Prof. Marcello Tonelli Department of Medicine University of Alberta Edmonton, Canada. Slide 1. Thank you very much. Slide 2

Transcription

1 Does phosphate qualify as a cardiovascular risk factor in CKD? Marcello Tonelli, Edmonton, Canada Chairs: Pablo Urena Torres, Saint-Ouen, France Carmine Zoccali, Reggio Calabria, Italy Prof. Marcello Tonelli Department of Medicine University of Alberta Edmonton, Canada Slide 1 Thank you very much. Slide 2

2 Thank you for the invitation to speak here. I've learned quite a lot from the previous two talks but I've learnt something very important from Professor Zoccali, which is that if you get a question that you don't want to answer, then you can just ask someone in the audience to answer it for you. I'll see if I have the chance to use that as well. I'm going to be talking about something quite different. I'm going to take a different approach than the previous two speakers. Slide 3 I'm going to be talking about the role of phosphorous as a risk factor and in particular in the context of a potential intervention that could be used and that intervention is really dietary modification. So, that's where a lot of my prospectors are going to come from and I'd just like to acknowledge the leadership of the National Kidney Foundation in the United States for setting up a conference a little more than a year ago in which a lot of these issues were discussed and which certainly has stimulated my thoughts on this issue. A publication will be coming out from that shortly and the leaders of that work are Geoff Block and David Wheeler, I'd just like to acknowledge them for their leadership. Slide 4

3 I'm going to start very briefly going over some of the observational data and professor Jadoul has already shown you some of this data so we can go quite quickly through it. Slide 5 If we look back to the 1990s, Geoff Block's group was working hard on showing a relationship

4 between phosphorous and adverse outcomes in dialysis patients and you're all familiar with those data. In 2004 he replicated his earlier work in a much larger cohort of prevalent US dialysis patients. A little more than 40,000. He found a U-shaped relationship between levels of serum phosphorous and all-cause mortality. Slide 6 But when he did further adjustment for illness severity, he found not a U-shaped relationship but this monotonic relationship. You can see that for levels of phosphorous above 9 mg/dl the risk is about two-fold higher than people in the reference range. Now that has been an important finding, it certainly reinforced a lot of our thinking about how to treat phosphorous in dialysis patients. Slide 7 If we move- forward a few years at the ASN in 2004 Bryan Kestenbaum presented this abstract which was based on the data that doctor Jadoul has already shown you from a cohort of US veterans. He showed that in very early stages of kidney disease there remains this monotonic relationship between phosphorous and the risk of death with a very consistent relative risk of 2 with this same category of 9 mg/dl. So very similar looking and if we look back to that point in time, at that ASN there was a lot of enthusiasm for phosphorous

5 binders that did not include calcium. So what I was hearing from a lot of people was, look we are not aggressive enough with phosphorous binders in dialysis patients and these are the beginnings of the data that we need to start using phosphorous binders in the early stages of kidney disease. Now I personally had some difficulty with this concept so I had a long flight home, many different connections and so I was able to work on this other study that doctor Jadoul showed you. Slide 8 This is a cohort of patients who had previous myocardial infarction. I initially was hoping that Bryan's findings wouldn't come out to be significant in care and would not be significant in the subgroup of kidney disease. I was very surprised when not only were the findings seen in those with kidney disease but also in those without kidney disease. So when we look, most of these people had normal levels of kidney function. Almost all of them had normal levels of phosphorous but as you can see, there is a direct independent and graded association between these levels of phosphorous and the risk of several outcomes not just all-cause death but also coronary outcomes. Some outcomes like new congestive heart failure might not be related to contrary disease per se, they may have other pathogenesis other than vascular effects. So this was quite though provoking at this time. I was uncertain how to explain this and we went down a few different pathways.

