Severe sepsis represents the leading cause of mortality and morbidity in critically ill patients worldwide. The sepsis syndrome is associated with an overwhelming, systemic overflow of proinflammatory and anti-inflammatory mediators, which leads to generalized endothelial damage, multiple organ failure and altered cellular immunological responsiveness. Although our understanding of the complex pathophysiological alterations that occur in severe sepsis and septic shock has increased greatly as a result of recent clinical and preclinical studies, mortality associated with the disorder remains unacceptably high, ranging from 30% to 50% 1234.

The cornerstone of therapy continues to be early recognition, prompt initiation of effective antibiotic therapy, source control, and goal-directed haemodynamic, ventilatory and metabolic support as necessary. To date, attempts to improve survival with innovative, predominantly anti-inflammatory therapeutic strategies have been disappointing, with the exception of physiological doses of corticosteroid replacement therapy 56 and activated protein C (drotrecogin alfa [activated]) 7 in selected patients.

'Renal dose' haemofiltration rate of 2000 ml/hour has successfully been used to treat acute renal failure for years 8. This dose suffices for renal replacement therapy and can remove inflammatory mediators; however, it does not alter plasma levels of these mediators, suggesting that its ability to clear inflammatory mediators is suboptimal 9. This was reflected in one study 10 by failure to demonstrate an improvement in organ dysfunction and survival. Hence, the indication for its use in septic patients was abandoned, beyond its function to provide renal support in the presence of renal dysfunction 11. However, the theory that underpins increasing plasma water exchange or higher dose haemofiltration seems reasonable.

Ronco and coworkers 12 demonstrated survival benefits by increasing the haemofiltration dose (35 ml/kg per hour) beyond the conventional renal dose (20 ml/kg per hour), but no further benefit was achieved, even at higher doses (45 ml/kg per hour), in the overall studied population. Nevertheless, there was an improvement in survival at the highest haemofiltration doses in that study for the subset of patients with sepsis. Additionally, benefits have been demonstrated in several animal models of sepsis. Improvements in cardiac function and haemodynamics were replicated in these animal studies using ultrafiltration (UF) rates up to 120 ml/kg per hour 13141516. Septic dose haemofiltration, or high-volume haemofiltration (HVHF), was thus conceived and applied in human sepsis. Findings of improvements in haemodynamics with decreased vasopressor requirements 171819 and trends toward improved survival 1920 are evidence that HVHF may be efficacious. Because HVHF technique requires high blood flows, tight UF control and large amounts of expensive sterile fluids, we proposed a new technique, namely 'pulse HVHF' (PHVHF) 2122. PHVHF is application of HVHF for short periods (up to 6–8 hours/day), providing intense plasma water exchange, followed by conventional continuous venovenous haemofiltration (CVVH).

We hypothesized that daily 'PHVHF' may have beneficial effects in severe sepsis by unselectively removing of proinflammatory and anti-inflammatory mediators, and hence improving patient outcomes. The present study evaluates the feasibility of PHVHF and the effect of this treatment on haemodynamics, oxygenation and 28-day all-cause mortality.

This is a prospective interventional study conducted in the intensive care unit (ICU) of St. Bortolo Hospital, Vicenza, Italy. Fifteen patients with severe sepsis receiving continuous renal replacement therapy (CRRT) were enrolled in the study. Patients were included in the study if they had severe sepsis or septic shock, as defined using the criteria reported by Bone and coworkers 23, and if they fulfilled one of the previously reported criteria for initiating renal replacement therapy in critically ill patients 24. Exclusion criteria were age less than 18 years, death imminent within 24 hours, and very high weight (>140 kg). All patients were treated using the same, recently developed management guideline for severe sepsis and septic shock 25. All except one patient were receiving mechanical ventilation because of respiratory failure. Broad spectrum antibiotics were given to all patients and were altered according to blood culture and sensitivity findings.

Eight out of 15 patients received activated protein C (drotrecogin alfa [activated]). The drug was not used in seven patients: one had underlying ruptured abdominal aortic aneurysm; the second was admitted because of multiple fractures and severe head trauma; the third had an Acute Physiology and Chronic Health Evaluation (APACHE) II score less than 25 at admission; and the remaining four had severe thrombocytopenia (<15,000/mm³) and/or impaired coagulation (international normalized ratio >3.0). The use of activated protein C (drotrecogin alfa [activated]) in approximately 50% of the patients included might therefore have contributed to any improved outcome identified. Clinical data are summarized in Table 1.

The APACHE II score, Simplified Acute Physiology Score (SAPS) II, and Sequential Organ Failure Assessment score were calculated from physiological measurements obtained during the first 24 hours of ICU admission. Expected mortality rates for APACHE II and SAPS II scores were computed using the logistic regression calculations suggested in the original reports 2627. The study protocol was approved by the hospital ethics committee.

