Introduction
Continuous renal replacement therapy (CRRT) represents the most widely used modality of renal replacement therapy in intensive care units [1]. A critical determinant of effective CRRT delivery is sustained circuit patency, which directly influences treatment efficiency while simultaneously reducing healthcare costs and staff workload [2,3]. Among the proposed quality indicators of CRRT performance, filter lifespan has been consistently recognized as one of the most clinically relevant measures [3-5]. Premature filter clotting interrupts solute clearance, increases blood loss, and imposes additional financial burden [3]. Therefore, a target filter lifespan of at least 48 hours has been recommended as a benchmark for adequate CRRT delivery [3].
Critically ill patients frequently exhibit a procoagulant state, creating the need for effective anticoagulation while carefully limiting bleeding risk [2,6]. Systemic anticoagulation with unfractionated heparin (UFH) remains the most used strategy worldwide [2,7,8]. Nevertheless, UFH may be inappropriate for patients with severe illness and underlying coagulation abnormalities, and bleeding complications associated with its use are well documented [7-9]. Nafamostat mesilate (NM; Futhan, SK Chemicals) has been utilized in Korea and Japan as an alternative anticoagulant for patients at high risk of bleeding [10]. NM is a synthetic serine protease inhibitor that inhibits activation of both coagulation factors and platelets [11]. Owing to its extremely short biological half-life of approximately 8 minutes, NM exerts its anticoagulant effect predominantly within the extracorporeal circuit and, unlike UFH, does not produce systemic anticoagulation [12-16]. Despite these pharmacological advantages, clinical evidence supporting the efficacy and safety of NM in pediatric CRRT remains limited [11]. This evidence gap is largely attributable to the geographically restricted use of NM, primarily in East Asian countries such as Korea and Japan. In 2014, Lee and Cho [2] proposed a stepwise anticoagulation protocol incorporating NM for pediatric CRRT (Fig. 1). Under this protocol, patients without a high risk of bleeding received conventional UFH anticoagulation, whereas patients at high risk were initially managed with anticoagulation-free CRRT. NM was introduced when hemofilter lifespan fell below 12 hours in the absence of anticoagulation. The starting NM dose was 0.25 mg/kg/hr, with subsequent dose adjustments guided by activated clotting time (ACT). Implementation of this strategy resulted in a significant prolongation of hemofilter lifespan from 7 to 27 hours [2].
The objective of the present study was to evaluate the real-world effectiveness of the NM-based protocol introduced in 2014. We hypothesized that NM would demonstrate comparable efficacy and safety to UFH in pediatric patients undergoing CRRT. We further explored whether, in selected pediatric patients, anticoagulation-free CRRT could be feasible during the initial treatment period.
Methods
Participants
This single-center retrospective cohort study included pediatric patients who underwent CRRT at Samsung Medical Center between January 2018 and December 2022. Eligible participants were pediatric patients aged 0–18 years, and informed consent was waived per protocol. Patients who received CRRT for ≥24 hours during hospitalization were included. The exclusion criteria were defined as follows: (1) use of more than two anticoagulation strategies during a single CRRT course; (2) death or discontinuation of CRRT within 24 hours of initiation; (3) inability to accurately determine the start or end time of hemofilter use; or (4) change in CRRT modality during treatment.
Between January 2018 and December 2022, 107 pediatric patients who underwent CRRT for at least 24 hours at Samsung Medical Center met the eligibility criteria and were included in the final analysis. Medical records were retrospectively reviewed to extract data on sex, age, underlying diagnoses, duration of CRRT, anticoagulation strategy, hemofilter lifespan, and relevant laboratory findings.
CRRT equipment and protocol
CRRT was performed using the Prismaflex system (Gambro Healthcare). Hemofilters were selected according to each patient’s body surface area (BSA): HF-20 for BSA 0.2 m², ST-60 for BSA 0.6 m², and ST-100 for BSA 1.0 m². The CRRT modality used was continuous veno-venous hemodiafiltration. Vascular access was established via the internal jugular or femoral vein. Anticoagulation during CRRT was managed according to the NM protocol proposed in 2014. Standard CRRT settings included a blood flow rate of 3–5 mL/kg/min and a combined dialysate and replacement fluid rate of 2,000 mL/1.73 m²/hr. The routine hemofilter exchange interval was ideally 72 hours.
