Gain an understanding of the current concepts in managing acute heart failure and understanding the pathophysiology of heart failure, focusing on the relationship to kidney function.
MARVIN A. KONSTAM: So I'm going to talk about current concepts, current and novel concepts in the management of heart failure and understanding the pathophysiology of heart failure a bit and to a great extent, emphasize the cardio-renal syndrome. These are my disclosures on the bottom. Really, the only ones that are of any significance-- Otsuka funded all the work with tolvaptan. I'm on the Dated Safety Monitoring Board of the RELAX-AHF-2 trial. That is a Novartis product. And Merck funded some of our secondary analyses of the heel project. And those are the only things relevant to what I'm going to talk about. So this is the future and this came from Paul Heidenreich's writing on behalf of the American Heart Association a few years ago. About a year or two later, we collaborated and came out with a similar set of projections specifically focusing on heart failure, but this is his initial work, and his colleagues, looking at the future of cardiovascular disease and I pulled out the data relevant to heart failure from there. And what you can see is that the projection, largely based on changing demographics, is that between 2010 and 2030, we're going to see a 25% increase in the prevalence of heart failure in the United States, up to 3.5% of the adult population. And the graph on the right, I have to say I'm pretty sure is based on practicing medicine the way it was practiced in 2010, which is already changing. And if that practice continues as it is with the growth of the population, that the projected cost, direct cost for the care of heart failure, they expect it to rise from about $25 billion to close to $80 billion, a 215%, increase, staggering, which has a lot to do with why so many people outside of our club of caregivers are interested in this and policymakers have zeroed right in on this. The one thing you can be sure is this actually won't happen. And we see the beginning of it now, but there's going to be a lot of forces in play to have us care for patients differently and keep the costs from going up that fast. Hospitalization is the single biggest cost driver. So there's a lot of attention on heart failure hospitalization for lots of reasons, cost being one of them, but the prognostic impact of being hospitalized for heart failure is substantial. So in this analysis, it was estimated that in patients hospitalized after heart failure, about 33% are dead within 12 months, which is much higher than any of the clinical trial data, certainly the chronic clinical trial data that we have. So it's a marker of badness. Maybe it's even what we do to them in the hospital that drives this. Who knows? But this is another reason to pay a lot of attention to this. We say there are about 1.1 million hospitalizations for heart failure in the United States. For the first time in the last couple years, that number is actually declining, as is hospitalization across the board, as is hospitalization for cardiovascular disease. In the Boston metropolitan area, eastern Massachusetts, between 2010 and 2012, there was a 14% decline in hospitalizations for cardiovascular disease just right under our nose. And so this is really happening and the things that people are doing are actually making a substantial impact on the way we practice and it's just the beginning. As Churchill would say, "it's not the beginning of the end, but it's the end of the beginning." So this is a construct, very simple construct, that I'll put up here and there are two points on this slide. One is that heart failure is not a monolithic clinical entity. It's actually a bunch of different clinical syndromes that we all call heart failure. And that's a problem, because "heart failure" is such a lousy term to begin with. But also, we kind of lump a bunch of things, like acute pulmonary edema and gross right heart failure with anasarca, right heart failure, left heart failure-- we lump it all together and call it "heart failure" and I think we have to sort of do better than that. The three major clinical syndromes are reduced functional capacity, systemic congestion, and pulmonary congestion, and there's some overlap. To a great extent, systemic congestion and reduced functional capacity go together very strongly and pulmonary congestion not quite as much. So there are different syndromes that we have to focus on. And then, the other thing is that heart failure is not just a disorder of the heart. It's a systemic disorder characterized by systemic inflammation and with participation by multiple organs, notably the vasculature and the kidney. And so we'll focus in a bit more on the kidney during this presentation. So we go back a few years and this is an analysis that one of our fellows, Amin al-Ahmad, did and published back in 2001 from the SOLVD database. And the SOLVD didn't let study didn't let people in if their renal function was really bad, like creat-- I don't remember what the creatinine cut-off was. It was something like 2.5. And so you couldn't get in with creatinine over that. And nevertheless, within that population, we were able to show discriminating value of estimated GFR across that little bit narrower population. And so a GFR less than 60 portended a much worse prognosis than higher GFRs. And since then, there have been numerous analyses of this sort. This is a relatively recent one by McCalister, who divided into heart failure with reduced ejection fraction, preserved ejection fraction, and