The changing prostate cancer pathway – new challenges,new pitfalls

The President: I am delighted to welcome Professor Mark Emberton. Professor Emberton is Director of the Clinical Effectiveness Unit at the Royal College of Surgeons, and he has been involved in prostate cancer throughout his professional career. He has been very interested in non-invasive diagnosis of cancer of the prostate, especially the use of MRI imaging, which does not require biopsy directly; he does do a lot of biopsy work with it, and he has become the leading specialist in the world in treating with high frequency focused ultrasound. If you have any interest in the prostate gland at all, this will be a really interesting lecture. I know that half of our audience will be very interested, because we have a prostate, and the ladies, I hope, have a keen interest in their menfolk. Professor Emberton.

Professor Emberton: Thank you very much. It is a real pleasure to be here.

I think this title is relevant because the pathway in one of the most common cancers that we have, i.e. prostate, is changing dramatically, and whilst change affords opportunity, it also creates risk. And I think it would be very useful to try and identify where those areas of risk might lie as we all struggle to adopt the new pathway.

But before we go into the new pathway I thought it might be worth exploring why it is that the existing diagnostic pathway for men at risk of prostate cancer is flawed and not a week goes by without one being able to read something negative about the prostate cancer diagnostic pathway. The errors in the pathway are legion; in fact, you would wonder why anybody really bothers.

There is the issue of over-diagnosis (the sin of commission), where we diagnose men, tell them they have got prostate cancer, but these men are not destined to die of the disease just attributed to them. In this way we give them a diagnosis of cancer but don’t confer any benefit. All they get is harm, anxiety and worry as a result of that, and obviously the harder you screen for something (it is like an early weather report) the more likely you are to have information that doesn’t prove to be true, and, in the sense of a man with prostate cancer, the further you go back or upstream you go, the less able you are to predict a prostate cancer-related death. So over-diagnosis is a difficult thing to grasp, and a very, very difficult thing to explain to a patient, because everybody believes that they are going to suffer from their cancer. In prostate cancer probably 50% of the men that are diagnosed really do not benefit from that diagnosis and would have been better off not being diagnosed, because no good will come of the diagnosis. So over-diagnosis rarely would end up taking the time of a solicitor. It is probably one of our most common sins, but because it is a sin of commission it is not deemed to be a medical error, although it is, because if we can’t confer benefit, all we can do is confer harm and incur cost, and obviously in healthcare there are opportunity costs, so that means less money that you can spend on something else.

Over-diagnosis occurs because we biopsy a lot of men unnecessarily. I have about a 30% chance of having prostate cancer cells in my prostate tonight. If somebody were to approach me with a set of needles and interrogate my prostate, there is a good chance they might find it. So the unnecessary biopsy is the agent for over-diagnosis. A million men will be biopsied in Europe this year and about half of them will have biopsies that probably should not have been done, had we known better.

Missed diagnosis you know all about. That is easy; a sin of omission; and if the patient is told that he is all clear and he subsequently finds that he is not, he will be very unhappy and will often seek recourse with the legal profession to see if there was some element of malpractice.

If that weren’t enough, we also get it wrong in terms of telling the patient the risk associated with their prostate cancer. Prostate cancer has been divided or subdivided into kind of “pussycats”, the cancer that most men will get (me included) that will not affect our quality or quantity of life, and so-called tigers, which rapidly accelerate and probably are incurable, and of course in the middle there is a big grey zone, and it is interesting that a recent paper, from Cambridge actually, showed that the reclassification [error] – in other words, men were told they had a certain disease by virtue of the biopsy, then went on to have their prostates removed, but that original diagnosis was wrong, in terms of risk, in 50% of the patients. So 50% of the patients went on to their surgery with an imprecise risk stratification. In other words, we got the pussycat/tiger/something in between links wrong in half of all the patients, and that reclassification error is a measure of your diagnostic imprecision. So men are going to big surgery, big radiotherapy with a lot of imprecision around that, and obviously, if there is imprecision, you might get the treatment allocation wrong. So “Sir, you have very low risk disease. We can watch this”, but half the time you will be wrong and that patient will have worse disease that might progress if left untreated, and so you get delayed treatment as a result.

