David Burger
With drug concentrations in paediatrics being so variable, and virological response being so closely linked to these levels, there are enormous practical benefits to using TDM. Every HIV-positive child in Holland receives these tests as part of routine care, including a full PK curve when they begin treatment.
TDM is available for routine adult care in the Netherlands and we believe that its use in paediatric care is essential. This presentation will focus on the practical benefits that we have been able to provide through this widespread use of TDM.
While there remain concerns about assay validation as laboratories broaden the availability of this technology, the relationship between plasma levels of protease inhibitors and virological effect is certainly apparent. The following studies also show that achieving optimal drug levels cannot be taken for granted for some of the key protease inhibitors currently used in clinical practice today.
The traditional view of TDM is that pharmacologists and clinicians look at the drug and its characteristics and decide that the drug needs therapeutic drug monitoring. A more modern view of TDM, proposed in a paper two years ago, is that it is the patient rather than the drug that determines TDM, so you need to look at the patient using that drug in a specific indication.
In practice TDM is inappropriate for RTIs due to the difficulty of measuring intracellular phosphorylation, but for the PIs and the NNRTIs the situation is different. These drugs are active in the plasma compartment and can be measured by HPLC or by LCMS. These assays are generally 'Home Brew' methods developed by laboratories in different countries, and so validation and quality control has become an important issue.
Therefore, an international quality control programme was established to standardise and validate the results from these laboratories. This involves an ongoing 6-monthly evaluation of blinded samples and the results from the first assessment are shown in Table 1. It is clearly important that we are sure that these labs are measuring drug levels accurately. Results from a second round of this programme, including nineteen laboratories, will be reported at the Retrovirus Conference in February 2001.
Drug levels and pharmacological response
Three years ago we looked at a group of 65 adults using indinavir in the original 800mg TID regimen. We found three risk factors that were independently related to virological failure: a very low indinavir drug level in plasma, high baseline viral load when starting treatment and whether previous treatment included a PI. [1]
Table 2 shows the results from subdividing this group by risk factors in order to determine the influence of the drug levels. For people most likely to respond to treatment - those who were PI-na•ve with a baseline viral load <100,000 - drug levels made little difference to the risk of non-response (at around 10%). In a similar way, drug levels made little difference for the people who were least likely to respond. All of the patients who were PI-experienced and had a high baseline viral load failed to respond, irrespective of drug levels.
In this study, the importance of adequate drug levels became significant for people in between these extremes. With patients who were PI-na•ve but had a high viral load, adequate drug levels made a difference in risk of non-response of 9% compared to 56% with suboptimal levels. Optimal drug concentration in people with PI-experience and low viral load levels lead to a 20% rather than 50% risk of failure. With treatment success depending on other variable factors, accurately identifying patient groups most likely to benefit is therefore important.
Some of the same relationships were shown in a paediatric study conducted in Rotterdam, Utrecht and Amsterdam. This study with indinavir showed a clear relationship between virological response (defined as viral load <500 copies/ml at 6 months) and AUC. There was a 55% response rate in children with AUC <20mg/L/h. The response rate was 100% in children with higher AUC levels >20 mg/L/h. This lead us to set a minimum target level of 20 mg/L/h in this population.
| Table 1. QA and QC assay programme |
Table 2. Risk factors related to PK-PD |
Table 3. Variability in IDV PK in children |
 |
 |
 |
Click on slide for a larger view |
Click on slide for a larger view |
Click on slide for a larger view |
Drug levels and side effects
The same study highlighted the relationship between drug levels and risk of side effects. The six children who had renal side effects in this study had a much higher AUC compared to the children without (mean 40.6 mg/L/h vs. 21.6 mg/L/h). Similar results associated risk of renal side effect with both AUC and Cmax were shown by Gatti and colleagues in a group of 11 children (aged 9-13.6 years) given indinavir at 500mg/m2 TID. AUC and Cmax levels in the children with and without renal side effects were 53.6 vs. 30.2 (AUC) and 15.3 vs. 9.8 (Cmax) respectively. [2]
Nelfinavir can produce similar PK difficulties, and this was shown in a recent presentation from the ATHENA study. After a median follow up of eight months we had about a 30% virological failure in a group of 48 treatment-na•ve adults using nelfinavir 1250 mg BID. If a patient has a relative concentration below 0.90 (ie 90% of the average data) we saw 50% failure in that patient population, whereas there was only a 17% failure in patients with higher drug levels. The relative risk in this group was 3.0. [3]
Using this threshold of 0.9, we also analysed the sensitivity and specificity necessary to predict whether a patient will fail therapy. We determined this to be 64% sensitivity and 75% specificity, which may look low for a diagnostic test. However virological failure is multi-factorial and there are other important factors such as adherence to consider.
