Pro/con debate: Continuous versus intermittent dialysis for acute kidney injury: a never-ending story yet approaching the finish?

The question of whether renal replacement therapy should be applied in an intermittent or continuous mode to the patient with acute kidney injury has been the topic of several controlled studies and meta-analyses. Although continuous renal replacement therapy (CRRT) has a theoretical advantage due to offering the opportunity to remove excess fluid more gradually, none of the several outcome studies that have been undertaken in the meanwhile was able to demonstrate its superiority over intermittent renal replacement therapy (IRRT). In the present article, therefore, questions are raised regarding which are the specific advantages of each strategy, and which are the specific populations that might benefit from their application. Although several advantages have been attributed to CRRT - especially more hemodynamic stability allowing more adequate fluid removal, better recovery of renal function, and more efficient removal of small and large metabolites - none of these could be adequately proven in controlled trials. CRRT is claimed to be better tolerated in combined acute liver and kidney failure and in acute brain injury. IRRT is more practical, flexible and cost-effective, allows the clinician to discontinue or to minimize anticoagulation with bleeding risks, and removes small solutes such as potassium more efficiently in acute life-threatening conditions. Sustained low-efficiency daily dialysis is a hybrid therapy combining most of the advantages of both options.


Introduction
Few topics in nephrology have been the subject of so many randomized controlled trials (RCTs), meta-analyses and reviews than that of extracorporeal renal replacement in acute kidney injury (AKI). Since the introduction of hemodialysis as a valid treatment for renal failure by Kolff in the early 1940s [1], intermittent renal replacement therapy (IRRT) was off ered as a bridge until recovery of kidney function; fi rst in a low-effi cient and therefore protracted version, later becoming progressively shorter. In the 1980s, Kramer and colleagues introduced con tinuous renal replacement therapy (CRRT) as an alternative, allowing blood purifi cation 24 hours per day -at least in principle [2].
CRRT originally applied a simple concept without pumps or technology (continuous arteriovenous hemofi ltration). Since this approach often lacked effi ciency, however, machines containing blood pumps soon made their appearance (continuous venovenous hemofi ltration). Whereas solute removal with IRRT at the origin essen tially made use of diff usion -that is, gradientrelated molecule shifts in a liquid milieu from higher to lower concentration gradients -CRRT started as a convective strategy, driven by removal of solute-containing ultrafi ltrate through large pores and its replacement by substitution fl uid. With time, diff usion was also implied in CRRT by introducing additional pumps to the machines, while convective strategies became more widely applied in IRRT. Characteristics of CRRT and IRRT tended to con verge further at the beginning of this century in a concept named sustained low-effi ciency daily dialysis (SLEDD) [3], by applying IRRT mostly at lower blood and dialysate fl ows but at prolonged dialysis times. Th e term low effi ciency is, however, in many cases a misnomer [4,5]. Sometimes, this strategy is also named prolonged inter mittent renal replacement therapy.
It is diffi cult to fi nd a uniform defi nition of SLEDD in the literature. In fact, one of the advantages of SLEDD lies in its fl exibility both in terms of duration and of intensity. In the present text, the term SLEDD refers to any hemodialysis treatment performed with conventional

Abstract
The question of whether renal replacement therapy should be applied in an intermittent or continuous mode to the patient with acute kidney injury has been the topic of several controlled studies and metaanalyses. Although continuous renal replacement therapy (CRRT) has a theoretical advantage due to off ering the opportunity to remove excess fl uid more gradually, none of the several outcome studies that have been undertaken in the meanwhile was able to demonstrate its superiority over intermittent renal replacement therapy (IRRT). In the present article, therefore, questions are raised regarding which are the specifi c advantages of each strategy, and which are the specifi c populations that might benefi t from their application. Although several advantages have been attributed to CRRT -especially more hemodynamic stability allowing more adequate fl uid removal, better recovery of renal function, and more effi cient removal of small and large metabolites -none of these could be adequately proven in controlled trials. CRRT is claimed to be better tolerated in combined acute liver and kidney failure and in acute brain injury. IRRT is more practical, fl exible and cost-eff ective, allows the clinician to discontinue or to minimize anticoagulation with bleeding risks, and removes small solutes such as potassium more effi ciently in acute life-threatening conditions. Sustained low-effi ciency daily dialysis is a hybrid therapy combining most of the advantages of both options. dialysis machines over a longer time lapse than traditional intermittent hemodialysis (usually ≥5 hours).
Already from the early days, the question was raised regarding which of CRRT or IRTT was related to better outcome. Th e general perception was that the continuous approach, due to its slow protracted nature, would result in better outcomes. At least seven published RCTs and three meta-analyses, however, were unable to demonstrate a diff erence in outcome between both approaches [6][7][8][9][10][11][12][13][14][15], with a reported relative risk of 0.99 in the most recent meta-analysis [15].
Some authors have pointed to fl aws in the design of these RCTs [15]. Several of these biases were logistic, however, and in that case inherent to the very nature of the strategies implied [16] -such as the incapacity to enroll patients into continuous protocol arms due to unavailability of appropriate devices [10], or the impossibility to reach the preset exchange volume [11]. Logistical factors should thus also be taken into account when deciding on CRRT or IRRT [16]. Other biases skewing these RCTs are related to study design, conduct and reporting fl aws.
In specifi c subpopulations and/or based on arguments other than outcome, however, one of these two approaches might still be preferable over the other. In the present pro/con debate, both the advantages and disadvantages of CRRT and IRRT will be reviewed.
Of note, all therapeutic strategies available should not be considered as competitors, but rather as alternatives, each of which might be applicable within the same unit and even the same patient, depending on the practical options at hand at a given moment and on the metabolic or the fl uid balance needs of the patient.

