Slide 1

Mr Chairman, Ladies & Gentlemen, thank you for this invitation to speak. Today, I shall explore with you the complexity of the relationship between inflammation and CVD in CKD patients and show you that an integrated, global, multi-technique approach (including proteomics) can help us make progress in understanding this relationship.

Slide 2

My talk will follow the following sequence. Firstly, I shall show a couple of slides to remind you that CKD is a micro-inflammatory state. I shall then discuss the complexity of the relationship between micro-inflammation and CVD and, lastly, seek to demonstrate the value of using global approaches (such as the CV risk index and proteomic evaluation) in exploring this complexity.

Slide 3

I hardly need to remind you that a wide range of studies have shown that CKD is a micro-inflammatory state, regardless of the marker you take. If you look at cytokines, you find a high level of cytokines, if you look at acute-phase proteins, you see high levels of markers such as CRP and fibrinogen. Then, if you look at negative acute-phase proteins like albumin, fetuin-A and so on, you see low levels. So, we know for sure that CKD is a micro-inflammatory state but one question is when do we start to see this in our patients?

Slide 4

There are a few studies on the appearance of inflammation over time. Here is a study with CRP. What is clear is that once the GFR goes below 60, you have a micro-inflammatory state. A second point is that this micro-inflammatory state does not increase gradually as the GFR falls. Therefore, even a slightly uremic state can increase inflammation and this will be stable over time. All our patients already have this micro-inflammatory state.

Slide 5

So, how does dialysis affect this state? Almost not at all, in fact! Here again is a CRP study from the initiation of dialysis onwards. You can see that 1 year later, there was no real change in the CRP level - dialysis did not correct the inflammatory state. This means that when the GFR decreases, you have a micro-inflammatory state which even persists in dialysed patients.

Slide 6

So, bearing in mind that we have this micro-inflammatory state, what is the correlation between micro-inflammation and cardiovascular disease in these patients?

Slide 7

Many studies have been performed in this field and you have already seen lots of graphs over the last two days. I’ll now show you an initial study by Christoph Wanner’s group demonstrating that elevated CRP is associated with increased mortality and increased cardiovascular morbidity. This has since been confirmed by many authors.

Slide 8

So, if we simplify the story, we can say that chronic kidney disease is a micro-inflammatory state and that micro-inflammation may then have a direct effect in the vascular wall - resulting in cardiovascular disease.

Slide 9

However, the story is not as simple as we’d like it to be! For example, inflammation can modulate a number of different factors, including cardiovascular factors. I’ll now show you one example of that: you heard from Peter this morning that chronic kidney disease is insulin resistant and we know that insulin resistance is associated with cardiovascular disease.

Slide 10

So, on one hand, we know for sure that inflammation modulates and amplifies insulin resistance. Hence, the relationship between inflammation and cardiovascular disease might be direct but there may also be indirect amplification of certain cardiovascular risk factors. I don’t want to repeat the previous talk but I'd just like to remind you that fat tissue can produce cytokines (chemokines which attract macrophages) and then you have a vicious circle that will be associated with insulin resistance.

Slide 11

However, on the other hand, inflammation can mask other parameters of cardiovascular disease. The most important one is LDL but I could say the same thing for obesity and for homocysteine. This inflammation has a direct effect on the level of LDL cholesterol and we also know that there is an obscure relationship between total cholesterol and LDL cholesterol in our patients.

Slide 12

You know that when you look for cardiovascular mortality in the general population, the curve rises gradually, like this: high cholesterol, high cardiovascular mortality. When we look at our patients, you see a new curve: low cholesterol with increased all-cause mortality and cardiovascular mortality, and then higher cholesterol does the same. So inflammation really obscures this effect. If you exclude patients who have this inflammation/malnutrition, the remaining inflammation/malnutrition-free patients become like the general population, with a gradual increase in cholesterol or total cholesterol and a proportional increase in cardiovascular morbidity and mortality. However, in the patient who still has micro-inflammation, the relationship is absent.

Slide 13

Therefore, we are faced with indirect effects which may amplify or mask other changes. It’s difficult to explore this situation using a single approach. We need to use an integrated approach to understand this relationship. To answer this question, I’ll show you what we did for the cardiovascular risk index and a proteomic evaluation, with a view to making progress in this area.

Slide 14

The first question is as follows: if we take historically well-known risk factors like age, gender, hypertension, diabetes, tobacco smoking, hyperlipidemia, left ventricular hypertrophy and so on (all included in what is known as the Framingham index), can we predict cardiovascular disease in CKD patients in the same way as in the general population?

Slide 15

What is the additional value of a chronic inflammation marker for better predicting cardiovascular disease in our population? Here, we looked at a cohort of 225 stage 3 and 4 CKD patients. We selected the 100 subjects who were free of coronary heart disease at the start of the monitoring period. We then followed up all the patients for between 4 and 10 years, which is similar to what was done in the Framingham studies.

