The Convertibility of Online Clearance Measurements

The automatic measurement of effective clearance K is now available with new hemodialysis and online hemodiafiltration machines, which greatly simplifies the surveillance of adequate dialysis delivery with every treatment.1, 2 Taking advantage of components and processes that are essential for dialysate preparation, clearance in these new machines is determined for electrolytes, mostly sodium (Na+) rather than urea, and measured in the dialysate using conductivity cells. Even though Na+ and urea are quite different with respect to their electrochemical and physiological characteristics, in terms of clearance across current dialyzer membranes, both solutes show a close correspondence and the clearance of Na+ can be taken as a surrogate for urea clearance.3

The calculation of clearance in these machines requires the measurement of dialyzer inflow and outflow conductivities before and after a step change in dialysate conductivity. This simple description of the method belies some subtleties in its use. As with any other clearance measurements, timing is important.3 For example, when measuring dialyzer urea clearance during hemodialysis, arterial inlet and venous outlet samples have to be taken simultaneously. An excessive delay in taking the arterial sample will spuriously decrease the calculated urea clearance because arterial line concentrations continue to decrease with ongoing urea removal from the patient. The same effect comes into play with online clearance measurements where extended conductivity profiles and measuring times cause substantial salt loading (or unloading), thereby changing blood inflow and dialysate outflow conductivities so that the calculated clearance will be spuriously low. This effect has been recognized as the major reason for the discrepancy in online clearance measured by different technical embodiments.3

Online clearance was introduced 15 years ago.4, 5 It was recognized to be lower than dialyzer clearance and to account for effects caused by access and cardiopulmonary recirculation, and therefore it was identified as an effective clearance.6, 7, 8, 9 This is fortunate as recirculation represents a major uncertainty in dialysis delivery when dialysis prescription is focused on blood flow and dialyzer clearance.10 The discussion about the validity of online clearance measurements is ongoing, but now that these measurements are provided by 2 manufacturers using proprietary technology, there finally is an opportunity to compare the different and competing approaches in the same patient.

Maduell et al set out to treat 31 patients with 4 different machines, 2 of each from the same manufacturer, and to measure clearance either by the Diascan Monitoring System or the online clearance monitor (OCM) under otherwise identical treatment conditions.11 Urea reduction ratio and Kt/Vu for urea (where t and Vu refer to time and urea distribution volume, respectively) were comparable when dialysis was delivered by machines from the same manufacturer but systematically different when delivered by machines from different manufacturers. As patients served as their own controls, the effect of V could be eliminated so that the difference in Kt/Vu could be translated into a 5% difference in delivered Ktu. Even though the delivered clearance could have been affected by compartment effects, which are known to depend on hemodynamic stability and regional perfusion, this is improbable because of the design of the study.12 The difference in Ktu must have been related to differences in delivered clearance K and/or effective treatment time t.

The uncertainties in delivered clearance related to extracorporeal blood flow can be substantial. While the revolutions of the peristaltic pump are exactly known, the actual stroke volume pumped per revolution depends on the elastic recoil of the pump segment and the negative arterial line pressure or pump preload.13 To minimize the risk of delivering a spuriously low blood flow and to account for the variability in elastic properties of pump segments, the stroke volume used for the calculation of blood flow for a given preload is usually assumed in the low range so that true blood flow is always higher than the blood flow displayed by the machine (Fig 1). Notice that uncertainties in blood flow depend significantly on the quality of blood lines and on characteristics of the pump segment, which usually change over time and which also depend on storage conditions. These uncertainties are important when control of treatment efficiency is based on achieved blood flow. When control of treatment efficiency is based on online clearance measurements, uncertainties related to actual blood flow, recirculation, and dialyzer fiber clotting are automatically accounted for.