Over-treatment is the consequence of over-diagnosis. So if you over-diagnose and you over-treat you commit two errors, and obviously if you over-diagnose and you can recognise it and you avoid treatment and just reassure the man, at least you mitigate some of the error. But that man is still left with the diagnosis. He will be changed forever; people will perceive him differently and very often they get treated anyway.

And all this is an extraordinary amount of waste. That could be time, expertise and money that could be spent elsewhere. It is often the PSA that is blamed for this, but in fact we are learning now that the greater contributor to all this error is in fact the biopsy.

Here is a prostate that was removed from a man. For the last 110 years, ever since Hugh Hampton Young removed the first prostate at Johns Hopkins University Hospital in Baltimore, the only way of really finding out what was in your prostate was to have it removed. But you don’t benefit from the diagnostic information if you have a one-size-fits-all strategy. So “PSA elevated; we’ve found a bit of cancer; we’ll remove your prostate”, and that is the way we have managed prostate cancer actually for the last 100 years. It is a one-size-fits-all approach and you get your prostatectomy irrespective of your overall risk, irrespective of your tumour volume, grade and multiplicity of location. It would be the same as treating all women with breast cancer with a bilateral mastectomy, and so the idea of observation or lumpectomy, and indeed treatment of one breast and observation of the other, has really yet to permeate to the prostate, although that is changing.

So where does the error come from? Well, I blame a chap called Stamey. He has died now, but he was a very, very eminent figure in urology at Stanford and in 1979 published a paper indicating that random biopsies were superior to targeted biopsies, and in those days a targeted biopsy was “I can feel a lump and I am going to stick this needle into it (and avoid my finger)”, and actually people got very good at doing that. Indeed there will be people in the audience who remember urologists doing that, with a rather large needle. What Stamey did was suggest that actually random interrogation of the organ was better, and ensured that the random interrogation was better, not because it was better but because there was so much prostate cancer around that the more needles you stick in the more you will find, and he showed superiority for the random approach against the guided approach. What he did was to take us from a situation where we were measuring disease, so the disease was becoming clinically manifest, to a new world where the disease wasn’t clinically manifest. We were interrogating the prostate randomly. You know, it would be like doing that to a woman suspected of breast cancer, just seeing if there was cancer present. So you move away from clinical manifestation of disease to a non-measurable disease entity and that was a huge shift. It was at the time thought to be a huge advance, and it has taken us 40, maybe 50 years to realise that it was probably an error, and I am going to share that error with you.

This is what happens in most of the million men that will be biopsied in Europe this year; 1.4 million will be biopsied in the States this year; and obviously, if you are biopsying the prostate blindly, you don’t know where the tumour is. You can see the tumour superimposed here. Most tumours occupy 1 cc and most prostates are 40 cc, so, just in terms of sheer probability, you are more likely to miss the tumour than to hit it. We have been taught to spread those needles widely, and, even if you do hit it and you don’t know where it is, you are unlikely to get a bullseye, and this is the misclassification. So the red bit in the middle of this tumour is the high-grade disease that will probably dominate the biology and progression of this lesion, but because you don’t know where it is, you are reliant on this needle that has just skimmed the edge to tell you that there are a few millimetres of cancer, when in fact there are 10, 12, 15 mm of cancer, but you would know no different. So this man would be told that he has relatively low disease, when indeed he did not. At least he has been diagnosed, and in the old world at least he would get treated, for the wrong reasons, and they would find out eventually that there was worse disease there, and then the doctor of the patient would say “Isn’t a good job that we removed your prostate”. But all that will happen without prior knowledge of the next step in terms of restratification.