Pro continuous renal replacement therapy
Several theoretical advantages have been attributed to CRRT over IRRT: more hemodynamic stability allowing more adequate fl uid removal; better recovery of renal function; and more effi cient removal of small and large metabolites. None of these assumptions, however, could consistently withstand the test of controlled clinical trial conditions.

Hemodynamic stability and fl uid removal
Several controlled trials fail to consistently demonstrate better hemodynamic stability and/or superior vital parameters for CRRT [6,7,10-12,17]. In a meta-analysis from the Cochrane group published in 2007, mean arterial pressure was the only clinical hemodynamic parameter that was signifi cantly higher with CRRT than with IRRT; the number of hypotension episodes was not diff erent, however [14]. Another systematic review showed no nominal diff erences [13]. A third review found a suggestion that CRRT was superior with regard to episodes of hemodynamic instability (P = 0.03) [15], based on four studies -with the major eff ect coming from the study by Augustine and colleagues, in which the diff erence between both strategies was signifi cant but the fall in mean arterial pressure amounted to only 2.6 mmHg versus the start of treatment [6].
Overall, it can be concluded that if there is a hemodynamic benefi t for CRRT, this nevertheless is not translated into diff erences of survival. Data also seem to suggest that part of the observed hemodynamic advantages of CRRT could be attributed to heat loss and hypothermia [12], improving venous return and blood pressure [18]. A similar eff ect can be obtained in IRRT by cooling dialysate, which has now become current practice in chronic hemodialysis [19] but applies to the AKI setting as well.
One problem potentially blurring the results of RCTs comparing CRRT and IRRT is the reluctance for including patients with major hemodynamic problems out of fear of instability in case of randomization to intermittent dialysis; this might result in the exclusion of the most unstable patients, reducing the diff erences among therapies.
A protracted treatment should allow removing fl uid at a larger cumulative volume. CRRT allowed markedly more negative fl uid balances in one RCT [6] but not in another [12].
In view of all of the uncertainties mentioned above and because of the physiological plausibility, fl uid-overloaded patients are among those with the highest potential to benefi t from CRRT or from IRRT in the SLEDD mode. CRRT has also been suggested to off er more possibilities for the administration of parenteral nutrition fl uids [20]a suggestion, however, not confi rmed in a prospective study [10].

Preservation of renal function
One of the major potential advantages of preserving hemo dynamic stability is a positive eff ect on recovery of kidney function. When autoregulation is lost due to AKI, each new hypotensive episode decreases glomerular perfusion [21], causing recurrent focal ischemic injury and postponing recovery of kidney function according to some studies [22]. Each condition such as IRRT causing more hypotension might thus theoretically emanate in a slower recovery of kidney function and a larger number of renal deaths (nonrecovery of renal function resulting in chronic dialysis), and might also aff ect perfusion of other organ systems, such as the heart. Nevertheless, all controlled studies [6,8-11] and meta-analyses [13,14,23] devoted to this aspect failed to demonstrate superiority of CRRT in this regard. For the sake of completeness, three observational trials suggested less evolution into chronic kidney disease stage 5 on dialysis (formerly endstage renal disease) with CRRT [24][25][26]. In view of the lack of a diff erence in fi ve RCTs and three systematic reviews [6,[8][9][10][11]13,14,23], however, the evidence base off ered by these uncontrolled trials is insuffi cient to overrule the controlled data.

