CME Slides Forum - F. Kronenberg

A joint Congress by ERA-EDTA and ISN

DETERMINANTS OF LONG TERM PROGRESSION - RESULTS FROM THE MMKD STUDY

Florian Kronenberg, Innsbruck, Austria

Chair: Giuseppe D'Amico, Milan, Italy

Raghu Kalluri, Boston, USA

kronenberg

Prof F. Kronenberg
Division of Genetic Epidemiology
Department of Medical Genetics
Molecular and Clinical Pharmacology
Innsbruck Medical University
Innsbruck, Austria

Dear Doctor D’Amico, Ladies and Gentlemen, first of all, I would like to thank the organisers for inviting me to present you our data on the mild to moderate kidney disease study.

So the mild to moderate kidney disease study is a prospective longitudinal observational study to identify markers for the progression of CKD in patients with primary non-diabetic CKD. It’s a relatively small study compared to the Greek study with a long-term follow-up. In contrast to many other studies which were done more in the general population it does not look at the ageing effects of the kidney because the general population often looks at the ageing effects and here we have a primary chronic kidney disease.

So the MMKD-Study consists of 227 patients with primary non-diabetic kidney disease recruited in Austria, Germany and in South Tirol. We excluded for the analysis patients with nephrotic syndrome because there were too low numbers. Those patients were recruited in the mid 90s and were relatively young with a mean age of 46 years. The characteristic feature of that study is that we had a very exact determination of GFR at baseline by iohexol clearance and then we followed those patients for 7 years. The endpoint for progression of kidney disease was defined as doubling of serum creatinine and/or the necessity of ESRD or the renal replacement therapy. So we had 177 patients who completed the follow-up and 65 of them reached an endpoint which was in 36 cases doubling of creatinine from the beginning of the study and in 29 cases ESRD.

So here are the parameters which I’m going to talk about now as a short overview and let’s start immediately with what triggered us to do that study.

At the time of the mid 90s there were almost no studies available on lipids and progression of kidney disease and most of the studies either had included a low number of patients or they had only a short follow-up and what you can see from that table here is that the results were very heterogeneous concerning the various parameters measured from lipoprotein metabolism and progression of kidney disease.

Our main focus at that time was lipoprotein A. So here are the results concerning lipid metabolism from the MMKD-Study, the Cox regression analysis here, in the middle here is the unadjusted and here is the adjusted analysis. From the unadjusted analysis you can see that HDL cholesterol was only of borderline significance. ApoA didn’t show anything, triglycerides also borderline but as soon as we adjusted the analysis one factor came out very impressively that was Apolipoprotein AIV and I can imagine that most of the people here in the room haven’t heard anything about Apolipoprotein IV before. You can see from the Cox regression analysis that an increase of 10mg/dL was associated with a 60% higher increase in risk for progression of kidney disease.

So what is ApoAIV? ApoAIV is an apolipoprotein which activates the lipoprotein lipase and LCAT, it stimulates the cholesterol efflux from peripheral cells and it contributes to the reverse cholesterol transport. It has antioxidative properties and shows a strong association with kidney function which you can see here.

When we looked at the baseline data from the MMKD-Study, we found a strong association between ApoAIV and GFR and as well between ApoAIV and creatinine levels.

You can see here that ApoAIV starts to increase already in very early stages of kidney impairment. So here are the Kaplan-Meier plots and we divided the patients in 2 groups, those with ApoAIV levels above the median and below the median which was 26 mg/dL. You can see that those with ApoAIV levels above the median had a much worse outcome in terms of renal survival compared to those with ApoAIV below the median and if you look for example, at 4 years of follow-up, the progression was observed in 12% of the patients with ApoAIV below the median compared to 51% in patients above the median.

Here’s the ROC curve and you can see that ApoAIV in green did also a good job compared to GFR. Remember the Cox regression analysis was a multivariate analysis which corrected already for the GFR values. So it contributes significantly besides the GFR values.

