Understanding clinical efficacy of drugs (2) - variability in a population

This is a another way of looking at the same plot that was shown in the previous post. A plot representing the chance of a beneficial effect (blue) and a similar plot representing the chance of a detrimental effect (red). The difference here, is that the plots are now a sample of a simulated population with a variation in sensitivity to the drug effect. Likewise, the toxicity profile. In this plot, the therapeutic range is defined as being between an empirical 'average' threshold for the beneficial effect (on the left), and the unacceptable 'average' level of toxicity (on the right). The understanding here, is that potentially you can continue to increase the dose from the left boundary of the therapeutic range, if a stronger drug response is needed. The downside to this is that there will be an increased risk of toxicity. The right boundary to the therapeutic range basically limits the dose increase as any further increase in toxicity risk becomes unacceptable.

As in the previous post, the clinical efficacy plots can be generated from the simulated population. It is shown here with the therapeutic range superimposed. As can be seen, there is an optimal zone where clinical efficacy is maximum. Here is concentration where you can expect maximum benefits with minimum risk of toxicity.

But it should be recognized that this only an expectation of the 'average' response within a population. What should be specifically noted here is the variability and wide scatter of response types within the population sample. For any specific patient within this simulated population, the clinical efficacy is unique, and may look totally unlike the population 'average'

The question is, how do you recognize and deal with this response variability?

The issue of clarithromycin and increased cardiac deaths #3 - Does the usual dose or concentration-response relationship apply clinically?

Very often students are taught models of the dose- or concentration response relationship that do not seem to apply clinically. This is because the traditional model of that relationship was derived from in-vitro experiments, and based on the early experiments using G-protein coupled receptor systems. Invariably these receptor systems were superficially sited on the membranes of effector cells. In the simplest of these models the binding of ligand to the receptor is reversible and competitive.

For clarithromycin however, the drug acts by the inhibition of ribosomal RNA within microbial cells. While this binding to the 50S sub-unit of the ribosome may be reversible, the effects are not. Unless the bacteria carries resistance genes, the bacteria either stop growing or dies. So the concentration-response relationship must incorporate elements of growth, death and resistance. Even so, the model will only work if we know the intra-cellular concentrations of clarithromycin. And we do not know that. For clarithromycin, we are ignorant of not only intra-microbial concentrations of the drug, but we are also not even sure of the concentrations of the drug in the fluids bathing the bacterium. We only see plasma concentrations. Unbound drug concentrations in the plasma vary between individuals, and have an unpredictable relationship with interstitial fluid concentrations.

Taking all these into consideration, it is clear that we will not be able to construct any meaningful concentration response relationship according to the traditional model. Instead, we have a model of drug response that is based on the minimum inhibitory concentrations (MIC). For S. pneumoniae, the clarithromycin sensitivity breakpoint occurs at approximately 0.25 ug/ml. Correspondingly, clinical efficacy will depend on how much of the concentration time profile sits above this concentration. More, correctly though, this refers to the concentration in the interstitial fluid, and we can only guess-timate this from plasma concentrations.

Another complication is the fact that, clarithromycin is not the only active molecule against Gram-positive bacteria. The main 14-OH metabolite has activity, albeit lower. Hence, concentrations of clarithromycin alone will under-estimate the anti-bacterial effect.

There is of course, a flip side of this story that has to do with the cardiac toxicity. Macrolides such as clarithromycin have effects of the cardiac HERG potassium channel, leading an inhibition of the delayed rectifier potassium current during the cardiac action potential. This results in a prolongation of the QT interval of the electrocardiogram, which predisposes to potentially fatal ventricular arrhythmias such as torsades de pointes. The concentration response relationship for this effect is quite different from that discussed above for antibacterial effect. The IC50 for clarithromycin on the HERG channel is approximately 30 ug/ml which is about 100 times higher than the MIC.

This therefore sets a therapeutic window for the use of clarithromycin where the physician would need to ensure that clarithromycin concentrations in the inter-cellular space will be higher than the MIC but not so high as to inhibit the HERG channel.

Think about how we best can do this. See this in the context of the Danish study where there was an excess of 37 cardiac deaths per million doses.

(To be continued)

The issue of clarithromycin and increased cardiac deaths #2 - Pharmacology

Clarithromycin is a macrolide bacteriostatic antimicrobial that came onto the market in 1991. It enjoyed considerable success as an orally administrable macrolide, being relatively lipophilic and having a slightly longer elimination half-life. Came off patent about 10 years ago.

It acts by inhibiting bacterial protein synthesis by blocking the ribosomal RNA. Resistance develops as bacteria acquire various resistance genes, such as the plasmid erm (A) gene that confers an ability to methylate the adenine in the binding site.

