Abstract
Background and Aims
Extracorporeal cytokine removal may be desirable. We sought to assess extracorporeal blood purification (EBP) techniques for cytokine removal in experimental animal studies.
Methods
We conducted a targeted, systematic search and identified 17 articles. We analyzed cytokine clearance, sieving coefficient (SC), ultrafiltrate (UF) concentration, and percentage removal. As this review concerns technical appraisal of EBP techniques, we made no attempts to appraise the methodology of the studies included. Results are in descriptive terms only.
Results
Applying predicted clearance for 80 kg human, high volume hemofiltration (HVHF) techniques and plasmafiltration (PF) showed the highest rates of cytokine removal. High cutoff (HCO)/HF and PF techniques showed modest ability to clear cytokines using low to medium flows. Standard hemofiltration had little efficacy. At higher flows, HCO/HF achieved clearances between 30 and 70 ml/min for IL-6 and IL-10. There was essentially no removal of tumor necrosis factor (TNF)-alpha outside of PF.
Conclusions
Experimental animal studies indicate that HVHF (especially with HCO filters) and plasmafiltration have the potential to achieve appreciable IL-6 and IL-10 clearances. However, only PF can remove TNF-alpha reliably.
Introduction
The role of cytokines in the pathophysiology of various acute inflammatory states is continuously being defined (1). The stimuli for cytokine release are many and may come in the form of pathogen-associated molecular patterns (PAMPs) in the case of microbial sepsis, or damage-associated molecular patterns (DAMPs) in the case of systemic inflammatory response syndrome (SIRS) induced tissue injury (2, 3). Cytokines are not normally present in large amounts in the circulation, so the presence of excessive amounts may herald the development of multiorgan dysfunction and high mortality (4). Additionally, imbalances between pro-inflammatory and anti-inflammatory cytokines occur at different stages of critical illness and are believed to contribute to a poor outcome (5).
Over the last few decades, attempts have been made to develop therapies that will help to regulate the level of cytokines in the circulation and restore cytokine balance. Such therapies should ideally be able to modulate several cytokines at the same time and decrease cytokinemia to levels associated with optimal outcomes rather than aim for complete elimination (6). As the timing and degree of cytokine release may vary under different circumstances even in the same patient, ideally these therapies should be continuous and self-adjustable to cytokine levels.
In theory, extracorporeal blood purification (EBP) techniques approach this ideal (7). Several characteristics of most cytokines make them a logical target for extra-corporeal removal. They are of middle molecular weight; they are water soluble; they are often found free of protein binding (8). Because of these characteristics and the belief that restoration of cytokine homeostasis is desirable, several investigators have sought to develop techniques of EBP which can achieve cytokine removal (9). We have recently assessed the efficacy of different EBP techniques at cytokine removal in ex vivo conditions (10). The relevance of such data to removal in vivo, however, remains poorly understood. In particular, no studies have systematically assessed the efficacy of different extracorporeal techniques in achieving such goals in the in vivo setting of experimental animal studies. The additional assessment of EBP techniques in preclinical studies is vital to the development of a logical approach to the selection of the optimal technique of cytokine removal in humans. Accordingly, we conducted a systematic review of all experimental animal studies of EBP techniques for cytokine removal.
Methods
We conducted a systematic search using the Pubmed and Embase databases for relevant articles on in vivo animal experimental studies on cytokine removal using known modalities of EBP. We then systematically assessed the efficacy of all EBP techniques previously reported in the literature using these data.
Our approach at identifying relevant articles for analysis is outlined in Figure 1.
Flow diagram summarizing manuscript review process.
