CME Slides Forum - M. Pepys

DECIPHERING THE ROLE OF C-REACTIVE PROTEIN IN HUMAN DISEASES

Mark Pepys, London, United Kingdom

Chair: Jorge Cannata-Andía, Oviedo, Spain

Luis Piera, Barcelona, Spain

pepys

Prof M. Pepys
Centre for Amyloidosis and Acute Phase Proteins
Royal Free and University College Medical School
London, United Kingdom

Mr Chairman thank you very much. I’d like to thank the organisers and you also for so kindly inviting me to speak at your meeting. Just to show you that I’m not completely out of place here, my first ever publication was something to do with nephrology. I may have strayed from the true path after that but I did at least start off in the right way.

I’m going to talk about what the role of CRP might be in human disease. There are really two questions: is it useful to measure CRP in our clinical practice and does CRP contribute to the pathogenesis of disease?

In order to answer these questions we need to really understand something about the biology of CRP and its behaviour in disease. Two very important points are, first, the major difference, the huge difference, between baseline C-reactive protein levels in healthy people and people without obvious disease and, second, the great confusion that has arisen lately particularly from the cardiologists, whom we all know aren’t quite as intelligent as the nephrologists. They confuse association and causality. That’s a very important thing to avoid if you can. When we’ve gone over these areas here I’ll then try to answer for you these following questions: is measuring CRP useful as a risk marker to predict disease? Could it be a pathogenic causal risk factor in disease? Does it actually contribute to disease in people who already have tissue injury? And finally could it therefore be a valid therapeutic target?

C-reactive protein, CRP for short, is the classical acute phase plasma protein made in the liver. It’s a very robust analyte and there are very well standardised assays based on the WHO international reference standard for CRP which I produced in the early 1980s and subsequently there have been secondary reference standards for which my laboratory has also provided the CRP. It’s very easy to measure. It’s an exquisitely sensitive but completely non-specific and therefore, non-diagnostic, marker of almost any tissue injury, infection, inflammation or disease. It has an enormous dynamic range and can increase from 50 micrograms/L to more than 500 mg/L, which is greater than 10.000 fold. With regard to function, its most notable function is that it binds to particular ligands in a calcium dependent way and these ligands in the body are things like dead or damaged cells, particular specific ligand proteins and some other things. When it binds to macromolecular ligands it can then activate the complement system and thereby induce both host defence and potentially also tissue damage and inflammation. But I have to tell you that its in vivo function is not known despite what you might read in the many articles about CRP, that it does this and its function’s that. To be honest we do not know what its in vivo function is yet. I’ll come back to that as we go along.

CRP is an interesting homopentameric molecule. It has 5 identical subunits each of 206 amino acid residues which are non-covalently associated. These subunits are non-glycosylated and a very important point about the structure of CRP is that it’s extremely stable under physiological conditions and in the presence of calcium these subunits do not fall apart and they also are very insusceptible to proteolysis. So we can probably dismiss straight away the substantial literature that’s built up about subunits of CRP and altered forms of CRP or fragments of the molecule which are supposed to happen in vivo. They almost certainly don’t. So, I’m not going to deal with that area further. This is a very, very stable molecule in the body.

It is produced and broken down almost exclusively by liver cells. There’s a literature about the synthesis and the secretion of CRP by other cells and certainly all somatic cells in the body obviously contain all the same genes. There’s no reason why other cells should not make CRP but the liver’s contribution is overwhelmingly great. It’s not at all clear and it’s a rather controversial area whether significant amounts of CRP are made by any other cells. In the liver its synthesis is regulated at the level of gene expression by the proinflammatory cytokines and very importantly from a clinical point of view the clearance, the half life of CRP in the plasma, is exactly the same in everybody under all conditions. It doesn’t matter whether you’re a healthy person with 50 micrograms/L of CRP or somebody who’s just had a heart attack whose CRP is 450 mg/L, the clearance of CRP from the plasma is always exactly the same, so the only thing that determines the plasma concentration of CRP is its synthesis rate.