6 Slide 9 These are not my thoughts; these are the thoughts of others. We've already heard some discussion and some good questions about what could be the mechanism that links phosphorous cardiovascular outcomes. We've heard a new hypothesis from professor Zoccali but here are two general classes of thought. One is that phosphorous somehow has an action of its own, which is toxic to endothelium or to other targets in the vasculature. Another possibility, which has been put forward by professor Kuro-o and recapitulated by professor Eberhard Ritz is that phosphorous is not the smoking gun on its own that sometimes it marks for abnormalities in other pathways that are themselves harmful. One putative example could be that since the klotho and FGF23 system exists to defend phosphorous levels that people with higher levels of phosphorous are those that have more abnormalities with klotho and FGF23 pathway and that they somehow have the phenotype that we associate with that such as vascular calcification. So I'm not going to go any further into the mechanism. Slide 10

7 What I'm going to do is now try and sum this up in epidemiological terms. If we are thinking that phosphorous is harmful and we want to do something about it, should we not go on and first assess what the strength of the case for causation is. Slide 11 So whenever I'm looking at a cause and effect problem, I return to these Bradford Hill criteria first described in You read about them a lot and that's because they're so useful and in my opinion, they really are a helpful way to think about these things. Now the first criterion that Sir Bradford Hill put forward was the strength of association. Slide 12

8 Now I think it's important for us to recognise in the modern day that we do not do this properly. When Sir Bradford Hill was speaking about strength of association, the example he gave was scrotal cancer in chimney sweeps and the relative risk there is 200. So, we can be pretty sure that these people got cancer because they were chimney sweeps, there is not a lot of likelihood of residual confounding. Now today we often will see 'oh we can't rule out the possibility of residual confounding. The relative risk is 1.2' and then we go on and pretend that we never said that. I think it's just very important that there is a strength of association that we see for phosphorus, it is present but it certainly is not a 200-fold increase, which always brings us back to the issues that Professor Zoccali talked about, the possibility of residual confounding. Slide 13 The second criterion is consistency. Now, we're all very familiar with this. Intuitively we want to see confirmation in different settings. Sir Bradford Hill puts it very well, different purses, places, circumstances and times. Now we do have quite a number of studies that show an association. Many of them but not all have been done in the United States but they have been done in different times and in different populations. So in my opinion this criterion has been mostly met, although there is room for improvement.

9 Slide 14 Specificity is another important concept and it's one that we do not do well in nephrology. We often think that there is something unique and different about our patients and we don't often confirm whether the findings that we see in kidney patients are actually something that no one thought of in the general population and phosphorous falls into this category as well. There are some additional things we can do to get certainty about this issue; there is this concept of an indicator condition. So if somehow people with higher levels of phosphorous are just sicker, they're poorer, they're sicker, they don't take care of themselves, then that certainly is a competing hypothesis that there's something unique about vascular disease. If this were true, if it were illness severity and not a unique effect of phosphorous on vascular events, then people with higher levels of phosphorous might be inclined to have other outcomes like pneumonia or they may fracture their wrists or get ---. This is something that was the so-called negative control or indicator condition. We haven't done this yet for phosphorous and maybe we should. If the adverse outcomes are limited to cardiovascular, this would support this line of reasoning that we currently have that there's something between phosphorous and vascular disease. Slide 15

10 Temporality is important. In cross-sectional studies one is never certain whether it is the people that have high phosphorous that get cardiovascular disease or the other way round. Fortunately, in this case, we're in good shape, we have several longitudinal studies, several prospective studies where the phosphorous was measured and the vascular events occurred afterwards. So I think we're in good shape there. I think the big issue here and I'm going to be talking about diet and as I'll come back to the question that we have at this point is although we know in dialysis patients that if we restrict their phosphorous, their serum phosphorous levels will fall, we're not as certain about that in people in the general population. Slide 16

11 The biological gradient, this is a so-called dose effect, we have this for phosphorous, it's not just one level of phosphorous that's harmful and everything else is fine. All of the studies that have shown an association between phosphorous and adverse outcomes show a graded response. So just like Bradford Hill observed reporting on -- work in lung cancer we have met these criteria as well. Slide 17 Now I'm just going to pose a possibility. This is just my observation. I have very limited strength in basic science. I learn almost everything I know from meetings and talks like this one but it seems to me that if we find something in an epidemiological study and then we need a mechanism, it's usually pretty easy to come up with one even if it's not very plausible. You can look in PubMed; you can usually find something that you can put down. So I'm not sure about plausibility all of the time. I don't think it's always met in our hypothesis but I do think that in this case we do have a lot of experimental evidence that links phosphorous outcomes to adverse outcomes on several different levels. So I believe that this criterion is met as well. Slide 18

