Expanding the Spectrum of Extracorporeal Strategies in Small Infants with Hyperammonemia

Hyperammonemia is a life-threatening condition mainly due to the neurotoxicity of ammonia. Mortality and neurological outcome are strongly related to the duration and severity of hyperammonemia; therefore, a rapid decrease of the ammonia level is strongly recommended [1]. When ammonia level is very high or rising rapidly, ammonia scavengers may be insufficient and extracorporeal treatment may be required [2]. The ammonia generation rate observed in infants with inborn errors of metabolism (IEM) is sustained; therefore, high dialysis clearance is required [3]. The most intensive treatment is intermittent hemodialysis (HD) which can achieve enormous clearance in a short time [4]. However, intermittent treatments involve some critical issues in these patients including rebound hyperammonemia, rapid fluid and osmotic shifts, and a higher risk of hemodynamic instability. On the contrary, continuous dialysis treatments allow to remove ammonia with less hemodynamic risk, avoiding rebound hyperammonemia and reducing the severity of electrolyte imbalances. Continuous renal replacement therapy (CRRT), specifically high-dose continuous veno-venous HD (CVVHD), is the recommended first-line treatment for hyperammonemia when possible [5]. High-dose CRRT treatment must be prescribed to ensure ammonia depletion fast enough to reach a concentration <200 umol/L in a few hours; a dialytic dose up to 8,000 mL/kg/1.73 m² is therefore required [6].

Many of the children who need this type of treatment are newborns and small infants, who physiologically have a much lower total blood volume than an older child or an adult. The majority of the CRRT systems have been developed for use in adult patients with large extracorporeal priming volumes and ultra-filtrate (UF) control systems that are inadequate for low body weight children [7, 8]; therefore, they are used off-label on such small patients. Recent strides have been made in the development of CRRT systems more suitable for young children with smaller sets and filters to use in adult machines and dedicated monitors for neonates and infants [8‒10].

Investigating the landscape of CRRT options for young children (<10 kg) with hyperammonemia, it becomes clear that there are 2 options, both with pros and cons. The first is to use adult CRRT systems with pediatric sets, and the second is to use monitors entirely developed for small infants. Table 1 summarizes some technical characteristics and treatment parameter ranges of some conventional CRRT monitors with pediatric lines and two specifically designed pediatric devices.

Overall, the main advantage of the machines for adults is the possibility of using higher dialysis fluid flows (Qd), allowing treatments with high dialysis doses. However, in small children, this choice brings greater hemodynamic risks due to the larger extracorporeal volume and the lower precision of the UF control system. The smallest pediatric set developed for use in adult machines is the Prismaflex/PrisMax HF20^® (Baxter) with an extracorporeal volume of about 60 mL. A 3.5 kg infant has a blood volume of approximately 300 mL; if the HF20 set is used, the priming would be approximately 20% of his body volume.

The extracorporeal circuit priming volume ideally should not exceed 10% of a patient’s circulating blood volume in order to avoid acute blood depletion from the circulation at the start of the extracorporeal treatment which may increase cardiovascular instability. In cases where the use of sets with priming volumes greater than 10% is necessary, priming with albumin or blood is recommended in order to prevent hypotension and hemodilution anemia. Despite the use of protocols to make the blood prime more physiological, this exposes the patient to the risk of potassium and citrate overload, thrombocytopenia, transfusion reactions, and infections. Therefore, the use of large extracorporeal circuits in small children should be minimized.