We are now learning, because we have methods other than biopsy, that actually we get it wrong quite a lot, and this is something that really didn’t present to us, and therefore didn’t present to you, but is doing so much more frequently now. These are men that have had several biopsies and they have been told they are all clear on numerous occasions, and then we do an MRI and we find a lesion, as I will show you now.

So here is a prostate. So this man is head in, feet out; he is lying on his back and he is having a salami slice through his pelvis. This is an MRI. Here is the prostate outline. The white bit is the prostate that this man was born with; we call it the peripheral zone of the gland and behind is the rectum. So it is front, back, head in, left, right. Does that make sense? It is as if I was lying into the screen like that. This is the transition zone and this is a prostate that has been growing since the man was a teenager. So the surge of testosterone results in prostate growth to the middle of the prostate but it eventually grows so much that it makes the outer bit of the prostate thinner and thinner as a man ages. A prostate will go from 20 cc maybe to 100 cc as a man ages, and that process is a benign one. You may be wondering what this thing is. Well, that is one of this man’s tumours, and it was missed because he had random biopsies. This tumour is at the kind of bow of the ship, what we call the apex, and therefore if you are biopsying through the rectum, which is how most of these biopsies are done, it is very painful to go round the corner, as you can imagine, with a probe, and so the apex of the prostate tends to get under-sampled. He had two sets of biopsies and was told he was all clear on two occasions, and yet his PSA was rising.

This is another way of looking at the prostate. Now the man is standing up, and that is the left, and that is the right, and the bladder is above; so it is as if I was standing like that and this is a cut down through me; and you can see just the wonderful precision of imaging. MRI is completely passive, there are no x-rays involved, no radiation. And on the right-hand side you can see that lesion that we saw earlier. It is at the extreme apex, pretty hard to hit if you didn’t know it was there, and that would be slightly worrying if you had missed that. I don’t know if there are radiologists in the audience, but some of you may be noticing the asymmetry in the gland, and this massive egg-shaped, huge cancer occupying most of the left-hand side of the prostate, which is very anterior, was also missed by two biopsies, and almost certainly this cancer was growing, and indeed may have spread during that period when he was being told quite confidently that he was all clear.

So these are events that we see increasingly commonly now because we have got a test that we can apply that can challenge the biopsy. Where a biopsy is negative and the PSA rises and you worry that you may have missed something, you repeat the same test, so you commit the same errors in terms of the diagnosis, and then you do it a third time and you commit the same errors again, so reapplying the test, which is what most of us have done over the years, is a very inefficient strategy of identifying men with clinically significant disease.

I suppose the other error that we made without MRI was in relation to the over-diagnosis issue. This man had a PSA of 5. If he wasn’t offered a biopsy some of you would be challenging the issue of medical negligence as to why that prostate wasn’t interrogated. Well, we can now do an MRI and interrogate the cause behind the rise of PSA. PSAs typically range between 0 and 4. There is no absolute cut-off. Five is above the reference range and therefore would be read on a GP printout and at least you would have to have discussion about a biopsy. But here is a triage test that allows us to look at the PSA and say “Wow, this man has got a huge prostate”. This is the hip joint, and hip joints typically measure 50 cc. This prostate is at least 100 cc, and therefore his PSA density – the more prostate you have the more PSA you will make, and therefore you can use size to make an adjustment – is 0.1, which is an extremely low PSA density, which would put him at very, very low risk of having prostate cancer.

So we have got a non-invasive test that allows us to look at a prostate and also say:

“This is a beautiful gland. It is enormous, but it is beautiful. The peripheral zone is shining out at us, the transition zone is huge, but that is just a benign process. It always looks like organised chaos, but it is symmetrical. There is nothing there to worry about.”

This bit of the MRI actually measures the movement of water, and it does so passively by just measuring the spin of electrons; it is really quite extraordinary; and when you have cancer you have lots of cells and water can’t move very far and so the vectors are tight and small and there is less water movement. When you don’t have cancer the water can move quite freely, like in the bladder, where you can see it white, where you have maxim infinite water movement. I will show you probably some later on with what we call restricted diffusion.