In normal healthy subjects CRP levels are very low and heavily skewed towards the low end. There is a tail of slightly higher levels but in normal healthy people the levels are always very low, generally around or less than 1 mg/L.

Here you see the distribution in 468 healthy volunteer blood donors. I’d like to point out that we published these results in 1981 not recently and there’s been a big hype, particularly in the cardiological literature, about so-called high sensitivity or high sensitive CRP, hsCRP, which is written about as if it’s some sort of different marker, something novel, specific for cardiovascular disease, risk assessment and so on. This is complete hype. An analyte that is measured with high sensitivity is exactly the same protein as is measured in any other way. There’s nothing new about it, there’s nothing specific about it for cardiovascular disease.

In the acute phase response the CRP concentration increases dramatically. We’re talking about normal levels around 1 mg/L but if you get really sick, and bacterial infection is one of the strongest stimulants for CRP production, you get hundreds of mg/L. You can see here a patient on peritoneal dialysis who developed an infection and the CRP level rises steeply and progressively until eventually the correct diagnosis is made and the correct antibiotic is given and then the CRP falls extremely rapidly, which is very reassuring objective evidence of successful treatment.

Just about anything that’s bad for you makes the CRP go up in a major way at that sort of level which I’ve shown you there. So most infections, nearly all infections, inflammatory diseases, cancer, necrosis of tissue, ischemic and other forms of necrosis and trauma, all these things make the CRP rise dramatically.

There’s a small number of conditions, listed here, in which the CRP, however, does not rise very much. In any of these diseases you can have a patient who is dying of it with extensive tissue damage and the CRP may remain normal or only modestly elevated. We don’t really understand the reasons for that but it’s very important clinically because intercurrent infection in these diseases does trigger a typical acute phase response. So, CRP becomes a very useful marker for testing for intercurrent infection in people who are immunocompromised by the disease or by the treatment of their disease.

We can summarise the clinical applications of serum CRP measurements in these 3 main categories. It’s a very useful screening test for organic disease. Once you know the diagnosis, it’s extremely useful for monitoring disease activity whether it’s infection, inflammation, tissue necrosis, malignancy. The third category is the detection of intercurrent infection which I’ve already mentioned.

Here’s just an example of these uses of the CRP in a context which will be familiar to many of you. Patients with systemic vasculitis who have got major renal involvement. You can see here they’re in renal failure these 2 patients and the CRP is extremely high and they’re very aggressively treated and the CRP falls very rapidly and reflects exactly the clinical improvement, whereas the erythrocyte sedimentation rate here remains grossly elevated, much slower to respond and not nearly so useful clinically. Here we see how non-specific the CRP response is. At this point the CRP is high because of systemic vasculitis and here it rises again when the vasculitis is under control because the patient gets a respiratory infection, a urinary infection, a herpes infection, just emphasising that CRP on its own is never a diagnostic test.

You can only interpret CRP values at the bedside when you know absolutely everything about the patient, the history, physical examination, the results of all the tests that have been done and in that context it becomes a very useful test. The idea that you could use a single CRP in a doctor’s office to give somebody a risk assessment for cardiovascular disease flies completely in the face of what we’ve known for 50 or 60 years about the clinical behaviour of CRP in disease.

What are the functions of CRP? Well, as I told you before we don’t know and there are several good reasons why we don’t know. The first one is that nobody’s ever been found who doesn’t have CRP and the gene and the protein are invariable, the coding areas of the gene are invariable so there’s no polymorphism. The mouse which is our usual experimental model has very little CRP anyway and so nobody has bothered until very recently to make a mouse CRP knockout. We are now doing that and there’s another group that’s also doing it and hopefully that may give us some information. CRP is a very ancient molecule in phylogeny, it goes back several hundred million years in evolution. So it’s been around a long time and just about every animal that’s been looked at from very primitive animals onwards has got a CRP or a CRP-like molecule. But although it’s so conserved in the evolution it’s also different sometimes subtly, sometimes less subtly, even between the usual experimental animals, mammals that we study in the laboratory, and humans. One has to be aware of that when one’s trying to use experimental models. The final point to make is that if you put human CRP into a mouse either by injecting it or by making a transgenic mouse, you have to realise that you’ve got a xenogeneic protein in a heterologous model and the effects it has are not necessarily physiologically relevant to what human CRP does in humans.