So the next parameter I would like to focus on is ADMA that is asymmetric dimethylarginine. It’s a potent long-lasting endogenous inhibitor of the NO synthase which results in less NO production. NO is a potent vasodilator and regulator of the vascular tone and blood flow. ADMA is markedly increased in renal patients and high ADMA levels are related to atherosclerotic complications. This work was done in collaboration with Doctor Fliser who at that time was in Hannover.

So here you can see a cartoon on how that works. By these high ADMA levels in renal patients, less NO synthase is produced and therefore, we can observe a decrease in renal plasma flow and increased renovascular resistance and increases in blood pressure and a further consequence is also endothelial damage.

So doctor Kielstein did a couple of years ago a very nice study where he showed that when he looked at patients or at young normotensive and elderly hypertensives that by this ADMA levels increased dramatically and at the same time the renal perfusion decreased. So ADMA was an independent determinant of the renal plasma flow and blood pressure.

So when we look in our data in the MMKD-Study then we can see that progressors had significantly higher levels of ADMA compared to non-progressors of kidney disease. Then we separate again the group by the median of the ADMA levels and we can see that those with ADMA above the median had a much worse outcome in terms of renal survival compared with those with ADMA below the median. In COX regression each increase of ADMA of 0.1 µmol/l was associated with this 47% higher risk for progression of kidney disease.

At the same time a very nice study from Ravani and colleagues from Italy came out which showed very similar results. A group of 131 patients, much older than our group, a lower GFR and a shorter follow-up but it impressively showed that ADMA or high ADMA levels were associated with risk for ESRD and also for risk of death.

So now let me talk about adiponectin. I think everybody knows about adiponectin, it’s produced in adipocytes, it has a central role in glucose and lipid metabolism, it seems to be an insulin sensitizer and has anti-inflammatory, anti-atherosclerotic and vascular protective properties. The large number of studies showed an association between low adiponectin levels and negative outcomes.

So we were quite surprised when we looked at our data we observed that progressors for kidney disease had markedly elevated adiponectin levels compared to those who did not progress during the follow-up of the study.

It was also quite interesting when we did a sex stratified analysis we couldn’t see any influence of adiponectin levels in women but we found a strong influence on adiponectin levels and independent of how intensively we adjusted our data, it remained pretty much the same that low adiponectin levels which are usually expected to be associated with a worse outcome, those patients had a much better outcome than those with adiponectin levels above 4 ml for example.

So this is not really totally new or at least in the last couple of years many data came out which has shown that high adiponectin levels in other studies and in other populations predicted a bad outcome. For example, high adiponectin levels were associated with a lower risk for cardiovascular events in ESRD or for mortality in CKD stage 3 and 4 in the MDRD study or for the mortality in patients with chronic heart failure and also for all-cause and cardiovascular mortality in coronary artery disease patients and in a couple of other studies more.

So if I would like to summarise on that, I would say that the association of adiponectin might turn into the opposite direction in the presence of severe disease conditions, as I mentioned in these studies here and it seems to be possibly a compensatory attempt to attenuate the endothelial and vascular damage. It could also be that we have in such situations adiponectin resistance. Let me come now to the peptides, brain natriuretic peptide, BNP and NT-proBNP which are secreted by cardiomiocytes during mechanic and neurohormonal stimulation of the heart. They antagonise the anti-natriuretic and anti-diuretic effects of the activated RAAS system and they are markers of heart failure and left ventricular dysfunction and correlate strongly with GFR.
We did that study in collaboration with Arnold von Eckardstein from Zurich.

Another study, a very recent one we did on A-type natriuretic peptide or in short ANP and adrenomedullin. Both are potent hypertensive, diuretic and natriuretic peptides involved in maintaining cardiovascular and renal homeostasis and they are produced in the kidney. We can see that increased concentrations are associated with cardiovascular disease and renal disease.
They seem to have renoprotective properties and a compensatory increase in CKD. This study was done in collaboration with Benjamin Dieplinger.

