OPTIMAL HEMODIALYSIS zyxw zyxwvuts Optimal Hernodialysis-The Role of Quantification Laurie J. Garred, Bernard C Canaud and William G.McCready Depament of Chemical Engineering, Lakehead Uni.jers* (LJG)and McKellar Hospital (WGM) Thunder Bay, Ontario, Canada and Lapeyronie Hospital (BCC). Montpellier, France Mathematical modeling of urea dynamics is the most common method currently used to measure and deliver the optimal dose of hemodialysis for each patient. The modeling concept was popularized by Gotch, Sargent, and coworkers (1, 2) in 1974 and became known as urea kinetic modeling (UKM). Since then, hundreds of papers have been published by a myriad of workers, but two landmarks stand out: the U.S.National Cooperative Dialysis Study (NCDS)published in 1983 (3, 4) and Gotch and Sargent's proposal in 1985 (5) of KtIV as a measure of the quantity of dialysis delivered to each patient. Roughly a decade has passed since the final report of the NCDS and the formulation of KtIV, yet quantification of patient therapy is not universally practiced among dialysis centers. Whether and how to use UKM remain controversial. In 1994, we find ourselves at a promising crossroads in the field of dialysis quantification. Emerging technology for automated urea sensing and a shift from blood-based to dialysate-based measurements offer a realistic expectation of improved accuracy and less complexity. This article will review the application and limitations of classical UKM, and will include simplified methods for measuring KtIV and the patient's protein catabolic rate from blood concentrations. It will then focus on methods based on measurement of urea in spent dialysate and explore current developments in urea sensing that offer the prospect of automated UKM. ficiency. The biood urea nitrogen level (BUN)is therefore a good indicator of the general accumulation of protein metabolic products and has been adopted as a surrogate uremic toxin. Historically, the BUN has been used to judge the degree of uremia and to determine the appropriate dose of dialysis. Many clinicians continue to prescribe dialysis treatments solely by observing the BUN. At first, this approach seemed reasonable because clinical observations supported and the NCDS confirmed the concept that high BUNS were associated with poorer patient outcome. Patients maintained at a mean BUN of 100 mddl in the NCDS had a dramatically increased incidence of withdrawal from the study due to death or medical complications than the group whose BUN was held at 50 mg/dl(6). The critical importance of maintaining BUN Ievels within the guidelines outlined by the NCDS continues to be emphasized. Nevertheless, urea levels considered alone may be misleading. While a low or moderately elevated BUN may be the consequence of appropriate treatment in a wellnourished patient, it may reflect underdialysis in a patient ingesting lesser amounts of protein. Diminished appetite is commonly observed in hemodialyzed patients (7). Anorexia may compensate for underdialysis by reducing protein intake, causing the BUN to appear the same as in better nourished and better dialyzed patients (8, 9). For the patient who is neither markedly catabolic nor anabolic, net protein catabolism (PCR),as measured by urea appearance, is approximately equivalent to dietary protein intake (DPI).Inadequate DPI was shown in the NCDS to be the second strongest factor associated with patient morbidity and mortality (10). The generally accepted target range for dietary protein intake in hemodialyzed patients is 0.8-1.4 glkglday (1 1) with 1.1 g/kg/day commonly adopted as a specific target. A primary objective of UKM is to provide the clinician with an accurate measure of daily protein consumption to help counsel the patient regarding diet. PCR is estimated from urea generation assuming 6.49 g protein is catabolized to produce each gram of urea (12). The additional protein converted to other end products, lost in feces, or sloughed from skin is unmeasurable but is reasonably constant at an estimated 0. I7 g proteinldaykg dry body weight (BW): zyxwv zyxwvu zyxwvut zyxwvuts Urea Modeling Basic Theory While dietary carbohydrates and fats are metabolized entirely to CO, and H20that are eliminated by the lungs, many of the metabolic products of dietary protein must be eliminated by the kidneys. Much of the symptomatology associated with renal insufficiency is attributed to the accumulation of these metabolites of protein catabolism. Since every 6-7 g dietary protein contain 1 g nitrogen for which urea is the metabolic end product, urea accumulates in abundance in patients with renal insuf- (a, 236 zy zyxwvutsr zyxwvutsr zyxwvu zyxwv zyxwvu zyxwvuts 237 QUANTIFICATION G PCR = 9.35 - + 0.17 BW G in Equation 1 is expressed in mgfmin. Only after PCR is measured can the appropriate amount of dialysis be established to maintain the BUN within the specific target range for the patient. Disagreement persists regarding which measure of BUN should be targeted, time-averaged (TAC) or peak BUN. Depner (7) argues that TAC should be maintained within the 43-63 mg/dl limits established by the NCDS. The mechanistic approach advanced by Gotch and Sargent (5) focuses on the more easily measured midweek predialysis BUN which approximates the average of the three peak BUNS in a week. Their prescription map ( 5 ) . derived from the NCDS data, shows that predialysis BUN should be held between roughly 60 and 80 mgfdl. The upper limit is allowed to rise as high as 90 mddl for the patient with high protein intake (see Ref. 7 for further discussion of this controversy and other aspects of dialysis quantitation). Quantifying the Amount of Dialysis Delivered: K t N It is clear that average BUN in a patient will depend directly on the rate of generation (G) and inversely with the amount cleared by artificial or native kidneys. For patients with native kidneys