And this is the injection that men get during an MRI, and this shows the blood supply to the prostate, and prostate cancers, they are like vultures, they scrounge blood supply from other areas, and that blood supply is abnormal, because the vessels are abnormal, and we can detect that when we see it. This might be drawing your eye. This is nice and round. It is benign and it is linked to this nodule here, which looks like a benign nodule; it doesn’t have an abnormal signal on here; so we can, through Boolean logic, rule out prostate cancer. The chances of this man having clinically important prostate cancer is about one in 100, or even less, I would suggest.

So an interesting story has been emerging, and it is really only this month that we have been able to piece this together and to blame Dr Stamey at Stanford.

This paper was published last month. These are the results of a big UK study. £52 million of taxpayers’ money was invested into screening men with prostate cancer and treating them, and it got in the news; it got on to Radio 4 on the morning it was published; and they said the risk is very low of prostate cancer and it doesn’t matter what you do to them. Some of you will remember that story. It is interesting to look at the three screening studies, because this is the last one. These have been attempts to identify men and treat them and then compare the treated men to men who had been observed to establish whether treatment is superior to doing nothing or to doing very little.

We have three studies to compare, and indeed this was published with the last study I showed you, and they are fascinating, because the first one was a Swedish study which diagnosed men that presented clinically in the way that men used to present, and we can see there that the death associated with prostate cancer if you did nothing over 10 years was one in five. So 20% of those men died of prostate cancer during the period of follow-up.

Then it is an American study that took place at the beginning of the PSA era, and this identified men who presented clinically but also – it was in America and it was a VA study, Veterans Administration study – some of the men within that study also presented via PSA biopsy route. So this was a mixed group of patients and the mortality in the men that were watched was 8% over 10 years.

In the English study this was a perfect application of PSA biopsy. It was a screening study, what we call population based. So men were identified in the community, they were minding their own business, somebody knocked on their door and said “We want to screen you for prostate cancer, and, moreover, if we find cancer, we are going to treat you”. So this is the perfect application of the test, and, surprise, surprise, (£50 million; 10 years), the risk of dying of prostate cancer, if you were selected by these means, was only 1%, and that risk didn’t vary whether you were treated or not treated. If you had such a low risk of dying of prostate cancer at that time, much less than dying of other things, then it is going to be very hard to prove that a treatment is going to improve upon that, and this is fascinating, because it tells us actually that the PSA biopsy, the thing that a million men are going to have this year, the better it is applied the worse it performs. So, in other words, the more men it identifies who don’t need treatment, the more men it must miss who do; you know, all the men that do need treatment were not included in this study, and clearly they are out there.

So I think we are developing this hypothesis, which is very hard to communicate to people, that actually we have been getting it wrong for the last 50 years, and actually incredibly wrong very recently. So what we need is fewer men told they are all clear.

Here is another one, sadly. On three occasions he was told he was all clear, and maybe some of you can see that lump right at the top, which is way beyond the needles which are inserted randomly.

And you can see it there. This is a diffusion scan and this is showing that the movement of water in that area there – this is a kind of negative of the one I showed earlier – is much greater. You can all see that there is a problem up there, there is asymmetry.

So what we want to get away from is that random approach towards something that tells us where a cancer is so that we can fire our arrows at it. We have been doing that for thousands of years; we are good at it; the eye is very good at firing an arrow when there is a clear target – and doing just this, so that all our arrows hit the target; one hits the bullseye and the other ones also hit the target. Together they give us the whole picture of the risk that this man has, so we can look him the eye and say “This is your prostate cancer. This is the prognosis and this, we think, is the best treatment for you”.

And that story is emerging, and this study that we undertook in several centres in the UK will very soon be in the public domain. It is not in the public domain as yet, but any day now you will wake up and, hopefully, there will be a good news story about the change in the diagnostic pathway for men with prostate cancer. It is called the PROMIS study (Prostate MRI Imaging Study). You have paid for it; again, it is a publicly funded study, not as expensive as ProtecT, and I think it is going to change things forever, and this is the evidence that we need to back up the assertions that I have just made.