So, the final point about the functions are that if you want to speculate about function, and there’s a lot of speculation going on at the moment, any proposed function must be compatible with the speed of the CRP response and its enormous dynamic range. It’s inconceivable that CRP is a molecule which regulates vascular tone in my opinion or cytokine production. This is a protein which changes in concentration 10,000 fold, and it can’t be a fine regulator of physiology, it just can’t be. Secondly, if the proposed functions are going to be phylogenetically conserved, that means the same in man and experimental animals, they have to be compatible with the remarkable interspecies differences in concentration, response and properties between human CRP and CRP in other species.

I’ll just show you here a little summary which I won’t go through in detail but essentially you can see that the levels of CRP at rest and in the acute phase response are radically different between humans and the experimental animals here, as are various other aspects of CRP function.

OK, so it’s an ancient protein and the question which we’re trying to answer today, which is does CRP have a role in disease, are not new, it has been around for a long time. More than a quarter of a century ago when CRP had its 50th birthday, I sent it this birthday card in the Lancet and I said here CRP can activate complement and may be able to initiate an exacerbated inflammatory lesion even though we think probably it’s been conserved in evolution to be protective. So this is not a new question that we’re asking.

The recent explosion of interest in CRP in relation to disease was triggered I think by this paper which we published together with Attilio Maseri and his colleagues in the New England Journal in 1994 where we showed that measuring CRP and also another non-specific acute phase protein, serum amyloid A protein, had prognostic value in patients with severe unstable angina. That was a really surprising observation and was taken up with great enthusiasm by the cardiologists and by others. Interestingly, you’ll note that they concentrated on CRP because that’s very easy to measure and although the results with serum amyloid A were exactly the same as with CRP, nobody really bothered with that because it’s much more difficult to measure. All the focus has been on CRP and inappropriately I think in this context as I’ll tell you as we go along.

So there’s been too much focus on CRP for the reason that it’s easy to measure and it’s a robust analyte and so on. The questions that really arise in relation to inflammation and cardiovascular disease, which is where this discussion started is, is it useful to measure CRP for predicting who’s going to have a heart attack? Does it have pathogenetic significance, the inflammation in coronary heart disease and strokes and so on? There has been a major confusion and conflation of association and causality. Just because 2 things are associated doesn’t mean that one thing causes the other. Because the cock crows every time the sun comes up in the morning it doesn’t mean that the cock crowing makes the sun come up. It could be that the sun coming up makes the cock crow, that’s reverse causality and that’s probably what’s going on in this relationship, as I’ll say as we go along.

It turns out that many systemic markers of inflammation are predictive of future cardiovascular disease both in general population cohorts and in patients who already have cardiovascular disease, and in patients like your patients with chronic renal failure with such an excessive risk of premature cardiovascular disease. It’s been assumed from the outset without any evidence that these systemic markers of inflammation reflect local inflammation in the atherosclerotic plaques, but there isn’t really any strong evidence for that. An alternative explanation could be that the systemic markers reflect focal inflammatory lesions elsewhere in the body or they could just represent individual genetic responsiveness to the adverse stimuli which we all encounter all the time. The final possibility is that what we’re measuring with small increases in baseline CRP and other markers of inflammation is not classical inflammation but metabolic tendencies which predispose to premature cardiovascular disease. It’s quite clear that obesity and the presence of the metabolic syndrome, which predispose to cardiovascular disease, and chronic renal disease which predisposes to it too, are associated with slight increases in inflammatory markers. Then we have a completely different category of event which is arterial occlusion, when you’ve actually got occlusion of an artery and ischemic necrosis of tissue, that’s a tremendously powerful acute phase stimulus and there we’re dealing with the very high levels of CRP not the very low baseline levels.

