|Online Slide/Audio Symposium|
Nordic Nephrology Days |
University of Lund, May, 1997.
Nils Alwall Lecture.
Urea Kinetics in Hemodialysis. Practical Implications of the Regional Blood Flow Model
Part Two of Two
Dr. John T. Daugirdas
Dr. Daugirdas is Professor of Medicine at the University of Illinois at Chicago. He is the author of numerous papers on hemodialysis hypotension and hemodynamics, urea modeling, and dialysis adequacy, and is an editor of the "Handbook of Dialysis".
|Listen to this Lecture using the Real Player|
|You can scroll
through the talk while listening to the audio. You won't be
able to listen to these files
if you have not downloaded and configured the Real Player
version 5.0 (earlier versions
will not work!).
The three phases of urea rebound
One issue that has concerned people about this is the effect of urea rebound. The observation that has been made that if we draw the post-dialysis BUN in some patients who have access recirculation, there may be a marked increase in the post-dialysis BUN within the next 15 or 20 seconds. There is maybe a further increase in about 1 to 2 minutes, and the last increase takes place over a time course of 30 to 60 minutes. there are three phases of urea rebound that occur after dialysis.
The first phase occurs only in patients who have access recirculation. Now in the past, we used to think that access recirculation was occurring in many patients; now we know that access recirculation occurs in only about 5 to 7 percent of patients on average. Of course, it depends upon the blood flow rate and the vascular access. It is higher in venous catheters, especially femoral venous catheters. This is a vascular access of blood flow going this way. Normally blood going through the dialyzer is filtered and none of the outflow blood will go back retrograde through the access that goes on to the heart. In the case of access recirculation, you can inject some saline here, and some of this will go back in and dilute the blood coming into the dialyzer. It is this dilution that has an important effect in terms of urea modeling.
Dead space in the arterial bloodline and its importance
The devil is in the details, and it's important how one draws the blood. If one looks at a dialysis or a set up, you have a needle with a dead space of about 2.6 ccs and usually there is a dead space of 7 or 8 mls before you have the arterial sampling port, which is where blood is drawn from. If we add up this total dead space, it's usually about 10 ccs.
Why and when recirculation occurs
At the end of dialysis if we look at the blood flow going full speed, if there is access recirculation, then the blood in the arterial line is diluted relative to the blood in the patient. So if we just put a needle in here and sample the blood at the end of dialysis, maybe the patient's post-dialysis BUN is 20, but we may only get 16 here because of this access recirculation. If we then compute either a urea reduction ratio or Kt/V, we will overestimate the amount of dialysis.
Now some technicians think it is okay if they stop the pump and draw the blood. The problem with that is that it is just like a freeze-frame video. When you stop the pump, all you've done is you've frozen in time exactly the same situation that was present several seconds ago when the blood pump was going full speed ahead. And so you still have this blood line that is filled with recirculated diluted blood. In one of the units that we worked with, this was what they were doing. It had various severe consequences because in some patients, therapy was being dramatically overestimated.
The basis for the slow flow technique of drawing post-dialysis blood The correct way to stop this is: The reason you have recirculation is because the blood pump is drawing more fluid than is going through the access. Now the average access has a flow of about 1 liter per minute (Depner and Krivitski, ASAIO J 1995;41:M745-M749). And usually the blood pump is set, even in the United States, it's rarely higher than 450 ml/min. Some people use 500 ml/min. So under normal circumstances whenever the access blood flow is higher than the blood pump flow, you get no access recirculation. However, some accesses deliver only 400 ml/min. This can fall further in the course of dialysis. Once the pump starts to pull more than the access can deliver, it starts to pull retrograde. So to stop access recirculation, all you have to do is lower the blood pump flow rate to 50 or to 100 ml/min, such that you no longer exceed the access blood flow rate. Then all of the blood, once again, starts to come from the upstream part of the access. Then, once again, the blood line begins to fill up at this reduced blood flow rate with blood that has the same concentration as the patient. The proper technique now is to slow the pump to 100-120 mls per minute and for about 10 seconds (Kapoian et al., J Am Soc Nephrol (abstract). 1996;7:1409) ... we said (50 ml/min for) 20 seconds -- it appears that now the correct time is closer to 10 to 15 seconds. Then stop the pump. Now you have blood in the blood line that reflects the actual blood that is in the patient and you will not overestimate dialysis in this way.
