Thread by @KRHornberger

🧵By request, here we go with another PK tweetorial, this time on the free drug hypothesis. 1/

References up front. A great one on how to properly use plasma protein binding (PPB) to interpret PD/efficacy data and NOT use PPB as an optimization parameter. This paper is a true classic. I’ll never be able to fully repay the debt I owe to it. 2/
nature.com/articles/nrd32…

And another. This paper has one of the best graphical abstracts in history. 3/
pubs.acs.org/doi/10.1021/jm…

Let's start with this question: what defines an efficacious dose, and the corresponding in vivo exposure, of a drug? As I've said on occasion, pharmacokinetics (PK), pharmacodynamics (PD), and efficacy should fit together to tell a sensible story. 4/

Consider an oral PK curve. There are different kinds of "exposure" that can drive PD and efficacy. Which one is in operation is dependent on the drug, its mechanism of action, the target, and the disease, among other things. 5/

Efficacy can be driven by the time the drug concentration is above a certain minimum efficacious concentration (time over MEC), by total exposure (AUC), or possibly by maximum concentration (Cmax). The latter is especially relevant for tox. 6/

Time over MEC is a common determinant of PD & efficacy and the easiest to grasp conceptually, so let's stay with that one for this discussion. Next question: how does one define the MEC? 7/

Short answer: by in vivo experimentation -- data trumps everything. Ideally it should relate to a measurable in vitro parameter (e.g. IC50, EC50, GI50). If an in vitro assay doesn't help frame in vivo outcomes, it's perhaps not the best SAR-driving assay! 8/

Now the critical complication: in the body, drug compounds bind nonspecifically to all kinds of proteins, lipids, and other biomolecules in all tissues -- not just to the intended target of interest. 9/

In human plasma, the major protein present (~50% of all plasma proteins) is human serum albumin (HSA). It's present in ridiculously high (saturating) concentrations - 3.5-5 g/dL. It binds up drug molecules like crazy -- 90+% bound isn’t atypical. 10/

Intuitively, if your drug is bound to HSA or other proteins, it's NOT available to bind to your intended target. The challenge: in a typical PK experiment, you're measuring total drug concentrations in plasma – bound and unbound together. 11/

That total drug concentration that you measure, by itself, cannot directly relate to the MEC -- because it doesn't reflect the pool of drug that's not bound up by plasma proteins and thus available to "do work" on the desired target. 12/

The free drug hypothesis (FDH) shows us the way out of the woods. I've inscribed it on two stone tablets below for your reference. (Language directly quoted from the reference in post 2.) 13/

Alt text: FDH Part 1: at steady state, the free drug concentration is the same on both sides of any biomembrane. FDH Part 2: free drug concentration affects pharmacological activity. 14/

So now let's look at this in a more chemical way. The setup for the FDH says that there are a series of drug binding equilibria in the body: to plasma proteins, to the therapeutic target, to other proteins and biomolecules in tissue. 15/

Alt text: (in plasma) Plasma Protein Bound Drug <-Kppb-> Free Drug <-Keq-> (in tissue) Free Drug <-KTB-> Tissue Bound Drug or <-Kd-> Intended Target 16/

This is good for chemists because we readily understand chemical equilibria. FDH Part 1 says that Keq = 1, always. Free drug concentrations are the same everywhere in the body. Total drug concentrations, not necessarily. 17/

FDH Part 2 says that when trying to correlate to an in vitro parameter such as IC50, the relevant concentration is the free drug concentration, not the total drug concentration. 18/

So we've established that for our MEC, it's the free drug concentration that matters. Now, how do we determine what the free drug concentration is in the tissue where the intended target is found? 19/

We can easily determine free drug levels in plasma. There are numerous off-the-shelf assays to measure plasma protein binding (PPB): equilibrium dialysis, ultracentrifugation, ultrafiltration, etc. I won't start a fight with anyone over their favorite method. 20/

All that remains is to multiply the obtained unbound fraction (fu) by the total plasma concentrations (Cp) from a PK study, and voila! We have unbound plasma concentrations (Cp,u), which are (via FDH part 1) the same as unbound tissue concentrations. 21/

Not too complicated. But that said, the literature is replete with what can best be described as "hot garbage" where the FDH is completely ignored. Let's now run through some of the biggest logical errors/fallacies. 22/