functioning continuously with a clearance, K,.,,,, the BUN (0is expressed: (11). Equation I1 can be factored for V , the patient’s urea pool volume, to reflect the fact that G and K depend on patient size: (Ha). Equation IIa can be used to estimate the clearance required of a continuous dialysis therapy to maintain a patient with a DPI of 1.1 g proteidkgfday at an average BUN of 50 mg/dl. The urea pool volume in most patients (in 1) is between 50% and 60% of dry body weight (in kg). Taking a value of 55% and inverting Equation I, suggests that a patient ingesting 1.1 g proteidkglday generates urea nitrogen (UN) relative to urea pool volume (GIV) at about 0.17 mg/min/l. Substituting this and 50 mg/dl as the desired average C in Equation IIa implies that a KIV of about 0.343 ml/min/l is needed to maintain this C . Over an entire week ( = 10080 min) a total of 3.46 1 clearance is required for each liter pool volume. If the above concept is naively transferred to intermittent therapies such as hemodialysis, each session of length t will provide a clearance volume Kt which becomes KrIV when expressed relative to urea pool volume. To achieve the same 3.46 total liter clear- ance per liter urea pool over a week with thrice weekly hemodialysis would imply one-third of this total or KtIV = 1.15 per session. Equating continuous clearance requirements to an equivalent intermittent therapy is not theoretically valid. This exercise demonstrates how KtIV arises as a measure of therapy and that the value of KtIV = 1.0-1.3, commonly considered the goal for thrice weekly hemodialysis, appears reasonable for a PCR = 1.1 gfkdday. The above example also demonstrates the interrelationship among average BUN, PCR and KtIV, illustrated schematically in Figure 1. In the previous section, PCR and BUN were advanced as the primary indicators of patient status; KtIV was presented as a secondary factor relating BUN to PCR. In more recent years, the major focus has shifted to KtIV, ignoring PCR and BUN. In principle, any of the three parameters can be calculated after the other two are measured; e.g., PCR may be determined from accurate values of KtIV and average BUN (or predialysis BUN) using UKM. Unfortunately, as will be discussed below, precise determination of KtIV is often difficult, and, indeed, its actual definition and interpretation in light of multipool urea kinetics is subject to debate. It is our contention that dialysis quantification should focus on PCR rather than KtIV and that average or predialysis BUN be regarded as the primary indicator of whether the patient is receiving adequate dialysis treatment relative to his or her protein intake. Independently determined KtIV would serve primarily to identify technical problems due to access recirculation, slow intercompartmental transfer, etc. In the following sections, we examine various approaches to evaluating PCR and KtIV in hemodialyzed patients. Classical UKM In classical UKM, BUN and body weight are measured at the beginning and end of hemodialysis, typically during the midweek session, and again before the start of the following session (Fig. 2). With zyxwvutsrq K1 V zyxw FIG. 1. Schematic illustrating the interrelationship among dialysis patient urea level, dietary protein intake and KrIV. The center panel is a reminder that this relationship is analogous t o the familiar C = C/K(BUN = urea generatiodurea clearance) relationship for functioning kidneys. Reprinted with permission from Garred LJ, DiGiuseppe B , Chand W , McCready W , Canaud B: Artif Organs 16:249, 1992. 238 zyxwvutsrq zyxwv zyxwvutsrq zyxwvu GarrecA et al. . " G v10.58 PCR, = 9.35 -+ 0.17 a - h l h R l Y - (111). PCR, reflects protein catabolic rate normalized not to actual patient body weight but rather to a hypothetical weight chosen such that the patient's actual urea space represents 58% of the hypothetical weight. PCR, is fairly insensitive to the K value used but is not equivalent to PCR normalized to actual patient weight as in Equation I. PCR (and DPI for the patient in protein balance) will be greater or less than PCR, depending on whether the urea volume to weight ratio of the patient is greater or less than 0.58. For example, the urea volume to weight ratio for a patient with considerable adipose tissue may be only 0.50, in which case PCR, may overestimate actual PCR by up to 15%. This magnitude of error might cause noticeable differences between PCR, and dietitian estimates of DPI from patient dietary recall. For such patients, the error may be reduced by replacing the 0.58 in Equation 111by a more reasonable value of the urea volume to patient weight ratio. For most patients, however, PCR, can be reasonably used to approximate PCR and DPI. A variety of reasons can be invoked to explain why classical UKM has not been used routinely in dialysis centers for determination of PCR and KrIV but two practical factors stand out. Firstly, there is the inconvenience of blood sampling over two dialysis sessions coupled with the need to limit the quantity of patient blood drawn for diagnostic purposes. Secondly, the requirement for computer calculation deters many clinicians. Both of these drawbacks have been countered by development of simple formulas to compute KtIV and PCR from the predialysis and postdialysis BUN during a single treatment. zyxwvutsr zyxwvutsrq m h U W FIG.2. The single cycle urea mass balance period for classical UKhf compared with the seven-day (or 14-day) period for dialysate-based UKM illustrated for a patient in steady-state protein balance (DPI = 1.0 & / d a y ) receiving 3-hr dialyses (KfIV = 1.0) on Monday, Wednesday, and Friday. the assumptions listed in Table 1, the differential equations governing the mass balance for urea during and between hernodialyses can be solved to obtain the complicated equations (7, 13) listed in Table 2. These are coupled, nonlinear relations