The way forward for clinical research

pharmafile | October 23, 2003 | Feature | Research and Development |  NICE, clinical research, healthcare 

Clinical investigators in drug development have made enormous contributions to the advancement of healthcare over the last 25 years, and I pay tribute to that, but I think we need to move on.

England's Chief Medical Officer Liam Donaldson is very fond of quoting his Irish grandmother who wrote to him after he did particularly badly at school: 'If you always do what you did, you will always get what you always got.'

Increasingly, health providers, whether they are HMOs or centralised structures like the NHS and individual prescribers, want to know three things about new medicines and other health technologies:

• does it work?
• if so, is it as effective, better or worse than current standards practice?
• is it good value for money?

I believe it is essential that together we learn and develop ways of answering these questions.

Does it work?

This question concerns a drug's efficacy, its safety in relation to its efficacy and its clinical effectiveness. We have a well-trodden pathway in studying drug efficacy, and that, of course, is randomised controlled trials (RCTs) starting with phase I, moving onto phase II and III. Clinical investigators can design studies like this in their sleep, but I think we have to very realistic about the problems surrounding current practice.

The first problem is that these trials are placebo-controlled. This is an elegant method of demonstrating the principle of drug action, but of course it doesn't tell us how good a new product is in relation to others. And that's what health providers need to know – not just whether a new antidepressant works or not, but how well it compares with others. Comparator trials are becoming more common but the comparator used often varies across the EU and North America too.

Secondly, surrogate markers are another useful but problematic area. Surrogate markers are very important in phase I and phase II but I believe we sometimes fail to adequately develop markers that are predictive of either intermediate outcomes or ultimate outcomes. We have to work harder at developing more robust surrogate markers, and even have them defined before we have an active drug.

We have to be very careful about intermediate endpoints as they vary widely across Europe. For example, if one is looking at hospital admissions as a surrogate for mortality or potential of mortality, one also has to accept that admission rates across the EU are very diverse because of the differences in the patterns in healthcare.

The fourth problem in this area is the question of quality of life. There is an increasing need for trials to demonstrate changes in quality of life and I believe we currently suffer from bad quality of life instruments. Most of them have been developed by people like you and me, when in reality we ought to take much more cognisance of the views of patients – many of our quality of life instruments do not capture the important and integral aspects of patient aspiration and what patients want. A very moving account was recently described by Rethink, the schizophrenia association, who asked both psychiatrists and patients what they would like from drug therapy.

Number one on the psychiatrists' list was to get rid of the voices. But with the patients it was much more subtle – the ability to sustain family relationships and to hold down a job. And when it came to the voices, all they wanted was for the voices to be kind. I think we do have to realise that so many of our instruments do not capture what are necessarily important things for patients.

Another problem for quality of life measurements is that many of them are non-linear, or measures that may not necessarily map very well the utility. Utilities are an important and critical part of economic evaluation and so our quality of life measures need to match with utilities so that we can compare the health gains in quality of life between technologies.

My final worry is patient resistance. I acknowledge your competence but I worry that patients will not necessarily continue to participate in RCTs, particularly placebo-controlled trials for diseases that have a nasty progression. For example, if someone developed a treatment to stop the progression of Alzheimers, I am not sure I would be very keen on taking part in a placebo-controlled trial where I had a 50% chance of receiving a placebo. And if I am not keen on that, then I think other people won't be either.

I think we need to look very carefully at how we construct clinical trials in order to tackle these sorts of problems. It came to the particular forefront in some of the AIDS trials, and a lot of work has now been done around how to encompass patient preferences for treatment in conventional clinical trials.

So what might we do differently? Firstly, it seems to me we need to merge the phases into a seamless development programme. Current practice often sees trials completing phase II only for the data to sit around for a year or two.

Comparative efficacy

The problem of providing data on comparative efficacy is another major challenge for all of us. Health providers need the information, and if they don't have it, they are likely to distance themselves from taking up new drugs.

The potential of pharmacogenetics or other biological techniques to allocate treatments to patients will be helpful but we shouldn't imagine it is going to solve all our problems. We need to consider much more interactive allocation of regimens where the allocation of the next patient in the trial depends on the outcome of the previous one. That's not possible for all types of conditions – for instance, the progression of Alzheimer's – but in many cases it is. And interactive allocation regimes and consequential analysis is something we are going to have to take account of much more in the future.

I believe we also have to look again at historical controlled trials [trials in which there is no placebo control] There are objections to historical controls but in many cases they appear to be reasonably accurate.

Historical controls can be very useful, particularly where one is investigating otherwise untreatable conditions where there is a biologically plausible basis for the treatment, and where the outcome untreated is homogenous and either very disabling or fatal. When I was at the CSM, for example, we agreed to licence valganciclovir [Roche's Valcyte] for CMV retinitis in AIDS patients. That was done entirely on the basis of historical controls.

At the time the FDA refused to accept historical controls and insisted on a placebo-controlled trial, which two years later came out with the same answer. So I think historical control studies are something we will to have to consider again.

The problem is we start collecting historical control trial data too late – we need to start to collect them now. For example, work is now being done on variant CJD. The investigators are collecting very careful historical controls because of the virtual impossibility of randomised controlled trials.

Safety/efficacy trade-off

The toxicity of new medicines is really looked at in RCT adverse event rates and the other is biotester marker.

Adverse event rates were an important advance in identifying safety problems and have made a major impact on how we assess the safety of a clinical trial. However, I am less sanguine about biochemical markers. The liver function testing that is sometimes used is probably almost useless – and a gross misuse of a test that was never designed for this purpose.