Definition of hemofilter lifespan
Hemofilter lifespan was defined as the duration from filter initiation to filter removal due to circuit clotting, scheduled exchange at 72 hours, or removal at the time of patient death. The mean hemofilter lifespan was calculated by dividing the total CRRT treatment time by the total number of hemofilter sets used.
Anticoagulation strategy
Patients were assessed for bleeding risk before CRRT initiation. A bleeding tendency was defined by the presence of one or more of the following: prothrombin time–international normalized ratio > 2.0, activated partial thromboplastin time (aPTT) >60 seconds, ACT ≥200 seconds, platelet count <50,000/mm³, major surgery within the preceding 7 days, or intracranial hemorrhage or neurosurgical procedures within the preceding 14 days. During CRRT, anticoagulation was managed according to the NM protocol (Fig. 1). Patients at high risk of bleeding initially received no anticoagulant therapy and were monitored using circuit ACT and hemofilter lifespan. NM was initiated when the hemofilter lifespan was <12 hours. The starting NM dose was 0.25 mg/kg/hr and was titrated to maintain an ACT of 180–220 seconds. Outside the protocol, NM was also initiated at the start of CRRT in patients clinically suspected of having a hypercoagulable state despite a bleeding risk, such as those with disseminated intravascular coagulation or neonates with small-caliber catheters. UFH was administered to patients without a definite bleeding risk. An initial bolus of 20 U/kg was followed by continuous infusion at 10 U/kg/hr. Plasma aPTT was measured every 6 h, with a target range of 45–60 seconds.
Comparison of anticoagulants
To evaluate the efficacy and safety of different anticoagulants, patients were classified into three groups according to their anticoagulation strategy: (1) Group 1 (no anticoagulation); (2) Group 2 (NM); and (3) Group 3 (UFH). Within Group 2, patients were further stratified into two subgroups: those initially managed without anticoagulation who later received NM due to early filter failure, and those who received NM from the initiation of CRRT. The efficacy of each anticoagulant was assessed based on hemofilter lifespan.
In this retrospective cohort, the decision to initiate NM was based on a combination of protocol-defined bleeding-risk criteria and clinical judgment. The major indications for NM use were: (1) inadequate hemofilter lifespan during anticoagulation-free CRRT in patients initially managed without anticoagulation because of bleeding risk; and (2) clinically suspected hypercoagulability despite bleeding tendency, such as disseminated intravascular coagulation or anticipated circuit vulnerability in neonates with small-caliber catheters. Because these indications were frequently overlapping in actual practice, treatment allocation should be interpreted as reflecting multifactorial clinical decision-making rather than a single uniform criterion.
Statistical analysis
Statistical analyses were performed using SPSS software (version 25.0). Comparisons of hemofilter lifespan before and after NM initiation within the same patients were conducted using the Wilcoxon signed-rank test. Differences in baseline characteristics and hemofilter lifespan among the three anticoagulation groups were analyzed using analysis of variance. A two-sided P-value <0.05 was considered statistically significant.
Results
Population characteristics
The mean age of the study population was 7.8 years, and the male-to-female ratio was 1:0.9 (Table 1). Among the underlying diagnoses, solid tumors were the most prevalent, followed by hematologic malignancies. Other conditions included inborn errors of metabolism, congenital heart disease, and chronic kidney disease. The most common indications for CRRT were thrombotic microangiopathy (TMA) or veno-occlusive disease (VOD), followed by sepsis. Additional indications included metabolic derangements, heart failure, and tumor lysis syndrome. The mean duration of CRRT was 11.7±11.9 days, and the mean hemofilter lifespan was 30.6±14.4 hours. Overall survival was 61.7% (66/107).
Comparison of anticoagulation strategies
Among the three anticoagulation groups, 42 patients were managed without anticoagulation (Group 1), 38 received NM (Group 2), and 27 received UFH (Group 3) (Table 2). Within Group 2, 11 patients initially underwent anticoagulation-free CRRT and subsequently received NM due to early filter failure, while the remaining 27 patients received NM from CRRT initiation according to institutional practice and individualized clinical judgment. In these non-rescue cases, NM use was generally considered in the context of bleeding risk and clinical concern for circuit vulnerability or coagulation-related complexity; however, per-patient indications were not systematically collected.
Hematologic and oncologic conditions accounted for a significant proportion of underlying diseases across all groups. TMA/VOD and sepsis were the most common indications for CRRT in Groups 1 and 2. In contrast, a considerable proportion of patients in the UFH group had congenital heart disease and required CRRT due to heart failure. This difference is likely attributable to the frequent use of CRRT in conjunction with extracorporeal membrane oxygenation and systemic heparinization.
Platelet count was the only baseline laboratory variable that differed significantly among the three groups. This finding is clinically important because thrombocytopenia was one of the predefined criteria used to assess bleeding risk and to guide anticoagulation strategy. The lower platelet count in the no-anticoagulation group, relative to the NM and UFH groups, supports the presence of protocol-based baseline differences among groups. No significant differences were observed in survival rate or hemofilter lifespan between the anticoagulation groups.
Hemofilter lifespan and complications
The mean hemofilter lifespan was 32.1 hours in the anticoagulation-free group, 26.7 hours in the NM group, and 33.9 hours in the UFH group. No statistically significant differences were observed among the three groups (P=0.092) (Table 2). The relatively shorter hemofilter lifespan in the NM group may be attributable to the higher clotting tendency within this cohort. Within Group 2, the 11 patients who initially underwent anticoagulation-free CRRT exhibited a mean hemofilter lifespan of 19.8 hours during the anticoagulation-free period. Following the introduction of NM, hemofilter lifespan increased to 27.0 hours, representing a statistically significant improvement (P=0.030) (Fig. 2). No major bleeding complications were reported in any of the three groups. One minor episode of oozing at the catheter insertion site was observed in the UFH group.
Discussion
The most clinically meaningful finding of this study was a significant increase in hemofilter lifespan after NM initiation in a subgroup of 11 patients who transitioned from anticoagulation-free CRRT due to early filter failure. Hemofilter lifespan increased from 19.8 to 27.0 hours after NM initiation, supporting the potential clinical utility of NM as a rescue anticoagulation strategy in pediatric patients with bleeding risk. In the overall cohort, mean hemofilter lifespan did not differ significantly among the non-anticoagulant, NM, and UFH groups. Although a numerically shorter mean filter lifespan was observed in the NM group, this finding should be interpreted cautiously because NM selection reflected both protocol-based and individualized clinical considerations, including bleeding risk and concern for circuit clotting. In particular, the subgroup finding after NM initiation is informative, because it reflects a within-patient change following inadequate circuit survival during anticoagulation-free CRRT.
In this study, NM was associated with a mean hemofilter lifespan of 27 hours, whereas the UFH group showed a slightly longer mean lifespan of 34 hours. Nevertheless, hemofilter lifespan in the NM group was not significantly different from that in the UFH group. Among the 11 patients in whom NM was added in response to circuit clotting, hemofilter lifespan increased significantly from 19 to 27 hours, indicating an effective anticoagulant effect. This finding aligns with previous reports, such as a 2014 study demonstrating a prolonged filter lifespan of 27.4 hours after NM administration [2]. Overall, bleeding complications were rare across all groups. No bleeding events were observed in the NM group, while one minor bleeding episode occurred in the UFH group. These results suggest that NM may offer efficacy comparable to UFH in selected pediatric CRRT patients at high risk of bleeding, while maintaining a favorable safety profile. Although the mean hemofilter lifespan in our cohort was below the commonly suggested 48-hour benchmark, this finding should be interpreted in the broader clinical context of pediatric CRRT, in which circuit survival may be affected by patient size, vascular access, disease severity, bleeding risk, and institutional practices regarding filter replacement. In patients with sepsis, more frequent filter exchange may be performed to optimize cytokine adsorption, independent of circuit clotting. Also, the 48-hour target applies to individual filters, whereas our analysis used the mean lifespan across all filters; therefore, the mean value is inherently influenced by shorter-duration filters and may not reflect the maximal achievable lifespan. Some filters in our cohort were maintained beyond 48 hours, but filter-level data on the proportion of circuits maintained for more than 48 hours and the specific reasons for filter replacement were not reliably available. Therefore, the observed mean hemofilter lifespan should be interpreted as a real-world measure influenced by both circuit performance and clinical practice patterns, rather than as a direct measure of maximal circuit patency.
Our findings should be interpreted in the context of previous pediatric studies of NM during CRRT. In pediatric cohorts, a previous study by Lee and Cho [2] reported prolongation of hemofilter lifespan after NM initiation in children at high risk of bleeding, and Miyaji et al. [16] found that NM showed efficacy comparable to citrate anticoagulation. More recently, He and Zhang reported similar or slightly longer filter lifespan with NM compared with other anticoagulation strategies in pediatric acute liver failure [17]. Compared with prior studies, the present study adds clinically relevant information by highlighting the subgroup of patients in whom NM was introduced after inadequate filter lifespan during anticoagulation-free CRRT. This within-patient improvement may be particularly informative in real-world pediatric practice, where anticoagulation decisions are often individualized according to bleeding risk rather than standardized across all patients.
Interestingly, a relatively preserved hemofilter lifespan was observed in the no-anticoagulation group, suggesting that anticoagulation-free CRRT may be feasible in selected patients. This observation likely reflects the clinical practice of reserving anticoagulation-free CRRT for patients in whom circuit patency remains acceptable despite bleeding risk, as well as the influence of patient-specific and circuit-related factors (e.g., catheter characteristics and underlying disease-specific coagulation profiles). However, rather than supporting a generalized strategy, this finding highlights the potential for a more individualized approach to anticoagulation during CRRT. Circuit survival during CRRT is influenced by multiple patient-related, circuit-related, and practice-related factors [15,18]. Therefore, this finding should not be interpreted as evidence that anticoagulation-free CRRT is equivalent to anticoagulated CRRT. Further studies are needed to better define the clinical settings in which anticoagulation-free CRRT can be safely applied, thereby minimizing unnecessary exposure to anticoagulants and reducing the risk of related complications.
This study has several limitations inherent to its retrospective design. First, its retrospective design resulted in a small number of patients receiving NM. Second, a substantial proportion of patients received NM from the outset based on clinical conditions, with only 11 patients switched from no anticoagulation to NM following the protocol. In the long term, protocol amendments may be warranted to clarify the indications for initiating NM as first-line anticoagulation. Third, treatment allocation was not randomized but determined according to a predefined institutional protocol combined with clinical judgment. As a result, baseline differences between groups, such as platelet count, reflect protocol-driven risk stratification rather than random variation. Patients with higher bleeding risk were more likely to receive no anticoagulation or NM, whereas UFH was preferentially used in patients with lower bleeding risk. Therefore, between-group comparisons should not be interpreted as reflecting causal differences in efficacy between anticoagulation strategies. Instead, these comparisons primarily reflect differences in underlying patient characteristics and clinical indications. Given this design, the more clinically relevant finding of this study is the within-patient improvement in hemofilter lifespan following NM initiation in patients with inadequate circuit survival during anticoagulation-free CRRT, which directly aligns with the intended use of the protocol. This limits causal interpretation of between-group comparisons, particularly for hemofilter lifespan. Furthermore, the calculated mean hemofilter lifespan may not fully represent the maximal achievable circuit patency. In our real-world clinical setting, filters could be intentionally replaced before complete circuit failure according to the patient’s clinical condition and local practice patterns, even when the circuit remained functional. Thus, the use of mean hemofilter lifespan as the primary efficacy outcome may have underestimated circuit survival in some patients. However, because no case reached the routine 72-hour exchange point, the overall circuit performance in this cohort should still be interpreted cautiously.
In conclusion, NM may be a feasible and safe anticoagulant option for pediatric CRRT patients at high risk of bleeding. Its efficacy appeared comparable to that of UFH in this cohort, but these findings should be interpreted cautiously given the retrospective design and baseline differences between groups.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Print