Here we can see that for all 3 peptides these peptide levels increased dramatically with decrease in kidney function.

When we look at progressors and non-progressors, we can see markedly elevated levels in the progressors compared to the non-progressors.

In the multivariate Cox regression analysis we can see that all 3 predict progression of kidney disease but as soon as we adjust for age, sex, GFR, proteinuria, then we can observe that mainly ANP and adrenomedullin remain highly significantly associated with the progression of kidney disease.

So here are the ROC curves and we can see form the black line here, this is GFR and the blue line is adrenomedullin and adrenomedullin is even doing better than GFR and also ANP is doing a good job but less good is NT-proBNP.

So finally about the calcium-phosphate metabolism. When we started the study, there were only animal models available about the progression of CKD by high phosphate, high PTH and lack of active vitamin D and then fibroblast growth factor 23 came out which is called a phosphatonin, it is a regulator of the phosphate balance and high concentrations result in hypophosphatemia and it’s a potentially uremic toxin.

Here again, you can see that with decrease in kidney function you can see a strong increase in the FGF23 levels. We did it with 2 different measurements, 2 different assays but the picture was pretty much the same

and then when we look at progressors and non-progressors, we can see dramatically elevated levels in the progressors compared to the non-progressors but also differences in the other parameters of the calcium-phosphate metabolism.

Finally, the COX regression model showed us that FGF23 was a very good predictor of progression and if an increase of FGF23 of 250 units/ml resulted in a 70% increased risk for progression or if we adjust for other variables at least of 35% increased risk.
But also the other parameters did well as long as we did not adjust to FGF23.

Here the Kaplan-Meier curves and we can see that FGF values below 104 showed after 4 years only a 10% progression compared to 60% in those with levels above 104.

So a couple of further considerations, a combined analysis of all the parameters which I have shown you, showed independent associations at least for ApoAIV, ADMA, ANP, adrenomedullin, FGF23 and adiponectin at least in men.

So they were all independent from each other. So the limitations of our study are clear. We have no detailed functional cardiac assessment at baseline. Of course, replication studies are required and I hope that studies like the CRIC study will do that. Causality is somehow supported by a prospective design but this is not proof.
Animal models or interventional approaches which demonstrate the causality for CKD progression are wishful.

If I want to summarise ApoAIV and ADMA, ANP, adrenomedullin, FGF23 and adiponectin are independent predictors of CKD progression. They are independent from GFR and proteinuria and they are also independent from each other. Some of them might also reflect a cardiorenal syndrome.

Finally I would like to thank all the people who were involved in the MMKD-Study from the various study centres where patients were recruited and also from the lab where all the measurements were done and also I thank all my statisticians involved in that project and finally all renal units and those who gave us the money to do this study. Thank you very much for the attention and I’ll be happy to take your questions.

Question: Since all your variables are independent and they show a good ROC, could you make a formula out of all these variables in order to boost the ROC?

Dr. Kronenberg: Yes, we could do that but you have always to bear in mind that we have only a sample size of 177 patients with a follow-up. The confidence interval would not be so good to be really sure to say to the clinicians for example you have to measure that and that parameter. So therefore, I’m hoping for other larger studies which have much better power and then we’ll get much clearer estimates.

Question: Locatelli, congratulations for the very important study. But you underlined the role of FGF23 and adjusting for GFR, the importance of the -- was reduced to 35%. So did you do any time dependent approach in order the adjusted marker or is it a real factor that affected progression?

Dr. Kronenberg: So it would be really nice to have the possibility to do a time dependent modelling but this would mean that we would have to have several measurements over time which could easily be done, for example, by the CRIC study because they have the samples and it might be really interesting to do that and I think at the moment the field is going more and more in the direction to do time dependent analysis at least as long as more measurement points are available and I think, some of the results in the future will change our field a bit. But unfortunately we cannot do it in the MMKD Study.