looked at a population base analysis of mortality based on estimated GFR. In HF-REF-- by the way, just footnote-- I despise these terms. So call me back for another lecture and I'll tell you about it. But anyway, so let's go with it for the moment. So as you raise GFR, there's a progressive worsening of survival. Interestingly-- maybe Dr. Joseph can explain this to me. I can't really figure this out-- but in HF-PEF, you don't see quite the same thing. In fact, the best survival is with modestly reduced GFR relative to above 90 and then it sort of goes back. There's something interesting going on here. I've heard some theories. I'd love to hear others. But the major take-home message of this slide is that if your kidney function is bad, it's a bad prognostic indicator in chronic heart failure. But it's also a bad prognostic indicator in acute heart failure, which is what we're talking about today. There have been many analyses on this and basically, renal functional characteristics are a strong predictor-- at baseline admission to the hospital with heart failure, are strong predictors of outcome both in the hospital and longer-term outcome. And change in that GFR or serum creatinine in that population during the course of treatment, an increase in it, also goes along with a worse survival. This is an analysis done by Mihai Gheorghiade from our EVEREST trial and demonstrates that both creatinine and BUN are independent predictors of longer-term mortality and recurrent heart failure hospitalization in a population with acute heart failure and low ejection fraction and some evidence of volume overload. And so this is progressive increases in BUN, progressive increases in creatinine, with the endpoint being mortality or heart failure hospitalization. So there are long-term implications to identifying abnormal kidney function in the hospital and changes in EF function also portend adverse outcomes. So what's going on and what is it that is the basis of this type of the cardio-renal syndrome? By the way, we're talking about only one element of an array of circumstances that are referred to widely as cardio-renal syndrome, like renal failure begetting heart failure or the coincidence of heart disease and kidney disease. What we're talking about is kidney dysfunction or kidney Injury in association, presumably as a consequence of heart failure. So that's really what we're focusing on here. So what is the pathophysiology of the abnormal kidney function in acute heart failure or chronic heart failure? So you think back to medical school and what did you learn about this? You learned about the concept of prerenal azotemia, "prerenal" meaning it has to do with forward flow. And so cardiac output goes down, renal blood flow goes down, and kidney function deteriorates. And that's illustrated by this limb here. By the way, this is the up-to-date article on the subject of cardio-renal syndrome. And this was our group that did it, but Mike Kiernan deserves most of the credit for this-- and so reduced stroke volume, reduced cardiac output, neurohormonal activation, vasoconstriction, sodium or water retention, decreased renal perfusion, and may or may not result in acute kidney injury but certainly has the effect of reducing kidney function. But that is no more than half of the story. It may not even be half, but it's certainly not more than that, because the other limb of this that's critically important and much more forgotten or maybe never known by folks-- and as I attend on the wards, I find there's really almost no recognition of this among the house staff. So it's not really being taught-- but it is that increases in systemic venous pressure resulting in increased renal vein pressure are a major cause of kidney dysfunction in the setting of heart failure. It's not just forward output. It has a lot to do with systemic ingestion and increases in renal vein pressure. And that will cause reduced kidney function and sometimes, presumably, acute kidney injury. So Doug told me-- for weeks, he kept sending me emails about how sophisticated this audience is and how proud he is of the group. And they're up-to-date on all the literature. So I better come with new studies that they won't have seen yet. So this is the best I could do, 1931. So I just love this. So this is The Journal of Physiology. And I have to admit, I missed this issue. I usually read this religiously, but I happened to be on vacation when this came out. So I missed it, but fortunately, there was this review article in 2007 that cited this paper. I went back and found the data or found the paper and replotted it from the information. And this is a very simple point. So this is a canine kidney and the only thing they did here is they increased renal vein pressure. And you can see it going up here from seven to 14 to 21. When it hits about 20 millimeters of mercury, there's a sudden drop in urine output and in this experiment, urine output ceased while renal vein pressure was in this range. And then, they relieved it and there's a marked diuresis that occurred. So people actually, by the way, are rediscovering this. You'll see papers about this and say, hey, we discovered that central venous pressure is important. But this guy, Winton, I love him. He knew this in 1931. Now, it's much more complicated than that. And we keep learning more about the specifics of the complexity and how this really works and what it's associated with. And the more we understand about it, the less we understand it, because we still haven't gotten our hands around it in a way that can clearly direct us, in terms of managing