Clarithromycin can be administered orally with a bioavailability of about 50%. Its permeability across biological membranes is only due in part to its lipophilicity. A significant part of the process depends on a complex interplay between influx and efflux transporters expressed on various membranes. Consequently intra-cellular, and tissue concentrations do not correlate with circulating unbound drug concentrations. Interestingly, tissue interstitial fluid concentrations are lower than free drug concentrations in plasma, but intra-cellular concentrations are to a variably extent much higher than plasma free concentrations.

The protein binding of clarithromycin is about 60-70%. The Volume of Distribution is about 10 L/kg, which is consistent with significant permeability into tissues. Again this increased permeability results not only from lipophilicity but from the complex interplay of influx and efflux transporters, in this case clearly favouring influx.

Clarithromycin is eliminated by both hepatic metabolism and renal elimination. It is extensively metabolized by CYP3A4 (which it also inhibits), to a principal metabolite 14-(R) hydroxyclarithromycin, which is also pharmacologically (less) active. The pharmacokinetics is not linear, and the elimination half-life increases from 3-5 hours at lower doses, to 5-7 hours at higher doses. Tissue concentrations persist for much longer.

Clarithromycin produces a range of adverse reactions, but the one that concerns us for this discussion is with respect to cardiac death. Like many of the macrolides, clarithromycin has an effect on the myocardial delayed potassium rectifier current, leading a prolongation of the QT interval of the ECG. This prolongation of the QT interval is associated with risk of torsades de pointe and a fatal ventricular arrhythmia.

The usual adult dosage is 250-500 mg 12 hourly for 7-14 days.

Clarithromycin is a drug with very interesting pharmacological properties. Give a thought as to how these properties contribute to variability in the clinical response and the risk-benefit ratio particularly with respect to the problem of cardiac death.

(To be continued)

Optimization? What's that?

Optimization refers to the situation where you can adjust the inputs into a system according to output functions, and eventually arrive at the best solution. In therapeutic terms, it is the process of calibrating the dosage of a drug according to the clinical response so that the best dosage for the patient can be arrived at depending on the therapeutic targets and the patient's individualized response.

This is really no different from the engineering concept of a control system.

When there is an input into a particular process, and no feedback is received about the outcome....this is called an 'open loop control system'. This is probably the least ideal of all control systems. It fundamentally assumes you know everything there is to know and that the decision taken about the initial input is already adequate. This happens in therapeutics when you have fixed dose regiments, and there is no feedback about the outcome. Consider the situation in most cancer chemotherapeutic regiments. Dosage regiments are pretty much determined at the outset, and the only feedback received is if the patient has obvious toxicity which requires cessation of therapy, i.e. switch off the system! This is essentially the pharmacogenomic approach towards 'personalized medicine'.

In other scenarios, there is possibility for the operator to make adjustments to the original input based on feedback received, but this takes place independent of the original model which determined the input. This is called an "open-loop feedback control system'. Most therapeutic scenarios are of this sort, where an initial decision about a starting dosage regiment is made based on starting knowledge and assumptions about the patient. Subsequently minor adjustments to the dosage regiment can be made by the physician depending on feedback received about the patient's clinical drug response. Where there is poor ability to receive feedback about clinical drug response, the system starts to flounder and approximates the simple open-loop control system.

The open-loop feedback control system can operate with varying degrees of sophistication. It can be empirical, where the response to feedback received is relatively intuitive and based on clinical judgement. Or it can be highly deterministic where the response is determined by precise mathematical (PK or PK-PD) models.

A more sophisticated model takes into consideration the uncertainty in the system. This is called a stochastic approach.

The ideal system is that of a closed loop system where the original models determining the original input is linked to, and continually modified by new feedback received. The following diagram represents a closed loop control system with warfarin as an example.

Modified from Applied pharmacokinetics & pharmacodynamics: principles of therapeutic drug monitoring. 1992 Michael E. Burton et al.

At top-right there is a population PK-PD model of warfarin. This represents what is known about how warfarin behaves in the population in which the patient exists. Together with the clinical model at top-left, of what the desired or target INR response is, a decision about the starting dosage regiment can be made.

Subsequently, feedback is regularly received about warfarin pharmacokinetics (bottom right) and INR response (bottom left). These feedback into the original models at top right and left, and continually adjust the model so that decisions are continually taken about how the dosage regiment can be adjusted.

This is optimization... and represents what personalized medicine ought to be.

Can it be done for all drugs? Yes, it can. But it requires that there are good population PK-PD models for the drug, and good biomarkers of response that can be used as feedback. It requires resources and effort.

Above all, it will require that physicians be prepared to put in the extra effort to optimize their therapy according the the patient's real requirements.

