Quantification of Acute Renal Replacement Therapy

Clinical Dosing Targets for Acute Intermittent Hemodialysis

A number of studies have suggested a relationship between small-solute control or clearance and patient outcomes during acute IHD. In the 1950s and 1960s, it was demonstrated conclusively during the Korean and Vietnam wars that IHD saved lives, although subsequent underpowered and clinically outdated studies arising from that experience fell short of proving the case for “early” IHD (initiated when the blood urea nitrogen [BUN] level was <100 to 150 mg/dL) or for “intensive” IHD (maintaining BUN <60 mg/dL and serum creatinine <5 mg/dL).

Since then, three multicenter randomized controlled trials (RCTs) evaluated the effect of IHD dose in AKI patients. compared daily with alternate-day IHD in 146 ICU patients with AKI. The daily arm received an average single-pool fractional clearance (spKt/V) of 0.92 on 6.2 occasions per week (i.e., a weekly Kt/V of 5.8). The alternate-day arm received an average spKt/V of 0.94 on 3.2 occasions per week, with a weekly Kt/V of 3.0. However, it has been observed that in this study the randomization was inadequate, and the dose in the control group was very low. Moreover, the low overall mortality in the study (34%) suggests that the results may not generalize. Furthermore, as noted by Palevsky, the higher rates of altered mental status, gastrointestinal bleeding, and sepsis in the alternate day arm potentially could be due to the low dialysis dose per treatment.

The Veterans Affairs/National Institutes of Health Acute Renal Failure Trial Network (ARFTN) study compared intensive to less-intensive RRT in 1124 ICU patients with AKI. Within each randomization arm patients were switched between IHD and CRRT or slow-efficiency extended hemodialysis (SLED), based on their hemodynamic status. Intermittent treatments were prescribed at a Kt/V of 1.4, with a delivered Kt/V averaging 1.3, and were performed three (less-intensive arm) or six (more-intensive arm) times per week. Consequently, the weekly Kt/V was approximately 6.5 in the intensive and 3.9 in the less-intensive arm. Mortality at 60 days was similar in both groups (53.6% and 51.5%) as was the percentage of patients recovering kidney function (15.4% and 18.4%).

The Hannover Dialysis Outcome Study randomized 148 ICU patients with AKI to two different doses of SLED: a standard-dialysis arm dosed to maintain plasma urea levels between 120 and 150 mg/dL (20–25 mmol/L), or an intensified dialysis arm dosed to maintain plasma urea levels less than 90 mg/dL (<15 mmol/L). No significant differences in either survival at day 28 (39% vs. 44%) or recovery of kidney function (63% vs. 60% of survivors) were found.

There are no data supporting a relationship between larger-solute clearance and outcomes for acute IHD. Intermittent hemodiafiltration (IHDF) and high-flux IHD for critically ill AKI patients have demonstrated no clinical or laboratory advantage over low-flux IHD. This likely is due to the low clearances of larger solutes afforded by these modalities. Whereas low-flux IHD clears approximately 3 mL/min of β ₂ -microglobulin from blood water during the course of treatment, high-flux IHD clears only about 35 mL/min, and even IHDF clears only 50 to 150 mL/min, depending on the substitution fluid rate. Given the short duration over which these modalities are applied, a meaningful clinical effect seems unlikely. The effect of IHD on solute control is therefore, for the most part, restricted to small solutes.

In summary, data suggest that the dose of IHD in AKI requires a weekly Kt/Vurea of 3.9 ⁹ . There are no published data relating clearance of larger solutes to outcome for acute IHD.

Clinical Dosing Targets for Continuous Renal Replacement Therapy

The first study to link solute clearance to outcomes for CRRT was performed over 25 years ago. However, it is only recently that clinical dosing targets have been refined. In a prospective, RCT from Ronco et al., mortality was lowest for patients receiving postdilution CVVH with an ultrafiltration rate (UFR) of 35 mL/kg/hr or greater (indexed to patient premorbid weight), provided it was applied more than 85% of the time. The external validity of this study is perhaps questionable because the participants were relatively small (average weight, 68 kg) and had a low incidence of sepsis (12%). A subsequent trial failed to confirm these findings, but it was underpowered and also was performed in patients who had undergone cardiosurgical procedures, for whom factors other than solute control were likely to be relatively more important as determinants of outcomes.

It is uncertain from the study of Ronco et al. whether the superior survival with higher UFR was related to clearance of small or larger solutes. Small-solute clearance would be equal to the UFR, but clearance of larger solutes was not reported. This issue was addressed in a later study from Saudan et al., which showed that the addition of approximately 18 mL/kg/hr of diffusive clearance using CVVHD to a basal amount of approximately 24 mL/kg/hr of convective clearance using CVVH resulted in superior patient survival. This finding demonstrated that at least some of the benefit of a higher dose of CRRT in Ronco's study was the result of increased small solute clearance.

Even higher doses of CRRT may benefit those with septic shock and a high predicted mortality risk. In Ronco's study, there was a trend to lower mortality for septic patients receiving a UFR of 45 mL/kg/hr or greater. These findings were supported by observational data from Honore et al., who found that the dose of high-volume CVVH was greater (average UFR, 132.5 mL/kg/hr) in those patients whose hemodynamic parameters improved during treatment than in those whose parameters did not improve (average UFR, 107.5 mL/kg/hr).