Now, if we’re going to mention cardiovascular disease, which obviously I am and I’m going to continue on that theme for a while, we have to know what we’re talking about. It’s a catchall phrase which is used to describe at least 3 completely different types of pathology. The first type of cardiovascular disease pathology is atherosclerosis, a disease which everybody in this room suffers from because we all eat our unhealthy Western diet and it’s a disease caused by LDL cholesterol which accumulates in the arteries and causes eventually, if it’s bad enough, arterial insufficiency. Then about 30-40% of us will go on to have a catastrophic complication of atherosclerosis which is an atherothrombotic event where there’s actual occlusion of the artery and that causes ischemic necrosis in the heart and in the brain and in the peripheral arterial circulation. There are different risk factors and markers and responses to treatment for those events in the different territories. So two completely separate things: the life long disease of atherosclerosis, and the catastrophic atherothrombotic event. And there’s a third category which is what happens after you’ve had the atherothrombotic event when you actually have ischemic necrosis of a chunk of tissue. That’s a very powerful acute phase stimulus and once again a different sort of pathology. So we have to be clear what we’re talking about when we mention cardiovascular disease.

The first question is: is measuring CRP a useful risk marker and predictor of atherothrombotic events?

There’s no doubt that if you measure CRP in renal patients and you look to see who does badly as far as cardiovascular disease is concerned, the patients whose CRP is more than 10 mg/L have a vast excess of cardiovascular events compared to patients whose CRP is less than 10 mg/L and the same is true in every population that’s been looked at. But is it really useful in clinical practice? That’s the important question. We have to start with understanding what happens with a normal serum CRP. So mostly each individual has a level which is typical for that person but if you measure people’s CRP serially, even very healthy people who are coming to work and have got no complaints will occasionally have a higher value without any overt signs or symptoms. So, if you want to measure somebody’s CRP for risk assessment, you have to know that you’re really measuring their true baseline. It means you’ve got to measure it several times and see what the real level is for that individual. The main things that determine baseline levels are genetic factors; we inherit genes which encode either slightly higher or slightly lower baseline CRP but there’s a major effect of adiposity and particularly central obesity which determines how high your CRP is. Finally, there are major ethnic differences, so the Japanese, for example, have much lower baseline CRPs than white Caucasians.

We have to recognise all of that if you’re going to use CRP as a risk marker test. The early studies of baseline CRP in general populations suggested that there was a very strongly increased relative risk of a future coronary event, if the CRP was in the upper third of the distribution rather than the lower third of the distribution. We published here back in 2000 a metanalysis of all the studies that have been done together with the British Regional Heart study which had about 500 cases in it, a total of about 2000 cases and it looked like if you had a slightly higher CRP compared to a slightly lower CRP, you had a doubled risk of having a heart attack anytime in the future. But subsequently we published this much larger study which is the Iceland heart study in 20,000 subjects with nearly a 20 year follow up, a huge study with about 2500 events, and a metanalysis of all the studies that have happened subsequent to this one here and you can see that the relative risk is greatly attenuated. It’s now only about 1.5 fold, so that is much less and the conclusion is that this highly statistically significant association that you see epidemiologically is almost certainly not useful at the level of individuals. The association is not strong enough, it’s too variable and so on and furthermore, it’s not at all specific for CRP. You get exactly the same associations, if you look at any other systemic marker of inflammation.