Pediatric monitors have been designed to reduce the volume of the extracorporeal circuit and to increase the precision of the treatment with the aim of accommodating the unique characteristics of newborns and infants. As shown in Table 1, there are currently available two CRRT machines that are licensed to treat neonates, small infants, and children, the CARPEDIEM™ (Bellco-Medtronic) and the NIDUS^® (Allmed). Priming volumes are significantly smaller than those of the pediatric sets in adult CRRT monitors (the largest set on the CARPEDIEM^™ has a volume of 41 mL). In a recent retrospective cohort analysis, the CARPEDIEM^™ monitor has been shown to be a safe and efficient out-of-body method for infants with better outcomes compared to adult-adapted machines [11]. Although the study was carried out prior to the approval of the HF20 set, the conclusions remain valid, suggesting the use of the pediatric monitor in newborns and infants. However, both pediatric monitors have substantial limitation in clearance performances as they are mainly conceived for slow and continuous treatments. The dialysis fluid pump in CARPEDIEM^™ provides a maximum flow of 600 mL/h. NIDUS^® monitor has a syringe pump which allows an even lower Qd with a maximum of 400 mL/h. Comparing these data with those of some pediatric sets in adult monitors, HF20 on Prismaflex/PrisMax^®, AVPaed on MultiFiltrate/MultiFiltrate Pro^® (Fresenius), and Aquamax HF03 on Aquarius^® (Nikkiso), the great inferiority of clearance of pediatric CRRT monitors clearly emerges. To treat an infant with AKI admitted to the neonatal intensive care unit, the Qd provided by CARPEDIEM^™ and NIDUS^® is usually sufficient, given that the standard dialysis dose is 2,000 mL/h/1.73 m². However, as mentioned above, in neonates and infants with IEM, the ammonia generation rate is tremendous, much greater than that of urea, potassium, and phosphate in AKI, making it necessary to triple or quadruple the dialysis dose up to 8,000 mL/h/1.73 m² [6].

In order to use a pediatric dedicated CRRT machine that has been shown better outcomes compared with adult-adapted systems and to minimize the use of blood priming, we have investigated a way to make the CARPEDIEM^™ more performing. Ammonia is a small molecule (17 Da), hydrophilic and not bound to proteins, which is highly clearable by diffusive methods. This is why the CVVHD is the recommended treatment. However, it is also well removed by convective techniques, as all the small molecules, with an almost one sieving coefficient. Therefore, if a convective component is added to the diffusive method, the ammonia clearance should increase. Since, for technical reasons, CARPEDIEM^™ does not have the continuous veno-venous hemodiafiltration (CVVHDF) function, we have applied a system to add a convective component to the CVVHD treatment.

