Early hemoperfusion with an immobilized polymyxin B fiber column eliminates humoral mediators and improves pulmonary oxygenation

This study was approved by the hospital ethics committee, and proper informed consent (oral or written) was obtained from each patient or their family. Patients with a clinical diagnosis of sepsis according to the criteria proposed by the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference were enrolled in the study [1]. Before the start of PMX treatment, global oxygen metabolism was measured. A thermodilution catheter (Edwards Life Sciences LLC, Irvine, CA, USA) was used to determine the oxygen delivery index and oxygen consumption index as parameters of oxygen metabolism. A thermodilution catheter was also used to monitor hemodynamics, and fluid balance was managed to maintain the central venous pressure in the range 7–10 mmHg. Criteria for inclusion in the study were the following findings within the previous 24 hours: signs of SIRS caused by infection (including fever or hypothermia [temperature > 38°C or < 36°C, respectively], tachycardia [> 90 beats/minute], tachypnea [> 20 breaths/minute], or an arterial carbon dioxide tension < 32 mmHg or mechanical ventilation, and a white blood cell count > 12.0 × 10⁴/l or < 4.0 × 10⁴/l or at least 10% immature neutrophils); mean arterial pressure > 60 mmHg irrespective of the use of catecholamines; stable global oxygen metabolism (oxygen delivery index > 500 ml/minute per m² and oxygen consumption index > 120 ml/minute per m²); and a diagnosis of ALI or ARDS according to the criteria established by the American-European Consensus Conference [13].

The diagnostic criteria for ALI and ARDS of the American-European Consensus Conference are as follows: acute onset of lung injury, diffuse bilateral infiltrates on chest X-ray film, a PaO₂/FiO₂ ratio < 200 mmHg for ARDS and < 300 mmHg for ALI, and a pulmonary artery occlusion pressure < 19 mmHg or no clinical evidence of congestive heart failure. Exclusion criteria were age < 18 years, mean arterial pressure ≤ 60 mmHg irrespective of the use of catecholamines, and impaired global oxygen metabolism.

The APACHE II score [19] was employed to assess of the severity of each patient's condition before DHP-PMX was performed. Survival of the patients was assessed at 1 month after treatment.

The three humoral mediators measured were IL-8 as a chemokine, PAI-1 as an index of vascular endothelial cell activation, and NE as an index of neutrophil activation. These mediators were measured before DHP-PMX treatment, and at 24, 48, and 72 hours after the start of therapy. Blood samples for the measurement of these mediators were collected from a peripheral vein. The neutrophil count was also determined. The PaO₂/FiO₂ ratio was used as an index of pulmonary oxygenation, and was measured before institution of DHP-PMX, and at 24, 48, 72, 92, and 120 hours after the start of treatment.

DHP-PMX was performed in 36 patients who had sepsis caused by Gram-positive or Gram-negative bacteria, and who fulfilled the inclusion criteria given above.

Because they had ALI or ARDS, all of the patients included in the study needed mechanical ventilation; the tidal volume and plateau pressure were set at 6 ml/kg [20] and ≤ 30 cmH₂O, respectively. The initial ventilation mode was volume-targeted assist control, which was changed to pressure-targeted assist control if the plateau pressure increased. The positive end-expiratory pressure (PEEP) was set at 5–15 cmH₂O and the ventilation rate at 10–35 breaths/minute.

The ventilator protocol at our hospital involved FiO₂ initially being set at 1.0 and PEEP at 10 cmH₂O. Then PEEP was increased in 2 cmH₂O increments up to 15 cmH₂O if PaO₂ remained below 55 mmHg on arterial blood gas analysis. When PaO₂ was 80 mmHg or greater, FiO₂ was decreased by 0.1. If PaO₂ was 100 mmHg or greater, even after decreasing FiO₂ to 0.6, then PEEP was decreased by 2 cmH₂O. PEEP was discontinued at a PaO₂/FiO₂ ratio of 300 or greater, and PEEP under 5 cmH₂O was not used.

Both ventilation and weaning from the ventilator were performed according to standardized protocols.

DHP-PMX was performed using a column of polystyrene fibers to which polymyxin B was covalently bound at a weight ratio of 0.5%. Binding of polymyxin B to the fibers was confirmed to be strong [7]. Blood access for DHP-PMX was obtained via a double-lumen catheter inserted into the femoral vein using Seldinger's method. Treatment was carried out for 3 hours at a flow rate of 80–100 ml/minute, and DHP-PMX was performed twice within 24 hours. Nafamostat mesilate (Torii Co., Ltd, Tokyo, Japan) was used as the anticoagulant. This is a serine protease inhibitor that exerts its anticoagulant effect primarily by inhibiting thrombin. The half-life of nafamostat mesilate is 8 minutes, and so its anticoagulant effect was confined to the extracorporeal circuit.