Solute removal
Prolonging dialysis, even if dialyzer blood fl ow and dialysate fl ow are decreased proportionally, promotes solute removal due to better mobilization from extra-plasmatic compartments [4]. In line with these fi ndings, it has been suggested that slow strategies result in more effi cient removal.
A mathematical study compared the possibilities of removing the small solute urea with CRRT and IRRT. With CRRT, the threshold urea concentration could be maintained by increasing the fl uid exchange volume in patients of all body weights up to a volume of more than 45 l/day. For IRRT, it became impossible to reach the lowest threshold (blood urea nitrogen 60 mg/dl) for a body weight in excess of 90 kg [27] but the treatment time was not allowed to exceed 4 hours per day in spite of blood fl ows of only 200 ml/minute. Of note, the way the modalities were introduced in the calculations (high volume for CRRT vs. a fi xed limitation to 4 hours and an intermediate blood fl ow for IRRT) is more important for the result than the modality per se: for IRRT, the target could easily have been reached by assuming longer treatment times and/or higher blood fl ows.
Real-life comparisons of small molecule removal are scarce. In one study, day-by-day urea and creatinine levels were lower with CRRT than IRRT [9]. In other studies, daily urea clearances or concentrations were the same with both approaches [10,11]; while in a third study, only creatinine and not urea decreased more with CRRT [12].
Removal of cytokines might be more clinically relevant than removal of urea or creatinine in a population that is in general very sick and infl amed. At least two studies confi rmed this cytokine removal by CRRT, by adsorption on the mem brane, and/or by transmembrane elimination [28,29]. In one study, TNFα could be found in the ultrafi ltrate, but there was no signifi cant decrease in plasma concentration for this compound, as well as for all nine other cytokines or cytokine receptors under consideration [28]. In another study, removal was rapidly overwhelmed by generation once the membrane surfaces were saturated -and removal aff ected proinfl ammatory cytokines as much as their anti-infl ammatory soluble receptors or antagonists [29]. It is conceivable that the same risk of indiscriminating removal applies also to anti-infl ammatory cytokines such as IL-4, IL-10 or 1L-18. In addition, since removal is essentially by adsorption, fi lters need to be changed regularly, increasing the cost and decreasing the continuity of the treatment. Of note, removal of cytokines and other large molecules can be obtained just as well, if not better, with IRRT or SLEDD, under the condition that open membranes with large pore size (so-called high-fl ux membranes) are applied.
Th e impact of increasing solute removal above currently applied levels can be questioned, as at least two large multicenter studies [30,31] and one meta-analysis [32] failed to demonstrate a survival advantage of more effi cient removal over standard removal. A potential reason for this failure could be that the higher intensity of solute removal also has a downside, such as greater removal of drugs resulting in inadequate drug concen trations (for example, of antimicrobials) or more electrolyte disturbances [11].
One factor negatively aff ecting removal with CRRT is the frequent necessity to interrupt the procedure -for example, because of fi lter clotting, which occurs more frequently in CRRT than in IRRT. Average delivery of treatment per day with CRRT was reported to be only 19.5 hours [3,33], with observed individual values as low as 13.4 hours per day [3].

Specifi c patient populations benefi ting from CRRT
Although the evidence is contradictory (see above), the application of CRRT in combating severe fl uid overload can be defended. Other specifi c conditions in which CRRT has been proposed as the preferred option are combined acute renal and hepatic failure because of a bene fi cial impact on cardiovascular stability and intracranial pressure [34,35], and acute brain injury because of prevention of cerebral edema [36].

Peritoneal dialysis
Peritoneal dialysis is an often neglected continuous modality in AKI, although it can especially be of help in hemodynamically unstable and fragile patients, in those with enhanced bleeding risk, and in children [37], especially neonates and small children with postcardiac surgery AKI and hemolytic uremic syndrome. Of note, up to now only two RCTs have compared peritoneal dialysis with hemodialysis or related strategies in AKI. In a Vietnamese study on infected patients, continuous hemo fi ltration was superior -but the applied peritoneal dialysis strategy was not comparable with what is currently state of the art [38]. Th e other study, in a supplement issue, showed no diff erences [37] -here also, the peritoneal dialysis regime was unlikely to be suffi cient to obtain satisfactory solute clearance.

Pro intermittent renal replacement therapy
From the above, it appears that very few arguments based on controlled clinical studies suggest superiority of CRRT above IRRT or vice versa. Except for a number of specifi c indications, each one of these two strategies as well as any intermediate possibility lying in between (that is, SLEDD in all its forms -see below) can clinically speaking be considered a valid option for dialysis treatment of the average patient with AKI [16].
When the arguments in favor of IRRT are to be summed up, therefore, one might consider practical elements such as user-friendliness or limitation of expenses, as much as clinical factors benefi ting its use in specifi c subpopulations. Consequently, the following benefi ts of IRRT will be discussed: practicality and fl exibility of application; limitation of expenses; restriction of bleeding complications; and small solute removal in acute lifethreatening conditions.

