In hemodialysis (HD), the conventional practice for

VIIth International Course on Peritoneal Dialysis May 23 26, 2000, Vicenza, Italy Peritoneal Dialysis International, Vol. 20, Suppl. 2 0896-8608/00 $3.00 +.00 Copyright 2000 International Society for Peritoneal Dialysis Printed in Canada. All rights reserved. CREATININE IS THE BEST MOLECULE TO TARGET ADEQUACY OF PERITONEAL DIALYSIS Peter G. Blake Division of Nephrology, Optimal Dialysis Research Unit, University of Western Ontario, London Health Sciences Centre, London, Ontario, Canada In hemodialysis (HD), the conventional practice for many years has been to use urea as a yardstick for assessing adequacy of clearances. Urea is used because the urea kinetic model was validated by the National Cooperative Dialysis Study (1). In peritoneal dialysis (PD), adequacy of clearances has conventionally been assessed using both urea and creatinine as measures. The popularity of urea relates in part to its common use in HD, and that of creatinine dates back to early studies that used it as a measure of clearances in PD (2). In recent years, the use of both urea and creatinine clearances in PD has been, to some extent, validated in clinical studies that show correlation with patient outcomes (3 6). Thus, recent guidelines for PD prescription have defined both creatinine and urea clearance targets, and this situation has led to some confusion among practitioners (7,8). This paper argues that urea and creatinine clearances give different, though often complementary, information and that both should be measured. However, it will be suggested that, when interpreted appropriately, creatinine clearance is probably the superior measure. WHY DO THE TWO CLEARANCES DIFFER? Before discussing the relative merits of urea and creatinine clearances, it is important to highlight how and why they differ when applied to PD (Table 1) (9). The most notable difference between clearances of the two molecules is the relative effect of residual renal function. It is well known that, in the failing kidney, a large portion of urinary creatinine content results from tubular secretion rather than glomerular filtration. Hence, creatinine clearance greatly overstates the true glomerular filtration rate (GFR). KEY WORDS: Creatinine; urea; adequacy; Kt/V. Correspondence to: P.G. Blake, Division of Nephrology, London Health Sciences Centre (Victoria Campus), South Street, London Ontario N6A 4G5 Canada. In contrast, urea is resorbed to a significant degree in the renal tubules, and resorption continues to be a major factor even in the failing kidney. Resorption is particularly marked in the presence of volume depletion, poor cardiac function, and diuretic therapy, all of which are common in patients with advanced renal failure. Thus, urea clearance tends to significantly understate true GFR. The overall result is that, in a patient with substantial residual renal function, the renal contribution to clearance will be proportionately far greater when creatinine clearance is used as a measure than when urea clearance is used. Stated another way, creatinine clearance will look disproportionately high compared to urea clearance. In contrast, in the anephric patient, creatinine clearance will be much lower relative to urea clearance (Table 2). It can therefore be said that creatinine clearance gives more weight to residual renal function than does urea clearance. To get around this situation, most recommendations suggest that, when calculating total creatinine clearance in PD patients, the average of urea and creatinine clearance be used to represent the residual renal contribution (7,8). This approach partly reduces the weight given to residual renal function by creatinine clearance, but creatinine clearance modified in this manner still remains disproportionately greater than urea clearance while GFR is still substantial. TABLE 1 Factors Differentially Affecting Urea, Relative to Creatinine Clearance 1. Residual renal function (the greater the function, the disproportionately greater the creatinine clearance) 2. Peritoneal transport status (the lower the transport status, the disproportionately greater the urea clearance) 3. Average dwell time (the shorter the dwell time, the disproportionately greater the urea clearance) 4. Body size (the greater the body size, the disproportionately greater the creatinine clearance) S65

