Case study: Adopting an adaptive approach to help HIV patients on antiretroviral therapy (cART)
The NIHR research infrastructure supports the delivery of complex and innovative clinical trials.
The adaptive dose-ranging model
The TAILoR trial was a non-commercial trial funded by the Efficacy and Mechanism Evaluation (EME) Programme - an MRC and NIHR partnership. It was a phase IIb (2b) multi-centre, randomised, open labelled, dose ranging trial which used an adaptive design. This approach allows multiple doses of a drug to be tested simultaneously and is ideal for trials which aim to find the optimum dose of a treatment, which was one of the aims of the TAILoR trial.
Patients with HIV treated by combination antiretroviral therapy (cART) are at risk of developing side effects such as reduced response to insulin (this is called insulin resistance), abnormal body fat distribution (HIV lipodystrophy) and high cholesterol levels. All of these can lead to an increased risk of diabetes and heart attacks with insulin resistance being a key underlying predisposing factor.
Telmisartan is a drug that is commonly used for the treatment of high blood pressure. However, it has also been shown to reduce insulin resistance and improve cardiovascular health in patients that do not have HIV. The aim of the TAILoR trial was to find out:
- Can telmisartan help reduce insulin resistance in HIV-positive patients?
- What is the optimum dose required to counteract the side effects caused by anti-HIV drugs?
The TAILoR trial, now complete, recruited patients between February 2013 and July 2015. Professor Munir Piromohamed, Chief Investigator, explains how the TAILoR trial was conducted:
“TAILoR used a novel adaptive trial design to compare three different doses of telmisartan with a control group (participants who did not take telmisartan) to determine the effect on insulin resistance over a period of 48 weeks. We recruited 377 HIV-positive patients from 19 different specialist centres across the UK. Patients were randomised to one of the four arms:
- Group A – No treatment
- Group B – 20mg Telmisartan daily
- Group C – 40mg Telmisartan daily
- Group D – 80mg Telmisartan daily
“Patients attended a baseline visit followed by a visit at 12 weeks, 24 weeks and 48 weeks. Those patients allocated to group C attended one additional visit and those patients allocated to Group D attended two additional visits in order to increase the dose of the medicine. Blood and urine samples were taken during each visit to monitor blood sugar and insulin levels and other indicators of cardiovascular and kidney health.”
“The results were analysed half way through the study. The study steering group worked with statisticians to analyse the data and make a decision on if or how the trial would continue. It was clear that 20mg and 40mg showed no effect and these doses were stopped. They decided to continue the 80mg dose and all participants were then randomised to either 80mg telmisartan or the non-treatment arm for the remainder of the study.”
“At this point the statistics also concluded that we needed more patients to achieve the required statistical power and the recruitment target increased by seven patients to 377. In other words, the study adapted and then continued.”
The adaptive approach was important for a number of reasons. Professor Pirmohamed continues:
When we are looking at a dose/response relationship the adaptive model offers a number of advantages.
“When we are looking at a dose/response relationship the adaptive model offers a number of advantages. It allows us to stop ineffectual doses and focus more where a dose appears to be having an effect. It also allows us to look at more than one dose at a time rather than having to do one trial after another or set up several parallel trials, which inevitably uses more time and resources. Where a drug is being repurposed for another indication, as was the case for telmisartan in TAILoR, it is wrong to assume that the dose-response profile will be the same as in the primary indication of the drug.”
Trial delivery and NIHR impact:
Back in 2013 trials using an adaptive design were less commonplace than they are now. Despite the new approach, Professor Pirmohamed says the study was well received and delivered efficiently by the clinical teams across the 19 sites:
“The model of the trial itself was not a challenge. Yes, there is a lot of complicated mathematics behind the trial design, but when I talked to the clinical teams they all fully understood the need for the different doses and the adaptive model. Paracelsus is credited with explaining how ‘a dose can either be a remedy or a poison’, it is very important to identify the correct dose of any therapy. Everyone was keen to support it, after that the delivery was much like any other trial.
“There were some delays at some sites but that was because of the way that NHS Excess Treatment Costs (ETCs) used to be distributed for academic studies back then, this was in no way related to the design of the trial. The NIHR Clinical Research Network was very supportive in terms of working through the ETCs issues at each site to get the study underway. The Network also assisted the study by allocating research nurse support at all 18 English sites (one was in Scotland) to help with patient identification and recruitment, and monitoring and sample taking during clinical visits.”
“The only factor we had to consider relating to the model was around patient safety. Telmisartan is used to control blood pressure so we could not start patients on a high dose immediately, just in case their blood pressure dropped. Instead we had to escalate the dose slowly, which meant slightly more frequent clinic visits for some patients initially. This didn’t seem to be a problem, in fact, the main barrier to recruitment was actually that we had to ask patients to fast (not eat) for 12 hours before coming to the clinic!”
The results of the TAILoR study were published in the journal Clinical Infectious Diseases, as an open access article, and make for interesting reading. Prior to the TAILoR study, telmisartan was commonly thought to help reduce insulin resistance in HIV patients receiving cART. However this hypothesis was anecdotal and there was no supporting clinical evidence available, hence the need for the TAILoR trial. Professor Pirmohamed summarises the results:
“Our primary outcome measure in the HIV patients was a reduction in HOMA-IR, an index of insulin resistance. The trial showed that after 48 weeks of taking 80mg telmisartan there was no difference between the telmisartan arm of the trial and the control arm in HOMA-IR. However, the trial also looked at some secondary outcomes and the findings suggest that telmisartan may have an effect on some aspects of insulin resistance in a select group of patients, but this will require further investigation.
“As a result of the TAILoR trial, we have access to a rich biobank of samples which we are now studying and looking for biomarkers which might be able to tell us in which patient groups telmisartan will help to reduce insulin resistance. This will require further studies in the future adopting a stratified approach.”
As a result of the TAILoR trial, we have access to a rich biobank of samples which we are now studying and looking for biomarkers which might be able to tell us in which patient groups telmisartan will help to reduce insulin resistance.
More information on the TAILoR trial is available on the NIHR Journals Library website.
More about adaptive trial designs
There are a number of different approaches to delivering a trial which are categorised as ‘adaptive’ designs. These include platform, umbrella, basket, and MAMS (Multi-Arm Multi-Stage) designs, among others. All of these approaches have the same underlying principle of flexibility built into the study protocol which allows the trial to ‘adapt’ as the science evolves. This is achieved by analysing the data, usually at set points, as the trial progresses. The findings are used to modify the course of the trial according to pre-specified rules and parameters set out in the study protocol.
Example modifications may include the drug or combination of drugs being used, the dosage, the patient sample size, or the patient selection criteria. In some cases the trial becomes an ongoing process, adding new arms for new therapies and/or new patient groups (such as patients with different biomarkers) as more data is acquired and analysed. Similarly, arms of the trial that fail to yield positive results can be closed down. In this way, adaptive trials are seamless - they do not need to stop and start a new trial for each modification.
The aim of this approach is to more quickly identify where a drug or device has a therapeutic effect and zoom in on the patient populations who are experiencing positive outcomes. Trials with an adaptive design are considered more efficient than trials with a fixed design. They have the potential to save both time and resources required for study set-up, and they often require fewer patients to be screened. In the long term, this approach could also help to speed up the overall drug development process and bring new treatments to patients faster.
- Watch more related videos our Complex and Innovative Trial Design and Delivery playlist on our YouTube channel: NIHRtv.
- Visit our Delivering Complex and Innovative Research web page to learn more about NIHR support available for complex and innovative trials.