This chapter answers parts from Section B(vii) of the 2023 CICM Primary Syllabus, which expects the exam candidate to "explain clinical drug monitoring with regard to peak and trough concentrations, minimum therapeutic concentration and toxicity". It has remained unchanged between editions of the syllabus because it is already biblically perfect and cannot be improved. Question 4 from the first paper of 2008 was the only time this had ever appeared in the exams. However, it is a commonplace topic in the intensive care workplace, and one of the major triggers for severe pharmacist whinge. Ergo, one can expect this question to resurface, in unpredictable and sadistic ways. "Discuss the factors which increase the error of drug level measurements", they might ask. "Explain the rationale for therapuetic drug monitoring" or "describe the desirable characteristics of drug level assay methods". Examples might be asked for. For the 2008 cohort, the SAQ wanted to know about the "factors that are important when interpreting plasma drug concentrations." The pass rate was 0%. In short, this topic is worth knowing about.
Reflective of this is the fact that the official college text on this subject (Birkett, 2009) contains an entire chapter (Ch.13, p. 113) dedicated to Therapeutic Drug Monitoring (here's a link to the 1996 version, when it came out in the Australian Prescriber). The majority of this summary was written using this canonic reference. If one is for whatever reason unwilling to review this important resource, one may instead read Ghiculescu (2008) or Kang & Lee (2009). LITFL also has an excellent list of drugs which are frequently monitored, individual caveats of monitoring them and an excellent point-form summary of the Birkett chapter.
Rationale for the monitoring of drug levels
It is possible to generate a list of valid-sounding reasons to justify the obsessive monitoring of drug levels; however the most important underlying principle of therapeutic drug monitoring is that there is always going to be a stronger relationship between plasma concentration and effect than between dose and effect. Anyway, those valid-sounding reasons:
- The drug may have unpredictable pharmacokinetics.
- The therapeutic index may be narrow (and one wishes to avoid toxicity).
- There may be significant variability in pharmacogenetics, i.e many patients may metabolise or eliminate the drug in some sort of unusual way.
- There may be a well-defined therapeutic goal level which the drug needs to achieve to be effective
- The therapeutic effect of the drug may be subtle, or somehow difficult to monitor (in which case it might be a good idea to monitor drug levels and from these infer that the drug must be doing its job). A classical example of this is lithium, where clinical titration to effect would require serially re-determining how stable the patient's mood is.
- The drug is being administered prophylactically, i.e. rather than titrating to positive effect one is titrating the dose to the absence of some unpleasant outcome
- The drug is being taken .. irresponsibly, and there is a need to monitor drug levels in order to monitor compliance. Classically, one might make this argument for monitoring of cocaine and cannabinoid levels.
Arguments against therapeutic drug monitoring
Some experts view the practice as flawed, and have offered the following arguments:
- There is little evidence that targeting plasma concentrations improve the therapeutic outcome (Vozeh, 1987)
- It is often difficult or impossible to define appropriate therapeutic ranges; they are usually derived from small groups and point measurements of questionable reliability
- The greatest benefit from monitoring is to specialised or at-risk populations, but therapeutic ranges are frequently designed without these groups in mind
- Obsession with monitoring may lead people to treat drug levels rather than the patient (Sjoqvist,, 1985), i.e. the clinical monitoring of effects may give way to remote monitoring of plasma concentrations
- The availability of monitoring may give rise to pointless routine monitoring, which would be irresponsible and costly.