The sepsis syndrome is associated with an overwhelming, systemic overflow of proinflammatory and anti-inflammatory mediators, which leads to generalized endothelial damage, multiple organ failure and altered cellular immunological responsiveness. The complex inflammatory network involved is redundant, synergistic and acts like a cascade. It includes mediators with autocrine and paracrine actions, as well as cellular and intracellular components. A large number of proinflammatory mediators, including tumour necrosis factor-α, interleukin-1, interleukin-6, platelet-activating factor and nitric oxide, play important roles in the cascade, but attempts to improve survival in human trials using innovative, predominantly anti-inflammatory therapeutic strategies have been disappointing 29. Almost paralleling the surge in proinflammatory mediators, there is a rise in anti-inflammatory substances, resulting in induction of a state of immunoparalysis or monocyte hyporesponsiveness 30. Both proinflammatory and anti-inflammatory factors become upregulated and interact with each other, leading to various rises in mediator levels that change over time. Neither therapies directed at single mediators nor single-dose interventions therefore seem appropriate, in part because of a discrepancy between the biological timing of the syndrome and the clinical timing of symptoms.

CRRT has made extracorporeal treatment possible in septic patients even when they are haemodynamically unstable; such treatment is given to balance hypercatabolism and fluid overload. In addition, 'high volume' and convective modalities have the advantage of removing higher molecular weight substances, which include many inflammatory mediators. Multiple animal studies 13141516 have shown a beneficial effect of HVHF on survival in endotoxaemic models. Recent studies in humans 171819 have demonstrated that HVHF improves haemodyamics, with decreased vasopressor requirements.

A daily PHVHF regimen was utilized as the intervention in the present study for the following reasons. First, the very high UF volume requires very close surveillance, which is difficult to maintain over 24 hours. Second, solute kinetics may render high volumes useless after a few hours because of saturation of membrane adsorption 1731. Third, standard CVVH (UF rate 35 ml/kg per hour) may help to maintain the effect of pulse therapy and prevent post-treatment rebound from sudden changes. Instead of using a fixed dose (i.e. UF rate 6 l/hour), we applied a dose of 85 ml/kg per hour during pulse treatment because body size is the main predictor of patient outcome 1218. Additionally, 'continuous' removal of soluble mediators may be the most logical and best approach to a complex and lengthy process such as sepsis; we therefore performed PHVHF on a daily basis and terminated treatment when haemodynamic variables improved. We hypothesized that beneficial haemodynamic effects would be achieved during PHVHF, and that these effects would be perpetuated after cessation of the pulse treatment by standard CVVH. We also hypothesized that they would be accompanied by improvement in oxygenation and reduction in mortality. The present pilot study in patients with severe sepsis/septic shock was conducted to test our hypotheses.

Overall, PHVHF was well tolerated by critically ill patients and appeared to offer many of the benefits conferred by continuous HVHF 19 while avoiding its drawbacks. Six to eight hours PHVHF during the daytime was widely accepted by the ICU nursing staff because it reduced the labour intensity of the protocol during the night shift. No treatment was prematurely discontinued because of extracorporeal circuit clotting or high pressure problems. It appears that PHVHF is a feasible modality and can safely be performed on a daily and prolonged basis. The greatest duration of treatment in any patient our study was 9 days. The most clinically relevant finding that emerged from our investigation is that adjuvant PHVHF in septic shock patients is associated with improvement in haemodynamic parameters (Fig. 1), permitting a significant reduction in vasopressor requirements as soon as halfway through and at the end of the PHVHF session, and this was maintained at 6 hours and 12 hours after treatment. The haemodynamic benefits of short-term HVHF regimens were recently demonstrated by Cole and coworkers 17 (UF rate 6 l/hour, duration 8 hours) and Honore and colleagues 18 (UF rate 8.75 l/hour, duration 4 hours). We proved that this beneficial hemodynamic effect can be maintained after HVHF by continuing with standard dose CVVH (UF rate 35 ml/kg per hour). Indeed, in practice our regimen could be adjusted on the basis of the individual patient's clinical response.

Several mechanisms are potentially responsible for the reduced need for pressor therapy with PHVHF. For mediator-independent factors, we were unable to demonstrate any differences in body temperature and arterial pH before and after PHVHF, including 12 hours after treatment. It is clear that cooling-induced vasoconstriction and correction of severe acidosis cannot account for this positive haemodynamic effect. The daily fluid balance on the day before initiation of PHVHF and that on the day of intervention were similar. Based on these findings, we argue that PHVHF permits continuous removal of soluble vasodilatory mediators or molecules identified in sepsis by either convection or adsorption, resulting in reduction in vasopressor requirements.