patients and new drug discovery. So Testani and his colleagues have done a lot of work kind of dissecting this out, studying cardio-renal syndrome and looking at worsening kidney function and what does it mean under different circumstances. So one factor that is really important, apparently, is the development of hemoconcentration. That is, you're reducing circulating plasma volume and so therefore, your hematocrit goes up and serum albumin goes up and certain proteins go up and you can measure it. The most common thing that people measure is just hematocrit and looking for subtle changes. And so this slide divides people with acute heart failures to whether they hemoconcentrated during their hospitalization or they did not hemoconcentrate, meaning presumably a reduction in plasma volume. And as you might have expected, the patients who hemoconcentrated had a greater evidence of reduction in GFR. So that sort of makes sense, the prerenal thing, the prerenal part of the story. But where it gets really interesting is that despite the worse kidney function you see in that group, that's also a group that is characterized by better survival. And that's shown here in this slide from their analysis and we had-- when I say "we," it's really Steve Greene, again, analyzing the EVEREST data. It basically replicated those findings from the EVEREST study, just looking at hematocrit and showing just the same way. Hematocrit going up is associated with the greatest probability of GFR going down, but these are quartiles of change in hematocrit, Q4 being the greatest change in hematocrit. And we looked at all-cause mortality subsequent to the trial. You can see the stepwise effect. So here's a way that we're looking at this a little bit more deeply and saying, OK, it's not just that every time your kidney function deteriorates, it's bad. Overall, it is, but we've got to be smarter than that and dissect this out, because here's something that tells us, if we succeed in reducing plasma volume in somebody who is congested and has high filling pressures, presumably, yes, that'll bring the creatinine down. But if you've done that, you've also improved survival-- so really fascinating association. So how does it work? What is this association between kidney function and survival prognosis? So I drew this schematic that sort of lays it out. So heart failure can directly lower GFR through hemodynamic effects that I described to you or it can also cause kidney injury, which can worsen GFR. So when we just look at measures of kidney function, we really don't know for sure if that's all we look at, whether this is just a functional change that we're looking at and it's correctable or whether we're looking at some actual kidney injury. But then, worsening heart failure is associated with worse survival, but worsening GFR, in general, is also associated with worse survival. So what's going on here? Is it that reduction in GFR somehow directly influences survival or is it just a covariate? Is it just that worsening heart failure is associated with worse survival and worsening heart failure causes a reduced GFR so they're just covariates and not directly causally linked or is it some combination? And treatment comes in here because treatment presumably can impact on kidney injury and can impact on the reduction in GFR if we know the right treatments to use. But over here, it shows that it really isn't as simple, again, as raising GFR is a good thing. Lowering GFR is a bad thing. And it's the well-known example of RAS inhibition, ACE or ARB. And so what does that do? We know that that's really good, in terms of improving survival in heart failure. We also know that it's really good at protecting kidney injury, at least in some forms of kidney dysfunction, particularly in diabetics and others with metabolic syndrome. So improved survival, reduction in kidney injury, but guess what? It raises GFR by a purely intrarenal hemodynamic effect on the efferent arteriole. And so the hemodynamic effect there, the mechanism by which that happens, has nothing to do with worsening survival. In fact, it may be a sign that the drug actually is working and that you've given the right dose and that then may be associated with better survival, perhaps. so this is disjointed and you really have to start thinking about what's going on here. No simple statement about this relationship will work. And in that light, I point out this paper that was recently published in genetic heart failure by Mike Kiernan. And this is from the HEAAL database. The HEAAL was a long-term study of low-dose versus high-dose losartan. And high-dose losartan, by the way, 150 milligrams of losartan beat 50 milligrams of losartan. So I hope you're all giving your patients 150 milligrams How many people give their patients 150 milligrams of losartan? He does. See, he reads the lit. Paul raised his hand, too. So the people who know the literature inside and out caught that paper. But anyway, what Mike did in this analysis is he looked at worsening kidney function as defined by a 0.3-milligram percent increase in serum creatinine and found looking at the overall population during the entire course of the study worsening kidney function was associated with a hazard ratio of about 1.45, an increase in mortality in this population associated with worsening kidney function. But when you dig in more deeply than that and look at the timing of the worsening kidney function-- in fact, when you look at the first four months after institution of ARB, sometimes high-dose ARB, there is, of course, a greater degree of kidney dysfunction