that can be solved by an iterative technique on a digital computer to obtain G and V, from which PCR and KtIV may be calculated. A major problem in determining G and V by classical UKM is their sensitivity to the value substituted for K in these equations. The appropriate value for K is effective patient clearance for single pool kinetics. This value is derived from in vivo dialyzer clearance that depends on the dialyzer membrane area, membrane permeability, extracorporeal blood flow and dialysate flow with an additional correction for any access recirculation. K is dflicult to evaluate accurately without careful measurement of flow rates and recirculation. Often manufacturers' published dialyzer clearances are used. These typically in vitro values usually overestimate actual patient clearance significantly, resulting in corresponding overestimates of G and v. This causes little error, however, in the ratio of K to V and so KrIV is relatively insensitive to the value chosen for K. Conversely, serious error in KtIV may result if the manufacturer's value for K is combined with a V estimated from anthropometric correlations such as those of Watson (14) or Hume (15). An erroneously high K,however, will inflate G and result in a proportional overestimate of PCR (Equation I). To avoid this error, the PCR equation can be modified as follows: zyxwvut KtN and PCR Determination from Predialysis and Postdialysis BUN The classical UKM equation applied during hemodialysis (see Table 2) is rendered complex by two factors: (i) the urea pool shrinks as excess fluid is ultrafiltered and (ii) urea is continually generated during dialysis. In the absence of ultrafidtration and urea generation, urea concentration, C,, in the anuric patient would fall exponentially during dialysis with constant K according to the equation, zyxwvutsrqpo TABLE 1. humptiolls assodatml rrith ClnSJieal Mmodeiing urea is dialyzed from a single, well-mixed body pool urea is generated at a constant rate throughout the test cycle patient clearance is constant during hernodialysis residual renal clearance is constant throughout the test cycle ultraWtration is constant during hernodialysis rate of weight gain is constant between treatments changes in body weight reflect changes in urea pool volume Were this true, KtIV could be calculated directly from the BUN at the start (Cpre)and end (CP,,,J of the dialysis period, where R is the CPos,/Cpre fraction. Equation V,however, is only approximate and requires correction for urea generated over the dialysis period, t, and for the volume of fluid removed as ultrafiltrate (ex- zyxwvut zy zyxw zyxwv zyxwvu zyxw zyxwvu QUANTIFICATION 239 TABLE 2. Classical urea kinetic modeling equations (variable volume single pool model) _ _ ~ On-Hernodialysis equation CPrc(K- K U F + KR) - G (-) 1 Cp0,,C~- KUF + KR) - G where ---. - ~ _ _ _ _ I pre and post refer 10 the start and end of a hemodialysis session = postdialysis (dry) urea volume V K,, = rate of weight loss during dialysis = (BW,,, - BWpoxI)/r K , = residual renal clearance Off-Hemodialysis equation where - pressed here as the predialysis to postdialysis decrease in patient body weight, MW). The following equation (16, 17) was developed theoretically from the urea mass balance equation and accounts for both of these factors: where t is hemodialysis treatment time expressed in hours. When tested in real (16) and in simulated (17) patients spanning a large range of clinical conditions, Equation VI was found to predict KtIV within an error range of less than kO.08. An alternative (slightly less precise) formula has been proposed by Daugirdas (18): Kt zyx p o s t and next refer to the start and end of a period between dialyses V = postdialysis (dry) urea volume B = rate of weight gain between dialyses = (BW,,,, BW,,,,J@ @ = time between dialyses PCR, = 0.018(1 - 0.162R)(l - R + ABWIV)C,,, 1 - 0.0003r + 0.17 (VIII), where t is treatment time in minutes. PCR, calculated from Equation VIII is relatively insensitive to the value substituted for V in the right side of the equation; setting V equal to 0.58 times dry weight is adequate. Derived from expressions for urea mass balance in anuric patients who are in protein balance, Equation VIIl was found accurate to less than 1% error when tested against 540 patient simulations spanning a range of KtIV (0.6-1.6), PCR, (0.61.6), t (2-4 hr), and interdialytic weight gain (22). A simpler but less accurate formula was also developed (22). zy PCR, = 0.0076(Kr/V)(Cp,, + C,,,,) + 0.17 (1x1. R - 0.008t - “W) BW (VII). Either formula permits calculation of KtIV from a simple algebraic calculation requiring BUN and body weights before and after a single treatment. The resulting KtIV is nearly identical to the KtIV calculated from 3-BUN classical UKM as described in the previous section. Other authors (19, 20) have proposed simpler KtIV equations linear in R; however, their use is not recommended since they can be quite inaccurate under some dialysis conditions (17, 21). Recently. an equation has been developed (22) to permit estimation of PCR, from (midweek) predialysis and postdialysis B U N and body weight: Equation IX is accurate to within a maximum error of 5% which is probably adequate for most clinical needs. Alternatively, the dependence of PCR, on KtIV and the sum of midweek C,,, and C,,,, is displayed graphically in Figure 3 (Fig. 3 IS slightly more precise than Equation IX). Although Equation IX or Figure 3 is simple to use, each requires an accurate estimation of KtlV, for example from Equation VI. The formulas and graph presented here provide the clinician with convenient tools for obtaining KtIV and PCR, essentially equivalent to those of classical UKM but requiring only simple calculator computations with a minimum of blood sampling and eliminating the need for careful measurement of blood flow rate, access recirculation, etc. Never- 1.7 zyxw zyxwvutsrqponm zyxwvutsrqpo Garred et 01. 