Slide 1

Dear Chairman, first I think I’ll phrase one sentence about the 4D study.

Slide 2

This was or is a randomised controlled trial testing the hypothesis that a statin in comparison to placebo reduces the cardiovascular complications in type 2 diabetics on haemodialysis. There was a primary composite endpoint, which I repeat here, and there were in the Atorvastatin group 243 events and in the placebo 226 events. This overall was a relative risk reduction of 8%. You see the confidence interval and to the disappointment of many people this was not significant.
Overall there were a lot of events and in the course of 4 years 617 people died. So you could do secondary analysis on this sample size.

Slide 3

The LDL cholesterol was lowered by 40% and this was impressive.

Slide 4

There was a 1 mmol difference lowering across 3-4 years. So, the goal was achieved and the question was why did this impressive LDL lowering not work in type 2 diabetics on dialysis?

Slide 5

We should go to review this in a Lancet paper from 2005 looking at the effects of statins on particular causes of death per 1 mmol LDL cholesterol difference in 90.000 patients with coronary heart disease. It became clear that first line, coronary death was reduced by statins by 19%, you see these impressive reductions, big numbers. Stroke was not significantly reduced and other vascular only by 7%.

Slide 6

So, I take these numbers now and bring it to the next table looking at the relative risk reduction; 19% for coronary heart disease, other cardiac, stroke and non-vascular.

Slide 7

And look at the proportional mortality in the general population and you see that most people in the general population die of coronary heart disease and non-vascular. Only a minor proportion of other cardiac and stroke.

Slide 8

So, I take these numbers of the CTT meta-analysis and compare it with the 4D group.

Slide 9

You see that only a minor proportion of type 2 diabetics on dialysis die of coronary heart disease ad most of them by other cardiac and non-vascular. So, if a statin is mainly working on coronary heart disease, you can understand why there was only a 9% risk reduction in the 4D trial.

Slide 10

Looking at the causes of death in the 4D trial we see that most of them were cardiac or cardiovascular, if you take the stroke. There was a bit more infection than in the USRDS. These were all prospectively examined endpoints by an endpoint committee.

Slide 11

If you look into causes of cardiac death, then you find that sudden death accounted for most of them. 25% could not be classified in detail because they died without observation.

Slide 12

So, the entire causes of death in the dialysis population are multiple and this may explain why one drug hits one endpoint that we need another approach.

Slide 13

Let’s further look at subanalysis and posthoc analysis. We did some biomarker studies, cardiac markers, we looked at the ECGs that were recorded and at biocompatibility and flux.

Slide 14

Overall the history of cardiovascular disease, coronary bypass, MI or coronary heart disease, all those people with previous history had a 37% increase of the risk. Heart failure and peripheral arterial disease was very prominent, a 75% increase. As in all studies, the age every year counted for a 3% higher mortality and albumin was a very impressive. Albumin as a negative acute phase protein every 1 g higher decreased the relative risk by 43%. CRP at baseline a positive acute phase response protein increased the risk by 24%.

Slide 15

If you go more in detail looking at the Kaplan-Meier survival curve for C-reactive protein and all-cause mortality, then you find in this group of 1255 patients according to the quartiles a very good separation and predictive value of 3 reactive proteins something that we already knew in the past and so confirmed in this big trial.

Slide 16

But unfortunately, there was no effect of atorvastatin on the left side, 612 patients on CRP before versus after 6 weeks. Impressive lowering of LDL but nothing on CRP. The placebo group had slightly higher values and they increased significantly over 6 weeks, so there was a difference between the groups but atorvastatin had no effect. So this is a new message.

Slide 17

Serum albumin was quite impressive, predictive if you set the serum albumin above 4. At the risk of 1 you see there is going down with the albumin in quartiles there is a more than 50% increase if you have a low serum albumin something that we already knew as well.

Slide 18

Looking at total mortality on the complex of phosphate, calcium and PTH on what we are currently working the phosphate increased the risk every mg/dL by 11%. There was no effect of calcium but the calcium-phosphate product was significant because phosphate was so strong. PTH increased the risk, the higher the PTH, the higher the risk, 12% from baseline.
We are currently working on other bone and vascular parameters such as osteocalcin, didn’t see anything. β-crosslaps a weak prediction in all-cause mortality but not in cardiovascular. Nothing for osetoprotegerin and fetuin is in preparation.

Slide 19

The baseline phosphate, if you look in detail and you set the lowest quartile at the risk of 1, then those with a phosphate above 2.26 had more than an 80% increase of relative risk to achieve here a primary endpoint, a cardiovascular endpoint.

Slide 20

Let’s look at the parathyroid hormone. You see the quartiles down there relatively low PTH in Germany in these type 2 diabetics and the first quartile was significant against the highest quartile for the primary endpoint and all-cause mortality as well.

Slide 21

But more impressive were the cardiac markers.

Slide 22

And we did analysis of Troponin T. The quartiles are down at the bottom and you see a very good predictive Kaplan-Meier survival curve for the primary cardiovascular endpoint as well as for all-cause mortality.

Slide 23

So Troponin T could be used as well as NTproBNP. Even a higher significance for the primary endpoint and you see these very high values a proBMP of below 500 is a normal value and dialysis patients for various reasons go up above 8.000 in the upper quartile and all-cause mortality and the primary endpoint could nicely be separated by the quartiles of NTproBNP.

Slide 24

Overall there were some other markers which were significant haemoglobin for each g/dL the risk decreased for each g increased the risk decreased by 6%. The quality of diabetes control per 1% HbA1 C a 10% increase of the risk.

Slide 25

So, we were currently analysing: what does this mean? Looking at the group of patients per se, if you look at the BMI and every kg decreased the risk, then you can understand what group of patients you have.

Slide 26

If you set the BMI here around 23.5, at the risk of 1 as has been done by many other studies, then there was a decrease of the risk but no increase below the BMI of 23, as seen, the red curve, in other studies from Kalantar-Zadeh, for example.

Slide 27

But I’m explaining at the moment the flat curve with the number of patients in the study. We only had 13-72 in the body mass range, in the ranges you see plotted below but there were many more patients, type 2 diabetics, in the upper range and therefore the missing increase at the low BMI may be explained.

Slide 28

There were ECGs recorded at baseline and at 6 month intervals mainly to detect non-fatal MIs which were silent. We also saw 56 fatal MIs and these ECGs were analysed according to a sinusrhythm, AV block, QRS Ax deviation, MI, re-polarisation disorders, left ventricular hypertrophy and bundle blocks and heart frequency. There was not much in ECGs. The sinusrhythm, as everybody can assume, was predictive and the loss of sinusrhythm, atrial fibrillation increased the risk or sinusrhythm was protective. Re-polarisation disorders increased the risk by 19% and a prior MI increased all-cause, as well as cardiovascular mortality. Something that we can assume but there was nothing in it on left ventricular hypertrophy, so ECG may not be sensitive enough or you couldn’t detect left ventricular hypertrophy by ECGs. They are recommended by guidelines at the moment to be recorded every 6 months but this is the first large scale analysis showing the value of an ECG.

Slide 29

If you look at the Kaplan-Meier survival curves for the primary endpoint, those without signs of MI, those with signs of MI, then you see the difference in mortality over time. There were more without MI at the beginning than with MIs.

Slide 30

Finally, at the end I want to only go a minute into flux and biocompatibility in 4D, later we get randomised controlled trial data. There were patients dialysed with different membranes over different periods of time during the 4 year observation period. We were looking at high flux, low flux synthetic and semisynthetic and cellulose and we identified groups of patients treated with exclusively one membrane over a median period of 4 years.

Slide 31

This is the Kaplan-Meier survival curve, at the first glance very impressive. High flux patients survived better than low flux synthetic and semisynthetic or cellulosic. Also there was a low number of patients in this cellulosic group.

Slide 32

We corrected for everything what we had in hand and these are the data for the primary cardiovascular endpoints, very predictive whether you are dialysed with high flux or low flux.

Slide 33

However, I must state that secondary analyses of randomised trials give only association but do not prove causality. Studies are in general observational, they are subject to bias, selection bias and careful adjustments have to be made. Secondary posthoc analyses are useful and important because they generate hypotheses or let’s wait for the randomised data.

Slide 34

My last slide summarises the exercise we do with this data. In the previous we have looked at calcification, pulse pressure, pulse wave velocity, endothelial function, left ventricular hypertrophy and intima-media thickness and have identified a number of acute phase proteins, which we tried to link with the disease, link to the pathogenesis of the disease in the previous years.

Slide 35

Then we tried to identify therapies, for example, statins that we have seen or in small randomised controlled trials a beta-blocker, vitamin E and acetylcysteine but there are other ideas out there and currently we need randomised trials with all these interventions. But let’s propose that maybe in the future we have also not only to look at these surrogate endpoint markers but also to look at the heart endpoints at which rate these endpoints are occurring and target or tailor the therapy to the endpoint. Statins so far through the 4D trial were identified to work on MI or congestive heart failure in coronary heart disease. Maybe one time we may use a polypill, which is under investigation in other countries in other diseases. Thank you very much.

Slide 36

Chairman: Thank you very much Christoph for that very interesting presentation. The paper is now open for discussion. There are microphones down at the front. Please come down and ask the question and identify yourself. While you’re thinking, I can ask you Christoph have you analysed your dataset for people who achieved all of the relevant guidelines or targets? So if you put in haemoglobin, calcium-phosphate, PTH, dialysis adequacy etc., you would have a group of people who did and a group of people who didn’t?

Prof. Wanner: Yes, if you take these groups out, then you come to lower numbers. Currently I’m struggling with our statisticians all the time. They don’t want to go in low numbers. Not all people achieved, for example, haemoglobin, the mean was 10.8. So let’s do first these analyses and then go to so-called what they say data mining or whatever.

Chairman; Data mining exactly. Well, we’ve got a question I think from Francesco.

Question: Yes, congratulations Christoph. It was very well done. I think were the data you suggested for the high flux effect related to all causes of the cardiovascular mortality or to specific causes?

Prof. Wanner: We only analysed it having the composite endpoint together to all causes. The same argument going into the various subcomponents of the endpoint brings us down to small numbers.

Chairman: Carmine.

Question: This is the time of secondary analysis to generate new hypotheses and you laid down several things to watch for. Could you tell us if you have planned or intend to plan secondary analysis to identify patients that may be responsive to the statin?

Prof. Wanner: So far we have been looking into the inflammatory axis, whether highly inflamed or low inflamed, well-nourished, high albumin, low albumin and there was nothing in it. We could see that the high CRP patients had a higher risk but there was no effect on the endpoint by atorvastatin unfortunately.

Question: Now, I’m suggesting a different type of analysis that is an analysis starting from mortality and doing the reverse to see whether there is a subgroup of patients which can benefit from statins. Of course this is a purely hypothesis generating exercise but it can be done.

Prof. Wanner: We may do this together.

Chairman: Ok that’s great. Thank you very much indeed Christoph.