PAI-1 was measured in duplicate by an enzyme-linked immunosorbent assay (Fuji Revio Inc., Tokyo, Japan). PAI-1 is known to have various forms in blood, including an active form, an inactive form, a latent form, tissue plasminogen activator/PAI-1 complex, and vitronectin/PAI-1 complex. We measured total PAI-1 in the present study to determine the total amount of PAI-1 produced by vascular endothelial cells.

The blood level of NE was measured by using a commercial enzyme immunoassay kit (Sanwa Chemical Institute, Nagoya, Japan). Two forms of NE (free NE and NE/α₁-antitrypsin complex) can be detected using the kit. The normal NE level, determined in 139 healthy persons using this kit, was found to be 29 ± 2.7 μg/l (SRL Inc., Tokyo Japan, unpublished data).

Four blood samples were collected from a peripheral vein. After centrifugation, serum was stored for a maximum of one month at -70°C. Then the serum level of IL-8 was determined using a commercially available specific enzyme-linked immunosorbent assay kit for IL-8 (Ohtsuka Pharma Co., Tokyo, Japan), in accordance with the manufacturer's instructions.

No standard criteria for the use of DHP-PMX have been established. It has been variously reported that an IL-6 level of 1,000 pg/ml or greater [21], an APACHE II score of 25 or greater [22], an APACHE II score of 30 or greater [9], and a cardiac index of 6 l/minute per m² or greater [23] are indicators of poor prognosis at the start of DHP-PMX treatment. The reported efficacy of DHP-PMX also varies among medical institutions, depending on the time of starting treatment. In the present study, DHP-PMX was performed before the deterioration of organ perfusion and global oxygen metabolism. Our use of criteria targeting global oxygen metabolism and the mean blood pressure was based on the concept that DHP-PMX should be started before the tissues and cells have been significantly affected, even if some organ disfunction is present. Once shock occurs and organ ischemia progresses, cellular disfunction at the molecular level becomes severe, and this makes it difficult to assess the effect of DHP-PMX. Therefore, we selected patients for this therapy according to criteria that excluded those already in shock. It is perhaps a result of the early introduction of DHP-PMX before oxygen metabolism had deteriorated that all of the patients remained alive after 1 month.

PAI-1 is a protein with a molecular weight of 50 kDa that is produced by vascular endothelial cells and plays a central role in the activation of fibrinolysis [24]. It has recently attracted attention as a marker of vascular endothelial cell activation [18]. We measured total PAI-1 in the present study to determine the total production of this mediator by vascular endothelial cells.

Proinflammatory cytokines such as TNF-α and IL-1β stimulate vascular endothelial cells, induce the expression of adhesion molecules, and cause activated neutrophils to stimulate PAI-1 production by vascular endothelial cells, thus inhibiting fibrinolysis [18]. In the present study, PAI-1 was significantly lower at 48 and 72 hours after the start of DHP-PMX then the baseline value. Rather than DHP-PMX directly inhibitiing PAI-1 production, it seems more likely that adsorption of pathogenic bacteria prevented the release of inflammatory cytokines and reduced the stimulation of vascular endothelial cells to lower the PAI-1 level indirectly.

NE is a glycoprotein with a molecular weight of approximately 30 kDa and it has three isozymes. Its physiologic actions are proteolysis of bacteria and foreign materials. Because of low substrate specificity, however, NE can also attack various host proteins such as plasma proteins, coagulation factors, complement, elastin, and collagen [25]. Intact cells can also be damaged by this protease [26]. Neutrophils are believed to be an important precipitating factor in the pathogenesis of ARDS [16, 17]. Indeed, neutrophils may contribute to the development of lung disfunction by releasing oxygen radicals and proteolytic enzymes that induce the parenchymal cell damage and connective tissue destruction characteristic of ARDS [16, 17].

We investigated changes in circulating NE following DHP-PMX and found that blood levels after 48 and 72 hours were significantly lower than before treatment. Adsorption of pathogenic toxins may have prevented the release of inflammatory cytokines [22], resulting in less stimulation of neutrophils, rather than DHP-PMX directly preventing cytokine production or neutrophils being cleared from the blood by the column, because the neutrophil count was not altered by this treatment. Thus, DHP-PMX appears to prevent the activation of neutrophils, albeit indirectly. The ability of DHP-PMX to prevent activation of vascular endothelial cells [27] and reduce the blood level of NE might help to decrease organ disfunction, and the same mechanisms are considered to be associated with improvement in PaO₂/FiO₂ ratio in the patients with ALI and ARDS studied here.

IL-8 is a 6.5 kDa cytokine [28] that belongs to a supergene family of host defense molecules characterized by potent neutrophil activation in vitro [29]. It can be produced by several types of cells in vitro, including macrophages, lymphocytes, neutrophils, fibroblasts, and endothelial cells, in response to a variety of stimuli such as endotoxin, viruses, and other cytokines such as IL-1 or tumor necrosis factor [29]. Similar to PAI-1 and NE, the mean blood level of IL-8 was significantly decreased from 48 hours after DHP-PMX compared with the baseline value. This observation suggests that removal of pathogenic toxins from the blood indirectly inhibited the release of IL-8 from IL-8 producing cells, as was observed for PAI-1 and NE. In fact, an in vitro study [30] has demonstrated that absorption of inflammatory cytokines by PMX is quite limited. However, there has been no previous report on the effect of this treatment on IL-8. The results of the present study suggest that the inhibitory effect of DHP-PMX on IL-8, which is a key mediator of inflammatory reactions [31], may contribute to improvement in PaO₂/FiO₂ ratio.