So I strongly believe if we are going to pursue this further we need to develop better tests of overt organ damage to our clinical trials. But I think we have a deeper problem. When one comes to putting in a marketing authorisation application, or to making a licensing decision, there is a trade-off between efficacy and safety, currently calculated first by the manufacturers before a marketing application, and then by regulatory authorities in considering the evidence.

I believe we have been to paternalistic on this issue. I think the notion that someone like me or you  who is not suffering from the disease – can really make a balanced decision is probably wrong. For the future we have to be asking patients what they think. This is not easy and it is particularly difficult across multi-cultural, multi-lingual organisation like the EU, but I think ultimately this is the direction we will have to take.

Surveys of patients with severe rheumatoid arthritis suggest those interviewed were willing to take risks far in excess of those that either manufacturers or regulators are generally prepared to accept. I think this should lead us to question whether we are denying patients drugs they would like to take, despite the fact that the drugs toxicity can be substantial.

Clinical effectiveness

By clinical effectiveness I mean performance under normal conditions. The problem of our RCTs, of course, is that they are on homogenous populations, small numbers, limited duration of treatment and the prescribers are usually, quite rightly, experts in that particular area. In contrast, when a drug reaches the market it is used in large heterogeneous populations for extended durations by inexpert prescribers. We must do more to facilitate this transition between clinical trials and use in the normal population – we are still not serious enough about how generalisable results from clinical trials really are. The case series can actually be very important in showing that that generalisability is occurring, and if it's not, why not.

Of course, drug development costs are a major factor in these considerations. Boston Consulting estimates that the cost of developing a new drug is around $800 million, but other estimates put it as low $250 million. Whatever the price is, it is nevertheless unsustainable, and if regulatory requirements continue to increase the number of new drugs available to patients will diminish. Quite a large proportion of this total is spent in clinical research and I believe we will have to construct ways of reducing the cost of drug development.

Economic evaluation of medicines

Health providers want to maximise the health gain delivered from the expenditure and there are broadly speaking three approaches to economically evaluating medicines. Cost minimisation is relatively simple in theory, but actually more complicated in practice. It is simply a matter of picking the least expensive of two equivalent products – which is fine but, of course, of demonstrating equivalence is not as easy as one might imagine.

The second approach is true cost effectiveness – that is to say the cost per actual unit of health gained. It may be the cost per millimole cholesterol lowered or it may be the cost per life year gained.

Cost effectiveness is relatively straightforward to assess but there are problems in extrapolating between one form of treatment and another. Because if you want to compare – as health providers do – whether to invest in treatment X or treatment Y for totally separate conditions one has to use comparative measures of effectiveness and cost of minimum cholesterol lowered is not something one can translate across disease states.

That is why it is very important for economic evaluations of medicines and other technologies and treatments to use cost utility, expressed generally speaking as cost per quality adjusted life year (QUALY).

There are other utility measures, disability adjusted life, but at NICE we use QUALYs, and in fact many other health economists around the world use it too. The reason for that is that it captures the quality of life as well as longevity, and also allows us, in principle, to be able to compare the value for money of totally different treatments for totally different diseases. And that is what health providers have to do – they have to balance the interests of patients with schizophrenia, against patients with cancer, against patients with heart failure and they can only do so if they have measures of cost utility.

A Euro-NICE?

There have been suggestions that an EU-wide equivalent of NICE could be set up. NICE has aroused considerable interest from healthcare providers all over the world. It is a rather unique venture and despite all our critics it seems we haven't done too badly over the first four years. So is a Euro-NICE possible or even desirable?

The first problem is that in any sort of economic analyses you have to undertake comparisons, particularly with current or sometimes best standard practice, and that varies enormously across the EU. Whether one is looking at equivalence or cost minimisation or superiority in cost utility, how does one accept and take current best practice? The second problem is that the direct acquisition costs vary considerably across Europe and while some member states aren't part of the Euro, then the differences become more complicated by exchange rates.

Indirect costs vary enormously across the EU, for example because of the differences in social services arrangements, and the differences in hospitalisation rates. In other words, the differences in the whole culture of medical care.

So the data requirements seem to me to be pretty tricky – is it possible? To make it work at all, all participants would have to be members of the Euro zone and there would have to be uniform pricing across the EU with no discounts – and I don't see either of those things happening in the short or medium-term. Politically, it would involve surrendering member states' responsibility for healthcare and, in effect, setting up a European government and an EU minister of health. Under the present arrangements what is acceptable in terms of costs varies considerably, and there is also diversity in healthcare priorities, so I find it very difficult at the present time to even conceive that this could happen. That we could achieve all of this within the present EU or indeed the enlarged Community seems to me extremely unlikely.

Healthcare spending varies considerably across the EU – in 1997 Germany spent the most with $2,700 per person a year, while Portugal, with $850 per person, spent the least. That's a threefold variation and that difference is inevitably going to be reflected in what is affordable for individual healthcare arrangements  if you are spending three times more then your room for manoeuvre is obviously very much greater.

When organisations like NICE look at acceptability of costs we don't do it in a threshold manner  we dont say that any manufacturer who comes in with a product under £30,000 per QUALY is OK and anyone over that is out. It would create a perverse incentive, with every manufacturer trying to come in at £29,999 per QUALY. In fact, what happens is that the probability of rejection increases as the cost utility rises. So overall I believe we do need new approaches to demonstrate safety and efficacy of medicines, and greater efforts to involve patients.

Finally, we need to show value for money. As Liam's grandma said: 'If you always do what you did, you will always get what you always got.'

This article is based on a speech given by Sir Michael Rawlins at the recent Institute of Clinical Research annual conference in Manchester.