View Large Image Download to PowerPoint

Figure 1. Extracorporeal blood flow. Left panel: Measured (crosses) and assumed (closed circles) stroke volumes Vs versus arterial line pressure Part described by linear regression functions (full and broken lines). Notice that for a given preload measured, Vs was always larger than that assumed by the machine. Right panel: Ratio of measured to displayed blood flow expressed as relative blood flow (Qrel) with (crosses) or without (closed circles) correction for variable preload versus Part described by linear regression functions (full and broken lines). Notice that with correction for preload (crosses), the measured blood flow was always higher than that given by the dialysis machine (100% line), whereas without correction (closed circles), measured blood flow only fell below that displayed by the dialysis machine (100% line) when Part was less than −150 mm Hg. Measurements were done at 37°C with flows ranging from 100 to 500 mL/min using a 4008 B dialysis machine and arterial line sets with an 8-mm pump segment (Fresenius Medical Care, Schweinfurt, Germany).

There also could be subtle differences between measured and effective treatment times. Even if a treatment is stable and blood and dialysate flows are not interrupted by alarms, not all the time on dialysis is always available to deliver a maximum clearance. For example, for the purpose of checking the integrity of high-flux dialyzers, some machines perform repeated pressure-holding tests during which dialysate is bypassing the dialyzer. This may account for up to 5% of total treatment time, comparable to the discrepancy observed between techniques in this study. What are the consequences for solute removal? Are short bypass times included or excluded from Kt calculations? After all, the time of a short dialysis bypass is not completely wasted: (1) there is continuing clearance in the dialyzer until the dialysate compartment is equilibrated, and (2) there is continued transport of solute from peripheral to central compartments, contributing to solute rebound, which would be expected to increase solute removal when dialysis continues.

The difference in online Kt was in the range of 10% and significantly larger than the 5% difference in delivered Ktu determined from urea kinetics. Thus, for the same Ktu there was only a small 5% discrepancy in online Kt measured by either Diascan or OCM technology. Since online Kt values measured by either technique were significantly correlated, the authors developed an equation to convert Diascan Kt into OCM Kt. The true value of these Kt currencies, however, remains unknown without reference to a gold standard. It cannot be said which one was true or superior.

The study is important as with the same online Kt delivered by machines using either Diascan or OCM monitors, the difference in Ktu will be in the range of 5% only, which is acceptable for clinical practice. Moreover, with the equation provided by the authors, the 2 measures of online Kt are now convertible.

The study has some limitations. Even though the differences between online Kt were small, there was a systematic difference between techniques and, for research purposes, it would be important to know which of the methods is closer to the true K or Kt. Such information may help to improve the measurement. Unfortunately, this cannot be decided with the information provided. The proper reference measurements are missing. To truly identify the difference in measured and delivered Kt, it would have been important to control for effective blood flow (or clearance) and effective treatment time using independent measurements for these variables. The focus on Kt makes it difficult to analyze separate effects of K and/or t, as the combination of any error in K and/or t could plausibly explain a particular difference in Kt. Errors can also be expected in both K and t measurements. Furthermore, the relationship provided by the authors to convert different Kt currencies only applies for machines using either Diascan or OCM monitors. However, the OCM manufacturer produces another online clearance (OLC) monitor for the US market, which is different from the OCM with respect to the duration and shape of the conductivity profile. Based on the importance of timing discussed above, there also could be systematic differences in clearance measured by OCM and OLC protocols. The relationships to convert between Diascan, OCM, and OLC currencies remain to be developed.

Whatever the exact reason for the small discrepancy in Kt measured between Diascan and OCM monitors, the differences are not alarming and do not give grounds to discuss the validity of Kt or Kt/V. Nor would it be justified to use the small difference for marketing purposes, especially because the exact cause for the discrepancy as well as the true Kt value remain unknown. The benefit of real-time and online surveillance of adequate clearance far outweighs the small differences measured between different technical approaches. The differences, however, need to be considered when comparing treatments using different dialysis technologies. The study of Maduell et al provides the basis for convertibility of Kt measurements when using different online clearance techniques.