BLAKE MAY 2000 VOL. 20, SUPPL. 2 PDI TABLE 2 Case History A 70-kg male of body surface area 1.73 m 2 and total body water 40 L initiates CAPD at a time when his residual renal creatinine clearance is 80 L/week and his residual renal urea clearance is 40 L/week. He is put on a prescription of four 2-L dwells daily, and his ultrafiltration is 1 L per day. The equilibration ratios for urea and creatinine on his 24-hour dialysate effluent collection are 0.95 and 0.66 respectively. On calculation, his peritoneal Kt/V is 9 7 0.95 / 40 = 1.5. His renal Kt/V is 40 / 40 = 1. His total Kt/V is 1.5 + 1.0 = 2.5 per week. His peritoneal creatinine clearance is 7 9 0.66 = 42 L per week. His residual renal creatinine clearance is 80 + 40 / 2 = 60 L per week. His total creatinine clearance is 42 + 60 = 102 L per week. Two years later, this patient has lost residual renal function, but is otherwise on an identical prescription with identical body size and peritoneal equilibration. His peritoneal clearance will be unchanged, but his total clearance will equal his peritoneal clearance. The drop in creatinine clearance during this time is from 102 L to 42 L per week, approximately 60%. The drop in Kt/V from 2.5 to 1.5 per week is just 40%. The point is that loss of residual renal function has a proportionately much greater effect on creatinine clearance than urea clearance. The second key difference in PD is that creatinine equilibration is slower and considerably more limited by peritoneal transport status than is urea equilibration (10). This effect relates to urea being a smaller molecule (molecular weight 60 D, compared to 113 D for creatinine). Thus, the average equilibration for urea with a 4-hour dwell is more than 90%, and even a low transporter is likely to achieve values in excess of 85%. In contrast, the average creatinine equilibration in a 4-hour dwell is approximately 66%. In a significant proportion of patients, it will be as low as 40% 45%. Furthermore, the spread of values is much greater for creatinine clearance. The difference between the lowest and highest transporter is frequently as great as 60%, compared to approximately 20% for urea. The practical consequence is that the ratio between creatinine and urea equilibration varies with transport status in that it tends to be much lower in low transporters and higher in high transporters (11). This situation in turn leads to creatinine clearances being disproportionately low relative to urea clearances in low transporters. The opposite effect is not seen to the same extent in high transporters. The reason is the relative loss of clearance in this group, on most prescriptions, owing to poorer ultrafiltration. A further consequence of the discrepancy in equilibration by transport type is that urea clearance tends to be less time dependent and more flow dependent than creatinine clearance. Thus, when prescriptions involve frequent exchanges and short dwell durations (as in cycled PD, for example), urea clearance will be disproportionately high relative to creatinine clearance. Conversely, when a patient on day dry auto- mated PD (APD) has a day dwell added, the proportionate increase in creatinine clearance is far greater than that in urea clearance. That is, the addition of the daytime dwell is relatively more important for creatinine than for urea clearance. For this reason, patients who are switched from continuous ambulatory PD (CAPD) to day dry cycler PD may see an increase in urea clearance, but a decrease in creatinine clearance, unless the prescription is appropriately adjusted (12). Again, because of the influence of peritoneal transport on the relative equilibration of the two solutes, the latter effect is particularly marked in low transporters. A further, but perhaps less significant, difference between the two indices is the manner in which they are normalized when applied to PD. By convention, urea clearance is normalized to a measure of total body water to give the well-known index Kt/V. In contrast, creatinine clearance is normalized to a measure of body surface area. Typically, body water and body surface area are calculated based on the formulas of Watson and DuBois respectively. One of the features of these formulas is that, as body weight increases, the rise in calculated body water is proportionately greater than that in calculated body surface area (13). The consequence is that, in larger patients, targets for creatinine clearance are relatively easier to reach than those for urea clearance. One final difference worth noting between the two molecules is their biological significance in relation to uremia. The central role of protein and urea metabolism in uremia has long been recognized and, of course, blood urea levels are affected by dietary protein intake as well as by renal and dialytic clearance. S66

PDI MAY 2000 VOL. 20, SUPPL. 2 CREATININE CLEARANCE IS BEST The main influence on serum creatinine levels, apart from renal and dialytic clearance, is lean body mass, although dietary protein intake also has some effect. Because dietary protein intake and lean body mass do not always closely correlate, a differential effect on serum levels of urea and creatinine sometimes occurs. It is thus important that serum levels of urea and creatinine be interpreted in the light of protein intake and lean body mass, as well as of delivered clearances. THE IDEAL INDEX OF ADEQUACY An ideal measure of adequacy in dialysis should show a number of key characteristics. First, and most important, it should be predictive of important clinical outcomes such as patient survival, well-being, hospitalization, and so on. Following on from this first requirement, the measure should also be capable of being altered with a subsequent beneficial