I don’t think the ASTRAL trial will help us to know in the future what best to do for our patients with atheromatous renal artery stenosis. I’m grateful for the chance to expand on my views. Ultimately, however, I suspect there is a fundamental belief that determines attitudes here — do you believe that a flawed trial is worse or better than no trial? Also, I’d like to make clear that I’m not against ASTRAL because I don’t believe in intervention or I already think I know what to do in every case. I just don’t see how the results of ASTRAL will be applicable to individual cases.

The chief problem is that ASTRAL is not a single trial, in which a clearly defined clinical entity is randomised to different specific reproducible standard treatments or no treatment, and some objective and worthwhile outcomes measured. ASTRAL is a group of trials in patients with varying clinical problems and significant other illnesses. There is no point in analysing these together, and I don’t think the trial will be able to analyse meaningfully the separate scenarios.

By analogy with heart disease, a good trial might be the effects of coronary angioplasty on heart attack and death rates in patients with severe coronary artery disease. The ASTRAL trial would be like agreeing to include patients with mild coronary artery disease (i.e. mild renal artery disease), including all grades of LV function (i.e. all grades of renal function), and including patients with other causes of angina and breathlessness such as anaemia or valvular disease but some coronary artery disease (i.e. poor function in both kidneys but unilateral renal artery stenosis).

Clearly, the overall results of such a trial would depend on the proportion of different scenarios included. Whatever the results, they would be impossible to extrapolate to individuals.

There are some specific scenarios which would lend themselves to a trial of intervention, which I will list. Undoubtedly, many such cases will be included in ASTRAL, but I fear that these subgroups have not been adequately identified in advance, and that numbers in some will be inadequate. Furthermore, ASTRAL allows investigators to not enter patients if they already think they know best.

With regard to preventing end-stage renal failure, there are at least three distinct scenarios.

1. One functioning kidney with a stenosis. Even this apparently discrete group can of course be subdivided by degree of stenosis and degree of function. Function in a single kidney with a tight stenosis is however not only susceptible to degree of ACE inhibition but also simple BP level — but as we’re probably chiefly trying to prevent end-stage renal failure (ESRF) by preventing occlusion, we have a good endpoint if follow-up is long enough. We’ve also got the problem that many more of those with successful intervention will go onto an ACEi than in the control group — yet another subgroup. And did we exclude all those with recurrent pulmonary oedema at the start or not?
   
2. Two good-sized kidneys, normal creatinine, unilateral RAS (again of varying severity). Well, if we’re focusing on hard endpoints like ESRF, we’d better follow these ones for at least 10 years (and of course hope that drug treatments, BP control etc remain equal in the groups over that time).
   
3. Gradually rising serum creatinine, now 200 — 400 (excluding ACEi, A2Ra effects of course). Symmetrical modest-sized kidneys, with unilateral or bilateral RAS (of varying severity). The theory here is that RAS produces ischaemic nephropathy i.e. gradual nephron loss due to reduced renal blood supply. The problems are that such a physio-pathological entity may not exist (kidneys beyond tight FMD stenoses don't shrink), and certainly in those with a creatinine >200 and unilateral stenosis, there must be another disease process damaging the kidneys.

Quite how anyone thinks a single trial will shed useful light on these complex issues is a mystery to me.

Even if ASTRAL clearly specified it’s subgroups in advance and recruited sufficient patients into each, it faces two further insuperable problems. Firstly, negative results are likely to be at least partially attributable to imperfect revascularisation — that is, the problem isn’t that revascularisation wasn’t the solution, but rather that revascularisation was incomplete, short lived, or produced unacceptable side effects. By the time the trial is complete, angioplasty and stent technology will undoubtedly have moved on (e.g. sirolimus eluting stents). If the results are negative, you’ll have to do it all again with the new technology.

Secondly, in some respects, the results will be uninterpretable. Take scenarios 1 and 2 above. If it were possible to magically dissolve the stenosis without risk to the patient, there wouldn’t be a trial. We need a trial because the intervention can do what we’re trying to prevent — kill the kidney or even the patient. All you will know will be that in this trial, intervention killed the kidney in a% and the patient in b% of cases. However, if the patient survived, successful intervention reduced risks by x and y% over z years. (You won’t know the natural history of true non-intervention, because not everyone was randomised.) In fact, you should already know the figures for a and b% in your own unit, and the literature suggests that even in bilateral RAS the rate of progression to ESRF annually isn’t very high. But how do you meaningfully compare a risk of disaster now with the risk of potential disaster in the vague future? (This isn’t like consenting for CABG where patients are usually very symptomatic or have a very high risk of early MI and death — our scenario 1 might be a bit like the latter, but we think we already know this which is why lots of you haven’t randomised in that setting!)

It is of course deeply unfashionable to suggest that randomised controlled trials might be not only futile but also a waste of precious financial and manpower resources. Nevertheless, I am convinced that the all-inclusive criteria for ASTRAL have sacrificed any hope of answering specific questions. ‘Atheromatous renal artery stenosis’ isn’t a disease — it’s an acquired anatomical pathology which co-exists with many other pathologies and specific clinically defined diseases, and can cause a variety of specific diseases or none at all. The concept of the worth of intervention in this non-disease is ultimately meaningless.