12 So I think this is what I'm going to sum up. I think that of these 6 that I have evaluated, I think we have met most of them to some extent. We have some concern about specificity and this is something that we could test in the future. So at this point we have a good candidate for causation but certainly not proven. Slide 19 What do we need? One thing that I have observed is that in the United States in particular there is a growing movement that makes this chain of argument. Ok so it says we know that high serum phosphorous levels are bad, we know that there's phosphorous in food and we all eat food. So the next step, we should restrict the amount of phosphorous that we eat and this will improve outcomes. Now this seems a little bit in some ways that I do this argument a disservice because it's much newer than this and more carefully made but this is what is gaining some steam. I've been very intrigued by this and that's what I'm going to discuss in the coming slides. Slide 20

13 The first problem is that association is not the same, it's not the same as causation and that's what we've been discussing in the previous slides. Slide 21

14 I just want to remind you this is not something that is unique to nephrologists, all medical disciplines have learned this lesson, we have learnt it as well here are just some high profile examples and we just need to keep this in mind. Eventually if we keep testing interventions that are null, eventually someone's going to stop letting us do trials and say you're finished with the trials. So we just need to keep that in mind. Slide 22 The second problem is quite striking to me and this was eluded to by some of the previous speakers is that we're not actually very good at measuring serum phosphorous. It seems like we can measure it but we don't always measure it properly and we're not always certain about what the level of phosphorous in the blood means for the amount of total body phosphorous, it's complex. Slide 23 Now it's been known for many years that there's a lot of diurnal variation in serum phosphorous. If we look at this graph, what it shows us is that at times of day there are significant fluctuations in the levels of serum phosphorous level. Some of it is at times of the day when people might be eating. So there are spikes in the postprandial period but there

15 are also substantial spikes in time sort of after midnight when people are not eating, they're sleeping. So there's a relationship to fasting status and there's a relationship to time of day. I think this is important. So how many of you in your dialysis units always measure fasting phosphorous? Raise your hands. Is that on your radar? I mean it's not on our radar. Is it always the same time of day that you measure it? So this is just a point. There are some fairly significant swings that we see here and it's something that we could potentially be thinking about. As a consumer of the studies that Professor Jadoul showed you not all of those studies standardise for fasting status. Not all of them standardized for time of day. So if you see a study that's null, doesn't show an effect that's a question to ask, did they standardize on those things. For those of us that are concerned about phosphorous in our dialysis units, maybe this is something we can think about as well. But the bottom line is right now there is this variation and it's something we're going to need to think about when we look into this problem further. Slide 24 Another interesting issue is that there's a whole team of experts that are wondering whether we should be looking at urinary phosphorous and not serum phosphorous to ascertain body status of phosphorous and that's because there's been a couple of studies now that have shown when you restrict dietary phosphorous, the serum level stays about the same. But there's an immediate change in the urinary phosphorous. So that was new information for me and it has got me thinking about what's the best way to mark, the best way to identify people who are overloaded with phosphorous.

16 Slide 25 The third problem is that we've got ourselves into a big mess with our food supply. This is not a problem that is unique to developed countries but it's most pronounced in developed countries. The bottom line is how things stand now even if we wanted to restrict phosphorous we would have a very, very difficult time. Now why is that? Slide 26 The first thing is that there's yet another complexity about phosphorous absorption. This is beautiful work done by Sharon Moe at her research facility in Indianapolis. Sharon took 7 patients with advanced CKD between stage 3b and 4 and used a cross-over design where they were fed food with no preservatives that was designed to provide a diet of a fixed amount of protein, nearly 80 g/day. In the first part, it was given from meat and dairy sources and in the second part it was from plants and grains. You can see here that these are the sequential measurements of phosphorus during the 24-hour period. The same amount of protein when given as meat leads to an increase in the plasma phosphorous level compared to the vegetable-based diet, a 0.3 mg/dl change. So this is just showing us that it's not just necessarily just how much you eat, it's also what you eat. This to me seems like if we think phosphorous is important, we should be looking further at this. So I said it's hard for us to

17 restrict protein. I just highlight this as an example which most people are not aware of which certainly our dieticians at least in Canada don't universally teach. It's something we would need to know more about before we could restrict phosphorous. Slide 27 But I think the much bigger problem is actually, what the food companies are doing. So you can't read the numbers on the slide. This is just the first 50 phosphorus-containing food additives that are available in the United States. There are many, many more, hundreds if not many hundreds. Why are there so many phosphorus products in food? Well the answer is phosphorous is very useful. If we look back to the 1950s which was when this trend began Jello pudding was trying to invent a product that busy housewives who had just starting working could rush home from work and put something on the table without needing to stir a pudding for four hours and they discovered that phosphorous-containing additives were the key to speeding this up and keeping a pudding that would be stable and could be prepared in a few minutes. So that was just the tip of the iceberg. Now you can see here these are all the things that phosphorus is doing in our food. Now a couple of other points about this is that the phosphorous, I don't have time to show data on this, but the phosphorous additives are much better absorbed than the phosphorous content in food. So gram for gram, milligram for milligram if you put sodium pyrophosphorous in food that phosphorous gets absorbed whereas the meat and the grain-based phosphorous is not all absorbed. The other important thing is that labels are currently useless. In most countries phosphorous is not reported. There is no plan to report it and furthermore the estimates that the dieticians use are completely wrong because they don't include the information on these additives. So really if we wanted to tell people, don't eat phosphorus, we couldn't. I'm going to come back to that in a moment. Slide 28

18 Now, here is an interesting randomised trial which was done in the United States. 279 patients at 14 facilities all had hyperphosphatemia. Now these people, these dieticians they just taught the patients to read labels, they gave them a magnifying glass so that they could read the tiny print and they gave them a little list of hundreds of phosphorous-containing ingredients. When the patient said 'I can't remember all those ingredients' they said 'ok just look for ingredients that have the word phosphorous in them or phosphate, pyrophosphate, sodium blah blah blah phosphate look for those and try not to eat foods that have those in them'. So here you can see intervention and control over the course of the study both the control group and the intervention group had a decline in their serum phosphorous but there's a reduction of 0.6 mg/dl in the education group compared to the control. Now I remind you all of us educate our patients about phosphorous, they all see the dietician but again in Canada it's not something that's on our radar to talk about additives specifically. This suggests to us that there's something that we can do. Slide 29 Now thinking about all of this has had a pretty dramatic impact in my way of thinking about things so the first is I make grounds in the dialysis unit and we nag the patients, we say 'don't eat phosphorus your phosphorus is high, don't eat phosphorous' and they say 'I haven't, I didn't eat any phosphorus'. Now sometimes clearly they're saying that as they're sitting there with a can of cola so you know they're not telling the truth but sometimes, half

19 the time, it seems like maybe they're not, it's just they've been avoiding foods that are known to be high in phosphorous but they're eating foods that have preservatives. I think that if we're going to keep doing this, we need to work as a community to get a database of how we can better explain to people where the phosphorous is coming from in the preservatives. The other thing has really been personal and it's been driven by learning about phosphorous and learning about the impact of what the food companies are putting in our food often without our knowledge. This is something that we're aware of or that some of us are aware of because we're nephrologists we deal with phosphorous. But reading about the chemicals, phosphorus one where there are putative and documented health effects had made me start thinking about other things that are in food that I don't pay any attention to and I think this goes beyond our patients I think it's potentially relevant for ourselves and for our families as well. Slide 30 I'm just going to leave you with one thought. So I was travelling in Australia and I saw this application I thought this is fantastic, this is what we need. This is an application it's put out by an academic institution, it runs on multiple smart phones. You can scan the barcode of a product and then it will say ok this is such and such a product and it will give you a traffic light score for the product in front of on its glycaemic content, its fat content, gluten content and sodium. It will offer you alternatives, so if the food is already green then great it says 'great it's green you're doing fantastic'. If it's red or yellow, it will offer you foods that are green or yellow and that are available in your area. Right there at point of service. Now, again we can't do this today for phosphorous because we don't know how much phosphorous is in the food so we couldn't make this app but if we decided to do it, I believe that patients don't actually want to eat unhealthy food, they would like to eat what is good for them and I think it's partly our job to help them do that. Slide 31

20 The last thought is that the last Bradford Hill criterion is experiment and does the experiment support our hypothesis? This just comes back to what the previous speakers have been saying if this is an important problem, we're going to need to do some randomised trials and I think those trials should use this new information about how we can restrict the amount of phosphorous in the food supply, change the source of phosphorous and get to have a look at these chemicals that are there and making things more difficult. Slide 32 So I think I'll close there, you've already seen the data on the adverse outcomes. The big learning point for me has been we're all eating more phosphorous, some of it is dietary but most of it is additives and although I don't think we're yet ready to regulate phosphorous in the food supply I think we should be taking a hard look at it and potentially a hard look at

21 some of these other chemicals as well. Slide 33 I'd just like to close by thanking again this group and especially Geoff Block for his leadership and the people who helped to pay for all this work. Thank you very much. Slide 34 questions Chairman: Thank you very much Marcello for this exciting and stimulating talk. It is now open for discussion but I want to say that I would like to keep tasty food with phosphate, there is another way to decrease phosphate intake or load by blocking its absorption in the gut. Question: Marcello, I enjoyed your lecture as usual, very lucid clear. But I think that your presentation in a way may be intimidating because you emphasise the difficulties we have had in doing trials intervening on phosphate. I would be careful in applying so strictly the Bradford Hill criteria. As you know better than me, most of this criteria are very weak. So I would not make judgement based on Bradford criteria. For example, one criterion strength you know better than me that the risks associated with smoking and cardiovascular disease are just two but no one doubts that smoking is bad. The end of the story I agree we need trials and I also emphasise trials but for the moment I would keep very well separated the

22 investigative part of the story. We need trials and these trials are doable from the public health perspective, because with the public health perspective with phosphate we will have difficulties which are even greater than those we are having with salt. Would you comment further on this? Prof. Tonelli: I think we actually agree Carmine. I put up the criteria because I often find that we claim causal relationships when they are not present. When I look at the phosphorous story, I see in my opinion most of the criteria are met or arguably met, so I think we agree on that. What I like to use them for is also to show where additional studies are most needed. So for me this is a helpful process. I totally agree with everything that you said about the need for intervention and how to select those interventions. Most importantly, with the last point that they are doable. I think we could do these trials if we wanted to. But for me the most startling thing about this whole area is not the science, it's not the molecular story or the biochemical story, the shocking thing is that we are agreed that our dialysis patients should limit the amount of phosphorous that they eat. We think that that may apply to earlier stages of kidney disease and we're uncertain about patients with normal kidney function but we cannot do it because we don't know what is being put into our food and that to me is the most shocking and surprising thing. We could fix that but first we would need to pay some attention to it. Chairman: Professor Ritz. Question: Doctor Tonelli, congratulations for a brilliant intellectually stimulating presentation. I have one specific comment. I fully agree with you that we do not know enough about the mechanisms involved and we are not able to tell definitely true causality in all contexts. But I think what we have to ask as a society and as a profession is that added phosphate which is absorbed between % should be labelled. There is immense, absolutely immense controversy and the food industry is indeed fighting the declaration of phosphate additives. We know from experience that it's the food additives, which are the most important culprit for increasing the phosphate load. Prof. Tonelli: Again, I completely agree with you. I would like to see more transparent labelling as a result of this for many of these things. I think the reason why it can be useful to go through this sort of history as I did in these slides is that others will do it to us. If we are going to make a case that we need to label phosphorous or chemical a or b, these arguments will be made and I think it's useful for us to say ok where's the gap what is the study we need, maybe we can close that gap and make a better argument. But we agree we both want to see better labelling for our patients, for ourselves and our children. Chairman: Please. Question: A very nice talk Marcello. I would like to insist again on your last point. How important it is to do prospective randomised trials? I would like to remind the audience of our conviction 10 years ago- the higher the haemoglobin the better. Then we did prospective randomised trials and showed that when we just normalized the haemoglobin in our patients, they were worse off not better off. So we don't know whether decreasing serum phosphorous to a certain level will improve or actually worsen the outcomes of the patients. We need to do these studies; we cannot just extrapolate from maybe non-causal relationships. I think you would agree. Prof. Tonelli: I thank you for that and I think that's why if feel so good about the labelling of the chemical side of things. I think that it is much safer to restrict people from eating sodium pyrophosphate than potentially to restrict them from eating dietary phosphorous, which might cause malnutrition and so on. I totally agree that there are unintended consequences of a lot that we do thank you. Chairman: Thank you very much Marcello.