effect on outcome. In addition,the measure should be easy and inexpensive to determine. An ideal index would also reflect a range of renal functions in addition to small-solute clearances. Such an index does not exist in practice, but in theory might reflect middle-molecule or large-molecule clearance, renal volume status, and endocrine and metabolic functions. Finally, an ideal adequacy index should not be confounded by other factors with consequent potential for misinterpretation. Using the above standards, we can evaluate urea and creatinine clearance. Both can be said to be easy and inexpensive to measure, and both have been shown to be somewhat predictive of clinical outcomes. No proof yet exists, however, that they can be prospectively altered in PD to improve clinical outcome. Furthermore, neither clearance has much ability to reflect middle-molecule and large-molecule clearance, volume regulation, or renal, endocrine, and metabolic function. The issue of potential confounding of urea and creatinine clearances will be discussed below. Alternative, more accurate measures of GFR such as iohexol clearance would obviously be better at reflecting true GFR and clearance of larger molecules (14). However, the advantage of simplicity is lost, data predicting clinical outcome are not available, and such indices would still fail to measure important nonclearance-related functions. ABILITY TO PREDICT CLINICAL OUTCOMES Over the past decade, literature has accumulated showing some association between both urea and creatinine clearances and subsequent patient outcomes in PD (3 6). An important point needs to be made about these data, however. In all of these studies, the bulk of the variation in total solute clearance has been related to a variation in residual renal function, and in those that have looked closely at the relative effects of peritoneal and renal clearance, it has been apparent that only the residual renal clearance has been shown to predict outcome (15). This fact may reflect a lack of significant variation in peritoneal clearance in these studies, but the important point is that uncertainty still exists about the true relationship between peritoneal clearance per se and clinical outcome. Against this background, it is worth looking at how urea and creatinine clearance perform relative to one another as predictors of clinical outcome. At least four studies have looked at the two indices in the same cohort of patients. The most notable is the CANUSA study, which showed that both indices were predictive of survival. A 6% fall in the relative risk of mortality was associated with every extra 0.1 unit of Kt/V per week, and a 7% fall in the relative risk of mortality was associated with every additional 5 L of corrected creatinine clearance per week. But, in the same period, only creatinine clearance was able to significantly predict technique failure and days hospitalized. For every extra 5 L per week of corrected creatinine clearance, a 5% reduction in the risk of technique failure and a 1% reduction in the relative risk of spending time in hospital was seen (3). The CANUSA study would thus suggest an advantage for creatinine over urea clearance in terms of prediction of clinical outcomes. In a small study by Brandes et al, based on 18 patients at a single center and published in 1992, only creatinine clearance and not Kt/V was able to predict appropriately whether patients fell into good, intermediate, or poor groups on the basis of clinical outcomes (16). In a larger study published by Genestier et al in 1995, of more than 200 patients with an average of two years follow-up, both clearance indices predicted survival, as was the case in CANUSA. But again, just as in CANUSA, only creatinine clearance predicted technique survival. The relative risk for technique failure was more than three times greater in patients with creatinine clearances below 50 L per week as compared to patients with clearances above 50 L per week (5). The only exception to this trend was a study by Selgas et al of 56 patients followed for three years (6). Both indices were shown to predict hospitalization equally well, but Kt/V showed a better ability to predict patient survival. From these four studies, it might be concluded that, overall, creatinine clearance showed a slightly better ability to predict outcome than did urea clearance, and that the advantage tended to be seen in measures other than patient survival (that is, patient wellbeing and technique survival). S67