in the group who gets the higher dose. But guess what? Across the population, there is no association between rise in creatinine and survival. If you increase your creatinine in those first four months after starting an ACE or an ARB, it does not demonstrate or predict a worse outcome. And presumably, it's separating out what I mentioned a moment ago, the acute intrarenal hemodynamic effect of these drugs, which do not have an adverse long-term impact. In fact, they may signal that the drug is really working. But then, if you keep going and they're still on that drug and you start to see kidney deterioration, it's not due to the drug. It's not hard to imagine that ACEs and ARBs will do this, but it's probably something else going on. But I'll tell you, one of the mechanisms for sure how a worsening kidney function is associated with worse mortality is that if the serum creatinine goes up, people stop the drugs. And so if you stop the drugs, that's going to be bad. And so I'm certain that part of it is that people get scared by the elevated creatinine and stop the drug, whether rightly or wrongly-- so interesting set of relationships. We have a lot to learn and one thing is sure. We can figure it all out by looking at serum creatinine. So this was a review by Brisco and Testani and looked at a variety and just demonstrated or tabulated all of the different biomarkers that are available and useful in this condition. And so there are the markers of GFR in kidney function. Particularly, obviously, creatinine is and cystatin C also is and maybe a more accurate indicator of what's going on with kidney function than is serum creatinine. And then, there are markers, importantly, of tubular kidney injury. And more work has to be done with this to dissect out which patients have functional derangement and which patients actually have kidney injury, NGAL being the one that's been most studied to this point, but there are others coming along. And then, there are these markers that they title neurohormonal activation and sodium avidity, meaning that they're diuretic unresponsiveness. And that's the BUN creatinine ratio, which in the setting of high creatinine or low GFR has a bad prognostic indicator and then something called diuretic efficiency, which says how much do you urinate in response to a given dose of a loop diuretic. And then, these are just general related heart failure biomonitors. So we have a growing toolbox here and going forward, we really need to employ these more to keep trying to understand what's going on. So now, let's just move to therapy. And there are not going to be any clear answers. Just to lay the groundwork, in chronic heart failure, at least in HF-REF-- it sort of rolls off the tongue, doesn't it? HF-REF, HF-PEF-- anyway, but in HF-REF, we have chronic. We have all these drugs and we know what we're doing because there have been 100 trials you with a number of drugs, a new one just about to be handed to us maybe, LCZ696 drug-- but so many trials showing if you give this drug to a population, the population will do better. We don't have anything like that in HF-PEF or in acute heart failure. Here are the drugs that are approved, this one just in Europe. And this was the regulatory basis for the improvement, always hemodynamic, with the one little caveat that nesiritide was helped along the way by a three-hour differential in a dyspnea score, which was gone by 24 hours, I think. But here, we can say, well, there was something clinical associated with the regulatory approval of nesiritide, but clearly no benefit of survival outcome demonstrated with any of these. And in some cases, in the case of milrinone and the case of Dobutamine, the possibility is that the possibilities have been raised along the way perhaps worsening mortality. Some of that's probably true. Some of that is probably spurious. So not a great picture-- I guess what I tell the house staff and the fellows, look, we've got to give them these drugs, these sort of semi-poisons that don't do the heart any good to get them over the hump so that we can give them the drugs that actually make them live longer, get to that point. That's the goal of treatment and using these drugs in simple terms. So a lot of studies going on now in acute heart failure, a lot, and we still haven't figured out how to do them. But I want to point out the NHLBI-sponsored Heart Failure Network and to point out a very important point from this slide. This was Heart Failure Network 1.0, the first round, the first five years. And then, there was a re-application for another round which we're in the midst of. It's lasting seven years and I think Doug'll agree that HF Network 2 is so much better than HF Network 1. And there are two reasons for that. One is that Tufts University joined and WashU joined. So it has to be better and maybe there's some relationship there. I don't know if it's coincidence or not. We weren't talking, but the brain waves were going back and forth. Anyway, that's the-- oh, the other thing interesting about this is, look. There's sort of this right upper quadrant deviation of these sites-- absolutely nothing west of St. Louis. St. Louis is the dividing line between heart failure research and nothing going on. And so I'd be proud of the fact that I'm in the furthest frontier of relevant clinical research. There's just no point in going any further. So let me point-- a lot of the attention of the network has been placed toward acute heart failure. The first paper to be published from HFN1 was the dose trial that Mike Felker ran from Duke and was a factorial design, high-dose