Slide 16

We sought to evaluate what the influence of this Framingham index in our population was and so we monitored these patients: some developed cardiovascular/coronary heart disease during the follow-up period and others did not. When we compared the respective Framingham indices, patients who developed coronary heart disease during the follow-up had a higher value than those who remained disease-free. However, if you look at what the real calculated index was, it was below what we would have expected in these patients. When you develop coronary heart disease during the follow-up, your index should be initially at 20. Therefore, the first conclusion is that Framingham underestimates the risk in our patients. We wondered whether this was related to additional markers and so we looked further at this cohort, where we have fibrinogen as a marker of inflammation. When we included fibrinogen with the Framingham risk factors, we saw that it increased (or improved) the detection but we are still below 20.

Slide 17

If you express this relationship with ROC curves, you see that there is little effect of adding fibrinogen to the Framingham risk factors in terms of the area under the curve.

Slide 18

Therefore, adding an inflammation marker to traditional risk markers did not improve the prediction of risk in our patients. Perhaps someone will stand up and tell me that fibrinogen is not a good marker for inflammation..! That is true. There are studies from Carmine Zoccali’s group and others showing that CRP and IL-6 are better predictors of cardiovascular disease than fibrinogen is.

Slide 19

Here is another result, taken from Doctor Shlipak's JAMA paper. The researchers included IL-6, CRP and fibrinogen with other traditional Framingham risk factors like lipoprotein levels, thrombogenic status and anaemia. However, you can see here that addition of these three inflammatory markers did not improve prediction of the risk.

Slide 20

So, inflammation markers do not help predict the risk in our patients. What’s going on? What is the gap that we have to bridge here? You know that if you want the best ROC curve, you should have it like that. We probably do have other risk factors. And, in our population, maybe these other risk factors are better predictors of cardiovascular disease than of inflammation or even anaemia! Some of these additional factors are oxidative stress, ADMA, hyperhomocysteinemia, insulin resistance and abnormal Ca-Ph metabolism.

Slide 21

We don’t yet have data on all these new factors in a global, integrated project in order to understand their respective roles. However, I’ll show you a study which does not feature inflammation markers but which uses the Brisbane score rather than the Framingham score. The Brisbane score includes 4 clinical variables; previous cardiac events, time since prescription of dialysis, the BMI (which reflects insulin resistance) and phosphate levels. This is the Framingham risk factor index and this is the Brisbane score. The end result is that the Brisbane score predicted risk better than the Framingham index did. This means that when you include insulin resistance and phosphate abnormalities, you get better prediction - even though the Brisbane score does not include inflammation.

Slide 22

But, in fact, the situation is even more complicated. We know that there are many uremic toxins in our patients: over a hundred have been well defined over the years and I’ll also show you some new ones which have been discovered in proteomic experiments.

Slide 23

These factors may be present in our patients, either alone or in combination. All could play a role in inflammation.

Slide 24

Here is one example for the uremic toxin para-cresol. This compound is known to be capable of modulating leukocyte function in vitro . You can see here that high free p-cresol was a better predictor of all-cause mortality in patients than low free p-cresol was. We don’t test routinely for free p-cresol and we don’t include it in our global evaluation, so we are missing the role of this uremic toxin.

Slide 25

The story is also more complicated. Let's focus on the plasma: if we use proteomic analysis to look at what the difference is between haemodialysis patients and controls, we find that we have 33 different polypeptides, of which 25 polypeptides are present in higher levels in haemodialysis patients than in controls. Conversely, 8 polypeptides are higher in most of the controls and rare or absent in haemodialysis patients.

Slide 26

So, there are new polypeptides present in haemodialysis patients. We don’t yet know what they are - we need to sequence some of them but we and Ray Vanholder are working on cardiovascular complications in our patients in order to understand the relationship between atherosclerosis and global approach involving proteomics. I’ll show you some work from his group performed under the umbrella of the European toxic uremic group. Here you see patients with atherosclerosis disease and here are patients without atherosclerosis disease. If you look at the proteomic analysis, you can see that some polypeptides are present in patients with atherosclerosis and absent in patients free of atherosclerosis, while the converse is true for other polypeptides. The important point about proteomics is that when we go to sequence some of these peptides, we find new markers in haemodialysis patients. I’ll show you one of them.

Slide 27

This peptide is present in haemodialysis patients but is absent or found at much lower levels in controls. We found that this peptide is C3f - a complement peptide. Normally, C3f is produced by the cleavage of C3b: in an old study, C3f was associated with vascular permeability and, in the general population, showed some relationships with cardiovascular disease. C3f is present in our patients and is an inflammation marker. We don’t evaluate this inflammation marker in routine clinical practice but it could play an important role. This global approach involving proteomics will no doubt enable us to discover more and more markers involved in this complex situation.

Slide 28

Let me finish by reminding you again that our patients are in a micro-inflammatory state. The relationship between micro-inflammation and cardiovascular disease is very complex and we cannot hope to understand it by using a single approach; we need to consider the overall cardiovascular risk index, proteomic evaluations or maybe genetic evaluations to explore this complexity.

Slide 29

This story will be continued in a forthcoming meeting in Amiens, where we’ll have more data (I hope) from the proteomic evaluation, and so I invite you to come and hear more about this topic. Thank you very much for your attention.