The following search terms were used: ‘cytokine’ AND ‘continuous renal replacement therapy’; ‘cytokine’ AND ‘hemofiltration’; ‘cytokine’ AND ‘hemodiafiltration’; ‘cytokine’ AND ‘high volume hemofiltration’; ‘cytokine’ AND ‘adsorption’; ‘cytokine’ AND ‘plasmapheresis’; ‘cytokine’ AND ‘bioartificial kidney’ and ‘cytokine’ AND ‘coupled plasma filtration adsorption’. To ensure no relevant articles were missed, further searches were conducted using the term ‘interleukin’, ‘interleukin-1’, ‘interleukin-2’, ‘interleukin-6’, ‘interleukin-8’, ‘interleukin-10’, ‘interleukin-18’ and ‘tumor necrosis factor alpha’. All these terms were combined using the term ‘AND’ with the terms ‘hemofiltration’, ‘hemodiafiltration’, ‘adsorption’, ‘plasmapheresis’, ‘renal replacement therapy’, ‘bioartificial kidney’, ‘high volume hemofiltration’, ‘CPFA’ and ‘coupled plasma filtration adsorption’. All the terms used were MESH and Emtree terms except for ‘bioartificial kidney’, ‘high volume hemofiltration’, ‘CPFA’ and ‘coupled plasma filtration adsorption’ which are keyword searches as neither MESH nor Emtree terms for these exists. ‘Renal replacement therapy’ is a MESH term but not an Emtree term. All MESH and Emtree terms were also searched as keyword searches. Non-MESH or non-Emtree terms were searched as keyword searches in the respective databases using both British and American English spelling.
Abstracts of articles retrieved were then screened for two inclusion criteria: experimental animal studies and the reporting of a numerical value of at least one of these measures of cytokine removal: clearance, sieving, percentage removal, or concentration in the filtrate. Two independent researchers performed the search and then manually screened retrieved articles for those which meet both inclusion criteria. Abstracts that did not include enough details as well as publications with no abstracts were traced using library resources and each paper was screened for inclusion criteria. We excluded review articles and articles published in languages other than English.
As this review is concerned with technical appraisal of each EBP technique in the terms described above and not clinical outcomes such as survival, we have not made attempts at appraising the methodology of each study identified. We sought to identify all articles that have been published on the subject so that a fair conclusion can be attempted without overlooking any particular technique. Studies are only excluded if the results appear to be a duplication. There were also real difficulties in appraising the quality of the individual studies as the animals, techniques, and operating characteristics that they used were rather varied. Data which were reported only in the form of graphs or figures had their numerical values estimated from the details given in the graphs. When more than one measurement was available, an average value was calculated. In an attempt to assimilate data from different animal experiments into a degree of homogeneity, a predicted CL in an 80 kg human was then calculated based on the CL and SC data. The CL values achieved using various devices and operating characteristics were then compared using this predicted data. The information was then analyzed to seek out techniques that offer the highest rate of cytokine removal based on experimental animal data. Where sufficient data were available, these techniques were then analyzed for operating characteristics that appeared to offer the best rates of cytokine removal.
In an attempt to derive meaning and a degree of uniformity that could be applied clinically to experiments that were conducted using a range of small, medium, and large animals, we calculated a “predicted clearance” in an 80 kg human. The calculation was done in the following manner: from the data given in the papers, the UF rate in ml/kg per hr was calculated for the animal subjects and then the equivalent UF rate for an 80 kg human was derived in ml/min. The product of the sieving coefficient (SC) from the animal studies and ultrafiltration (UF) rate in ml/min gave the value of CL in ml/min for an 80 kg human. Where data on UF concentration, pre- and postfilter plasma concentration were available, SC was calculated using the formula: 2 × UF concentration / (prefilter concentration + postfilter concentration) and CL derived from the calculated SC. For plasmafiltration or selective plasma filtration where data were only available in the form of CL, SC was first derived through CL divided by plasmafiltration rate in ml/min. The product of this derived SC and the derived rate of plasmafiltration flow (ml/min) in an 80 kg human equals CL. Data on percentage removal was not analyzed further because of insufficient data and insufficient shared features across experiments.
Due to the limited amount of data, we did not calculated means, standard deviations, medians, or interquartile ranges. Rather than make any statistical comparisons, we kept to descriptive terms.