Unlike recent studies conducted by Honore and coworkers 18 and Joannes-Boyau and colleagues 19, we was unable to demonstrate any benefit of HVHF on cardiac index. The possible explanation for this is that we recruited patient at an earlier time point in septic shock (i.e. during hyperdynamic state); the mean cardiac index of our patients was 3.4 l/min per m², whereas those in the other two studies were less (2.0 l/min per m² 18 and 2.9 l/min per m² 19). The aim of haemodynamic support in our sepsis patients was to maintain condiac index at 2.5 l/min per m² or above because the studies that attempted to maintain a supraphysiologic cardiac index of above 4.0 to 4.5 l/min per m² have not shown consistent benefit 3233. Interestingly, this indicates that the improved haemodynamics and decreased vasopressor requirement conferred by daily PHVHF are clinically significant even during hyperdynamic septic shock.

Although it is beyond the scope of this report to provide a full comparison of mortality rates between standard sepsis treatment and such treatment combined with PHVHF, it appears that PHVHF may have beneficial immunomodulatory effects with prolonged daily use, especially with respect to patient outcome. The 28-day all-cause mortality was 47%, as compared with 72% as predicted by APACHE II and 68% as predicted by SAPS II severity scores. This is consistent with the findings of another study 19, in which 96 hours of continuous HVHF was given to patients with septic shock (46% observed and 70% predicted mortality rate). In fact, of the seven deaths at 28 days in our study, only four were attributable to refractory septic shock. How long would it take for a clinically relevant benefit to manifest? Tailoring our daily PHVHF regimen according to clinical response should permit sufficient duration of HVHF. In addition, because absolute or relative contraindications were met in seven patients, only the remaining eight patients in the present study received activated protein C (drotrecogin alfa [activated]) – a drug that has shown the benefit in terms of 28-day mortality in recent trials 7. However, we can state that PHVHF is feasible and, as a treatment for severe sepsis/septic shock, can affect physiological end-points. In terms of mortality, the only way to demonstrate the effect of PHVHF in this population is to conduct a prospective, randomized, controlled study on a larger scale. Nevertheless, we can hypothesize that the use of activated protein C (drotrecogin alfa [activated]) in 50% of the population might have contributed to the improved outcome. If so, then the combination of activated protein C (drotrecogin alfa [activated]) and PHVHF might be particularly useful.

The present study is limited by the fact that the population was highly heterogeneous, relatively small and reflective of patients seen in a single center. We did not measure mediator levels in plasma and in the UF over time, which might have helped to explain the mechanisms of mediator removal. However, the nonselective, simultaneous removal of different mediators demonstrated by a reduction of the circulating cytokines or an increase their levels in the UF may not necessarily implicate as the gold standard of blood purification for sepsis patients. A more effective strategy would be to attempt to influence the functional responses of cells that are implicated in the pathogenesis of sepsis. Such approaches are under evaluation, and findings reported in a preliminary paper 21 are encouraging. Also, we did not evaluate removal of sedative drugs with vasodilatory effect, such as midazolam and sufentanil. For ethical reasons, we could not conduct the trial in sepsis patients who did not have acute renal failure. In the context of acute renal failure in sepsis, it is clear that metabolic compounds partly accumulate as a consequence of the loss of renal function. Uraemic toxins rapidly accumulate in tissues and plasma, and they may be responsible for the immune dysregulation associated with sepsis.

In summary, PHVHF appears to be feasible and is a promising technique for the treatment of severe sepsis. We demonstrated a clinically and statistically significant beneficial effect of this therapy on vasopressor requirements during treatment and after therapy. It may be a beneficial adjuvant treatment for severe sepsis/septic shock in terms of patient survival, and it represents a compromise between CRRT and HVHF. Further confirmation is required in large, properly designed clinical trials to establish the benefit of PHVHF.

• PHVHF represents a feasible compromise between CRRT and HVHF, in which HVHF is applied for short periods of up to 6–8 hours/day and followed by standard dose CVVH.

• PHVHF, when applied in patients with septic shock/severe sepsis, can achieve beneficial effects on vasopressor requirements.

• PHVHF applied on the daily basis and tailored according to clinical response may represent a beneficial adjuvant treatment for severe sepsis/septic shock in terms of patient survival.

APACHE = Acute Physiology and Chronic Health Evaluation; CRRT = continuous renal replacement therapy; CVVH = continuous venovenous haemofiltration; HVHF = high-volume haemofiltration; ICU = intensive care unit; MAP = mean arterial pressure; PHVHF = pulse high-volume haemofiltration; SAPS = Simplified Acute Physiology Score; UF = ultrafiltration;

The author(s) declare that they have no competing interests.

RR conducted the study, collected data, performed statistic analysis and drafted the manuscript. AB and ZR conducted the intervention in the study. PP and MD carried out the haemodynamic measurements. GS helped to collect the data. CR conceived the study, participated in its design and helped to draft the manuscript. All authors read and approved the final manuscript.