BLAKE MAY 2000 VOL. 20, SUPPL. 2 PDI WHAT MIGHT THE BASIS BE FOR THIS ADVANTAGE FOR CREATININE CLEARANCE? The most likely explanation is that, as previously discussed, creatinine clearance gives relatively greater weight to residual renal function than does urea clearance, and that residual renal clearance is more potent than peritoneal clearance as a predictor of patient outcome on PD. If the modest predictive advantage of creatinine over urea clearance in PD patients is attributable to residual renal function, then an obvious question arises. Is either index superior in patients who are completely anuric? Quite simply, no data exist to address this issue. Only a few studies have looked at the anuric PD population (17). It is to be hoped that more data addressing this question will be obtained in the future. CONFOUNDING FACTORS In recent years, it has become apparent that many confounders are involved in the use of urea and creatinine clearances to predict outcome in PD patients. The first and most notable of the confounders relates to the normalization of clearance. As mentioned earlier, urea clearance and creatinine clearance are conventionally normalized to total body water and body surface area respectively. These two measures of body size are, in turn, typically derived from nomograms based on a variety of factors, the most significant of which is body weight (13). It is now apparent, however, that body weight is itself an independent predictor of survival in dialysis patients. This relationship has been most clearly shown in the HD population, where it truly seems to be the case that the bigger the patient, the better the prognosis (18). Similar studies have not been done in the PD population, but some evidence exists that increased weight is, at least, not an adverse prognostic factor in this population (19). These findings introduce a degree of complexity. When, in clinical outcome studies, clearance is normalized to a derivative of body weight, the index has a numerator (that is, clearance) and a denominator (that is, body size) that may both, of themselves, be beneficial for patient survival. Dividing one by the other creates a potentially confusing situation. Such a situation may lead to an inability to differentiate the poorly dialyzed, poorly nourished patient from the well-dialyzed, well-nourished patient (that is, both patients may have the same normalized clearance values) (18). Attempts to get around this problem include the practice of normalizing to indices based on ideal rather than actual body weight. Still, the potential for confounding remains, and the validity of this approach is controversial. Alternative suggestions based on avoiding normalization altogether would certainly avoid this confounding, but might be considered biologically inappropriate. This problem needs to be resolved. A second confounder relates to the fact that, when urea and creatinine clearances are being evaluated in renal failure, both are being used as surrogate renal toxins. The proof that either of these molecules has any inherent toxicity is limited. Indeed, studies have tended to show that higher serum levels of urea and creatinine both tend to be positive rather than adverse prognostic predictors (20). These results likely relate to the fact that serum urea and creatinine are strongly influenced by protein intake and lean body mass, respectively, and that these latter indices have both been shown to be independently predictive of patient outcome to varying degrees (3). A clearance index based on a more clearcut renal toxin or at least on a surrogate, the serum level of which has no independent predictive effect on outcome would be preferable. With regard to the two confounders already discussed, urea and creatinine clearances have similar weaknesses. However, with regard to the third confounder, creatinine clearance is potentially more misleading. The third confounder relates to the issue of peritoneal transport status and its independent predictive effect on patient outcome. A variety of studies in recent years have clearly shown that high peritoneal transport status is associated with worse clinical outcome, in terms of both patient and technique survival in patients on CAPD (21,22). This effect is likely, at least in part, to be related to better ultrafiltration and volume control in low transporters. As previously mentioned, low transporters tend to have disproportionately poor creatinine equilibration (relative to urea), and so their creatinine clearances are disproportionately low relative to their urea clearances. Yet these patients have superior outcomes on CAPD. We therefore have a potential paradox, in that the patients who tend to have the lowest creatinine clearances tend to have the best survival. This finding tends to mask any independent effect of creatinine clearance on survival in PD. The use, therefore, of creatinine clearance targets in low transporters, especially when they are anuric, becomes very problematic. Indeed, in these patients, it is physically very difficult to reach the target of 60 L per week corrected creatinine clearance recommended by the U.S. National Kidney Foundation (7). In recognition of this difficulty, the recently published Canadian Society of Nephrology clearance targets for PD specify that a target of 50 L per week creatinine clearance is sufficient for low and low-average transporters (8). S68