Ventilation parameters may be another factor that influences circulating IL-8 concentration. Parsons and coworkers [32] reported that mechanical ventilation with a lower tidal volume of 6 ml/kg reduced the mean blood IL-8 level by approximately 12% in patients with ALI on the third day after the onset, suggesting that the decrease of blood IL-8 levels observed in our study was possibly associated with mechanical ventilation at low tidal volumes. However, the mean blood IL-8 level was decreased by approximately 72% compared with baseline on day 3 after the onset of treatment, which may represent a synergistic effect between mechanical ventilation at low tidal volumes and DHP-PMX.

During ventilation, PEEP and fluid balance are factors that are likely to influence pulmonary oxygenation, other than humoral mediators. The mean PaO₂/FiO₂ ratio was 244 ± 26.3 and the mean level of PEEP was 8 ± 1.3 cmH₂O at the start of DHP-PMX. Because PEEP was discontinued at a PaO₂/FiO₂ ratio of 300 or greater, its mean duration was approximately 81 ± 13.5 hours. The level of PEEP was gradually reduced during this period, and PEEP of 5 cmH₂O or less was not employed. Given that the PaO₂/FiO₂ ratio improved while PEEP was gradually reduced, it is not likely that PEEP directly contributed to the improvement in pulmonary oxygenation. A thermodilution catheter was inserted before the start of treatment to monitor oxygen metabolism and hemodynamics for 7 days. Because the central venous pressure was maintained between 7 and 10 mmHg in all patients, it is not likely that the fluid balance had a direct influence on differences in pulmonary oxygenation.

Tsushima and coworkers [12] reported that the PaO₂/FiO₂ ratio was significantly improved at 2 hours after DHP-PMX in all 20 patients with ALI or ARDS that they treated. They suggested that the observed improvement in PaO₂/FiO₂ ratio was related to inhibition of the production of inflammatory mediators and consequent suppression of systemic inflammation, both of which are caused by modulation of activated mononuclear cells and neutrophils by PMX, although the detailed mechanism of action remains unknown.

In the present study, although the PaO₂/FiO₂ ratio was significantly improved from 96 hours after DHP-PMX, no improvement was observed soon after treatment (24 hours). The patients included had definite ALI and/or ARDS and did not have cardiogenic pulmonary edema, according to assessment with a thermodilution catheter before DHP-PMX. The following mechanisms underlying the development of ALI and ARDS in patients with sepsis have been suggested. Bacterial substances such as endotoxin induce the production of IL-8 by stimulating pulmonary vascular endothelial cells and thus activate neutrophils. Production of adhesion molecules is then accelerated, causing neutrophils to adhere to pulmonary vascular endothelial cells and migrate into the extravascular space. Neutrophils accumulate locally in the lungs along the concentration gradients of chemotactic factors such as IL-8. Neutrophils then cause damage to pulmonary vascular endothelial cells and impair the barrier function of pulmonary capillaries, leading to the development of ALI or ARDS [16, 17]. Considering the above mechanisms, it is unlikely that PaO₂/FiO₂ ratio would improve immediately after DHP-PMX.

In the present study we examined correlations between the PaO₂/FiO₂ ratio and circulating levels of mediators that are considered to be associated with the development of ARDS (i.e. PAI-1, NE, and IL-8). Our findings indicate that there are significant negative correlations between the PaO₂/FiO₂ ratio and IL-8 and NE levels. Jorens and coworkers [33] reported a negative correlation between the PaO₂/FiO₂ ratio and the IL-8 level in bronchoalveolar lavage fluid as well as between the PaO₂/FiO₂ ratio and infiltration of neutrophils in patients with ARDS. Although the correlations demonstrated in the present study did not indicate such strong relationships, perhaps because we measured peripheral blood levels, the results of our study are consistent with the findings of Jorens and coworkers.

We used PAI-1 as an index of vascular endothelial cell activation, but no significant relationship with the PaO₂/FiO₂ ratio was identified, probably resulting from testing of peripheral blood samples. If we had measured the PAI-1 level in bronchoalveolar lavage fluid, then a significant correlation with the PaO₂/FiO₂ ratio might have been demonstrated. The reasons why a period of 96 hours is required to detect improvement in the PaO₂/FiO₂ ratio is that the patients had definite ARDS or ALI, and that 48 hours after DHP-PMX were required for decreases in PAI-1, NE, and IL-8 concentrations to occur.

Because this work is an uncontrolled observational study, observed changes can not be concluded to result from extracorporeal treatment. It will be necessary to conduct controlled studies in this area in the future.