Saye Khoo
In spite of compelling evidence for the benefits of TDM, and a world-class site at the University of Liverpool, only 2.5% of patients using antiretrovirals in the UK actually get TDM at the moment (including a very tiny number of children). Which is very small indeed compared to Holland or France.
Many doctors who believe that there are definite clinical benefits from TDM - and the evidence is compelling but falls short of being conclusive - often fail to use it as a regular clinical tool. Even if you accept that there are benefits there can be practical problems with choice of sampling strategy and how you define your therapeutic index and these vary slightly from lab to lab as well.
There are at least five different strategies that are used:
Most people regard measuring AUC as the gold standard. Indeed some of the best data relating drug levels to actual virological response are based on AUCs. The problem with the AUC is that it is quite laborious, difficult and expensive. So a lot of people use trough levels.
In Liverpool we have found that while the trough level gives you an indication of the AUC it is not as reliable as we would like, especially when there can be a lot of variability in the trough levels within the same patient on different days.
This variability in trough results is often due to an inaccurate recording of when they last took their medication. Coming to clinic itself is very disruptive for taking a trough. You have got to wait for the doctor, wait for the nurse, wait for someone to find a vein, and you may be twelve to fifteen hours since your previous dose. This is particularly true for children (who go to bed early, and where dosing may be more erratic), or people taking TID regimens (where trough levels at 6 or 7 o'clock in the morning does not easily lend itself to accurate sampling for TDM). Nevertheless, trough levels are important especially in relation to the IC50. We therefore use a trough and the peak 1-2 hours later to reduce some of the variability. This doesn't get round the problem that troughs can be difficult. They require training of both patients and clinic staff to take the bloods at the right time. Current clinics in the UK operating from 9am-5pm do not lend themselves to accurate TDM.
Various programs have been developed to allow full mathematical modelling. This can be performed from a sample taken at a random time point, with knowledge of the patient's exact dosing times for the past couple of days. From this one sample you can derive the AUC, the Cmax and Cmin, and the clearance for that patient. Although this approach is in its infancy this is probably the way forward. The difficulty of this approach is that you need different models for each combination of drugs, even when the same drugs are being used. The addition of ritonavir to indinavir at different doses (ie 100/800, 200/800, 400/400 etc) requires separate models because different doses of ritonavir have a different effect on indinavir clearance.
In the paediatric setting there is an additional problem because of different metabolic rates at different ages. Neonates may not clear PIs well but metabolic function is increased during infancy (to levels greater than that of adults) and this increased clearance may result in infants being under-dosed. Depending on the drug involved, metabolic function may approach that of adults by around 2 to 6 years. A large amount of data are required to build these different models, but much of this information is already available from different labs and industry, and we need to get together to pool this data.
| Figure 1. |
Figure 2. |
Figure 3. |
 |
 |
 |
| Click on slide for a larger view | Click on slide for a larger view | Click on slide for a larger view |
In the UK, TDM is largely done through the website at the University of Liverpool. We offer a service for PIs and NNRTIs. Although the uptake has not been as great as in Holland or France it has been steadily increasing over the last 18 months. At the end of 1999 we were doing roughly twenty requests per month. This rose throughout 2000, and may have in part been due to funding for TDM from industry (Roche with saquinavir and nelfinavir, and Merck with indinavir). Indications for TDM include altered liver function, PI plus NNRTIs, potential drug interaction and all paediatric patients. We currently receive over 80 requests per month and the majority of PIs analysed are saquinavir, nelfinavir and indinavir.
The most common reasons for requesting TDM are shown in Figure 3, and include use of a non-standard dose (40%). This is a very broad kind of definition of people who maybe had TDM done before and had insufficient levels which we have increased. It includes people who are on two PIs, or PI plus NNRTI, or anti-tuberculosis chemotherapy. PK enhancement with ritonavir accounts for 17% of the requests.
The next most common reason for seeking TDM is suspected treatment failure at 24%. You could argue that this is shutting the stable door after the horse has bolted, but certainly if you do it early on enough, or have the results of resistance testing, TDM may offer the option of treatment intensification (e.g. by increasing sub-therapeutic doses, or boosting with ritonavir). Understanding why a treatment failed is also essential if you are to prescribe an effective subsequent regimen.
The other reasons are fairly standard - suspected interactions or toxicity, and clinical indication, for example for people with hepatitis C co-infection or liver impairment. I had a patient who had a liver transplant about two months ago and prior to his transplant he was clearly not clearing efavirenz at all. We used TDM to find his optimal efavirenz dose which turned out to be as low as 200mg twice a week. This is one situation where TDM was essential for clinical management.
You will see that only a small amount of our work is with children. This perhaps largely reflects the relatively small cohort of children in the UK, but also the general uptake of TDM. Only 2.5% of the 12,000 patients in the UK who are on antiretroviral therapy actually get TDM at the moment, and compared to France or Holland this is very modest indeed.
Figure 4 shows pooled data from Liverpool suggesting that a substantial proportion of patients have sub-therapeutic trough levels of PIs. It is also worth pointing out our cut-off for both saquinavir and nelfinavir are much lower than in many European labs. Some people considered that ritonavir-boosted combinations would obviate the need for TDM. We found that when used with saquinavir or indinavir, ritonavir definitely elevated the plasma levels of those drugs, but that there were still patients who failed therapy.
Figure 6 shows data on sub-therapeutic levels from three different cohorts. The left-hand bars relate to the Athena Dutch cohort which includes 300 - 600 patients using concentration ratios and was presented by David Burger at Noordwijk in March 2000. The centre bars are from a study that Liverpool University has recently been involved with in Manchester. These are data from the first 50 of about 180 patients taken from random time points in an unselected clinic cohort, and using mathematical modelling in collaboration with Virco to derive the estimated trough levels. The right-hand bars are taken from trough samples sent for TDM to Liverpool. The Liverpool cohort may represent a selected group of patients who are most at risk of failure, but this bias was not present in ATHENA or the Manchester cohort.
Each of these three different methods in different cohorts has produced roughly comparable results. The high frequency of low ritonavir levels largely relate to Ôbaby' doses which haven't been excluded yet from the analysis. Around 30% of indinavir samples are sub-therapeutic (even with ritonavir), and 12-24% of nelfinavir, 30-40% of saquinavir, and 10% of nevirapine samples also fell below therapeutic levels. These kind of data are very worrying, and somewhat depressing given the high numbers of people who are working through an already limited number of treatment options each year. There is clearly a substantial number of patients not getting enough drug in their system to suppress even wild type virus let alone their own isolates.
Finally, although we give out advice with all drug levels we have no idea whether doctors chose to follow this - and that is clearly important information to know.
| Figure 4. Sub-therapeutic troughs | Fig 5. Compared results from 3 TDM cohorts | Figure 6. Results and recommendations |
 |
 |
|
| Click on slide for a larger view | Click on slide for a larger view | Click on slide for a larger view |
Inividualised TDM is recommended for all children currently using PI or NNRTI-based antiretroviral combintions.
For further information on drug level monitoring please contact Sara Gibbons, Saye Khoo or David Back at Liverpool University. All costs for patients using saquinaivr, nelfinavir or indinavir combinations are covered by the subsidised programmes supported by Roche and Merck.
A copy of the TDM request form for this service is included on page 40 of this report.
Dr Saye Khoo is Senior Lecturer in Pharmacology in the University of Liverpool and an Infectious Diseases clinician. Research interests focus on the role of pharmacology in HIV treatment failure. He has recently joined the PK team of the PENTA trials group.
 |
 |
 |
 |