Practicality and fl exibility of application
IRRT can be performed with the same technical infrastructure as that available in the unit for chronic intermittent hemodialysis, allowing more fl exibility for the treatment of the unpredictable and ever-fl uctuating number of AKI patients in need of dialysis [16]. Th is approach also allows treatment of several patients per day with the same device, in contrast to CRRT where one dedicated device is to be attributed to each single patient. IRRT also allows more liberty for patient care and investigations outside the treatment and monitoring unit, by off ering a dialysis-free period, without loss of dialysis time or adequacy. Th e extra free time also off ers opportunities to mobilize the patients during their time off dialysis, in contrast to more continuous strategies whereby the relative immobilization may result in more severe muscle wasting and a higher risk of nosocomial respiratory tract infections.
IRRT has also another advantage: machines can be used in an extended protracted mode when needed, and the treatment time can be decreased coupled to an increase in effi cacy when the condition of the patient improves. Th is contrasts with CRRT machines, which do not allow an increase of the intensity of the treatment to allow shorter treatments.

Limitation of expenses
Th ere is an increasing trend to take cost into account for clinical therapeutic decisions. Th is is more the case, if, as concluded from the data above, a clear distinction based on outcome studies is diffi cult.
In some early clinical trials in which cost estimation was not the main purpose, little diff erence in expense was demonstrated [9,39]. More recently, several specifi c cost analyses have been developed [33,[40][41][42][43]. Most of these studies were limited to one single center or a few centers, and in all of them IRRT was less costly than CRRT [33,[40][41][42]. In one study the cost of CRRT was more than double of that of IRRT [42]. Of course, analyses performed in single health service delivery settings may be biased because data may be context specifi c. A recent transcontinental multicenter and multinational analysis in the context of the Beginning and Ending Supportive Th erapy for the Kidney study, however, evaluated cost in the intensive care unit (ICU) and revealed substantial diff erences in cost by region, but in addition indicated a higher trend for global cost of CRRT for most of the locations where data were collected. Th e median diff erence was US$289.6 per day in disfavor of CRRT [43]. Th e main diff erences could be attributed to fl uid replacement and the extracorporeal circuit; reducing the substitution volume to <25 ml/minute/kg, as recently suggested [44], diminished cost only by US$67.2 per day [43]. Although the general trend for the median cost of nursing staff was slightly in disfavor of IRRT, results were highly variable and depended on whether or not extra staff were deployed for the application of CRRT. Overall, dialysate and replacement fl uid costs and extracorporeal circuit costs were in favor of IRRT worldwide, irres pective of the continent where the analysis was undertaken.
Overall, however, the data from all these studies taken together point to a cost advantage for IRRT. Of note, although the Beginning and Ending Supportive Th erapy for the Kidney study is an important step forward as it contains comparative analysis from 23 countries and fi ve continents, assessments were center-based rather than patient-based. Cost prediction was for that matter focused on general perception rather than on individual data. Further analyses might include patient-to-patient assessments of real costs comprising relative work load and related expenditures.

Restriction of bleeding complications
Owing to the continuous contact of the fi lter with blood, CRRT necessitates appropriate anticoagulation 24 hours per day -increasing the bleeding risk, especially in those who had a trauma or recent surgery or who suff er from medical diseases prone to bleeding. Th is drawback can be solved by regional citrate anticoagulation [45,46], but this option depends on skilled personnel and any mistake or technical problem may have grave consequences, such as life-threatening hypocalcemia. In units with dedicated personnel, however, outcomes with citrate might be benefi cial [47]. Owing to high diff usive clearance, citrate anti coagulation is more easily applicable and less dangerous in IRRT as compared with in CRRT.
In a large RCT, bleeding complications were more frequent in the CRRT group and were the major reason for switching modalities from CRRT to IRRT [11]. With IRRT, anticoagulation may be omitted or minimized, and does not take place all day long. Kumar and colleagues found that patients on IRRT versus those on CRRT had less need for anticoagulation, and that a much larger propor tion of patients could be treated without coagulation at all [48]. Additionally, and in spite of continuous anti coagulation, fi lter clotting tends to occur more frequently with CRRT than with IRRT [14].