Results
The data extraction process is summarized in Figure 1. In terms of definitions, the term “standard technique” was used to refer to the use of standard high flux hemofilters at standard doses of filtrate flow (<25 ml/kg per hr), while “high cutoff techniques” refer to the use of super high-flux hemofilters with a nominal cutoff point of more than 60 kDa (11). The term, “high volume hemofiltration” (HVHF) was used to refer to techniques of hemofiltration at doses higher than 50 ml/kg per hr. HVHF using HCO filters are labeled as HCO/HVHF and classified under HCO hemofitration. The term “plasmafiltration” was used to refer to techniques involving the passing of blood through a large pore filter that resulted in filtration of plasma, where this filtered plasma was discarded and replaced by another source of colloid/plasma (plasmaexchange). The term “selective plasma filtration” was used to refer to techniques that involve initial plasmafiltration, followed by passage of the filtered plasma through a second selective plasma filter which then separates plasma components according to their size prior to elimination and subsequent replacement. “Adsorption techniques” included all techniques where either whole blood or plasma was exposed to a sorbent. “Modified ultrafiltration” refers to a technique of ultrafiltration initiated upon the termination of cardiopulmonary bypass.
We identified only 17 articles that fulfilled our selection criteria and proceeded to detailed analysis and extraction of data using four main ways of expressing cytokine removal: clearance (CL), sieving coefficient (SC), ultrafiltrate (UF) concentration, and percentage removal.
Most of the studies were conducted on animals in the weight range of 20 kg to 40 kg with only three studies conducted in small animals (rats) and one on a large animal species (ponies). The majority of studies (nine) involved between 15 to 36 animals; one study each involved a sizable number of 48 and 84 animals and only three studies involved 10 or fewer animals. The models studied were mostly endotoxic shock (10 experiments); others were pancreatitis (3), fulminant hepatic failure (2), and cardiopulmonary bypass (1). Ten studies looked at hemofiltration using standard hemofiltration filters and only two were conducted using high cutoff filters. Out of these studies on hemofiltration, four studied hemofiltration at doses defined as high volume hemofiltration (1 of the high cutoff group and 3 of the standard hemofiltration groups). Only two experiments studied plasmafiltration techniques and only one experiment each studied adsorption (direct hemoperfusion) and modified ultrafiltration in relation to cardiopulmonary bypass.
In summary, we identified four main techniques: 1. standard techniques; 2. high cutoff (HCO) techniques; 3. adsorption techniques; and 4. plasma filtration techniques. Standard techniques and HCO techniques include both hemofiltration at standard doses as well as hemofiltration at high volume doses according to current definitions.
The cytokines measured were interleukin-1b (IL-1b), interleukin-6 (IL-6), interleukin-8, interleukin-10 (IL-10), and tumor necrosis factor-alpha (TNF-alpha). None of the studies included data on albumin loss. These studies were done on a mixture of small to large animals and the experimental conditions under which studies were conducted were acute pancreatitis, acute fulminant hepatic failure, gram negative sepsis, and cardiopulmonary bypass with sternotomy, all of which are known precipitants of hypercytokinemia. Tables I and II show data on clearance (CL) and sieving coefficient (SC) extracted from the animal experiments. One paper on HCO hemofiltration (HCO/HF) was subsequently excluded from analysis because the results displayed appeared to be a case of duplication (12).
Clearance (CL) Data From Animal Studies
HCO = high cutoff; CVVH = continuous veno venous hemofiltration; Std = standard; CAVH = continuous arteriovenous hemofiltration; VHF = high volume hemofiltration; TSI = thromboxane synthase inhibitor; Qb = blood flow; Qf = ultrafiltrate flow; Qp = plasma flow in ml/min; Pre/post = refers to predilution or postdilution respectively; Blood pressure+ = blood pressure dependent; No details = details not given in the paper; Filter code = refer to Tab. IV;
= reported only as IL-1; n = number of animals in the subgroup subjected to each operating characteristics.
Sieving Coefficient (SC) Data From Animal Studies
HCO = high cutoff; CVVH = continuous veno venous hemofiltration; Std = standard; MUF = modified ultrafiltration; CPB = cardiopulmonary bypass; Qb = blood flow; Qf = ultrafiltrate flow; Pre/post = refers to predilution or postdilution respectively; Filter code = refer to Tab. IV; Late = commencement of hemofiltration after decline of total peripheral resistance of 30%; Early = prophylactic i.e. immediately after the induction of pancreatitis using sodium taurocholate; Change/no change = refers to whether filters are changed every 12 hr. Therapeutic = after mean arterial pressure decreased 20% below baseline; Prophylactic = CVVH simultaneous with induction of insult;
= reported only as IL-1;
= according to criteria set in reference 22 regarding ‘late’ vs. ‘early’ commencement of CVVH; some data are reported as both CL and SC; n = number of animals in the subgroup subjected to each operating characteristics.