We treated four infants with hyperammonemia using the CARPEDIEM^™ set in CVVHD mode, implemented in CVVHDF modality, boosting the diffusive clearance with a post-replacement convective mechanism. To circumvent the technical limitations of the machine, the monitor was set in CVVHD mode and a replacement fluid (Qr) was administered directly to the patient through a second central venous line (CVL), setting the UF on the CARPEDIEM^™ monitor equal to Qr. The patients were treated in the pediatric intensive care unit and neonatal intensive care unit of the University Hospital of Padova, between November 2016 and April 2022. Informed consent was obtained for all patients in light of the off-label method. In all cases, an HCD 025 dialyzer was used. The CVVHD parameters have been chosen in order to maximize the ammonia clearance, taking into account all technical limitation of the pediatric monitor (CARPEDIEM^™ parameters are shown in Table 2). The Qb was set up to 50 mL/min according to the maximum allowed by patient hemodynamics and the catheter size. Qd was set up to 600 mL/h. Total weight loss was set to be equal to the total Qr volume or greater if a net weight loss was desired. If the total weight loss was higher than 2,000 g, the total treatment time would be less than 24 h, requiring the use of a new circuit. In Figure 1, it is showed the ammonia clearance of the first patient treated with this technique in 2016. The CRRT was started with CARPEDIEM^™ monitor set in CVVHD mode achieving disappointing results (dashed line in the graph), prompting us to look for a way to improve the effectiveness of the treatment. We, therefore, added the convective component giving the Qr to the patient through a second CVL and removing it as UF. This solution considerably increased the clearance, and an ammonia value of 200 μmol/L was then reached within 8 h (solid line in the graph). This setting has been repeated in the subsequent cases with positive results. Patient characteristics, treatment parameters, and ammonia clearance are shown in Table 3. It must be specified that our 4 cases had different types of hyperammonemia. In the first and in the last case, hyperammonemia was caused by IEMs characterized by a very high ammonium generation rate [3]. In the second and third case, the cause was instead acute liver failure, with different ammonium generation rates. Furthermore, the starting blood ammonium levels were very different: about 700 μmol/L in the first and fourth and around 400 μmol/L in the second and third cases. Consequently, the required clearance rate was different for the patients within the two diagnostic categories. Finally, the 2 patients with IEM were newborns and the two children with acute liver failure were infants, therefore with significantly different weights and body surface areas. This difference in size obviously has a significant impact on the technical limitations of CARPEDIEM^™, with higher Qb and Qd per kilogram available in smaller patients. In all patients, we set the CARPEDIEM^™ on CVVHD mode using the largest dialyzer (HCD 025) to ensure the best clearance possible, despite the fact that it was oversized for the 2 newborn patients. The Qb was set at the maximum allowed by the catheter, the machine, and the patient hemodynamics. We set the Qd to the higher value (600 mL/h) except in the first case where it was 300 mL/h. Although during the CVVHDF the parameters were the maximum allowed by the machine and the total weight loss was much higher than the standard, there were no hemodynamic issues and no major machine alarms. In the 2 patients with IEM, the blood ammonium level decreased under 200 μmol/L in about 8 h in the first case with a clearance rate of 64 μmol/L/h and after 6 h in the fourth case with a clearance rate higher than 89 μmol/L/h. In the 2 patients with acute liver failure, the time needed to reduce the ammonium under 200 μmol/L was 21 h in the second patient and about 8 h in the third child with a clearance rate of 7 μmol/l/h and 24 μmol/l/h, respectively. These findings are consistent with the dialytic doses, that is, the sum of Qd and Qr, administrated to the patients: 4,697 mL/h/1.73 m² for the first patient, 3,703 mL/h/1.73 m² for the second, 3,208 mL/h/1.73 m² for the third, and 6,119 mL/h/1.73 m² for the last. The rate of ammonium decrease in the two infants with liver failure was significantly lower than in the two newborns due to a lower dialytic dose.

By adding a convective component to the CVVHD mode, the treatment is implemented in a true CVVHDF. In this way, the dialytic dose could be increased from 600 mL/h to 900 mL/h (the maximum flow rate of the CARPEDIEM^™ effluent pump). However, the flow rate of the dialysis fluid pump and that of the effluent pump are not the only limitations of this machinery (Table 2). In fact, the CARPEDIEM^™ monitor has a UF limit of up to 2 L per treatment, a safety parameter introduced to avoid excessive weight loss in small patients. Therefore, after every 2 L of UF, the software interrupts the treatment and requests to change the set. Every set has a cost, and every treatment stop increases the risk of ammonia rebound. This is the reason we suggest a maximum Qr and consequently a maximum UF rate of 175 mL/h which allows to continuing the treatment for 12 h before having to change the entire dialysis set. As already mentioned, using this strategy, a dialytic dose of approximately 6,000 mL/h/1.73 m² can be delivered. In a recent case series, Spinale et al. showed a good ammonia clearance in two neonates with hyperammonemia due to IEM treated with high-dose CRRT using adult-based platforms [6]. They reached a dialytic dose of 8,650 mL/h/1.73 m² in the first patient and 7,700 mL/h/1.73 m² in the second newborn, suggesting a dialytic dose up to 8,000 mL/h/1.73 m² to decrease the ammonia concentration as fast as possible. In Figure 2, we compared the ammonia clearance achieved in these two newborns, with the median values of our 4 patients (median dialytic dose of 4,200 mL/h/1.73 m²) and with the clearance obtained in our fourth newborn, the one who was treated with a dialytic dose of approximately 6,000 mL/h/1.73 m². In the graph, we have entered the ammonia levels of Spinale et al. patients starting from values close to ours (about 800–700 μmol/L), although the initial ammonia values were higher in their original work. This was done to make the data more comparable as ammonia clearance is greater at higher ammonia levels and because if we also considered the high starting values, the time required to reach the 200 μmol/L threshold would be falsely longer. In the 2 patients described by Spinale et al., the time required to decrease the ammonia below 200 μmol/L would be between 3 and 4 h in the first and between 4 and 5 h in the second newborn. These results do not differ considerably from what we achieved in our fourth patient (between 5 and 6 h). Considering our median values, however, the time increases to 8 h, highlighting the worse clearance obtained in our other 3 patients treated with lower dialytic doses. Taking into account these limitations, according to our experience, the CARPEDIEM^™ machine used in a CVVHDF modality can provide an effective dose to infants weighing up to 3.5 kg. Up to that weight limit, in fact, it is possible to guarantee a dialysis dose of at least 6,000 mL/h/1.73 m² with a reasonable set change frequency; as an example, in a 3.5 kg infant (median BSA of 0.22 m²), setting the Qd to the maximum allowed by the machine (600 mL/h) and a Qr of 175 mL/h (which allows a set change every 12 h), the dialysis dose would reach 775 mL/h equal to 6,100 mL/h/1.73 m². In order to maximize the ammonia clearance, we suggest the use of the HCD 025 dialyzer regardless of infant size. Although this set has a much smaller priming volume (41 mL) than the pediatric sets useable with adult machines, when applied in infants weighing 3.5 kg or less, it results in an extracorporeal volume >10% of the patient's blood volume; the use of albumin as priming is therefore suggested [11].