Common scenarios for drug level monitoring
One may send drug levels in the following scenarios
- At the beginning of treatment, after steady state has been achieved (3-5 half-lives)
- After a dose adjustment has been made
- When clearance physiology is altered mid-treatment (i.e. the patient develps renal failure)
- When a drug interaction is suspected which is expected to alter clearance
- To confirm compliance or overdose
- When the treatment seems to be ineffective (i.e. you suspect that the levels might be subtherapeutic)
One needs to include the following information in the order (from Birkett, 1996):
- Time of sample in relation to last dose
- Duration of treatment with the current dose
- Dosing schedule
- Age, gender
- Other drug therapy
- Relevant disease states
- Reason for request e.g. lack of effect, routine monitoring, suspected toxicity
Factors which are relevant to the interpretation of drug concentration measurements
This section attempts to address Question 4 from the first paper of 2008, making use of the fact that the college answer spells out some of their expectations. The college decided to classify the factors into three major categories. In accordance with their suggestions, wherever possible examples of drugs are use to illustrate the point.
- Pharmacokinetic factors:
- Measurement assay: the measurement of the drug level may be
- Protein binding: the total plasma drug levels may not be reflective of the drug activity, as in the case of phenytoin and hypoalbuminaemia (i.e. with a low total level there is still enough free unbound drug around to have a significant therapeutic effect). Even if you manage to monitor free drug levels, the situation is made more complicated by the fact that the drugs bind variably depending on pH and competition for binding sites (which may not be well represented in the cooled and centrifuged blood sample)
- Relationship of plasma concentration to effect site concentration: if the drug penetrates variably and incompletely, it is difficult to relate the robust serum levels to effect site levels. One example of this is vancomycin being administered to treat ventriculitis, where high plasma levels are no guarantee of bactericidal CSF levels (Moelering et al, 1981)
- Factors which influence plasma concentration: these include the volume of distribution, tissue binding, sites of metabolism, rate of clearance, and organ-independent biotransformation (which my carry on inside the blood sample tube and then inside the measurment apparatus, leaving one to measure the metabolic breakdown products instead of the parent drug).
- Timing of the sampling in relation to the dosing: most drug levels are trough levels, which makes it important to collect them before the next dose rather than at some arbitrary point in the dosing regimen (particularly for drugs with short half-lives). The trough is generally viewed as the least variable point in the dosing regimen because the drug concentration is changing the least over time - in contrast, after the peak the concentration would be changing (falling) rapidly.
- Steady state concentration: most intermittently dosed drugs will achieve the steady state after 3-5 half-lives, and any drug levels collected before this point will need to be interpreted accordingly.
- Pharmacodynamic factors:
- Relationship of the plasma concentration to clinical drug effect: in some cases, there is no such relationship, making it pointless to measure drug level (eg. in the case of levitiracetam).
- Active metabolites: the presence of chemically distinct daughter molecules which have their own therapeutic (or adverse) effect makes it difficult to relate measured drug levels to the clinical effect.
- Individual variability in drug response: there may be groups in the population who have a satisfactory response to therapy at plasma levels which are below the expected "low" concentration threshold, and there may be those who experience toxicity below the "high" threshold.
- Clinical and pragmatic factors:
- Simplicity of the assay: there is often no point in measuring drug levels if the measurement requires such equipment and expertise as to take weeks.
- Convenience of sample collection, eg. whether the blood sample needs to be handled in some carefully ritualised manner before being transferred to the laboratory (on ice, on dry ice, in liquid helium, etc)
- Accuracy of the measurements: the assay may be accurate, but it may be interfered with by other factors, rendering it uninterpretable
- Cost and benefit of the assay, i.e. it may be inefficient to measure drug levels if they be expensive and will not alter the design of the dosing regimen and prevent the administration of the expensive drug, or if it cheaper and equally effective to titrate the drug to some sort of clinical effect.
- Therapeutic range misalignment when the population in whom the therapeutic ranges were first defined may be completely mismatched to the scenario in which the drug is being monitored - i.e. it may be inappropriate to relate adult therapeutic ranges to the drug levels measured in the neonate
- Relationship of the therapeutic range to the clinical situation, i.e. where the therapeutic range which is quoted does not represent the clinical scenario - for instance, the therapeutic plasma levels quoted for an antibiotic may be well above or well below the MIC for the organism being targeted