The importance of access recirculation and proper post-dialysis blood draw technique This is a complicated slide. I didn't have the right slide I wanted to show you. We screened 400 patients for access recirculation, and we found that 27 of these 400 had severe access recirculation. In these patients the blood was being drawn after simply stopping the pump. Many of these patients had severe overestimations of their dialysis therapy. Their actual dialysis therapy was in the range of Kt/Vs of less than 0.8 in many cases.
The next rebound is due to cardiopulmonary recirculation. This is an interesting phenomenon. It's a minor effect. It is due to the fact that during dialysis an arterial venous gradient develops. The size of this gradient depends upon the efficiency of dialysis. For low efficiency dialysis it's about 5 percent; for high efficiency dialysis it's about 8 percent. Depending upon where the dialyzer is, cardiopulmonary recirculation will exist. If you have a venous catheter, you have an arterial venous gradient, but you do not have cardiopulmonary recirculation. Cardiopulmonary recirculation only occurs when the dialyzer is being fed from an arterial access.
Now it's a little bit complicated... this word cardiopulmonary... and also to understand why is this a recirculation? If you think about it, you just have to look at the way a dialyzer is connected to the body. What does recirculation mean? Recirculation means that blood leaving the dialyzer goes back to the dialyzer without first going through the tissues to pick up more urea. That can be followed as a definition of recirculation.
Here you have the arterial circulation of the heart up top, the venous circulation on the bottom here. This is your access loop. You can see that when a dialyzer is hooked up to the arterial side of the circulation, some of the blood going from the dialyzer goes through the heart and then goes through the tissues, picks up more urea, and then goes back to the dialyzer. This is non-recirculated blood. However, a certain proportion of the blood will go from the dialyzer, go through the heart and then not go through the tissues but just go right back through the dialyzer.
So in a sense, we have a short-loop access recirculation here. The way the dialyzer is hooked up to an arterial access makes the possibility of long-loop recirculation as well. And that's why we call it recirculation. We call it cardiopulmonary recirculation because instead of recirculating through the access, the blood recirculates through the heart and lungs, which after all have very little urea to remove, and it goes back to the dialyzer.
Effect of cardiopulmonary recirculation on the timing of the post-dialysis blood draw
The interesting thing is that what happens after you stop dialysis, the arterial venous gradient dissipates because you're no longer... what causes the arterial venous gradient is that you're feeding the heart with cleared blood and this causes the arterial blood urea to be lower than the venous blood.
The practical implications are that within 1 to 2 minutes of stopping dialysis, you no longer feed to the heart cleared blood, and this arterial venous gradient rapidly closes, and so you have a second rapid rebound that takes place in a 1- to 2-minute period after dialysis. A lot of people have a hard time understanding this, but if you think about it a little bit and think about why we came up with this term and why we call it this, it will become clear to people, hopefully.
Again if the dialyzer... assuming that you have a dialyzer that's hooked up to the venous circulation, that it was hooked up here--in this case, all of the blood leaving the dialyzer goes through the heart and through the tissues and comes back to the dialyzer. There is no way for blood that leaves the dialyzer to get back to the dialyzer without first passing through the tissues. This is why with the venous access there is no such thing as cardiopulmonary recirculation.
Compartment/regional blood flow effects
The third effect is the standard rebound effect that is called a compartment effect, and everyone has heard of 1 pool and 2 pool. What does this mean? In the talk we had a 1-pool model. We had a lot of urea at that start; at the end of dialysis, your urea concentration has been removed, your red has become pink, and the concentration is equal in the pool. If we look at the decrease of the BUN over time, we had a nice monoexponential fall. If we plotted this on logarithmic paper, we would have a line.