In contrast with the results of these earlier single-center studies, two large multicenter trials have found that a much lower dose of CRRT could suffice. The first trial, the ARFTN study, compared standard-intensity predilution CVVHDF with a prescribed effluent flow of 20 mL/kg/hr to high-intensity CVVHDF at 35 mL/kg/hr. No differences in outcomes between the two study arms were found. Importantly, more than 95% of the prescribed dose of CRRT was delivered in the less-intensive group.

The second trial, the RENAL study, compared the effects of postdilution CVVHDF at dosages of 25 and 40 mL/kg/hr on 28- and 90-day mortality rates in 1464 AKI patients. The delivered dose was 88% and 84% of prescribed dose in the low- and high-dose groups, respectively. As in the ARFTN study, there was no difference in mortality between the two groups.

On these bases, it has been concluded that there are no benefits of increasing CRRT dosage in AKI patients above effluent flows of 20 to 25 mL/kg/hr. In clinical practice, to achieve a dosage of 20 to 25 mL/kg/hr, a greater dosage, in the range of 25 to 30 mL/kg/hr, should be prescribed. Moreover, based at least on a small single-center RCT, it is possible that a higher dose could be beneficial in some patients with septic shock.

There are no published data relating specifically to larger-solute clearance to outcomes during CRRT.

Calculation of Fractional Clearance for Intermittent Hemodialysis

To calculate Kt/V, UKM must be applied. UKM is based on the mass balance principle that “urea accumulation equals urea input minus urea output.” Practically, this principle is embodied in a model that estimates urea concentration based on three patient-dependent parameters—urea distribution volume (V), urea generation rate (G), and renal urea clearance (Kr)—and three treatment-dependent parameters—dialyzer urea clearance (K _D ), session length (T), and treatment schedule. A differential equation can be developed from this model, whose solution provides the general equations for UKM that are presented later.

Urea kinetics can be assessed through either blood measurements or direct dialysate quantification. The former option is logistically more feasible, although the role of partial dialysate collection or online urea and ionic dialysate monitors warrants further study in this setting. For the moment however, the standard approach is to use blood measurements. The blood urea nitrogen (BUN) and available estimates of UKM parameters are entered into the general equations for UKM. These equations are solved iteratively to impute UKM parameters that are not provided to the model (usually G and V) from those that are (usually K _D ). This allows the calculation of Kt/V.

Studies of urea kinetics show four major differences between the critically ill AKI population and the ESRD population. First, critically ill patients often have markedly increased values for G, attributable to a more catabolic state. Second, they often have markedly increased V at 65% to 70% of their body weight, compared with ESRD patients at 55% to 60% of body weight. Some of this increase is attributable to Na ⁺ and H ₂ O loading (and concurrent loss of lean body mass) in critical illness, although most of the increase is attributable to the dissociation of V from its usual anatomic correlate of total body water (TBW): V is between 10% and 30% higher than TBW in critically ill AKI patients. This discrepancy is not just a by-product of UKM but a literal one demonstrable with the use of radiolabeled ( ^13C ) urea and deuterium oxide. This discrepancy is not satisfactorily explained by intercompartmental urea dysequilibrium (i.e., delayed entrance of urea into the blood from body pools that have high resistance to solute transfer resulting from a low ratio of tissue perfusion to water, such as muscle or skin), which is, in fact, surprisingly similar to that seen in patients with ESRD.

Third, critically ill patients often have values for K _D that are lower than expected and specifically lower than those calculated by usual means (e.g., Michael's formula). Venovenous angioaccess leads to high recirculation rates, especially in short femoral catheters, where it can approach 25%, and this is exacerbated by the frequent need for line reversal in 25% to 50% of treatments. Fiber-bundle clotting also reduces K _D , especially in the absence of anticoagulation.

Finally, critical illness is associated with marked variation in all of these UKM parameters over time. The assumption of urea steady state underlies many of the UKM calculations in the maintenance IHD population and affords convenience (e.g., model fitting using two BUN points rather than three). However, urea steady state cannot be assumed for critically ill AKI patients or for the modeling of IHD dose in this setting.

Calculations using UKM equations provide a critical and extremely important benefit when dealing with the uncertainties and sources for error mentioned previously, in that they allow for the mathematic phenomenon, whereby erroneous UKM parameters are offset. In this manner, any error in the calculation of, for instance, K _D leads to proportional overestimates (or underestimates) of V and G and little or no error in the final value of Kt/V or normalized protein catabolic rate (nPCR). Such offsetting of error does not occur if UKM techniques are not used. Occasionally, we see Kt/V directly calculated from values of K _D and V that have been measured by other means (e.g., K _D from Michael's equation, V from bioimpedance analysis). We do not recommend this. As shown previously, values for K _D and V are unpredictable in critically ill AKI patients, and any error in their assessment will result in a proportional error in Kt/V during direct substitution. We therefore recommend that such measurements be used as input UKM parameters.

The most common UKM equations for formal iterative calculation of Kt/V are those derived from the variable-volume single-pool (VVSP) model developed by Sargent and Gotch. Alternatively, simplified (noniterative) calculation of Kt/V is possible using equations such as those of Daugirdas and Garred ( Box 146.1 and Table 146.1 ). Formal UKM calculation is preferable for accuracy, although some data suggest that the simplified equations may provide reasonable estimates of dose. All of these approaches calculate spKt/V. To obtain the equilibrated Kt/V (eKt/V), the Daugirdas rate equation (eKt/V = [spKt/V − 0.47] × [K/V] + 0.02) has been shown to be as accurate as complicated double-pool variable-volume modeling in this setting. The eKt/V undoubtedly provides a more realistic reflection of acute IHD dose; however, spKt/V defines dosing targets from the literature and should be used preferentially in clinical practice.