versus low-dose furosemide and continuous administration versus pulse administration of furosemide. And I have a problem with the design because we don't give loop diuretics this fixed dose. We're doing it and titrating the patient constantly. So it's a tough design. But nevertheless, a bit more interesting part of the question to me was continuous infusion versus pulse therapy. But neither of those factorial analyses showed anything positive. There was no improvement in the primary endpoints, here relating to patient global assessment. It's a little unusual way of displaying it in the area under the curve, but just this little area here represents a slight difference between the two strategies, in this case, high-dose versus low-dose, but basically nothing going on. There's no guidance here. To me, I think I'm intrigued by the fact that there was no differentiation between a continuous infusion and pulse because you could make reasonable arguments for both. I was very much looking forward to thinking that there would be a winner here, but there wasn't. In terms of kidney function-- so as we diurese the patient, kidney function can get better. I ask the house staff on rounds, as we diurese patients with heart failure, what's going to happen to kidney function? And somebody says, it's going to get worse. I say, right. Any other answers? Somebody's going to get better. Right, because we don't know what's going to happen. And I guess simplistically, the more degree of right heart failure-- and there is evidence for this-- that more degree of right heart failure at baseline, the more likelihood of the renal vein pressure being a key pathophysiologic element and greater the likelihood that if you believe that, that renal function will improve. The problem is it takes a long time to do that. If you have a patient with anasarca, you're not going to begin to see that until maybe you've removed the 20 kilograms worth of extra vascular fluid that the patient has. So it's so complicated. But anyway, if you looked at low-dose, high-dose, continuous infusion, or pulse therapy, there really was no difference. You look at the units here before you get excited about these differences in the bars-- really no difference between the two that was discernible and changes in kidney function, either. So not a blockbuster trial, but worthwhile data. The next study in this area that was published is the CARRESS-HF trial from the Heart Failure Network that looked at ultrafiltration. And it randomized patients to ultrafiltration if they were admitted to the hospital for acute heart failure with worsening renal function and persistent congestion, either continued pharmacologic care or ultrafiltration. And here's the answer. So it's a little bit a complicated slide, but the primary endpoint was an integrated analysis between change in body weight and change in serum creatinine. And don't ask me about the statistics here, but here was where the point fell with pharmacologic therapy and here's where the point fell with ultrafiltration. That is to say, you've got about the same degree of fluid removal in the same overall period of time, but ultrafiltration was actually associated with worse creatinine kidney function than was pharmacologic therapy. So it's a positive-negative trial, right? It's a small p value. So it's not a neutral trial. It's against the hypothesis that was being posed that we would see the opposite, but it's interesting. This has kind of killed the ultrafiltration business to a great extent. my own opinion about this is that this has everything to do with the rate of fluid removal. If you look at the time course of fluid removal with ultrafiltration compared to diuresis, it was very rapid early and then it sort of leveled off. And my own thinking is that-- and not really only my own thinking-- it really has to do with matching the degree of fluid removal to the plasma refill rate, because the first thing you're trying to do is eliminate extra vascular fluid and so you don't want to suddenly deplete intravascular volume. So if you go too fast, that's what's going to happen because you're going to exceed the plasma refill rate. And so I don't know whether this would have worked differently if a different algorithm for how the ultrafiltration was used had happened. But be that as it may, these were the results. The third study to point out-- recently-- well, not so recently anymore-- published the ROSE-AHF study, which was the theory being that low-dose nesiritide and/or low-dose dopamine would improve kidney perfusion, improve parameters of kidney function while making the patient better or allowing the patient to get better. So this was acute heart failure, estimated GFR between 15 and 60 and this double randomization. That is to say, all patients entered and then they were randomized to either go into the nesiritide treatment part of the study or the dopamine treatment part of the study. And then, they were randomized into the drug versus placebo-- a little complicated design, but I think a very good one. And this was the result-- neutral. It did not demonstrate what the hypothesis was that everybody was thinking. I certainly felt from other data that weren't as good and my own experience-- always a problem-- that low-dose dopamine had a role in this. I did not believe that nesiritide would do anything. And here's the dopamine results, the primary endpoints being 72-hour urine volume, changes in cystatin-C as the renal marker, but absolutely no difference between the dopamine and placebo group. And guess what? Absolutely no difference with nesiritide