PDI MAY 2000 VOL. 20, SUPPL. 2 CREATININE CLEARANCE IS BEST SUMMARY The overall conclusion is that neither creatinine clearance nor urea clearance is the perfect index for predicting outcome in PD patients. In the absence of indices that are better validated and more convenient, creatinine and urea are what we have to use. They are best seen as two imperfect, but potentially complementary, measurements. On balance, creatinine clearance is the better of the two indices, in that it gives greater weight to residual renal function, and residual renal function is probably a stronger predictor of patient outcome than peritoneal clearance per se. However, creatinine clearance has a particular weakness in low transporters; values have to be interpreted with discretion in this group. Furthermore, both indices are flawed because of the manner in which they are conventionally normalized. Research into more appropriate methods of normalization, or into whether normalization is required at all, would be helpful. The complex relationship between these clearance indices, protein intake, and lean body mass also needs to be kept in mind. While this paper argues in favour of creatinine clearance as the better index, it still suggests that both indices be used, together with a large measure of clinical judgment. REFERENCES 1. Laird NM, Berkey CS, Lowrie EG. Modelling success or failure of dialysis therapy: The National Cooperative Dialysis Study. Kidney Int 1983; 23(Suppl 13):S101 6. 2. Boen S, Haagsma Schouter W, Birnie R. Long-term peritoneal dialysis and peritoneal dialysis index. Nephrol Dial Transplant 1978; 7:377 8. 3. Churchill DN, Taylor DW, Keshaviah PR, and the CANUSA Peritoneal Dialysis Study Group. Adequacy of dialysis and nutrition in continuous peritoneal dialysis: Association with clinical outcomes. J Am Soc Nephrol 1996; 7:198 207. 4. Maiorca R, Brunori G, Zubani R. Predictive value of dialysis adequacy and nutritional indices for mortality and morbidity in CAPD and HD patients: A longitudinal study. Nephrol Dial Transplant 1995; 10:2295 305. 5. Genestier S, Hedelin G, Schaffer P, Faller B. Prognostic factors in CAPD patients: A retrospective study of a 10 year period. Nephrol Dial Transplant 1995; 10:1905 11. 6. Selgas R, Bajo MA, Fernandez Reyes MJ, Bosque E, Lopez Revuelta K, Jimenez C, et al. An analysis of adequacy of dialysis in a selected population on CAPD for over 3 years: The influence of urea and creatinine kinetics. Nephrol Dial Transplant 1993; 8:1244 53. 7. National Kidney Foundation Dialysis Outcomes Quality Initiative. Clinical practice guidelines for peritoneal dialysis adequacy. Am J Kidney Dis 1997; 30(Suppl 2):S67 136. 8. Blake P, Bargman J, Bick J, Cartier P, Dasgupta M, Fine A, et al. Guidelines for adequacy and nutrition in peritoneal dialysis. J Am Soc Nephrol 1999; 6(Suppl 13):S311 21. 9. Chen HH, Shetty A, Afthentopoulos IE, Oreopoulos DG. Discrepancy between weekly Kt/V and weekly creatinine clearance in patients on CAPD. In: Khanna R, ed. Advances in peritoneal dialysis. Toronto: Peritoneal Dialysis Publications, 1995; 11:83 7. 10. Twardowski ZJ, Nolph KD, Khanna R. Peritoneal equilibration test. In: Khanna R, Nolph KD, Prowant BF, Twardowski ZJ, Oreopoulos DG, eds. Advances in CAPD. Toronto: Peritoneal Dialysis Bulletin, 1987; 3:138 47. 11. Blake PG, Burkart J, Churchill D, Daugirdas J, Depner T, Hamburger RJ, et al. Recommended clinical practices for maximizing peritoneal dialysis clearances. Perit Dial Int 1996; 16:448 56. 12. Nolph KD, Twardowski ZJ, Keshaviah PK. Weekly clearance of urea and creatinine on CAPD and NIPD. Perit Dial Int 1992; 12:298 303. 13. Tzamaloukas AH, Malhotra D, Murata GH. Indicators of body size in peritoneal dialysis: Their relation to urea and creatinine clearances. Perit Dial Int 1998; 18:366 70. 14. Marx MA, Shuler CL, Tattersall JE, Golper TA. Plasma iohexol clearance as an alternative to creatinine clearance for CAPD adequacy studies. Kidney Int 1995; 48:1994 7. 15. Blake PG. Critique of CANUSA. Perit Dial Int 1996; 16:243 5. 16. Brandes JC, Piering WF, Beres JA, Blumenthal SS, Fritsche C. Clinical outcome of CAPD predicted by urea and creatinine kinetics. J Am Soc Nephrol 1992; 2:1430 5. 17. Oreopoulos DG. The optimization of continuous ambulatory peritoneal dialysis. Kidney Int 1999; 55:1131 49. 18. Chertow GM, Owen WF, Lazarus JM, Lew NL, Lowrie EG. Exploring the reverse J-shaped curve between urea reduction ratio and mortality. Kidney Int 1999; 56:1872 80. 19. Fried L, Bernadini J, Piraino B. Neither size nor weight predicts survival in peritoneal dialysis patients. Perit Dial Int 1996; 16:357 61. 20. Lowrie EG. Conceptual model for a core pathobiology of uremia with special reference to anemia, malnourishment and mortality among dialysis patients. Semin Dial 1997; 10:115 29. 21. Churchill DN, Thorpe KE, Nolph KD, Keshaviah PR, Oreopoulos DG, Page D, for the Canada U.S.A. (CANUSA) Peritoneal Dialysis Study Group. Increased peritoneal membrane transport is associated with decreased patient and technique survival for continuous peritoneal dialysis patients. J Am Soc Nephrol 1998; 9:1285 92. 22. Davies SJ, Phillips L, Russell GI. Peritoneal solute transport predicts survival on CAPD independently of residual renal function. Nephrol Dial Transplant 1998; 13:962 8. S69