (not the significance) of the trends used to establish it. For example, knowing that a coin lands heads up for 51% of a billion throws may tell us the coin is biased but it’s not a whole lot of help if you’re trying to predict the result of the next throw.

Let’s go back to the Flatland paper. Our correlation inflation article did a bit more than just question the strength of the correlation between Fsp3 and solubility. We actually stated the correlation coefficient and that was something that the authors of the study declined to do. Did they actually calculate the correlation coefficient and decide that the trend had to be shown in a more convincing manner? We will probably never know. However, this study does have the potential to become a text book example of correlation inflation because a non-proprietary data set was used. As an aside, you might want to take a close look at the chemical structures of the compounds in that data set to see whether or not you would regard them as representative of what people in drug discovery would be interested in.

That’s probably enough for now and I’ll leave you with a question. What would be the professional athlete’s equivalent of inflating correlations in data?

Dr. Teddy Z said...

As always you make excellent points. I agree we need to stop talking about rules. It's stupid and counterproductive. Rules are capricious and laws are arbitrary. Guidelines are based upon a consensus of the common knowledge, at best. Barring the type of data regression analysis that you have done and continue to do, that's the best we can hope for. Would I (and everyone else) like to see guidelines based upon trends in the data, absolutely. I however don't see this happening with internal data from Pharma (who wants to air decades of dirty laundry?) and I don't know of anyone who is going to do externally.
I continue to agree with you philosophically, and I will support you morally, but eventually, Sir, the windmill wins.

Anonymous said...

Arguments about ligand design metrics can be summed up as 'Sure it works in Practice, but does it work in theory?'

Pete said...

A consensus of the common knowledge is certainly a laudable objective although I don’t see that happening anytime soon when one self-appointed expert says we should use logD to quantify lipophilicity but a wannabe key opinion leader says we should use logP instead. A key problem for the drug discovery scientist is how to distinguish genuine insight (result of competent, honest analysis of relevant data) from opinion (which is supported rather than informed by data analysis).

The correlation inflation and LEM critiques were written primarily for two groups of people. The first group consists of drug discovery managers and these folk have duties that take them away from details like deciding what compound to make or how to analyze data. I want to help make these folk aware of the some of the questionable data-analytic practises that are used when ‘evidence’ is presented in support of what is essentially opinion (dare I say blind prejudice). Most of the correspondence that I received in the wake of the correlation inflation article was from people at the director+ level who certainly understand the implications of correlation inflation even if they haven’t got the time to explore the problem themselves.

The other group consists of drug discovery scientists who want to get maximum value out of the data available to them. I hope provide encouragement (and ammunition) for this group and let them know that it really OK to challenge widely-held opinions and ‘knowledge’ gleaned from observation of weak trends in structurally-diverse (how relevant to my series?) data sets. Although we suggested using residuals to measure extent to which activity beats a trend, the more important message is that we should analyze the activity data in the most appropriate manner rather than use generic, pre-packaged analysis (i.e. metrics) of questionable validity.

There are implications (and potential pitfalls) for those who seek to influence the direction(s) in which the drug discovery field should be headed and the molecular obesity article is a case in point. Would the molecular obesity article have reproduced the plot of promiscuity against median cLogP (or described the review in question as seminal) had there been a greater awareness of the correlation inflation problem at the time of writing? We will never know (although it would have been fun to a fly on the wall during any subsequent discussions).