Small solute removal in acute life-threatening conditions
Although in general the adequacy of IRRT and CRRT depends on the actual conditions under which the modali ties are applied, IRRT has a more effi cient immediate eff ect than CRRT when small water-soluble compounds are to be removed in an acute lifethreatening condition because of the high blood and dialysate fl ows that can be achieved, resulting in a superior clearance and mass transfer per time unit [49]. Th is is highly relevant for severe hyper kalemia, especially in the initiation phase of AKI and in patients with rhabdomyolysis, in whom potassium release from the compressed and necrotized muscle may last for several days [50]. For that reason IRRT has been used extensively in the aftermath of disasters [51,52]. Other specifi c indications profi ting from high solute removal are tumor lysis syndrome and certain cases of poisoning.

Specifi c patient populations benefi ting from intermittent renal replacement therapy
IRRT might be a useful strategy in any patient with bleeding or bleeding risk, including those after recent surgery, and is indicated for the acute treatment of hyperkalemia and rhabdomyolysis. For the global ICU population, there seems to be no clinical preference for either one of both IRRT or CRRT, but assets in favor of IRRT are its practical fl exibility and cost-eff ectiveness.

Conclusions -towards slow long-extended daily dialysis
Since both CRRT and IRRT are perfectly acceptable for clinical therapeutic use, by extrapolation the same might also apply to all intermediary solutions that lie in between.
SLEDD is a hybrid therapy, off ering advantages of both CRRT and IRRT [3, 48,53], combining protracted treatment with an intermittent time scheme, usually applying IRRT machines, and representing a high-tech return to the roots of dialysis as applied in the early days by Kolff [1]. Th e major advantages of this approach are the fl exibility of the system, the reduced costs as compared with CRRT, and the possibilities for application with low or even absent anticoagulation [54].
Comparative studies between SLEDD and CRRT resulted in similar indices of adequacy and a similar hemo dynamic response [3, 48,55,56]. In one study of 16 patients, acidosis was slightly higher and blood pressure was lower with SLEDD, but only at 2 hours after the start of treatment [57,58]. Of note, blood pressure was nonsignifi cantly lower by some 8 mmHg with SLEDD than before treatment onset. Th e appli cation of the Genius ® batch dialysis system (Fresenius Medical Care, Bad Homburg, Germany), whereby dialysate is warmed only before the start of the session and allowed subsequently to cool slowly, may have an extra positive hemodynamic impact [59]. Flexi bility lies both in the duration of and in the intensity of the treatment. In SLEDD, the blood fl ow, the dialysate fl ow and often also the rate of ultrafi ltration can be tailored to the actual needs of the patient -in contrast to CRRT, where, due to technical constraints, in practice the maximal intensity is limited. SLEDD can thus be performed as a low-intensive and prolonged treatment but also as a shorter highly intensive modality, for each option with the same machine.
Because of the adequate removal capacity, the possibility that drugs are more eff ectively cleared than with traditional short IRRT should be taken into account, so that the classically recommended drug doses might be not applicable [60,61]; especially, antibiotics might be aff ected substantially -if possible, concentrations should be measured regularly for therapeutic monitoring. Th e same might apply, however, for CRRT [62][63][64].
Most of these conclusions are extrapolated from compari sons between IRRT and CRRT (see above) with SLEDD as an intermediate strategy. Controlled outcome trials comparing SLEDD with the more traditional approaches in large populations are to the best of our knowledge lacking at this moment. Although such a comparison was not the primary aim of the study, however, indirect evidence from the Veterans Adminstration trial suggests SLEDD to yield similar out comes to CRRT and IRRT [30]. Further studies exploring potential benefi ts of each modality are awaited. In view of the heterogeneity of the ICU population and the strong infl uence of center experience with individual modalities, however, it is unlikely that a fi nal answer will ever be obtained.
In summary, CRRT and IRRT are equivalent dialysis strategies regarding outcome for the ICU patient with AKI, with a few exceptions for specifi c problems that are a direct indication for either one or the other strategy. Assets evoked in the present article in favor of CRRT are its potential for more fl uid removal in severely overloaded patients, its potential -however hardly proven -for better hemo dynamic stability in severely unstable patients, and its better tolerability in combined acute liver and kidney failure and in acute brain injury. IRRT is more practical, fl exible and cost-eff ective, allows the clinician to dis con tinue or to minimize anticoagulation with bleeding risks, and removes small solutes such as potassium more effi ciently in acute life-threatening conditions. SLEDD is a hybrid therapy combining most of the advantages of both options. All these options should not be considered as com petitors, but rather as alternatives that may be switched in the same patient depending on his/her condition and the ad hoc possibilities at a given moment in a given unit. From the practical point of view, among these modalities, SLEDD seems to off er the highest fl exibility to tailor treatment according to the individual needs of the patient.