Comparison of predicted clearance in an 80 kg human (Tab. III) showed that HF using standard filters at flows of 100 ml/kg per hr (labeled as Std/HVHF) achieved good clearance of IL-10 but no TNF-alpha clearance. Standard HF at standard flows (Std/HF) only achieved little CL of all studied cytokines; frequent filter changes or prophylactic versus late commencement of HF did not affect clearance. HCO techniques were able to achieve modest CL of IL-6 and IL-10 at moderately high flows of 45 ml/kg per hr. However, another paper on HCO filters reported low CL for IL-1b and zero clearance for IL-6 and TNF-alpha despite very high flows at 120 ml/kg per hr. Plasmafiltration (PF) techniques were the only techniques able to achieve reliable modest-to-good clearance of TNF-alpha. For both standard techniques and PF techniques, rate of clearance increased with increasing the filtration rate. Selective PF did not appear to confer additional improvement in cytokine CL when compared to standard PF.
Predicted Clearance (ML/MIN) in 80 Kg Human Using SC Data
Calculation was done in the following manner: from the data given in the papers, the UF rate in ml/kg per hr was calculated and then the equivalent UF rate for an 80 kg human was derived in ml/min. The product of the SC (from the animal studies) and UF rate in ml/min gave the value of CL in ml/min for an 80 kg human. SC is taken to be zero if CL is zero from CL data. For plasmafiltration/selective plasma filtration = SC derived from CL data as CL/Qf in ml/min. Product of SC and rate of Qf in ml/min in human equals CL.
HCO = high cutoff; CVVH = continuous veno venous hemofiltration; Std = standard; CAVH = continuous arteriovenous hemofiltration; HVHF = high volume hemofiltration; TSI = thromboxane synthase inhibitor; Qb = blood flow; Qf = ultrafiltrate flow; Qp = plasma flow in ml/min;
= derived value for 80 kg man; Pre/post = refers to predilution or postdilution respectively; Filter code = refer table 4; Late = commencement of hemofiltration after decline of total peripheral resistance of 30%; Early = prophylactic i.e., immediately after the induction of pancreatitis using sodium taurocholate; Change/no change = refers to whether filters are changed every 12 hr.
Filter/Device Code and Details
PMMA = polymethylmethacrylate; BMG = β2 microglobulin
There was only one animal study that assessed adsorption and measured the rate of cytokine removal. This study, which used a β2 microglobulin adsorption column (Lixelle) through direct hemoperfusion (DHP), displayed adsorption rate or percentage of cytokine removed over the course of 180 minutes. The values displayed were averaged from six readings and showed an approximate adsorption rate of 30% for IL-1b, IL-6 and TNF-alpha (data not shown).
Discussion
Key Findings
We performed a systematic analysis of animal experimentation involving different techniques of EBP to determine their efficacy in the removal of cytokines during animal experiments. There were limited data available, but a comparison of these techniques using data from animal studies suggested that appreciable clearances of IL-6 and IL-10 were achieved with very high volume hemofiltration using standard filters. High cutoff hemofiltration offered appreciable clearances of IL-6 and IL-10 at filtration rates of 45 ml/kg per hr. Std/HF at low filtration rates achieve negligible cytokine removal. None of these techniques, however, removed TNF-alpha. Plasmafiltration was the only technique to achieve clearance of TNF-alpha.
Relation to Previous Literature
There are no other reviews in the literature that have studied experimental animal cytokine clearance data for comparison. However, we have previously published a systematic review of cytokine removal by EBP techniques when assessed ex vivo (10). In this regard, our analysis of animal data reveals that the ex vivo findings may not apply to these techniques when used in experimental studies. In particular, while many studies of HCO/HF and some studies of Std/HF reported in vivo TNF-alpha removal with clearances up to a mean of 25 ml/min for HCO/HF, the animal studies could not confirm this efficacy. In fact, they showed that essentially only PF can be expected to reliably remove TNF-alpha. In contrast, the clearances achieved for IL-6 and IL-10 with HCO/HF were broadly similar in the ex vivo studies and in the animal studies. Additionally, while there were essentially no ex vivo data on IL-10 clearance with Std/HF, eight animal studies showed appreciable IL-10 clearance with both high-volume or low-volume Std/HF. In contrast, while IL-1 beta has been well studied ex vivo and removal has been shown to be substantial, no robust evidence for such removal exists in animal experiments.