240 KVV I 1.61.4 1.2 1.0 0.0 0.6 urea pool. No simple means is available for determining either entire pool K or V when multiple-pool urea kinetics prevail. Clearances measured in vivo from predialyzer and postdialyzer BUN in clinical practice are dialyzer clearances. Dialyzer clearances usually exceed the effective whole-body clearance often by a significant margin. Current practice is to calculate KtlV using a single-pool model, whether or not multipool dynamics are suspected. The resulting KtIV may be considerably higher than the actual dose of dialysis received by the patient. This uncertainty suggests that Kt/V, as currently measured, should be used with caution and not be relied on as a sole indicator of hemodialysis adequacy. Urea disequilibrium causes classical single-pool modeling to overestimate PCR as well. PCR, determined by classical UKM, by the two-point formulas, or graphically (Fig. 3) is significantly overestimated in these patients. Other concerns have been raised about the assumptions listed in Table 1 underlying the UKM equations. Patient clearance may fall during a treatment session from loss of dialyzer surface area or from increasing access recirculation as vascular volume contracts. Many dialysis sessions are interrupted by a variety of alarm conditions that cause the dialysis machine to automatically go into a period of dialysate bypass during which patient clearance is negligible. It is not uncommon as well for technicians to reduce blood flow during part of a dialysis session in response to a hypotensive episode. Two-BUN or three-BUN UKM is performed for a single dialysis session typically once a month in each patient. How representative are the parameters determined from an isolated UKM over an extended period? The UKM day may fall on a “bad session” due to dialyzer clotting, low blood flow associated with poor needle placement, or hypotens h e episodes which are uncharacteristic of the majority of sessions for that individual. Conversely, UKM on a “good day” will mask the effects of access or other problems that may otherwise frequently appear in the studied patient. In addition, PCR is known to fluctuate from day to day; a high value measured during a single session may not be representative. These limitations of classical UKM, particularly the problem of multipool urea kinetics during shortened high-flux dialysis, have prompted a new dialysate-based approach to UKM that will be outlined in the following section. zyxwvutsrqponm zyxwvutsrqponm zyxwvutsrqpon Cpr. + C-., tddU FIG. 3. Graph for determining patient protein catabolic rate (g proteidkg dry body wt/day) from the predialysis ( C,,) and postdialysis ( C,,,) BUN values (mddl) for the midweek dialysis and the KtlV for that session. theless these simple relationships are limited by the same assumptions as classical UKM (Table 1). Limitations of Classical UKM and its Simplified Equations In recent years a number of the classical UKM assumptions listed in Table 1 have been called into question. Of chief concern is whether urea obeys single pool kinetics. Urea is distributed across both intracellular and extracellular water pools. During hemodialysis intracellular urea must transfer into the vascular space of the extracellular compartment to be dialyzable. If this occurs at a finite rate, concentration gradients may develop between the two pools. Alternatively, regional variations in blood flow, in the absence of resistance to diffusion at the cell membrane, may have the same effect. Evidence for urea concentration gradients is found in the postdialysis urea rebound which is routinely and easily demonstrable. Rebound presumably is due to re-equilibration of the blood compartment where urea has been relatively depleted with compartments that retained higher concentrations during dialysis. This disequilibrium effect lasts for about 30 to 60 min following dialysis and may be accentuated by the high dialyzer clearances achievable in many dialysis centres today, sometimes >300 d m i n . In such instances where two or more pools are required to describe urea dynamics during and following hemodialysis, the classic KtIV ratio does not reflect the effective amount of therapy received by the patient. For KdV to reflect the true quantity of dialysis received, the denominator of the ratio remains the total volume of the entire urea pool under dry weight conditions but the “K” of the numerator then must be the effective clearance for the entire zyxwvut Dlalysate-Based UKM Dialysate-Based PCR Determination The UKM equations that describe urea generation and PCR are derived from simple expressions of urea mass balance, a common-sense accounting for the whereabouts of generated urea. The urea zy zyxw zyxwvut QUANTIFICATION mass balance may be expressed in words as follows: mass of urea nitrogen generated over period considered = mass of urea nitrogen removed by dialysis over period considered + mass of urea nitrogen eliminated in urine over period considered + mass of urea nitrogen added to body stores over period considered As noted earlier, in the classical blood-side approach to UKM this urea balance is applied moment by moment over a single dialysis cycle from the start of one session to the start of the next session (see Fig. 2). The differential equations describing mass balance during and between hemodialyses can be mathematically integrated, subject to the assumptions of Table I , producing the complex equations of Table 2. In contrast, using the dialysatebased approach to PCR, urea balance can be examined over a seven- (or 14-) day period (see Fig. 2). The advantage of considering a seven-day period is that any change in body urea stores is reduced to a small fraction of the total amount of urea generated, usually to less than 2% of the urea generated, proportionately less