Clopidogrel - variability in response

Indian Heart Journal. 2008 Nov-Dec; 60(6): 543-7

The use of clopidogrel presents another interesting challenge with respect to the variability in drug response.

Clopidogrel is a a platelet inhibitor, acting through irreversible binding to the P2Y12 purinergic receptor on the platelet membrane; though it is not clopidogrel itself that binds, but the active metabolite. The PK of clopidogrel itself is quite complex. Upon oral administration about 90% of clopidogrel is removed through the action of circulating and hepatic esterases to inactive metabolites. Only about 10-15% gets activated by CYP2C19 and CYP3A4 to the final metabolite that binds to the P2Y12 receptor. As the receptor inactivation is irreversible, the recovery of function is dependent on fresh platelet regeneration from megakaryocytes.

The way clopidogrel produces its action is therefore fraught with all kinds of problems which clearly contributes to the observed variability in therapeutic response. These are potential sources of variability:

a) high first pass and low active metabolite bioavailability

b) variability of CYP3A4 and CYP2C19 activities due to pharmacogenetics and food/drug interactions

c) irreversible binding to receptor

d) temporal delay in onset, as well as in recovery of platelet function

e] variability in rate of platelet recovery.

This extent of variability really points to a crying need for dosages of clopidogrel to be optimized according to some clinical measure of drug response. Unlike the situation with warfarin however, there isn't a universally accepted way of monitoring plate function. Nevertheless, platelet function test is shaping up to become a standard bedside test for this very reason. A recent review by Williams et al (Thromb Haemost 2010; 103: 29–33) is worth a read.

Drug level testing would clearly not be useful as it is not clopidogrel itself but the metabolite that is active. Furthermore the irreversible binding to the platelet purinergic receptor would not allow concentrations of the active metabolite to be useful in predicting the level of platelet inhibition.

Temporal considerations in understanding efficacy

This is one of the overlook areas for lab-based pharmacologists.

It is easy enough to think of a concentration-response relationship based on static concentrations of active drug in an incubation medium but drug concentrations in clinical practice are almost never static (unless administered as a fixed infusion regiment). Concentrations fluctuate over a dosage interval, and even in a pseudo-steady state situation simulated through multiple dosing, concentrations continue to fluctuate. Even 'average' concentration are seldom stable because patient's compliance vary dose to dose and over days and weeks. It is therefore almost never possible to identify drug effects with any drug concentration. For convenience, we refer to average (randomly obtained) concentrations, peak concentrations, trough concentrations or even areas-under-the-curve (AUC) if multiple sampling has been done over a dosage interval, but in reality we seldom know which concentration measurements best reflect the observed drug effects.

Different aspects of the drug effects may in fact to different types of concentration measurements. Toxicity effects may in fact relate more to peak concentrations while therapeutic effects may related better to trough concentrations. In some situations the converse may even be true. In various other situations drug efficacy may better relate to an index of exposure such as the AUC, or even the amount of time exposed to concentrations above a certain threshold concentration value.

Another aspect of the temporality of drug effects may be related to the involvement of down-stream effects of any drug action. Significant delays in drug effects 'coming on' and 'going off' will make the association between drug effects and concentrations less obvious.

Here's an interesting exercise for you:

We know that there is a sigmoidal log concentration-effect relationship. There is also a log decline of concentrations (assume simplest one compartmental IV model) over time. How would the Effect - Time relationship look like? Email me your answer.

Mechanisms of drug action - we need to rethink our models

When we think about the mechanism of drug action we almost always look to the traditional sigmoidal log dose or log concentration response relationship.

Students are almost always befuddled by the fact that once you go into the clinics, hardly anyone ever refers to this fundamentally important relationship. It seems to be important only at lab benches and do not seem to apply in the clinical context. Part of the reason for this is few drugs exhibit a range of actions that span the entire range of that sigmoidal curve. Many drugs either just operate close to the Emax or are limited in getting close to Emax because of toxicity, or compensatory mechanisms. One other constraint is that the estimates of concentrations are poorly representative of the actual concentrations at the effector site. So often we are reduced to just looking at circulating (fluctuating) plasma concentrations (which are very distant from the effector site) or a very crude estimate of the administered dose.

One other problem is that our ideas of drug response mechanisms are heavily influenced by receptor binding models shaped by earlier studies of G-protein type membrane receptors. The simplistic model assumes easily reversible competitive binding to a receptor with almost immediate effects. More and more drugs nowadays do not operate that way.

Here is a list of the top 50 prescribed drugs (as listed by IMS in 2007).

Of these only a minority can be regarded as operating according to that model of drug action. If you consider that many clinically useful drugs (antiinfectives, anticancer, etc) work through mechanisms more related to irreversible cell kill type models, I think you can readily appreciate the inadequacy of that simplistic traditional concentration-response model of drug action.