Significance of Study Findings
Through this review, we now provide some basis for the design of randomized, controlled, clinical trials. In particular, by demonstrating that high cytokine removal is achieved by using higher UF rates, we show that one logical approach to the treatment of cytokinemia by EBP techniques should involve the use of higher flows.
Flows of 100 ml/kg per hr if applied to humans (8 l/hr in an 80 kg human), however, would require the use of large amounts of replacement fluids, involve, technical difficulties, expose patients to other risks (e.g., hypophosphatemia) and require catheters that can deliver very high (400 ml/min) blood flows. This will significantly increase cost, nursing care burden, and large fluid shifts, with potential for electrolyte imbalances, especially if this therapy is provided around the clock. A pulse therapy at 100 ml/kg per hr might be more feasible but would result in intermittent high removal of cytokines with periods of no cytokine removal.
Ultrafiltration flow rates of 45 ml/kg per hr using high cutoff filters appear more feasible for continuous application but may result in lesser degree of cytokine removal compared to continuous therapy at 100 ml/kg per hr of ultrafiltration. Therefore, a combination of middle-level (40-50 ml/kg per hr) flows with the use of a more porous filter may be able to achieve similar clearances at a lower cost with little added nursing care burden and may offer a safer approach. However, even such techniques cannot be expected to effectively remove TNF-alpha. The findings of this review also further support the observation that standard hemofiltration using standard flows cannot achieve meaningful cytokine removal. Another technique that achieves good cytokine clearance is plasmafiltration. It is also the only technique that can reliably remove TNF-alpha. However, plasmafiltration is suitable only for short duration therapy and intermittent application, which may be inadequate for hypercytokinemia, a condition that requires therapy around the clock. The requirement for plasma replacement using donor plasma or albumin may also prove costly and invites added risk.
Strengths and Limitations
Animal experimentation is an important step towards future developments and we have performed the first literature review on experimental animal data on the subject of extracorporeal cytokine removal. This review adds a clinical perspective to a previous assessment of cytokine removal ex vivo and demonstrates that the ex vivo data, although valuable, are not reliably predictive of performance in the experimental animal setting. CL, SC, and percentage removal as measures of cytokine removal are directly related to the efficacy of the device being studied, as opposed to measurements of plasma concentration which are affected by other factors such as endogenous production and decay. Despite the above strengths, this analysis has many limitations, including limited data for analysis and lack of homogeneity in the performance of the studies. This makes pooling the data questionable despite our attempts at achieving normalization. We have excluded studies in language other than English for practical reasons and thus might have excluded some publications. Data on adsorption, a technique of extracorporeal blood purification that is gaining much prominence, are under-represented as most studies involving adsorption do not report CL, while measurement of SC is irrelevant for this technique. Live animal studies operate under a more complex system compared to ex vivo experimentation, but major differences may exist in the biological system compared to humans, such as differences in volume of distribution between species, that may jeopardize the conclusion. For these reason, the results may not be reproducible in experiments involving human beings.
Future Studies
Our analysis is limited to experimental animal work. It seems desirable to explore the data supporting such work by application of the techniques in humans. Systematic assessment of the human evidence may be useful.
Conclusions
In conclusion, experimental animal data support the view that cytokine removal by HVHF offers a substantial rate of IL-6 and IL-10 removal by extracorporeal therapy. HCO hemofiltration at moderate doses may achieve comparable clearances. However, neither approach will effectively remove TNF-alpha, which can only be reliably removed by plasmafiltration. Further systematic assessment of the efficacy of these techniques in humans seems desirable.
Footnotes
Appendix
The following search strategies were displayed by Pubmed (under ‘Search details’) for the following terms: