Analysis of Endotoxin Adsorption in Two Swedish Patients with Septic Shock

The AN69 bulk based and amplified with a positively charged poly-imine ethylene layer capable of adsorbing negatively charged endotoxin molecules, is in use, for patients with critical septic shock and adjacent acute kidney injury requiring continuous renal replacement therapy at the general intensive care unit at Skåne University Hospital, Lund, Sweden. The oXiris filter is a full blown dialysis filter capable of providing a complete dialysis treatment in addition to its adsorbing ability [1]. Lipopolysaccharide (endotoxin) from the outer Gram-negative bacterial wall can induce a harmful immunologic response, involving hemodynamic deprivation in Gram-negative septic shock, and is one important motor driving the septic cascade. Endotoxin levels have been correlated to the severity of illness, maximum organ dysfunction, and mortality. Endotoxin can also occur in Gram-positive septic conditions, for example, by originating from an ischemic gut [2].

Endotoxin molecules have a size of about 10 kDa but can form aggregates up to 1,000 kDa, consisting of a lipid and polysaccharide covalently bonded. They attach to the toll-like receptor 4, especially in dendritic cells, monocytes, macrophages, and lymphocytes, which promotes the secretion of pro-inflammatory cytokines. Endotoxins are also strong pyrogens. They are to be considered as completely detrimental and induce an escalating shock state in humans, which manifests as vasodilation, capillary leakage, relative hypovolemia, and hypotension.

The endotoxin load in the blood stream can exist in form of a short initial pulse, which is fast cleared by the intact immune system, but it can also exist in a tight balance between the clearance capacity of an overwhelmed failing immune system and a constant feeding by the infectious bacterium colony. Some foci might be more encapsulated and the endotoxin might not reach the blood easy, while others are more profoundly perfused and thus more prone to seed endotoxin [3]. Normally the immune system neutralizes the endotoxin within minutes; mainly in the liver. If the system is overwhelmed a harmful fraction remains in blood, which is available for an adsorption process when blood is pumped through the oXiris filter. Theoretically, the blood flow, the concentration of endotoxin, and the total inner surface of the filter are the most important factors determining the endotoxin reduction rate. After a while, the membrane will be saturated and a filter change must take place [3, 4].

In a former simplistic in vitro model [5], 15.8 μg was injected into the oXiris filter and 6.9 μg (43.7%) was removed over 6 h. We tried to find out the corresponding endotoxin reduction in 2 patients with severe Gram-negative septic shock including verified endotoxemia. Both patients showed KDIGO class 3 acute kidney injury, and continuous renal replacement therapy was started using an oXiris filter. Arterial blood samples were drawn at 0, 1, 3, 8, 16, and 24 h for endotoxin measurement in our research lab. Our updated Limulus Amebocyte Lysate assay was used [6]. Patient data and treatment data are presented in Table 1, and endotoxin data including calculations are presented in Table 2. For patient A, an estimated amount of 70,000 pg was present initially, and 45,000 pg (64.3%) was removed by the eighth hour measurement (Table 2). Correspondingly for patient B, an estimated amount of 35,000 pg was present initially and 20,000 pg (57.1%) was removed by the eighth hour (Table 2). For the calculations of the absolute endotoxin amount from the measured endotoxin concentration, the assumption of a blood volume of 5 L was made. In the in vivo situation, many actors come into play; a variable seeding rate from the bacterium colony into the blood compartment, degradation by the overwhelmed immune system, and adsorption on the oXiris membrane. How these actors divide between themselves of the total net removal is impossible to calculate.

A theoretical formula describing the in vivo endotoxin balance could be:

endotoxin removal =

(endotoxin seeding from infectious focus)

– (endotoxin degradation by the immune system)

– (spontaneous endotoxin degradation)

– (adsorption on the oXiris membrane due to blood flow and due to modified Langmuir system)

Because the blood flow brings new endotoxin in contact with the membrane continuously and that the input endotoxin concentration will vary over time according to (1) changing seeding of endotoxin amount into the blood compartment and (2) a changing degradation by the immune system, it is easy to realize that no calculated prediction of how fast the membrane will be covered with endotoxin molecules can be formulated. Also, the endotoxin concentration in an in vivo system is low and can even approach the critical concentration where adsorbtion/disorption rates equal meaning that no netto adsorption longer takes place. Although the endotoxin concentration in blood is low a septic cascade that has started will continue, and the septic cascade is not directly proportional to the endotoxin concentration. The Langmuir equation is strictly valid on a static volume from which adsorption on a membrane takes place and not on a fluid flow. In addition, the Langmuir mathematics assumes that the endotoxin–poly-imine ethylene layer is monolayer, which need not be the case on the oXiris membrane [7].

All this sums up in a conclusion that an in vivo reality differs from a simplistic in vitro reality, but still the data from our 2 patients suggest that adsorption on the membrane is a strong actor in the endotoxin removal.

Ethical permission was granted for this research by The Regional Ethics Board of Southern Sweden.

M.E.B. and M.B. have got a research grant from Baxter for endotoxin research.