Now the problem comes in that we have a constriction in the box. The idea is that we have urea that is being sequestered in some parts of the body, and we don't know exactly where that is. Now what happens is that at the end of dialysis and early on in dialysis, we have a gradient that develops. Unfortunately these colors have faded a little bit, but there is less urea early on in the proximal compartment and more urea in the distal compartment, such at the end of dialysis there is a dis-equilibrium and after dialysis, urea will pass from the sequestered compartment to the more open compartment, and you will get a rebound.
Intradialytic urea inbound and postdialysis urea rebound are due to the same cause
Now the key thing that most people don't understand is that this constriction not only causes a post-dialysis rebound, but it causes what I call an intra- dialysis inbound. What this does is it makes the average BUN concentration during dialysis lower than you would predict if there were no constriction in the box. Why is this happening?
Because the dialyzer is dialyzing this accessible compartment, which is smaller than this large compartment, so it's rapidly reducing the concentration that it sees. So the efficiency of urea removal is reduced. Even though the dialyzer is clearing the same amount of plasma, the amount of urea removed is the clearance of the dialyzer times the mean plasma concentration. So if there is no constriction, the mean plasma concentration is higher.
With the constriction, the mean plasma concentration during dialysis drops, and as a result, you have less urea removed than you think. So the constriction causes both the inbound and the rebound.
This, in fact, is the basis of the Smye technique measuring the inbound, you can predict the amount of rebound because the more constriction there is, the more inbound you will have and the more rebound you will have. Both are caused by the same problem, this constriction.
Urea sequestration in the standard model -- intracellular
The standard model, and again forget the equations, is due to... assume that the sequestration was due to the intracellular pool and there was some constant, this intracellular mass-transfer coefficient, that was the rate of clearance from cells to the extracellular pool, and then there was this other clearance, which was removal from the extracellular pool. These boxes are simply set at two-thirds and one-third of the body weight. This was the standard model.
The basis for the regional blood flow model
the standard model has no basis in physiology. If you look back at the old physiology literature, which we did, there is much more evidence that the removal of small solutes from the body is flow related. And so what we did, and other people have thought of this earlier, is we quantified the model that we call the regional blood flow model.
The basis of the regional blood flow model is the hypothesis that, the main barrier to urea transport is not across cell membranes, and the main area of urea sequestration is not in cells but in certain body organs. It is due to the fact that if you add up ...look at where the urea is in the body--the urea is in the same place where total body water is. Eighty percent of the total body water is in muscle, skin, and bone. So 80 percent of the urea are in these organs. But if you look at the flow from the heart, the flow pattern is reversed. During dialysis, only 15 to 20 percent of the cardiac output comes to muscle, skin, and bone. Most of the cardiac output goes to visceral organs--organs that have not too much total body water; they have a high flow.
During dialysis it's not enough to just remove urea from the dialyzer--you have to bring urea to the dialyzer to get urea out of the body. And the problem that you have is most of your urea is in a bank with a very small road to it. So our theory was that at the end of dialysis, the urea is sequestered not in cells but in certain organs. Again we could model this mathematically with a high-flow pool and a low-flow pool.
We also had a capillary transport coefficient in the model, permeability coefficient. However, this was pretty high. The mathematics turned out to be that the flow was the main factor that was limiting urea transport in this model.
This is not the case for creatinine and certain other sites where diffusion does become a problem. This general model can be applied to other solutes as well.
When we set up the model, we wanted to set it up not based upon some fictitious numbers. We went back to the physiology literature. These are old studies done in the 30s and 40s where they would take organs and cadavers and chop them up and desiccate them and see in which organs they had total body water and total body water potassium.
We added up on one hand the body water fraction. Again you can see that most of the body water fraction is in the muscle, skin, and bone. This is blood, bones, and skin. But if you look at where the flow is, most of the flow is to the kidneys, which is not a factor; but most of the flow is to the small organs--the heart, brain, portal system, and lungs. All of these organs contain very little water and very little urea.
Here you can see that the actual flow to muscle is a small fraction of the total pool. An interesting thing is that in patients, of course, with hypotension or with vasoconstriction, that the flow to muscle, the place where urea is, is reduced even further. So this model does have certain physiological implications that are quite different from the intracellular model.
This just shows that once we made all the mathematics, we could take a patient and look at the rebound. The initial rebound here was access recirculation. This rebound was due to equilibration of the different flow components. We could predict it quite well with the mathematical model.
The first part is Kt/V is proportional to urea reduction ratio. If we take the blood right away, without clearing the dead space in the line, we miss all three phases of rebound--access recirculation, if it exists; cardiopulmonary recirculation because we are sampling arterial blood, not venous blood--we're sampling while the AV gradient is still in place; and we are missing this two-pool rebound, which is due to the regional blood flow effect.
If we draw the blood after 15 seconds' slow-flow period, we will get rid of access recirculation. This is extremely important. This is probably clinically the only important thing to do. You will miss this other rebound, but you can compensate for this, as I'll show you. If you draw two minutes after dialysis, you will then correct for two of the rebound components, but you will miss this regional blood flow component.
The proper way to do this is to draw the blood 30 to 60 minutes after dialysis. Then most data suggests that all of these body compartments have re-equilibrated and you get what is called an equilibrated Kt/V.
Use of the regional blood flow model to predict urea rebound
One of the beauties of the regional blood flow model is that we can now use it to make predictions. Using average values of cardiac index and access blood flow, we wanted to see if we could predict the amounts of rebound in dialysis patients.
We found that the amount of rebound here, delta Kt/V, was equivalent to the rate of dialysis. That was simply Kt/V divided by time. This looks complicated, but I'll show you... the regression equation we got was 0.6, -0.6, plus about 0.03. I'll show you how this works in the next slide.
The "rate equation" to predict post-dialysis urea rebound
Basically what we found was that the single pool Kt/V, unequilibrated minus the equilibrated...this delta business... this is the rebound. And this delta was 0.6 times the rate of dialysis minus 0.03.
So say for example you're giving a Kt/V of 1.2 in three hours. The rate of dialysis then is simply 1.2 divided by 3, which is 0.4 Kt/V units per hour. So you multiply that by 0.6. You get 0.24. Subtract 0.03; you get 0.21. So your rebound is 0.21.
Now say in another patient you're giving the same dialysis rate in four hours. So now you're giving 1.2 Kt/V units in four hours. Your rate of dialysis is .3 per hour. Six times 3 is 18; subtract 3, that's 15; divide by 100. So your delta Kt/V is 0.15. The regional blood flow model allowed us to come up with this very simple formula for urea rebound.
Validation of the rate equation in the NIH HEMO Study dataset
This is data from the NIH HEMO study. In the Hemo Trial we were very worried about urea rebound. We measured the urea concentration 30 minutes and in some cases 60 minutes after dialysis. We compared the equilibrated Kt/V based upon this delayed sample with an early post-dialysis sample with a rate adjustment with this equation. And you can see that both in the standard-goal group and in the high-goal group that there is a very nice correlation, and the rate adjustment does correct for the amount of rebound, and it allows you to predict the amount of rebound.
There is still some effect of physiology. Here we took some patients who had a high cardiac output and a low cardiac output index. These were patients who were preselected for very high and very low cardiac index in a unit that was using a lot of vasodilators. We found that the rebound was... we could predict the rebound based upon the cardiac index. This is the observed rebound, the predicted rebound. In the patients with the higher cardiac index, they had a lower degree of rebound.
Here again you can see that based upon cardiac index, there is a nice correlation with rebound, expressed as delta Kt/V.
In conclusion, you can follow urea red uction or single pool Kt/V. There is no evidence that one is better than the other in terms of patient outcomes. I didn't have the time to talk about modeling in terms of the information you can get from following the urea distribution volume. I also didn't have time to talk about the advantages of looking at PCR (protein catabolic rate) and G (urea generation rate). That's for another lecture.
One needs to realize that rebound occurs and know that it is basically a function of the rate of dialysis and that an average is going to be about 0.20 (Kt/V units) for fairly rapid dialysis, less in a less rapid dialysis. You need to take special care in how you draw the post-dialysis blood to avoid access recirculation effects. This is the most important thing, and you can seriously overestimate therapy if you don't do this.
Blood vs. dialysate side modeling: Why there's a difference
You can do either blood or dialysate-side modeling as long as you correct the Kt/V properly. Again, from this inbound what happens is the reason single pool kinetics underestimates the amount of urea removed is, because you multiply the clearance times the average BUN level during dialysis to get the amount of urea removed. And you think it's here (upper line), but it's here (lower line due to urea inbound).
Blood vs. dialysate side modeling: You can correct for the discrepancy
There is data from the NIH Hemo Study where we compared urea removal indices with corrected Kt/V on the blood side, and there was no systematic error between the two measurements.
As a final slide, the regional blood flow theory predicts post-dialysis rebound fairly well, as does the standard two-pool model. Because the magnitude is small and fairly reproducible at any level of dialysis efficiency, these effects are of interest but they're not of great clinical importance.
1. Depner TA and Krivitski NM. Clinical measurement of blood flow in hemodialysis access fistulae and grafts by ultrasound dilution. ASAIO J 1995;41:M745-M749.
2. Kapioan T, Steward CA, Sherman RA. Validation of a revised slow/stop flow recirculation method. Kidney Int 1997; 52:839-842.
3. Daugirdas J, Burke MS, Balter P, Priester-Coary A, Majka T. Screening for extreme postdialysis urea rebound using the Smye method: patients with access recirculation identified when a slow flow method is not used to draw the postdialysis blood. Am J Kidney Dis 1996:28:727-731.
4. Schneditz D, Kaufman AM, Polaschegg HD, Levin NW, Daugirdas JT. Cardiopulmonary recirculation during hemodialysis. Kidney Int 1992; 42:1450-1456.
5. Smye SW, Dunderdale E, Brownridge G, Will E. Estimation of treatment dose in high- efficiency hemodialysis. Nephron 1994; 67:24-29.
6. Schneditz D, Van Stone JC, Daugirdas JT. A regional blood circulation alternative to in-series two compartment urea kinetic modeling. ASAIO J 1993; 39:M573-M577.
7. Schneditz D, Daugirdas JT. Formal analytical solution to a regional blood flow and diffusion-based urea kinetic model. ASAIO J 1994; 40:M667-M673.
6. Daugirdas JT, Schneditz D. Overestimation of hemodialysis dose depends on dialysis efficiency by regional blood flow but not by conventional two pool urea kinetic analysis. ASAIO J 1995; 41:M719-M724.
7. Daugirdas JT, Depner TA, Gotch FA, Greene T, Keshaviah P, Levin NW, Schulman G. Comparison of methods to predict equilibrated Kt/V in the HEMO Pilot Study. Kidney Int 1997 Nov;52(5):1395-405.
8. Depner TA, Greene T, Gotch FA, Daugirdas JT, Keshaviah PR, Star RA. Imprecision of the hemodialysis dose when measured directly from urea removal. Hemodialysis Study Group. Kidney Int 1999 Feb;55(2):635-47.
9. George TO, Priester-Coary A, Dunea G, Schneditz D, Tarif N, Daugirdas JT. Cardiac output and urea kinetics in dialysis patients: evidence supporting the regional blood flow model. Kidney Int 1996 50:1273-1277.
Back to Topic Index: