Slide 1

Thank you Friedo I would like to thank all speakers for staying in time so well so that I have my quarter of an hour to give my presentation. There’s not supposed to be any discussion afterwards but if you are interested, I will be happy to stay to take any questions. What I have been asked to talk about is how should we screen for chronic kidney disease in the elderly? The first question then that we have to address is how does renal function change with aging? Because what we all think is when you get older, your renal function gets lower and that’s normal but there are many editorials written that in completely healthy subjects there is no decrease in renal function. Now we want to know whether this is true.

Slide 2

When you look at data on studies looking at the association between age and renal function, you can see that there are only a limited number studies. In women there are 24 studies, in males there are 37 studies. However, the studies that included subjects that were older than 60 years of age, those studies are very, very limited. The most cited studies, this study by Wesson and what you can see is that they took healthy men and healthy women and they showed that indeed in these healthy men and women you see a decrease in renal function during aging.

I think it’s important to know that in this study renal function was measured with inulin clearance. So the gold standard. However, when we are interested in the elderly and for instance, here in healthy women when you look at a number of elderly in this study that is cited to show that renal function goes down in the elderly, it is remarkable to note that there is only a very limited number of elderly in this study.

Slide 3

Another well cited study is the Baltimore study of aging 450 subjects followed up over a period of 20 years of time and creatinine clearance was measured, so not inulin clearance. I think it’s the second best to use creatinine clearance and this study is most times cited and then they show you this picture that these are subjects with oedema and proteinuria and these are subjects with hypertension and diabetes but these are healthy subjects but in fact, when you look the study up of 1985 these are all representatives of healthy people. When you look at these healthy people, you can see there is a variation. Approximately 1/3 goes down with renal function rather steeply, 1/3 goes down but only slowly and 1/3 remains fairly stable over time but when you look at this study, the Baltimore study of aging, the average decrease in creatinine clearance in healthy subjects is 0.75 ml/min/year again indicating that indeed renal function goes down.

Slide 4

However nowadays, we don’t work with inulin clearances in our patients not even in many institutions with creatinine clearance, we work with eGFR. These are data from our own study, the PREVEND study and what I show here is age versus eGFR and healthy women and healthy women being defined as the absence of diabetes, the absence of hypertension and known renal disease or cardiovascular disease. What you can see that is there is a clear association between eGFR and age or better to say it the other way round a clear association between age and eGFR.

Slide 5

Why is this so important? What happens with renal function in completely healthy subjects? Because we have got our classification system of the KDOQI and in that classification system we call everybody as having chronic kidney disease when they have a renal function below 60 ml/min. What you can see is that rather a large number of women, healthy women especially octogenarians will have a GFR below 60. Is this eGFR below 60 really a risk factor for getting cardiovascular disease and getting ESRD? So this is that KDOQI guideline classification system that was used, that was designed to help for stratifying for renal and cardiovascular outcome. There you can see that below 60 you always have an abnormal renal function. When renal function is above 60, we need additional signs of CKD be it microalbuminuria, macroalbuminuria, erythrocyturia or abnormalities on an ultrasound. This stage III is getting a lot of attention at the moment in epidemiological research and also I think when you are practising nephrologists, these are patients you now get referred and it’s our question how to deal with these patients. GPs refer them to your nephrology practice and are these people sick? Should you control them? Should you treat them? Or do you just send them back to your practice? It’s approximately 5-7% in the population that has an eGFR between 30 and 60.

Slide 6

Why was that level of 60 chosen? That comes from studies like this one from Go in the New England Journal of Medicine 2004 based on data of the Kaiser Permanente study more than 1 million people and what they see is that when renal function is plotted on the x axis and on the y axis the adjusted hazard ratio for all cause mortality and also for cardiovascular events, that there is a nice graded response, a nice graded association that the lower your renal function is, the more cardiovascular events you have and the higher your chance of dying. This is an adjusted hazard rate; it’s adjusted for all kinds of things like sex, social status, blood pressure, dyslipidemia, diabetes etc. etc. It’s also adjusted for age. This suggests that the association between renal function and these events is the same in exactly every age group.

Slide 7

I think that can be questioned and this question was addressed by O’Hare in an article in JASN 2004 and she looked at the interaction between GFR and age. Here on this axis, GFR plotted on this axis, age plotted and the adjusted relative risk for mortality but you can also have the data adjusted relative risk for cardiovascular events. What you can see that especially in the young, the subgroup of 18-44 there’s good association between level of eGFR and risk for mortality but also for cardiovascular events and that starts already at that cut off of 60. However, in the elderly, so 75-85 or over 85 the association between cardiovascular mortality, cardiovascular events, total mortality and GFR is hardly present. You can only see it in the subgroup of elderly patients with very, very low renal function. This is the risk for cardiovascular endpoint and total mortality.

Slide 8

Two years later O’Hare also reported now in JASN 2007 on the endpoint of incidence of renal replacement therapy per thousand persons years. Again, you can see that the graded association between GFR and the risk to get end stage renal failure for which we start renal replacement therapies is especially there in the young. Whereas in the elderly there’s hardly any association and the increased risk is seen only in subjects with very, very low renal function.
So, one size, so one fixed cut off value defining abnormal renal function does that apply to all ages? I don’t think that’s the case.

Slide 9

In the group of Go that showed that nice picture in the New England between GFR and endpoints came back because these are again data from the Kaiser Permanente study but now they also looked at the interaction between age and endpoints. Endpoint is here the combined instance of death and ESRD by age group and here in blue the younger patients and in black the oldest patients. What you can see, is that the association between renal function and the risk for dying and getting end stage renal failure is even in the Kaiser Permanente study that showed that figures adjusted for age but when you look really at interaction, you can see that interaction again, in the elderly the association between GFR and risk for endpoints is lower.

Slide 10

That can mean 2 things, either the association is not there or the way we measure renal function is not a good way. Some people say because you get older, you lose muscle mass so you lose creatinine. It might be that in the elderly there is a misclassification. This is the reason why I’m showing you these slides because some people suggest that you should measure renal function with cystatin C especially in the elderly because cystatin C is not produced by muscles but by all nuclear cells and goes down less with aging than creatinine. In this study, this is a pooled analysis of 4 cohort studies true GFR was measured by io-thalamate and based on the true GFR they designed estimate equations for GFR. Here is the classic MDRD formula shown, here are the results shown for a GFR equation estimation based only on cystatin C. Here on based on cystatin C, age and sex and here they combine cystatin C, creatinine, age and sex. What you can see is that the MDRD formula, the bias overall in this study is only 0.3%. So the MDRD formula on average had a good estimate of true GFR. The P30 means that the GFR estimate was within 30%, so plus 15 or minus 15 of the true GFR and 85% of the GFR estimates with MDRD formula are within that P30. When you look then at the various subgroups and subdivide them according to age, you can see that bias of the MDRD formula versus age is not that great because most of the time we think in clinical practice that the MDRD formula does not really work well and that there is an underestimation of renal function. But this has been the largest study until now 3,000 patients with true GFR measurement and in this study the MDRD formula performs rather well even in the elderly. When you then look at cystatin C the bias and the P30 value are not that different from the MDRD formula neither in the overall group but also not in the elderly. When you look at cystatin C plus age and sex, a kind of MDRD formula but now cystatin C based and not based on creatinine again you see that yes, it doesn’t perform any better than a creatinine value. Only when you combine cystatin C and creatinine, and age and sex and then design an equation formula then you have a P30 that is better but the relative impact is limited.

Slide 11

So, is cystatin C really better than creatinine in estimating GFR? We can also use cystatin C instead of creatinine in predicting outcome in the elderly and this is a study by Shlipak in the Annals of Internal Medicine showing here on the left data with serum creatinine on the x axis. But what you can see here is the hazard rate for cardiovascular morbidity and mortality combined endpoint and there’s hardly any association between creatinine level and the combined endpoint of cardiovascular morbidity and mortality. When you look at cystatin C however, there’s a nice association between the level of cystatin C and risk. This is only creatinine. In the clinical practice of course we work with the MDRD formula and Shlipak also shows data for the MDRD formula and there’s hardly any association in this specific study between renal function and risk. However, this is a study performed in the elderly. That is one thing to remember but this study was also performed in subjects with a GFR over 60, so in clinical practice we don’t look at GFRs over 60 as a risk predictor for cardiovascular endpoints. We look at subjects with lower renal function. So, these are nice data but are not compelling evidence to use cystatin C in the future.

Slide 12

Other data come from Astor presented in abstract from the American Society of Nephrology 2007. What they show here is subjects from the NHANES study and here is estimated GFR and this is the histogram of the distribution of the eGFR when estimated with the MDRD formula. Here you see adjusted animal cardiovascular mortality rate and adjusted for age, race, sex, prevalent cardiovascular disease, blood pressure, etc. Here in red is the association of cystatin C values when they were used to estimate GFR versus cardiovascular endpoints. In green it’s creatinine. This data with subjects below 60, they suggest that there is indeed a better relationship between cystatin C, derived GFR formulas and cardiovascular endpoints. However, again in clinical practice we only look at subjects with a GFR below 60. So on the left side of this slide. When you only look at the left side of this slide, you can see a beautiful association between cystatin C and GFRs and creatinine and GFRs and endpoints and I do not see really a difference between the 2. So I think there’s no compelling evidence at the moment to go over to cystatin C.

Slide 13

Another step I would like to make because I showed you data that in the elderly there is less of an association between GFR with endpoints. I think that this maybe due to misclassification and that we can improve classification by looking at additional signs of chronic kidney damage and this is done in a study by Brantsma a PhD student in our centre. Again a three dimensional graph; GFR level, albuminuria level and hazard rate for cardiovascular endpoints. This is a crude analysis and what you can see is lower renal function, more chance of a cardiovascular endpoint but also higher albuminuria, more chance for cardiovascular endpoint. These are crude data and I think it’s always important to adjust for age and sex and then you see a completely different figure. In the GFR stratum over 90 there’s a beautiful association with albuminuria and that can be seen in every GFR stratum. However, in every albuminuria stratum you do not see a clear association anymore with cardiovascular endpoints. So I think it’s important not only to look at GFR but also to look at additional signs of chronic kidney disease.

Slide 14

This is a study we are going to publish next month or the month thereafter but there are also other data in the literature already present. Not that people really emphasise that point but when you read those articles, you can get that message from these studies. What you can see GFR over 60, GFR below 60. Here in yellow microalbuminuria negative, microalbuminuria negative. Then a GFR over 60 versus under 60. When age and sex are adjusted there’s a hardly an increase in risk for cardiovascular endpoints. That risk is only apparent in subjects with additional signs of chronic kidney damage. Especially when you’ve got the 2 lowering of function and albuminuria there you see the risk coming. That’s not only the study by Tonelli but also the study of Foster.

Slide 15

This is with respect to cardiovascular endpoints. In the overall population but Hallan in the HUNT study looked especially at elderly patients over 70 years of age. Again age, sex and cardiovascular risk factors adjusted. When there’s only low renal function, but no additional signs with low renal function no increased risk in the elderly. The risk is there especially in subjects with additional signs.

Slide 16

I showed you data for cardiovascular endpoints but when you look at the endpoint of ESRD, the same interaction can be found between proteinuria and GFR predicting the end point end stage renal disease. So again emphasising the need to look at additional signs of chronic kidney damage. This has led to the idea that we should redefine our present classification of chronic kidney disease.

Slide 17

Some have proposed that in the future we should not use an age independent cut off but that we should use age and sex specific normal values for eGFR as for instance proposed by Wetzels from the Netherlands from Nijmegen. I think this might be a solution but this has also a disadvantage because by definition 5% of the young subjects for instance, the subjects 20 years of age are labelled as having an abnormal renal function. Whereas in all probability those youngsters none of them or hardly anyone of them will have an abnormal renal function.

Slide 18

I think when we are going to use age and sex specific normal values that then we have first to study whether it’s really better in predicting outcomes. We did this on the basis of our own study looking at the 5th percentile of eGFR. So here on the left the good renal function. Here on an eGFR below the 5th percentile for age and gender. What you again can see when your GFR is below the 5th percentile but when there are no additional signs of chronic kidney damage, there’s no increased risk. That risk is only there when there are other signs and this holds true for the overall population but also for the elderly.

Slide 19

I think we are going to redefine our present classification system for CKD and because of the sake of time I’ll skip this last slide and come to my conclusions.

Slide 20

I think the routine use of cystatin C to estimate GFR has yet no place in clinical practise because of the costs. It has not been proven that it’s really a better estimate of true GFR and there’s no clear improvement for risk stratification of cardiovascular disease in subjects with low GFR in comparison to creatinine based GFR estimations.
During aging renal function declines also in the healthy and this has consequences for prognosis. Lower GFR in the elderly when compared to the younger is less associated with an increased risk fro cardiovascular events and renal function deterioration. That leads to my final conclusion that I think that we should redefine the present CKD classification system. I think there are 2 options; either go the way of introducing age and sex specific cut off values or go in the way to define only subjects with an eGFR below 45 always at risk. Why that level of 45? I think because at that moment the metabolic derangements come in our patients. So it’s not only the risk for cardiovascular endpoints and the risk for end stage renal disease but we should look at age --- and quality of life etc. At higher levels, so over 45 we should pay more attention to additional signs of chronic kidney damage be it microalbuminuria, macroalbuminuria, erythrocyturia or abnormalities on an ultrasound. Thank you for your attention.