for a 14-day period. Therefore for the anuric patient receiving three dialyses per week the urea mass balance over a seven-day period may be approximated as: mass of urea nitrogen generated over seven-day period = mass of urea nitrogen removed in three dialyses 241 150 I of spent dialysate generated during a typical treatment is clinically impractical. However, the same concentration can be obtained by diverting a small fixed fraction of the spent dialysate stream to obtain a convenient volume of the same composition as the total spent dialysate volume (23, 24). U is then determined from the urea concentration in the partial volume K,)multiplied by the total dialysate volume, V,, calculated as the product of dialysate flow rate (Q,) and session duration ( t ) : (XII). Thus, three dialysate samples are required, one for each of the three U values in Equation XI. For classical UKM, three blood samples from two dialysis sessions are required. This partial dialysate collection (PDC) technique has been used for several years in a satellite facility in one of our centers (Montpellier, France). Each of the eight Fresenius 2008E machines (Fresenius AG, Bad Homburg, Germany) has been fitted with a small pump which diverts 8-10 ml/min from the spent dialysate stream into a small collection container. An infrared sensor situated on the venous line bubble trap senses the first blood to leave the artificial kidney and signals the collection pump to start. The pump stops when the bubble trap clears of blood at the end of the session. A counter is activated over the same period to count the piston stroke cycles of the dialysate delivery pump; the total number of strokes provides an accurate determination of V,. Initially PDC was performed for three consecutive dialyses using Equation XI on a once-a-month protocol. More recently and for more than one year a partial dialysate collection has been obtained during every dialysis providing continuous monitoring of PCR. Figure 4 shows a 10-month period of continuous PCR values in a 68-year-old, 88.5-kg woman with analgesic nephropathy. Each open circle in Figure 4a represents the PCR determined by PDC over a 14-day period, i.e., from six consecutive hemodialyses. These define a smooth PCR curve showing the long-term trends in this poorly nourished patient. Dietary protein intake fell slowly over a five-month period reaching a low of 0.35 g/kg/day in May 1993, after which it rose steadily reaching 0.58 g/kg/day by mid-August. This PDCderived PCR curve appears to provide more frequent and reliable information for patient dietary counseling than the less accurate monthly determinations of PCR using classical three-BUN modeling (solid squares in Fig. 4a) even when dialyzer clearance was measured in vivo and corrected for access recirculation. While most of the classical PCR values are reasonably close to the PDC curve, occasional erroneous outliers occur (e.g., 12/92) probably due t o a misrecording or inaccuracy in one of the three BUN measurements. Surprisingly, reasonably accurate values of PCR can be obtained zyxwvu zyxwvu zyxwvuts zyxwvuts or where the factor 10080 represents the number of U2, U , symminutes in a seven-day period and U , , bolize the mass (mg) of urea nitrogen removed during the three treatments. Substitution of this relation into Equation I gives: PCR = 0.000928 u, + u, + BW 113 + 0.17 (XU. Determination of G or PCR therefore becomes a problem of identifying a convenient and accurate When U is means for routine measurement of 17. directly measured from urea appearance in spent dialysate the classical UKM assumptions of Table I are no longer required. For example, whether or not urea follows multiple pooi kinetics will have no bearing on PCR determined by Equation XI. In the past, U has been measured directly by multiplying the volume of a total dialysate collection by its urea concentration. Routinely collecting the 90- 242 1 - zyxwvutsrqp zyxwvutsrqp zyxwvu i.oI i B Garred et al. 1.10 . zyxwvu zy zyxwv . 1.Ooo . uUmD. -. . - u. 3B -- . 4-1 aim. LPI- LoI*. Ps f 1 f f 1 s ......-..............................-.--. FIG.4. The partial dialysate collection method of monitoring PCR is illustrated for a 68-year-old, 885-kg female subject with analgesic nephropathy. (a) Open circles represent PCR determined from six consecutive dialyses (14-day urea mass balance); solid squares represent monthly PCR from three-BUN classical UKM with in vivo determination of patient clearance corrected for access recirculation. (b) Solid line represents PCR determined by PDC from each sequence of three dialyses; open triangles represent PCR determined from midweek PDC approximating U ,+ U, + U, in Equation XI by 3.12 . Urnidweek (see text). from a single midweek partial (or total) dialysate collection assuming that this approximates the average of the three dialyses during that week. Figure 4b shows PCR determinations based on midweek partial dialysate collections, replacing U, + U2+ U,in Equation XI by 3.12 * Urnidweek. (The 3.12 factor rather than 3 has been found to compensate for the proportionately larger U for the first dialysis of the week following the three-day interdialytic weekend.) The PDC methodology offers a marked improvement in accuracy and convenience of PCR determination compared to blood-based classical UKM but it still requires manual sampling and central laboratory analysis. The recent development of urea sensors permits U to be measured directly from periodic or intermittent urea concentrations in the dialysate stream exiting the dialyzer (25-27). Figure 5 shows the continuous dialysate urea concentration (C,) from a prototype sensor (26)for a 3-hr treatment. U is determined as the shaded area under the curve (the integral of C , over dialysis time) multiplied by Q,. For most dialyses, C , falls exponentially, and U can be calculated analytically (25, 26) from