#1 (biggest): folks show a PK curve and draw a horizontal line corresponding to an IC50 across the graph. "Look, we're above the IC50 for 24 h!". No. Just no. TOTAL drug concentrations are above the IC50 for 24 h. FREE drug concentrations are what matters. 23/

Correction for #1: measure PPB (obtain fu) and then re-plot the PK curve so the Y axis shows unbound drug concentrations (Cp,u). Then and only then is it valid to draw an IC50 line across the graph. 24/

(One nitpicky complication here. There’s also often serum protein present in cell-based assays that are used to determine an IC50 that becomes the foundation of the MEC. Those values themselves often need correction to “zero serum” conditions.) 25/

#2: "I measured [plasma]=1 uM and [tissue]=3 uM. Therefore, drug is accumulating in the tissue and I should drive my PK/PD off of tissue concentrations." Again, a big nope because you measured TOTAL drug concentrations. FREE drug is the same everywhere. 26/

Correction for #2: The logic error stems from assuming PPB = tissue binding, but this would result in Keq =/= 1 and violate the FDH. The correct interpretation: tissue binding is 3x higher than PPB, and free drug is the same everywhere. 27/

#3: “By optimizing my compounds to have lower PPB (higher fu), I will achieve greater free drug levels.” This is one of the hardest ones to wrap your brain around, but it's wrong. The pubs at the beginning of the post spend a lot of time talking about this. 28/

Correction for #3: this may sound like a subtlety, but the correct optimization parameter is free drug concentration (or AUC), not fraction unbound. Math time! Consider the relationship of some basic oral PK parameters: AUC_total = (F * dose) / Cl_total. 29/

However, Cl_total = fu * Cl_int, where Cl_int is the unbound, intrinsic clearance. So, AUC_total = (F * dose) / (fu * Cl_int). Thus AUC_u = fu * AUC_total = dose / Cl_int. The takeaway: since the real goal is to optimize the unbound AUC... 30/

…the unbound AUC is *independent* of PPB (the fu term) and only dependent on Cl_int. Put another way, if you have two completely identical compounds (including Cl_int) that only differ in their PPB, they will have different AUC_total but the same AUC_u. 31/

(Totally punked you by inserting some unexpected algebra there btw. You lived.) 32/

Yet another way to think about this: just as only free drug is available to bind to the target, it’s also the only thing available to bind to metabolic enzymes. Higher PPB can protect from metabolism and lower *total* clearance, but w/ no change in underlying free drug. 33/

Thus, it's very misleading to look at any *total* drug PK parameter because these are not corrected for differences in PPB. Regardless of any change you achieve in fu, if there has been no change in Cl_int, you haven't optimized anything. 34/

Major logical errors out of the way, now let's turn to the exceptions, which are the favorite of med chemists everywhere. Folks have their stories where the FDH "didn't work", many of which I’d probably dispute on closer inspection. 35/

Exception #1: the system is not at equilibrium (steady state). If, for example, your compound has low passive permeability or your target's in a poorly perfused tissue, kinetics can be playing a role. FDH is fundamentally thermodynamic, so you gotta be at equilibrium. 36/

Exception #2: active transport. The FDH assumes drug equilibrates by diffusion only. If e.g. your drug is effluxed by P-gp or moved into tissues by a transporter, these scenarios drive a concentration gradient which will lead to Keq =/= 1. 37/

Exception #3: nonlinearity. If e.g. your drug is an irreversible inhibitor, then it's again a kinetic situation rather than a thermodynamic one. Complex mechanisms of action (degradation?) can also fall into this bucket. 38/

Closing thoughts: the FDH works in many cases. Very much so in my own experience. It's a framework on which to build the edifice of relationships between PK, PD, and efficacy. Assume the FDH is in operation until you have a reason to believe that it isn't. 39/

PPB data is one of the most important pieces of early ADME data that you can gather for your compounds. Potency and PK are equally important, but you can't fit all the puzzle pieces together without also understanding free drug levels. 40/

If the PK/PD/efficacy relationship isn't coming together to tell a clear story, this doesn't mean the FDH is "wrong". But it’s a clue that there's something you don't understand about your system and should trigger more experimentation. /end