Here’s a solid med-chem paper from Merck on the topic of extending half-life for small-molecule drugs. This obviously is most important (and can have the biggest effect) if your compound has a short half-life after dosing to begin with (and plenty of compounds do). As the paper notes, if you have constant clearance for the unbound fraction of your compound, then the relationship between dose and half-life is exponential, not linear (whereas most of the other combinations you can come up with are simply linear ones). That really gives you a handle to work with.
And of course, you don’t have to wait until human data to worry about such things. The allometric scaling factor between rats and humans is about 4.3, so if you’ve done your homework about what the mechanisms of clearance are (which enzymes are responsible and how well they correlate to human ones, for example), you’re ready to work from there. The paper notes, though, that these sorts of correlations have a much higher chance of going haywire on you for very short half-life compounds (less than two hours, say), so if you want to have any confidence at all in your dose prediction models, you really should get rat half-life out longer.
The group shows a plot of about 10,000 Merck compounds with rat measurements of unbound-clearance versus steady-state volume of distribution, and it’s (as you might figure) a pretty good correlation. The key, then, is to find changes in a molecule’s structure that will get it off that line, because otherwise you’ll end up in a similar situation to where you started. That is, you can find yourself reducing the unbound clearance and ending up with an effective half life that’s not really improved.
So what sorts of modifications actually improve the half-life? The paper shows the result of a large matched-molecular-pairs (MMP) analysis, where structures were varied at only one place. Increasing lipophilicity will do the trick, sometimes, if the partitioning of your compound into tissue increases more than just its binding to plasma proteins. The Merck data suggest that adding multiple fluorines to a compound’s structure tends to have this effect. The fluorinated compounds tend, apparently, to bind to tissue at an increased rate more than they bind to serum albumin. You may or may not be comfortable with this advice, since polyfluorinated compounds may also be harder to formulate, and some of that tissue binding may be to things that you’d rather not bind to. But it will increase your half-life, and it’s up to each med-chem team to see if they want to make that deal with this particular fluorocarbon-coated devil. The free lunch has (still) not yet arrived in the lobby.
Of course, another classic way to increase half-life is to block sites of metabolism. The Merck group noted this in a general way with matched molecular pairs that changed hydrogens to chlorines, hydrogens to methyls, and methyls to chlorines. H to Cl increased half life, presumably for similar reasons to fluorination. H to methyl did not, though: that does increase lipophilicity, but at the expense of adding new groups for the liver enzymes to chew on. But methyl to chloro did improve half-lives, on average – the lipophilicities are nearly the same, but you’ve battened down the hatches for enzymatic degradation. (As you’d imagine, H to CF3 is the extreme example of this effect).
How about replacing with more polar groups, rather than with nonstick coatings? On average, phenyl to pyridyl lowers half life, H to OH does too, and H to COOH really lowers half-life. It’s worth noting that the carboxylic acids, overall, tend to have higher binding to plasma protein, so that tells you that the increased half-life of the halogenated compounds isn’t coming about by that mechanism, at least not exclusively. It’s worth noting, though, that just greasing up your compounds is not the answer. Fluorination and chlorination seem to work, but as a look at the broader MMP universe shows, increasing lipophilicity in general does not necessarily help your half-life. Decreasing lipophilicity, on the other hand, has a much better chance of hurting it. It looks, overall, as if the steady-state volume of distribution is more sensitive to modifications in the structure than is clearance.
The paper has a list of groups that seem to help with increasing half-life, more often than not, and ones that seem to hurt. In addition to general halogenation, others in the first beneficial category include primary amines, NH pyrrolidines, piperidines, piperazines, and morpholines, 3,4-dihaloaryls, and (interestingly) 2-methyl-4-pyrimidones. In the likely-to-lower category, on the other hand, are pyridines and pyrimidines attached at the 2 position, N-attached pyrazoles (and other related 5-membered heterocycles that have that core), unsubstituted phenyls (and 4-pyridyls), primary alcohols, and simple cyclopropyls. Read and heed!