a log-linear fit of the C, versus t profile (see Fig. 6). These computations can be built into the software of the urea sensor to provide U as a direct output of the device. Such urea sensors permit PCR determination free of blood or dialysate sampling and free of most of the complex computations required by earlier methods. All that is required for PCR determination is the substitution of U values from three consecutive treatments into Equation XI;alternatively, an approximate PCR can be determined from a single midweek U as described above. zyxwvutsrqponm Dialysate-Based Monitoring of Patient Urea Level One goal of a dialysate-side focus for UKM is to minimize or eliminate the need for blood sampling: nevertheless, monitoring time-averaged or midweek predialysis BUN is an important element of patient management. How can this be accomplished from only dialysate side urea measurement? At any moment during hemodialysis, patient urea concentration, C,, is related to C ,, the urea concentration in the effluent dialysate stream, by the ratio of dialysate flow rate to patient clearance: (XIII). 0 30 60 SO 120 Dialysis Thna 150 100 (min) FIG. 5. Continuous measurement of spent dialysate concentration througb a 3-hr treatment using an in-line prototype urea sensor. Total urea removed (U = 11.5 g urea nitrogen) is determined as the product of spent dialysate flow rate and the shaded area under the C,, trace. Adapted and reprinted with permission from Gamd LJ, St Amour N, McCready W, Canaud B: ASAIOJ 39M339, 1993. Thus when C , is measured at a particular instant during dialysis the corresponding BUN may be determined from equation XIII if the QJX: ratio is accurately known at that moment. Predialysis (and postdialysis) BUN could be determined in this fashion. Similarly, TAC urea could be calculated from the set of predialysis and postdialysis BUNS obtained from dialysate urea monitoring over an entire week's treatment. However, as discussed earlier, K is difficult to estimate accurately and a direct in vivo measurement is cumbersome. Therefore, estimations of BUN from dialysate urea concentrations and Equation XI11 are at best approximate. QUANTIFICATION An accurate measurement of predialysis BUN may be obtained from the dialysate side without blood sampling either by isolated ultrafiltration or by recycling dialysate in a closed loop through the dialyzer for a short period before starting dialysis treatment. With either technique the urea concentration equilibrates quickly with predialysis BUN and can be measured by a resident urea sensor or by direct sampling. When single-pass dialysis is subsequently started, dialysate urea concentration falls quickly to a level determined by the patient’s clearance and the blood level at the start of dialysis. Measurement of C , (and Q,) at this point allows calculation of clearance using Equation XIII. 243 laboratory error in either C,,, or C,,,,7, will strongly influence S, the slope can be more reliably determined from several serial measurements or continuous measurement of C,. To obtain a more precise value for KtiV, whether based on blood-side or dialysate-side measurements, small corrections are required to account for generation of urea and urea space contraction during treatment. When S, is obtained as -In(R)t, these corrections lead to Equations VI or VII, presented earlier for KtlV determination from predialysis and postdialysis B U N . The dialysate-based equation corresponding to Equation VI is the following: zyxwvutsrq zyxwvutsr zyxwv zyxwvut zyxwvu zyxwvutsrq Dialysate-Based Monitoring of K t N Finally we demonstrate how the spent dialysate urea concentration profile during a dialysis can be used to monitor KtIV (25, 26). As discussed above, the decline in BUN (C,) during hemodialysis is nearly exponential (see Equation IV). Expressed equivalently, the natural logarithm of C , versus t is almost a straight line with slope = - K / V . Thus KtIV = -S, * t where S, is the slope of the straight line fitted to ln(C,) versus time. When C , at the start and end of dialysis (C,,, and C,,,,,) are used to estimate the In(C,) versus time slope, we obtain, SB = zyxwvut z In CpO,,- In Cpre ln(CposKprA- -In R t t t (XIV). S, can also be evaluated from dialysate-based urea concentrations. If patient clearance and dialysate flow rate do not vary, Equation XI11 indicates C, is a constant fraction of C, throughout the treatment and therefore S!, equals S,, the slope of ln(C,) versus time. This is illustrated in Figure 6. Since a n y Kt - -V -SDt 1 - 0.01786 30 60 90 120 150 Dialysis Time 180 (min) FIG.6. Illustration of dialysate-based KtIV determination with an in-line urea sensor. The spent dialysate tracing from Figure 5 is plotted here on logarithmic coordinates. Ln(C,) falls linearly versus time with same slope as In(C,) versus time (suggested by the broken line between predialysis and postdialysis BUN). KdV determined from predialysis and postdialysis BUN using Equation VI equals l . 18; KrIV determined from the In(C,) versus time slope using Equation XV equals 1.18. Adapted and reprinted with permission from Garred U,St Amour N , McCready W, Canaud B: ASAIOJ 39:M340, 1993. (XV,. A urea sensor for on-line measurement of C , can be programed to compute So as dialysis progresses. It is possible, with input of anticipated or actual ultrafiltration rate and software incorporating Equation XV (or an analogous formula based on Daugirdas’ Equation VII), for a urea sensor resident in the spent dialysate line to display the KrlV achieved up to any moment during dialysis as well as the additional time required to realize a final goal value of KtIV. It must be emphasized, however, that both blood-based and dialysate-based estimates of KtIV may be misleading when effective patient clearance changes significantly during dialysis whether due to multipool effects, access recirculation, or