First of all, really importantly, it represents UK practice. You will recognise hospitals there, and they are District General Hospitals, they are not Metropolitan Teaching Hospitals, and so in this study is the extreme of performance. We are particularly good at doing this at UCLH but there are hospitals here who are just starting, and so the results of the study will be conservative, if they prove to be positive.

We should also thank the patients, who went through a huge ordeal to participate in this study in which we were testing MRI, and in order to test MRI we asked or invited them to also have a standard biopsy, the useless one I told you about, but also a very, very detailed biopsy that I will tell you about in just a second.

The aim of this study was to see whether we could mitigate some of the more important errors I listed on the second slide. Can we avoid biopsies? Can we address over-diagnosis? Can we result in fewer big misses; in other words, can we improve the precision? And to do that for a diagnostic study basically you just need to get all the men to have everything. All the tests need to be reported independently of each other. In many studies you have a test because your first test is abnormal, and if your first test isn’t abnormal you don’t get your test. You can’t test the true negative.

So in this study everybody got everything. They had an MRI, a standard biopsy and this really, really, really detailed biopsy, and all the men, we put them to sleep and then we basically played battleships on their prostate and sampled the prostate every 5 mm at the apex of the prostate and the base, the front and the back, and this took sometimes an hour and sometimes meant that, if you had a big prostate, you had about a hundred biopsies taken from your prostate, and that is why I have the slide thanking the men who took part in this study, and there was extraordinary altruism. Men knew that they could probably manage to get an MRI if they went round the back and persuaded us, but actually they so wanted to help us change the pathway that they volunteered (maybe some of you did) to take part in this study and to undergo this extra biopsy so we could calibrate the accuracy of MRI, and this allowed us to exert extraordinary precision. Because we fired needles every 5 mm, we could rule out disease over a certain volume with extraordinary levels of precision hitherto unheard of. This study will never be repeated, and it was that ability and the men that allowed us to do it.

And this is the cancer that we are trying to identify or rule out. The details don’t matter. Basically it is a high-grade cancer that everybody would want identified if they had it and also cancer that exceeded a certain volume. The two drivers of risk in all cancer are always volume and grade, and they were two components of this study.Seven hundred and forty men volunteered for the study. As in all studies, some men weren’t eligible.

But the main reason for not being eligible is your prostate was so big that we couldn’t do the battleships thing on them because the bone got in the way, and if we couldn’t complete one of the three tests that man had to be got rid of, excused from the study and replaced by somebody else.

There was no difference between the men that withdrew compared to those that stayed in the study, as I am showing you now. So there was no bias that resulted from that.

The results are not in the public domain yet, you are getting them early and you will be able to nod when you are listening to Radio 4 in a couple of weeks’ time. This has been accepted by the Lancet and will hopefully very soon be out.

So how did TRUS-biopsy perform? This is the random biopsy that I have been criticising. This is the first time we have been able to ascertain its accuracy, because when you repeat it, all you do is repeat the test; you can’t test how good it has been. The sensitivity, so the ability of the biopsy to identify clinically significant disease when it is there, was 48%. So it missed half of all clinically significant prostate cancers, and it has been doing that for 50 years, so millions of men globally would have been told they are all clear when indeed they are not. If you are told you are all clear when you are not, you don’t get into these studies, and so the risk is under-represented. If you are told you are clear and you are not, if you are diagnosed downstream, your disease will be worse, or you will present with metastases and it has spread elsewhere.What about MRI? Sensitivity for detecting clinically significant disease was 93%. So we still missed some cancers, and I will show you which ones were missed; they weren’t too bad. There is no test that is 100% perfect, but here we have a test that is 100% better than the test that we are using every day, and indeed if you tell a man that he is clear, you will be right most of the time, 90% of the time. Now, remember, this was done on old technology. I hadn’t told you. It was done on weak magnets, 1.5 tesla magnets – today we do it on 3 tesla – and it was done with lots of learning curves within it, so the results of this are conservative, but yet we have got a test that is twice as good as the practice standard. Most studies in medicine are designed to detect a 10% difference between treatment A and treatment B, diagnosis A and diagnosis B. Here we have one that is not 10% but 100% different.

And this is what happens in your MRI score. The MRIs are scored 1 = normal, 5 = abnormal, 2 = nearly normal, 4 = a little bit abnormal and 3 = I can’t make my mind up. So these are the scores that the radiologists did. We classified 1 and 2 scores as normal, and very, very, very few men had clinically significant prostate cancer. So if you had a score of 1 your chance of having clinically significant disease was very low indeed. If you had a score of 5 it is almost guaranteed that you have clinically significant disease. Now, everybody worries about that. Remember, in our current standard practice 52% of men would be missed. Here it is a few percentage points, but nevertheless it is always worth focusing on the error.

And here are the misses that you get with your standard approach versus the new approach. TRUS-biopsy missed 100 men in our study with moderately clinically significant cancer: these are cancers you probably wouldn’t want to miss. MRI missed a few as well, 16 versus 99. But when you got to the really bad cancers, 13 were missed using the traditional approach, and, to our amazement, with all the conservative things I have said, MRI missed zero truly clinically significant cancers when we applied it to 573 British men, which is really quite extraordinary, and that is the kind of good news story that is going to emerge from this.

So how might this play out going forward? The old-fashioned route is that it is linear: PSA, biopsy, treatment. Now we will say “Look, your PSA is high. Let’s just do this scan and then we will discuss what to do next”. So this is what happens if you biopsy everybody. Obviously if you do an MRI you can avoid a biopsy, conservatively we have said, in 25% of men who have a normal MRI. It might be higher than that. So that is a quarter to half a million men a year that will be saved from having an unnecessary biopsy, and also, by avoiding biopsy, you reduce over-diagnosis. Because you are not sticking needles in randomly, you are going to tell fewer men that they have got disease that doesn’t matter, and despite biopsying fewer men you are actually detecting more men with clinically significant disease. So the efficiency of the whole process goes up hugely. So 37% of men are detected with clinically significant disease versus the 19% that would have been detected in the old-fashioned way.

These are other studies that are running at the moment; we don’t need to know that.

Obviously this transition is not going to be easy. None of us trained in using MRI; skills, quality control; so there will be lots of errors that will result from the new pathway as well. But let’s just see how this plays out.

There is a pool of clinically significant disease that exists out there. If you use MRI you will identify most of them. If you use the existing practice standard you will miss most of them, and this is even using 12 random biopsies. The ProtecT study, which is the study that I showed you earlier that was published last week, used six biopsies, so they will miss even more, because of the associative error. If you have got six instead of 12 opportunities you are less likely to hit cancer if it is present. And of course none of this takes any account of the error associated with PSA, which is there, but it is very, very small compared to this. So if you add PSA, which also has true/false positives, this will move a little bit in that direction there.

So that is probably how well we are performing at present. This is a kind of summary, if you like, of the error that we are committing day to day, week to week, year to year currently in diagnosing our men with truly clinically significant prostate cancer, prostate cancer that will benefit from being treated.

So I think the history is a bad one, but the story going forward is a good one, in that we have now got a mitigation to most of this error and, in a rather kind of parochial way, it is a kind of British invention. The MRI story is largely a London-led story. It is now happening elsewhere in the UK and elsewhere, but it is largely a story led by a few enthusiasts here and it will mitigate some of the error that I shared with you in the first slide.

I am just going to show you one more slide. Because the study was paid for by the taxpayer we have to do a health economic analysis to work out whether this new strategy is cost effective, and we worked with some very good health economists in York, and they sent this slide to me and I couldn’t really understand it. People talk about “pathways” in diagnosis and I thought prostate cancer had a straightforward pathway: PSA, biopsy. But they said “No, no, no, no, there are 399 ways to diagnose prostate cancer”, and I really couldn’t believe it, but there are, and I am going to show them to you here, and this is my last slide. I have just been told to finish.