either-- so again, a neutral trial, disappointing. But now I'm going to show you subgroups. So close your eyes. This is not important because subgroups are rarely important but they're fun to look at. So if you look at it and look at the subgroups within the dopamine analysis, there's one thing that pops out as a statistically significant treatment by subgroup interaction. And that is what the ejection fraction was, because patients with normal ejection fraction and patients with low ejection fraction were both allowed in. And it turned out that the ones with the preserved ejection fraction did very poorly with dopamine and the ones with low ejection fraction trended in the right direction. And again, a mystery-- what is it about this population that confounds us so? Again, if you ever invite me back, we can talk about that. But maybe we just missed it, because there's something about this population that screwed us up. But who knows? We have to do another study to look at it. I'll make an admission. I'm still using low-dose dopamine in certain circumstances but I'm just a smart clinician and I know I'm right, regardless of what the data show. So those are, I think, important contributions. We're very anxious to see some really blockbuster positive trials coming out of the network. We haven't quite seen that yet, but we've learned an awful lot as we've gone and we trudge ahead with a number of very interesting studies both in acute heart failure, chronic heart failure, have HF-REF, and HF-PEF. We've been very interested in vasopressin receptor antagonists. Vasopressin is another neurohormone that's subregulated in heart failure. Elevated vasopressin levels are associated with adverse outcomes. So why not try to block it? And the focus has been blocking the V2 receptor, which is the renal tubular receptor that influences prewater retention. And so if you block it, you get what's called an aquaresis, a removal of free water. This is an early study that John Burnett did to tie this into renal function. So he took this cohort and he gave half of them furosemide and half of them tolvaptan, I think trying to match the amount of urine output, and found that there was a difference in the renal hemodynamics, that the renal hemodynamics actually trended favorably with tolvaptan and unfavorably with furosemide in an equally potent dose from the perspective of fluid removal. So there's a little hint that maybe there's a preferential benefit for tolvaptan over conventional loop diuretics, something that has not really been borne out well by subsequent investigations. But that was an early thinking and there may be some truth to it. So this is the big trial that we did, the EVEREST trial and I'll just show you a couple of slides from it. This is the most important finding. Unfortunately, it was a secondary endpoint, not a primary endpoint, but the most obvious positive finding, which was dyspnea by a seven-point Likert scale and looking at the percent of-- and there were two different identical trials to meet FDA regulations for approval characteristics. So at the end of the tolvaptan one at 24 hours, there was about an average of a 0.8-kilogram greater fluid reduction in one day compared to furosemide alone. That was background therapy. And that was associated with statistically significant improvement, relative improvement in dyspnea versus placebo. People have looked at this slide and said, oh, there's nothing here. I look at this slide and say, wow, we got a significant result with dyspnea. So those of you who work in this field know that the endpoints are very messy with enormous amounts of noise and it's very difficult and a lot of changes in background therapy along the way. So my interpretation is that there really is something going on here. And as I'll show you, we're doing a study now to see if that can be replicated. The longer-term outcomes-- so patients were treated in the hospital and it continued for immediate follow-up at about nine months. And these were the dual primary endpoints for the combined long-term trial of the same patient population, an absolute overlap for all-cause mortality and for CV mortality or hospitalization. Actually, well, you can't go back. But the good news there-- there is good news on that slide. It didn't kill people. Now, you laugh, but that's real important. And there's a suspicion that a lot of the drugs that we use now are associated with worse long-term mortality. And so that's a key issue for the FDA and a key issue for clinicians. For my money, if I had a drug that I knew augmented relief of congestion, improved dyspnea to a significant point early on, and I knew it didn't kill you, I'd probably use that drug over some of the drugs that we have now. But let me be clear. It's not approved. This drug is not approved at this time for treatment of heart failure. It is, of course, approved for the treatment of severe hyponatremia, including in the high heart failure population severe or symptomatic hyponatremia. Close your eyes again, because this is a post-hoc subset analysis, which is worse than a regular subset analysis. Nevertheless, it's always fun, hypothesis generating. So when we looked at patients with very low serum sodiums-- and in this population, there were really relatively few patients with serum sodium less than 130-- a signal between tolvaptan and placebo appeared to emerge and particularly focusing on the time to cardiovascular mortality and morbidity. It very slightly reached nominal statistical significance. But notice, actually, the point estimate is a pretty impressive treatment effect. The