Table 3 shows the relationship of dose to concentration in children using indinavir and interpatient variability in pharmacokinetics with PIs. [4] Of the nineteen patients who received the same dose, in this case 100 mg/kg metabolic weight, only 50% came within the target range of AUC - with 40% below and 5% above. This variability in indinavir pharmacokinetics can be partly explained by the age of the children. Drug clearance reduces with age. It therefore appears that you have to increase the dose even more in the youngest children even when correcting for metabolic weight.
What about sampling times?
There are several important aspects for TDM that I would like to raise. Although it is reasonable to look at trough levels if you are looking at virological efficacy, it is important to realise that the trough level is not always the C min. After taking medication the drug level continues to decrease for the next 1-2 hours. There may also be some variation between trough levels in the evening and morning.
The lower quantitation range of assays can also be less accurate than in the middle or the highest range - so that is a problem in doing only trough levels. There are also practical issues with the patients coming to the clinic for a pre-dose level, which makes accurate recording of timing of the previous dose so important. Only if it is twelve hours (+/- I hour) for a BID regimen can you use it for a trough, or Cmin.
To check toxicity levels in children you will need to check the Cmax. The time for sample for this will vary from 1-4 hours depending on the Tmax for individual drugs. It may be rational to do a pre-dose concentration but what is actually the Cmax?
In both cases, the best thing to do would be a full PK curve to determine the AUC and all the other parameters. Although this is not possible in every situation, we do this in Holland for every child who starts treatment.
Case study 1
The first example is of three children that started treatment with a triple drug combination including indinavir. [5] They all had a good virological response but after one year they relapsed. Their indinavir trough level at that time was <0.1 mg/L. We clearly thought that the low drug level was the reason for the relapse and so we intensified the treatment in these three children by adding a low dose of ritonavir to boost the pharmacokinetics.
The effect of ritonavir in this situation was quite variable with indinavir drug levels increasing to 0.24, 0.7 and1.9mg/L, but this addition of ritonavir was sufficient for viral load to become undetectable in all three children. This is clearly a way of effective intervention when you see low drug levels and when you see virological failure.
Case study 2
The second example is of adults who suffered from urological symptoms when they were using indinavir 800 mg TID. All these patients were found to have indinavir drug levels that were at least double the expected level. We felt confident in decreasing the dose to
Figure 4. TDM intervention to increase
nelfinavir dose in 18 patients with low plasma nelfinavir levels |
 |
| Click on slide for a larger view |
600 TID. We reduced the indinavir concentration ratio to around 1.0 (0.63 - 1.37) - which is considered normal. Subsequently the patients all remained free of symptoms and maintained undetectable levels of viral load without future rebound. [6]
Case study 3
The last example is from eighteen patients using nelfinavir dosed at 1250mg BID who showed low nelfinavir plasma levels. We increased the dose to 1500mg BID - adding one extra tablet in each dose, but we saw a very variable response - see Figure 4.
About 50% of the patients had a big increase in their drug level, some remained stable and some inexplicably even went down. Although this intervention did not help all patients, at least half benefited from better levels.
References
Dr David Burger is currently working as a hospital pharmacist responsible for coordination of all clinical drug trials conducted at the University Medical Centre Nijmegen and as an AIDS research pharmacist, leading a research group focusing on clinical pharmacology of antiretroviral agents. He is co-author of more than 100 publications in this field. His topics of interest are drug-drug interactions, paediatric pharmacology, TDM and adherence.
 |
 |
 |
 |