According to our experience and given the technical limitations, CARPEDIEM^™ implemented in CVVHDF modality could be used to treat newborns and small infants with hyperammonemia up to 3.5 kg. Although effective, this approach is associated with a series of drawbacks. First, the administration of a convective dose requires a second CVL. However, this issue could be resolved by providing a pre- or post-filter convective component directly through the access or return line of the device. Second, the use of non-standard approach requires specific nursing and medical training that carries with it additional risks.

For children weighing more than 3.5 kg, CARPEDIEM^™ is in practice unable to deliver high dialysis dose; therefore, adult-adapted, but non-licensed, monitors should be used. Considered the CRRT machines available in our center, we built an algorithm useful to guide the physician in the choice of the CRRT set that best fits with patient’s size, taking into account the extracorporeal volume, the maximum Qd allowed, and the filter surface area (Fig. 3).

In 2021, Eloot et al. [3] proposed an algorithm to treat newborns with hyperammonemia in which the choice of dialysis method was based on ammonia levels. In particular, the use of intermittent dialysis with adult machines was recommended in case of ammonia >400 μmol/L and the choice of CRRT with CARPEDIEM^™ monitor in case of lower ammonia levels or as de-escalation. However, the use of maintenance HD machines in very small infants put them at high hemodynamic risk due to the intensity of the treatment and its relatively large extracorporeal volume. In our proposal, treatment modality is based on patient’s size and includes only CRRT, in light of the growing evidence of the aforementioned better hemodynamic tolerability, the lower risk of rebound hyperammonemia, and osmotic shifts of continuous treatments. Regarding the extracorporeal volume, our algorithm guides the choice for the smallest CRRT set able to perform high flow dialysis based on patient’s weight.

The solution we used to increase the efficacy of CARPEDIEM^™ was adopted to face the technical limits of the monitor. However, implementing a “high flow treatment” option in the machine software would make treatment even more precise and safer.

The use of CVVHDF with the CARPEDIEM^™ has been shown to be effective in treatment of hyperammonemia in infants, expanding the possible applications of this dialysis monitor, providing an alternative to the current need of adult-based devices. We suggest using this monitor for the treatment of infants up to 3.5 kg and choosing an adult-based machine for larger children, in order to find the best balance between guaranteed dialytic dose and treatment invasiveness.

The authors have no conflicts of interest to declare.

No funding resources were required.

Giovanni Ceschia analyzed the patient data and wrote the manuscript. Mattia Parolin, Germana Longo, and Enrico Vidal participated in the clinical care of the patients. Enrico Vidal and Claudio Ronco critically reviewed the manuscript. Enrico Vidal supervised the whole process. All the authors read and approved the final manuscript.