problem is there are so few patients in this group. Certainly hypothesis generating, that if we focus our attention on those patients who are likely to have elevated vasopressin levels, maybe we'd see more benefit. It's not-- there we go. I mentioned we're doing a small study of about 250 patients. We're, I think, about 2/3 through with enrollment. We call it the SECRET of CHF trial. Don't try to figure out what that stands for. But my colleague, Jim Udelson has been trying to use this name to name a heart failure trial for a decade and we finally let him do it and he figured out what it stood for. So that's all you got to know about the name. But the SECRET of CHF trial, a multicenter, randomized, double-blind, placebo-controlled trial looking for just 250 patients, randomize a patient to tolvaptan, 30 milligrams a day, which is the same dose that was used in EVEREST, and focusing on dyspnea in patients hospitalized with worsening heart failure. So we did two things differently here. One is although it's multi-site, it's really a fairly well-controlled multi-site. We're not outside the United States. We really focused on investigator training and nurse coordinator training. We have a specific set of questions that are being asked during the question about how they fit on the Likert scale. You can go into that in detail, but try to eliminate some of the noise of this measurement. That's one thing we're doing. And the other thing we're doing is trying to enrich the population with patients in whom there is some rationale for benefit with this drug, including patients with hyponatremia, renal insufficiency, or inadequate initial diuretic response. So we'll see how this comes out. The primary endpoint is dyspnea. And the last drug I'll mention is serelaxin because this could be really good, based on the results of the RELAX-AHF trial. So this was a study performed in patients admitted to the hospital with acute heart failure within a short period of time after hospitalization. And they had to have a blood pressure greater than or equal-- I don't remember-- greater than 125 millimeters of mercury systolic. So the mean blood pressure in this population was in the 140s. So just keep that in mind. That's the population we're talking about here, which I think is a population that may be, as you can imagine, particularly prone to benefit by vasodilators. But this is serelaxin. So it's a hormone upregulated to a great extent during pregnancy. It's an active vasodilator. It increases renal blood flow. It's also been shown-- I don't know the data directly, but what they say is it's been shown to be antifibrotic and anti-inflammatory, as well. So those may play a role. The primary endpoint of the RELAX-AHF trial was dyspnea measured two different ways, co-primary endpoints. It hit one of them. It didn't hit the other. The one it hit is the visual analog scale of analysis looking at the area under the curve and this was a highly significant finding. Hard to figure out from this particular metric how much of an effective this is-- that's where the Likert scale, I think, has an advantage. But it was positive trials, a positron trial. One of its primary endpoints was positive. So we can keep looking. One of the things we see when we keep looking is that the serelaxin group was associated with fewer renal adverse events, substantially fewer renal adverse events, 4.6% versus 8.6%. So there may be something here about it being renally protective. We have to learn more. The other thing is this popped out. It really popped out because this was not a prespecified endpoint. The prespecified secondary endpoint, I think, was 60-day combination mortality and hospitalization and it didn't hit that with actually a numerical worsening of hospitalization rates in that period of time for some bizarre reason. But when they kept digging and they came out with this endpoint and they demonstrated that it looks like if you look at 180 days and you look at all-cause death, it looks like something might be going on. This is really interesting. It's not definitive because I can cite several examples of apparent mortality benefits that popped out of trials that were not designed to look at mortality. And when they were replicated, it was not borne out, a lead versus a lead two, promise one versus promise two, first with losartin, the second with amlodipine. Both are examples of that. But you've got to go on. You see a finding like this, you go on. And so these investigators have gone on and are looking at cardiovascular mortality as the endpoint in a much larger study and seeing if this can be replicated, which would be real exciting in patients with blood pressure over 125 given this vasodilator, serelaxin. It certainly would be the first drug to demonstrate this kind of finding. In the last two or three minutes, I want to transition to look at heart failure differently, as a problem for us in continuity, a problem for the patient in terms of quality, a problem in terms of cost. And so the patient in the hospitalization event is a bad predictor for mortality, as I showed you, but a very high rate of readmissions. Now, readmissions have become very important to the federal government in CMS. And the reason is that they're expensive. That's the reason. The reason that we have the 30-day readmission endpoint is because readmissions are expensive. So across all diagnoses in a Medicare population, there are-- I don't know what year this is-- about two million patients re-hospitalized within 30 days of discharge per year among the Medicare beneficiaries. It represents 19% of Medicare