dialyzer clearance reductions associated with clotting or variations in blood flow. Urea Sensing Technology Real-time monitoring of urea kinetics during hemodialysis requires on-line urea measurement. The 1990s have witnessed a flurry of activity in developing urea sensing devices specifically for this purpose (2633). Two devices currently undergoing clinical trials may soon be available. Other sensors are in advanced stages of development. Measurements that are highly specific for urea involve enzymatic breakdown of urea by urease at neutral pH: CO(NH2)2+ 3H,O + 2NH4+ 0 ABW + 3 - BW + HC0,- + OH-. Urea sensors typically measure either the rise in solution conductivity associated with the above hydrolysis of urea or they measure the amount of ammonium ion formed. Ammonium ion may be sensed by an ion specific electrode or photometrically following a further reaction. The two sensors approaching market availability are based on ammonium ion measurement. The BioStato 1000 (Baxter Healthcare, McGaw Park, IL) (27, 28) utilizes an ammonium ion electrode fitted with a urease-impregnated membrane cap. This urea monitor, designed as an add-on device for 244 zyxwvutsrq zyxw zyxwvut zyxwv zyxwvu zyxwvut Garred et at. most current systems, draws samples from the spent dialysate line at 5- to 10-min intervals throughout a treatment session. With user input of dialysate flow rate, patient dry weight and expected ultrafiltration volume the device determines U and Kt/V for the session along the lines described in the previous section. The user may select an option to measure BUN from a dialysate equilibrating procedure prior to initiating hemodialysis. If this is done, the patient’s clearance can be calculated by the instrument via Equation XIII. This value of K together with measured session time and KtIV, determined from the serial dialysate concentration profile, are then used by the BioStata to backcalculate an estimated patient V. When the predialysis BUN measurement option is not done, V must be obtained by another method or from a previous study in the same patient, and K is calculated from KtIV and t. The BioStata makes bloodless, dialysate-based UKM feasible and convenient for any dialysis center. Another sensor undergoing clinical trials is the DianalyzeP (ParaDigm BioTechnologies, Toronto) (29). This is a blood-side device which draws periodic ultrafiltrate samples across permeable membranes from the dialyzer arterial line. Ammonium ion produced following enzymatic degradation of urea is measured colorimetrically using a pH indicator dye. Progression of KfIV is followed during treatment from the decline in (logarithm) arterial line urea concentration with time. The Dianalyzee compares calculated KrIV with a previously entered prescribed value to estimate time remaining. Although this device does not directly measure U (precise dialyzer clearance would be required) PCR is estimated from patient BUN at the start of dialysis and the machinedetermined value of KtIV. The BioStatm and the DianalyzeP represent major advances towards automated UKM. They will be followed by future generation devices more closely integrated with the dialysis machine. These will likely be in the form of small, relatively inexpensive, possibly disposable probes which plug directly into the effluent dialysate line. These anticipated instruments will be stable, robust devices which require no reagents and minimal calibration. A promising configuration is a differential conductivity electrode (30, 31) composed of identical miniature conductivity electrodes on opposing sides of a thin, narrow plate introduced into the spent dialysate stream. One of the permeable membranes covering each electrode contains cross-linked urease such that the associated electrode senses the elevated conductivity associated with ureasecatalyzed hydrolysis of urea. The opposing electrode is required to measure the background conductivity in the spent dialysate stream which fluctuates during a dialysis session. Urea concentration is a linear function of the difference in conductivity measured by the back-to-back electrodes. It is likely that future generation delivery systems will have built-in urea sensors as standard compo- nents. These machines could provide fully automated UKM with total dialyzed urea and up-to-theminute KtIV achieved available for display throughout the session. Provision could be made to capture completed treatment results into a linked computerized data bank providing the clinician with up-todate and comprehensive histories of patient UKM. In addition, future urea monitors will be capable of recognizing changes in the rate of (logarithm) urea concentration fall caused by a significant reduction in patient clearance associated, for example, with dialyzer clotting or an increased access recirculation due to a shift in needle positioning. A machine alarm could alert medical staff of such occurrences permitting immediate attention to the problem. These technological advances may be expected to better monitor quality assurance of dialysis treatments and bring about improved patient care. In summary, hemodialysis machines introduced within a few years time should offer complete automation of UKM for measurement of patient BUN, PCR, and KrIV free of any need for blood or dialysate sampling or user computations. Urea monitors, such as the two described above for use with existing machines, will become available even sooner. In the interim, patient urea level, PCR and KtiV can be monitored with monthly or more frequent predialysis and postdialysis BUN measurement using Equation VI or VII for KtIV determination and Equation VIII or IX,or Figure 3 for PCR, determination. One must bear in mind that multipool effects are not accounted for in these formulas so that calculated values may be misleading for some patients receiving high clearance dialysis. Monthly partial dialysate collection over a sequence of three treatments is an alternative means of determining PCR accurately even when multipool effects are present. Periodic midweek PDC alone can provide PCR data sufficiently accurate for most clinical needs. These improved techniques promise to simplify yet render more accurate the task of optimizing hemodialysis dosage. zyx zyxw zyxw zyxwvu References 1. Sargcnt JA. Gotch FA: The study of uremiaby manipulation of blood concentrations using combinations of hollow tibcr devices. ASAIO Tram 20:39M1, 1974 2. Gotch FA, Sargent JA. Keen ML, Lee M: Individualized quantified dialysis therapy of uremia. Proc Clin Dial Transplant Forum 427-35, 1974 3. Harter HR: Review of s w c a n t tiodings froin the National CooperOtivc Dialysis Study and recommendations. Kidney fnr 23(SuppI 13):S1U74112. 