The problem with the guidelines which have emerged in the United States suggesting that if your CRP is more than 2-3 mg/L, you’re at higher risk and so on is that an enormous number of people fall into that category. 50% of people in the US have got CRPs more than 2 mg/L and 33% of people have got CRPs in the range of 3-10 mg/L. Furthermore, this is far from being an independent risk marker. Nearly all the cardiology papers tell you CRP is a strong independent risk marker for coronary heart disease. This is just not supported by the evidence. Something like 70% of the variance in CRP is associated with obesity, smoking, metabolic syndrome, diabetes, high blood pressure, lack of exercise, greater age and low socio-economic status. It’s very hard to adjust for those things in epidemiological studies. Even if you ask all the right questions, it’s not possible to correct completely for those things which determine most of the variance in CRP. It turns out that CRP is slightly elevated in almost any medical condition. Anything disease or disease like that people have looked at, whether it’s depression, atrial fibrillation or anything else is associated with a slightly higher CRP. Probably this reflects metabolic distress if you like, anything that’s wrong with the patient, rather than classic inflammation. CRP is certainly not specific as a predictor of cardiovascular disease risk, it predicts global poor prognosis. Anybody in this room whose CRP is persistently more than 10 mg/L has got a significantly lower life expectancy for whatever reason than people whose CRP is less than 10 mg/L. Being sick predicts a bad outcome.

The real answer to CRP in cardiovascular disease risk is going to come in the near future. There’s a group called the Emerging Risk Factors Collaboration which John Danesh in Cambridge is running and they’re busy working now on a metanalysis of the individual data from over 100,000 subjects with about 9,000 incident coronary heart disease cases and they’re also doing a Mendelian randomisation study, which I’ll come back to in a moment, looking at the CRP gene in once again a huge number of patients and controls. This will give us the true nature of the association between CRP and cardiovascular risk.

Let me move to the second question which is, is CRP actually pathogenetic for atherosclerosis or atherothrombosis? Is it on the causal pathway?

This idea arose for several reasons, one is the epidemiological association which I told you about. Everybody jumped from a statistically significant epidemiological association to thinking CRP causes heart disease. There’s another reason why one might think about it, and I like this reason because I thought of it first and generally one likes the ideas one has oneself. CRP, we showed a quarter of a century ago or more, binds specifically to certain lipoprotein classes, particularly LDL which causes atherosclerosis, and subsequently there are a lot of other observations which associate CRP with possible elements of the pathogenesis of atherosclerosis, and you actually find CRP in atheromatous plaques.

So that’s quite attractive but does it really mean that CRP causes atherosclerosis? First of all, is this specific? Nearly all plasma proteins are present in the plaques. They leak in from the plasma, they get in the plaques. Once CRP is in the plaque, you can make just as good an argument that CRP would be atheroprotective as that it would be proatherogenic and actually there’s some very new data that’s about to be published in the PNAS shortly in an animal model which shows that CRP is atheroprotective and I’ll come back to that in a moment. Now, Mendelian randomisation is a very powerful epidemiological approach. If you’ve got genes which encode high and low CRP, and indeed there are genes which encode high and low baseline CRP, and if CRP is on the causal pathway, then people who inherit a gene associated with relatively higher baseline CRP should have a higher risk of cardiovascular disease and people with a low CRP gene should have a lower risk. This analysis is not contaminated or confounded by whether the people smoked or became fat or whether they were born in high or low socio-economic status and so on.

So there should be that same association. That’s Mendelian randomisation and it turns out that this so far does not support any pathogenic role. Here is the data so far, it’s only about 1000 events, not enough to be certain, but there’s absolutely no difference in the risk of having a heart attack when you inherit a gene for a high or low CRP. As I said the Emerging Risk Factors collaboration will extend this to some tens of thousands of cases and we’ll know robustly what the answer is very soon.

Here’s another bit of evidence that CRP probably is not on the causal pathway. This comes from a meta-analysis of all the statin studies, so an enormous number of patients who had LDL cholesterol lowered by statins, many, many events and so on. It turns out that the relative risk of having an event is lowered by exactly the same amount, about 20%, in all individuals whose LDL cholesterol is lowered by about 1 mmol/L. It doesn’t matter that you can’t read all of this slide because it’s too small, but it doesn’t matter whether you have high blood pressure or low blood pressure, whether you have diabetes or not diabetes, whether you’re young or old, male or female. All of these things are associated with differences in baseline CRP but the relative risk lowering is exactly the same regardless of which category the patients are in, which suggests that the benefit of statins is exclusively due to their LDL lowering. It doesn’t provide any support for the idea that it is due to any other effect that statins might have. Statins do lower baseline CRP by about 15%. Does it matter if your CRP goes down from 2.8 to 2.4? Probably not, almost certainly not; it’s very unlikely that that’s what protects you from having a heart attack. It’s lowering the LDL cholesterol that does it.

Now, the people who like the idea that CRP causes atherosclerosis have jumped into all sorts of experiments and they’ve taken CRP and they’ve put it on to cells in all sorts of models and many, many powerful effects of CRP have been claimed and I list the whole lot of them here. Many of these could be associated with increased risk of atherosclerosis and atherothrombosis.

The question is, are these effects that have been reported really due to CRP? The problem is that most of the studies have been done with commercial, usually bacterial recombinant, CRP made by Escherichia coli not by Homo sapiens. People buy this stuff, they stick it on to the cells, they don’t know where it comes from, they don’t know how pure it is. They don’t test it, they don’t remove the sodium azide which is present as a bacteriostatic preservative. It’s got endotoxin in it, they don’t do any controls. It’s an amazing phenomenon that these studies have been published in reputable journals but the fact is they have and so there’s a lot of confusion out there. Even if you exclude all these very obvious caveats there are some more problems that even natural CRP is sometimes hard to purify completely. Also if you try to get around problems with antibody detecting where CRP is binding by directly labelling CRP, covalently modifying it alters its binding properties, so it then binds to things which real CRP doesn’t bind to.

So one has to be very sceptical about a lot of these claims. I’ll just show you here how some of them are not born out. It’s been claimed that CRP has prothrombotic and proinflammatory effects, if you inject it in vivo. Well, this is completely not reproduced if you use pure natural authentic CRP made by human beings and not by bacteria. So here we’re injecting a small amount of CRP into mice and we inject the same amount of natural CRP or recombinant CRP, and 24 hours later we have the same concentration of human CRP in their blood and we measure the levels of mouse SAA, a very sensitive marker of inflammation in the mouse. You can see that the natural CRP does absolutely nothing, whereas the recombinant CRP, which does all those horrible things to endothelial cells in vitro, has a tremendous proinflammatory effect in vivo in the mice, even if you dialyse out the poisonous azide and so on. This is almost certainly due to endotoxin and other bacterial product contamination.

Another study which has been done now and repeated by us and several others is to look at the development of atherosclerosis in apoE knockout mice; these mice get atherosclerosis spontaneously. If you make them transgenic for human CRP, you can test whether human CRP is proinflammatory, proatherothrombotic, proatherogenic and the answer is it isn’t. There is no difference in atherosclerosis in mice with or without human CRP. Actually as I told you there’s a paper going to come out in the PNAS shortly in a much better model of atherosclerosis in LDL receptor knockout mouse which is expressing human apoB-100, so it’s like a humanised atheroma mouse, and in that mouse transgenic human CRP is atheroprotective, it prevents the atherosclerosis developing so fast and so extensively.

Finally, to put another nail in the coffin of association and causality, people with endothelial dysfunction are known to progress to atherosclerosis. Having endothelial dysfunction is the beginning of atherosclerosis and people with endothelial dysfunction have been reported to have slightly higher baseline CRP values. So everybody says well, the CRP causes endothelial dysfunction. Well, in this study which we published a little while ago, we induced transient endothelial dysfunction in volunteers by vaccinating them with typhoid vaccination and that produces transient endothelial dysfunction which you see here as impairment of flow mediated dilatation in the brachial artery. That happens before the CRP has changed at all and, at the time when the CRP is at its peak, endothelial dysfunction had disappeared completely. The thing that correlates with endothelial dysfunction, at least time wise, is the proinflammatory cytokines which you see in the bottom panel there. So association is not the same as causality.

Let me turn now to whether CRP actually causes tissue injury or exacerbates tissue injury when there is already damage, inflammation, ischemia or whatever. Can it be a pathogenetic factor there?

Well, there’s a very well recognised longstanding history in clinical practice of an association between infection, inflammation, anything bad that might happen to you, and a subsequent heart attack. Something like half of the people who come into the hospital with a myocardial infarct will give a history of having had some sort of process that could have given them an acute phase response in the preceding few days or week or so. This very important paper in the New England Journal from our department looked at hundreds of thousands of people in general practices across the UK. And it showed a very significant association between a preceding upper respiratory tract infection or a urinary infection and a subsequent acute myocardial infarction or stroke. Interestingly, having a flu vaccination didn’t seem to be a risk factor for that or risk marker, but these infections did. We know that in the acute coronary syndrome, the severe unstable angina patients who we first reported on 13 years ago, that systemic markers of inflammation are prognostic of outcome and they identify a group of patients who do badly even if you treat them aggressively with all the standard therapies. Thirdly, there is a strong association between the amount of CRP that is produced after an acute myocardial infarction and the outcome, death or subsequent infarct and so on.

I’ll just show you one bit of evidence about that. Here is a large study with several thousand patients in whom CRP was measured 25 days after an acute myocardial infarction and you can see the people who had a low CRP at that point survived much better than those who had a high CRP at that point.

What they died from was essentially cardiovascular death, the significant results here are associated with cardiovascular death rather than other outcomes.

So why would CRP be bad for you, if you’ve had an acute myocardial infarction? The answer is that there is a lot of evidence suggesting that this might be because the CRP actually gets involved in pathogenesis in the heart. It’s been known for more than 40 years that if you produce an experimental myocardial infarction, CRP is deposited on necrotic cells in the infarct. It’s been known for more than 30 years that complement contributes significantly to the extent of experimental myocardial infarction and ischemia reperfusion injury. If you look at all human acute myocardial infarcts, you find CRP and complement co-deposited there and you can extract from the heart complexes of CRP which activate complement.

So, everything is in the right place and at the right time but does it actually do the damage? Well, we need to see, is CRP actually pathogenetic? The way we explore that is to use the rat as a model.

Both human and rat CRP bind to dead or damaged cells but unlike human CRP, which activates complement both in man and the rat, rat CRP does not activate rat complement. So, if we inject human CRP into a rat, it’s a very good model for what human CRP could be doing in a human. So, our experimental model is to occlude the coronary or the cerebral artery in rats, then inject them with human CRP or a control protein and monitor the severity of the tissue damage.

It’s quite clear in a very reproducible way that rats treated like this which receive human CRP have much bigger infarcts, as you can see here in a typical experiment. It’s completely complement dependent and you find the human CRP and rat complement deposited in the infarct. So, this is strong evidence that CRP, human CRP which can activate complement, can make existing tissue damage worse.

Here’s an experiment where we looked at ejection fraction and at rats which are sham-operated and they’ve got normal ejection fraction. If rats are given either CRP or a control protein SAP which is very similar to CRP but doesn’t do the same things, if they’re only given this for two days after tying the coronary artery, there’s no difference but if you have a sufficient amount of CRP, equivalent to a patient who has a big CRP response to an infarct, then the ejection fraction is significantly lower in the rats that got CRP compared to the controls.

CRP treated rats who have had enough CRP are doing worse. When we kill them at the end of the experiment, they’ve got bigger infarcts.

The same thing is true in stroke. So here we’re occluding the middle cerebral artery and the rats that get CRP compared to rats that get human serum albumin as a control have got significantly bigger cerebral infarcts.

And we set out to make a drug to inhibit it and produced this drug which I rationally designed. It wasn’t from a high throughput screen but was based on knowledge of the structure and function of CRP.

It’s a bisphosphocholine, 2 phosphocholines linked by a hexane chain, and this was published just over a year ago in Nature, shown here on the cover of Nature.

We solved the crystal structure of CRP in a complex with this drug and you see here are two pentameric CRP molecules lying face to face cross linked by 5 drug molecules joining together the binding faces of the CRP. Can I have the video on now please.

This is a crystal structure once again. In green you see the CRP molecule and the space filling model shows the drug joining together the two pentameric CRP molecules. This drug therefore, blocks the binding of CRP to its ligands and it also accelerates the clearance of CRP from the plasma, so you get rid of this bad molecule.

Could I have the video here too please? This is a higher power view just showing one drug molecule cross linking two CRP molecules. The extensive atomic resolution knowledge of the structure of CRP and what it binds to enabled us both to design this drug and now to rationally improve it and make better drugs with higher affinity and better pharmaceutical properties, which we hope to get into man as soon as possible. Although, you all know how slow drug development can be.

This drug really works. If we put this drug into the rats, it completely abrogates the adverse effects of CRP. You can see here in group A there are rats that have an infarct, that’s the size of their infarct in group A. In group B it’s rats that have an infarct plus CRP. They’ve got bigger infarcts and in group C, CRP plus the drug completely abrogates the bad effect of CRP. If you give the drug on its own, it has absolutely no effect at all. So the drug is having its beneficial effect just by blocking the bad effect of the human CRP.

Here we’re looking at a more physiological model, perhaps a more clinical model I should say, of myocardial infarction because everybody now gives reperfusion in the developed world when they have an acute infarct and not a terminal occlusion. They get thrombolysis or a PCI procedure and once again, CRP makes this worse and the drug makes it better.

So, targeting CRP in cardiovascular disease could be a good idea. Inflammation definitely contributes significantly to the pathology of acute myocardial infarction and stroke. If we inhibit real things that really contribute to that, it could be better. It will be a proof of principle, if we could do it in acute myocardial infarction, that CRP could be bad for you potentially in other conditions.

Can I have the next slide please? So, is it just cardiovascular disease? The answer is no in my opinion. You have a huge acute phase response of CRP in many different diseases including, of course, the many renal diseases with which you are so familiar. Is it just a useful marker for measuring what’s going on? Or is it actually a mediator of tissue damage? Rheumatoid arthritis is a nice example. The evidence there that CRP could be pathogenetic is just as strong as it is in acute myocardial infarction. CRP values predict and reflect individual disease activity. CRP is in the joints and there are complement activating CRP complexes in the plasma. The same thing is true in many diseases.

Now CRP stands for C-reactive protein. Here’s another thing that CRP could stand for. It’s a very useful, very sensitive non-specific clinical marker of inflammation and tissue damage, so I think it’s empirically a wonderful test but it’s not a strong predictor of cardiovascular disease. It’s certainly not an independent predictor of cardiovascular disease risk. It’s not a clinically useful marker of cardiovascular disease risk in general populations.

CRP is not proinflammatory for cells in vitro, real pure CRP. It’s not proinflammatory in healthy animals, if you inject vast amounts of CRP into animals that are healthy, it does absolutely nothing and I think the same thing is true in humans as well. It’s not proatherogenic or proatherothrombotic in mice and it certainly doesn’t cause atherosclerosis, unlike the known risk factors. So, lowering cholesterol, stopping smoking, lowering your blood pressure, losing weight, getting rid of the metabolic syndrome, treating diabetes, all those lower your risk of having cardiovascular disease. They cause cardiovascular disease, CRP doesn’t do that.

Now, inflammation is significant in acute coronary syndromes and CRP might have a role there. The only way to know is to have our drug or a drug that blocks CRP in patients and see if it does them any good. We have shown that CRP can exacerbate ischemic tissue damage in vivo. We’ve shown that it can contribute to the severity of tissue damage in acute myocardial infarction and stroke in animals. We hope to show that in patients too when we get our drug into man. So, CRP may be a valid therapeutic target in many diseases where CRP is grossly elevated, not in the conditions where there’s a small increase in baseline CRP but when the CRP is very high. Specific inhibition of CRP will therefore, be informative and possibly useful in many diseases and also about what the physiological role of CRP is.

So, let me end with two beautiful CRPs, one is the molecule which you can see there. The other one is one of my 9 grandchildren who just by chance happens to have the initials CRP and she is also very beautiful. Thank you very much.