index hospitalizations, a high number. Annual estimated cost to Medicare, $17.5 billion-- that's worth building a road somewhere or something. You can do something with this. Build another ship. So let's try to reduce this. A really noble-- no, it really is worthwhile to reduce it. I'll talk more about the endpoint in a second. I will say that we have the tools to try to do this. This is the Cochrane analysis that I referred you to. It says 2011, but this was updated in 2012. And it's a meta-analysis of disease management programs divided into two types, telephone-based and a more technology tele-monitoring-based. And in either case, looking at this meta-analysis of numerous trials looking at heart failure hospitalization or all-cause hospitalization, significant benefits-- look at the hazard ratios here, more, understandably, for heart failure hospitalization than for all-cause hospitalization, but pretty impressive hazard ratios for reducing hospitalization. But get this, it seems to reduce mortality. So if you pull enough patients together and really look at it-- of course, meta-analyses can be flawed, but it really looks like a disease management-based approach to patients as they leave the hospital could really work and not just for the 30-day readmission, but for more of the benefit of the patient. So we have things that we can do. We can build on this and we have to use this. We're presently doing a study. I'm doing a study with Jenica Upshaw, who's one of our senior fellows and we got a grant from Verizon to study their tablets and use their tablets and software that's provided to us by Bosch. And with a three to one randomization, we're conducting a randomized disease management study to see if this higher-tech greater opportunity for communication and education approach will improve over more conventional disease management. So I said I had to give a dig to the 30-day heart failure hospitalization. And so here it is, a picture of the common housefly, Musca Domestica. And I wrote this as an editorial response to a paper that was published in another journal about impacting the 30-day readmission. And I got fed up with this and I said, I'm going to write something about it. So I Googled a lot and I found out that the common housefly, according to this, "Once the adult housefly hatches from the pupal stage, it has an approximate lifespan of 15 to 30 days." and if you don't believe me, there's a very reliable reference, kidzworld.com. And if you read my paper, this is referenced. Here, I give my references. And so I kind of say, it's great for the housefly because 30 days is really important for a housefly. But how important is it for the patient? Why are we just focusing on that 30 days? That's one of the problems with it. The other problem, which I think is a bigger problem, is that the 30-day heart failure readmission metric is not a quality metric. It's a cost metric or a deficiency metric. And the reason I say that is, oops, they forgot about death. If we were putting a clinical trial together, we would never discard the competing risk of death. We'd be shot out of the room by every statistician alive. But this 30-day readmission metric is just readmission. If you die, that's fine. So if we really wanted to do a quality metric, we would have to integrate mortality. And actually, there have been several analyses done that show an inverse correlation between death and re-hospitalization over time hospital by hospital, interestingly. And it creates perverse incentives, that 30-day. There's really a problem with that 30-day window, because that's all you care about. And so if you see a patient in the office and you care about the 30-day and there's two weeks out from the hospitalization and you notice that they're not on a beta blocker and they're not on an aldo blocker, on that two-week visit, are you going to start a beta blocker? Are you going to start an aldo blocker? Or are going to say, no, they're pretty stable. I'm going to wait. And maybe they'll never get on those drugs. So I think that there's sort of an perverse incentive. Anyway, I think these metrics that are handed to us by various folks, including the federal government, we have to get past this. And the answer that I say is, give us the money. Let's go into bundled payment, capitation, integrated delivery systems. So give us the money and let us drive efficiency and quality. Yes, publish our quality metrics, but don't give us financial penalties. Let the quality work in the marketplace. So for higher quality, people will choose us. If we're lower cost, people will choose us. So I embrace that. Damn the torpedoes. Let's transition and see if we can get past the yoke of government-based penalties for these different metrics. So conclusions-- heart failure is a major and growing worldwide health problem. Hospitalization rates are growing. This now is actually not true anymore. It's trending down now. The hospitalization's a marker of increased disease severity. Clinical trial evidence in acute heart failure lags behind that of chronic heart failure. Attention should be focused on important sub-populations, such as those with renal impairment and hyponatremia. Cardio-renal syndrome contributes importantly to the clinical expression of acute heart failure and therapy should be directed toward addressing this aspect of the condition if we could find any. And so that's a problem and so further investigation is ongoing to explore newer vasodilators, inotropes, and potentially renal protective agents. But there's a paucity in terms of direction. So thanks very much for having me.