1983 4. Lowrk EG, Techan B P Principles of d d i dialysis thaapy: Implementing rrcommendations for the National Cooperative Dialysis Study. K i h e y In# 23(Suppl 13):Sll34122. 1983 5. Gotch FA, Sargcnt JA A mechanistic analysis ofthe National Coop erative Dialysis Study (NCDS). Kidney Inr 28:5&534,1985 6. Lowrie EG, Laird NM, Parker TF, *nt JA Effect of the heddysis prescription on patient mortndity: Report from tbe National Cooperative Dialysis Study. N Engl J Med 305:11761181, 1981 7. DepnerT A Prescribing H e d ~ s i cA: Guide to Urea Kinetic Modeling. Boston, Kluwer Academic, I991 8. Lindsay RM,SpPaher E: A hypothds: The protein catabolic rate is dependent upon the type and amount of treatment in dialyzed uremic patients. Am J Kidney Dis 13:382-389, 1989 9. Lysaght MJ. Pollock CA, Hallet MD. Ibels LS,Famll PC: The nl- zy zyxwvutsrq zyxwvutsrq zyxwvutsrqponmlk QUANTIFICATION evance of urea kinetic modeling to CAPD. ASAIO ‘Trans 35:784-790. I989 10. Laird NM. Berkey CS, Lowrie EG: Modeling success or failure of dialysis therapy: The National Cooperative Dialysis Study. Kidnev Inr 7:S2315. 1975 11. Sargent J. Gotch F. Borah M, Piercy L. Spinozzi N, Schoenfeld P. Humphreys M: Urea kinetics: A guide to nutritional management of renal failure. Am J Cfin Nurr 31:16%-1702, 1978 12. Borah MD, Schonfeld PY, Gotch F A , Sargent JA. Wolfson M. Humphreys MH: Nitrogen balance during intermittent dialysis therapy of uremia. Kidney Int 14:491-500. 1978 13. Farrell PC. Gotch FA: Dialysis therapy guided by kinetic modeling: Application of a variable-volume single-pool model of urea kinetics. Trans Aust Conf Hear Mass Transfer 29-37, 1977 14. Watson PE. Watson ID, Batt RD: Total body water volumes for adult males and females estimated from simple anthropometric measurements. Am J C h Nufr 33:27-39. 1980 IS. Hume R, Weyers E: Relationship between total body water and surface area in normal and obese subjects. J Clin Pafhol24:234-238. 1971 16. Garred LJ, McCready W: A new simple formula for calculating KfIV based on urea kinetic modeling theory. (abstract) Blood Purf 10:W. 1992 17. Garred U.Barichello DL, DiGiuseppe B. McCready WG. Canaud BC: Simple KrlV formulas based on urea mass balance theory. ASAIOJ, in press 18. Daugirdas JT: The post:pre-dialysis plasma urea nitrogen ratio to estimate KrIV and NPCR: mathematical modeling. Inf J A r t y O r p 12:411-419. 1989 19. Jindal KK. Manuel A. Goldstein MB: Percent reduction in blood urea concentration during hemodialysis (PRU). A simple and accurate method to estimate K d V urea. ASAIO ‘Trans 33:286-288. 1987 20. B a d e C, Casino F, Lopez T: Percent reduction in blood urea concentration during dialysis estimates KrIV in a simple and accurate way. Am J Kidney Dis 1 5 : W 5 , 1990 2 1. Daugirdas IT: Second generation logarithmic estimates of single-pool variable volume K f I V : An analysis of error. J Am Soc Nephrol 4:1205-1213. 1993 245 22. Garred LJ. Barichello D. Canaud B. McCready W: Simple formulas for protein catabolic rate calculafion from pre- and po5tdialyvs urea. (abstract) ASAIO Ahsrr, 23:71. 1994 23. Garred LJ. Rittau M. McCready W. Canaud B: Urea kinetic modelling by partial dialysate collection. Inr J Arff Org 12:96102. 1989 24. Garred LJ, Rittau M. McCready W. Canaud B: Evaluation of the partial dialysate collection method of urea kinetic modelling. iahstract) Blood Purf 6:372. 1988 25. Garred LJ. DiGiuseppe B. Chand W, McCready W , Canaud B: Kr I’ and PCR determination from serial urea measurement in the dialysate effluent stream. Arff Orguns 16:248-255. 1992 26. Garred LJ, St. Amour N, McCready W. Canaud B: Urea kinetic modeling with a prototype urea sensor in the spent dialysate stream. ASAIOJ 39:M337-M341. 1993 27. Depner T. Collins A. Jindal K, Lazarus JM. Nissenson A, Hossli S. Luhring D. Ebben J, Kerhaviah P: Multicenter clinical validation of an on-line urea monitor. (abstract) J Am Soc Nephrol 4:343, 1993 28. Keshaviah P. Ebben J. Luhring D, Emerson P. Collins A: Clinical evaluation of a new on-line monitor of dialysis adequacy. (abstract)J Am Sor Nephrol3:374. 1992 29. Lindsay R, Heidenheim P. AbuNadar P. McCarthy G: On line urea kinetic modelling: Preliminary results. (abstract) ASAIO Ahsrr 22:90. 1992 30. Goggin MJ. Evans J . Agraharkar M. Pepper M: Development of an on-line system for the measurement of urea using conductiomerric electrodes. (abstract) Blood Purf 10:68, 1992 31. Jacobs P. Suls J , Sansen W, Hombrouckx R: A disposable urea sensor for continuous monitoring of hemodialysis eficiency. ASAIOJ 39:M353-M358, 1993 32. Smirthwaite PT.Fisher AC. Henderson I A . McGhee J. Moktar N . Simpson KH. Whitehead AJ, Gaylor JD: Development of a blood urea monitoring system for the closed loop control of dialysis. ASAIOJ 39:M342-M347. 1993 33. Kuhlmann U. Knaack J . Schindler J , Schindler M . Herna K. Schmidt B. Lange H: Continuous “on line” monitoring of urea concentration in blood and dialysis fluid by molecular selective electrodes in hasmodialysis. (abstract) EDTA Ahrfr 17. 1993 zyx zyxwvutsrqp zyxwvutsrqponm zyxwvutsr