And each dot here is a diagnostic strategy for diagnosing prostate cancer. So you can do TRUS-biopsy first, TRUS-biopsy second if negative; TRUS-biopsy first, MRI; then TRUS-biopsy if the MRI is negative. So you can imagine all the permutations, and I think for the medico-legal brains in the audience this is kind of “Goodness!” So this strategy would have 100% effectiveness; we get pretty close to that in some of the strategies; these strategies here would have a 10 or 20% efficiency, if you like, at diagnosing clinically significant disease. And here is cost.

So the best approaches to prostate cancer would be along this efficiency frontier that we can see here. I won’t break the code for you, but you won’t be surprised to hear that most of these studies, indeed all of them involved MRI in the pathway.

With that I will finish, and thank you very much for your time and the opportunity to speak to you here tonight. Thank you very much. (Applause)

Discussion

The President: Professor, thank you very much indeed for this introduction to diagnosis of prostate mishaps. I am very pleased that we are going to be able to be together after the meeting with canapés and a glass or two of wine, so shall we invite one or two questions now and then break up.

Mrs Brahams: My name is Diana Brahams. I am a retired barrister who used to sue doctors and hospitals, and I also edit the Medico-Legal Journal. Thank you so much for coming to speak to us; it was a really interesting talk. What strikes me is that you obviously feel you have got through to the powers that be and you are going to stop random screenings and save a great deal of money. I feel so depressed about the situation with breast cancer screening and I just don’t understand why we can’t get to the same point with this. I almost feel encouraged to say please take it over, please deal with it. Do you have any comments on why you have been successful and why they have not?

Professor Emberton: We haven’t yet been successful, and actually that is an appropriate policy response in retrospect because our tools weren’t adequate at the time. We now have a tool that is widely available, has no harm associated with it, unlike mammography or giving people a dose of radiation therapy, that is extremely efficient at identifying predictors of disease that you want to find and overlooking the disease that you don’t, because a positive MRI is influenced by tumour volume and tumour growth, the two things by progression. So I think now our task is to convince policy makers, who are now adopting a very minimalistic position. Their response to that paper that was published a few weeks ago is that men are at very low risk of dying of prostate cancer and it doesn’t matter what you do to them, therefore, we don’t need to invest, and they are quite comfortable with that, because it saves money. We would really like to say “Actually, no; we/you have got it wrong for the last 40 or 50 years and we need to start again”, which is a really tough thing to tell policy makers. So we have a challenge to get this out to the same extent that now we are going to introduce this into breast cancer. I don’t think we were ready before today; I don’t think we are quite ready yet actually. We need to do some piloting of MRI in a low-risk population. In fact, we are starting only next week to do just that, and we are going to ask high risk white men living in East London, who are slightly higher risk than the average population across the centre, and we are going to try and target those individuals who are very reluctant to come forward and have examinations and biopsies, etc. So they will be on the list.

Mr Baird: Peter Baird, retired orthopaedic surgeon. Is PSA negative prostate cancer a problem, or are you relying for any screening you do on anybody who presents with prostatic symptoms on PSA as a first hurdle?

Professor Emberton: I think PSA is brilliant, but it has been getting a hard time because on verification tests it has been so poor. So everybody blames PSA, but PSA actually is an incredibly useful predictor of whether you are going to die or otherwise of prostate cancer, and we know from some Swedish studies – they were pilot studies where they stayed still and watched a cohort of men die, and they found what they died of and colleagues went back and looked at the blood and found that PSA was a really good predictor, and of course we ruin that predictor’s accuracy by doing TRUS biopsy. So it is not PSA at all that is the problem. There is an error associated with PSA but it is tiny compared to the error associated with biopsy.

The President: Can we continue our questions over a glass of wine?

Professor Emberton: Certainly, yes.

The President: Thank you very much indeed. (Applause)