BACKGROUND The association between mortality and hypoalbuminemia has been observed in several diseases. Nonetheless, the efficacy of albumin on survival in critically ill patients is controversial. Several meta-analyses have reported either negative, neutral, or beneficial effects of albumin administration. To clarify this controversy, a large multicenter prospective study has been performed, comparing the effects of 4% albumin vs. saline for volume replacement in critically ill patients. Although no difference in the overall mortality has been observed, a predefined subgroup analysis has shown a trend of longer survival in septic patients treated with albumin. As fluid replacement has been shown to be critical in sepsis, and based on both its primary (oncotic) and secondary properties (anti-inflammatory), it is conceivable that the use of albumin for volume replacement and for treating hypoalbuminemia may have a beneficial effects on survival of septic patients.

OBJECTIVES Primary objective: to verify whether volume replacement with albumin (treated group) and its maintenance within plasmatic physiologic range (equal or above 30 g/l) improves survival of patients with severe sepsis of septic shock, as compared to crystalloids (control group). Secondary objectives: to verify the differences in organ dysfunctions, hospital and intensive care unit (ICU) length of stay between the treated and control group.

METHODS About 1350 patients with severe sepsis or septic shock will be randomized to receive either albumin or crystalloids as fluid therapy. Volume replacement will be performed for both groups according to the early-goal directed therapy. Treated group will receive 60 gr albumin infusion after randomization, and 40-60 gr albumin daily infusion to maintain serum album level equal or above 30 g/l. Control group will receive crystalloids for the entire study; albumin administration will be allowed only when daily serum albumin level will be lower than 15 g/l. Patients will be treated until the 28th day after randomization or until ICU discharge, whichever comes first.

EXPECTED RESULTS Primary outcomes: absolute risk reduction of overall mortality of 7.5% at 28th day, with a further control at 90th day, following randomization. Secondary outcomes: reduction of number and severity of organ dysfunctions (as assessed by the Sequential Organ Failure Assessment score), reduction of ICU and hospital length of stay.

BACKGROUND AND RATIONALE The association between mortality and hypoalbuminemia has been documented in several diseases, including liver cirrhosis, nephrosic syndrome, and others. Being responsible for about 80% of the oncotic power of plasma protein in humans, albumin is a key factor in regulating the fluid exchange between interstitial and intravascular spaces, thereby extensively affecting hemodynamics. Besides its oncotic properties, albumin presents other characteristics potentially important in critically ill patients: 1 - a binding capacity for several physiological molecules and drugs, 2 - a scavenging action on oxygen free radicals, 3 - a modulating activity on nitric oxide (NO) metabolism, and 4 - a buffer power for the acid-base equilibrium.

Regardless its theoretical usefulness, the efficacy of albumin administration on survival in critically ill patients is controversial. The critical review published by the Cochrane Albumin Reviewers in 1998, including 30 clinical studies, has shown an increased mortality of critically ill patients treated with albumin. A further meta-analysis performed in 2001, including 55 clinical trials, has concluded that albumin administration is safe although with no effects on survival. In a more recent meta-analysis, including 90 cohort studies, and, separately, 9 prospective controlled trials on correcting albumin level, hypoalbuminemia has been shown to be a negative prognostic factor in terms of mortality, morbidity and length of stay, with a dose-dependent effect. In particular, each 10-g/l decline in plasmatic albumin level has been observed to significantly increase the odds of mortality by 137%. Moreover, the analysis performed within 9 controlled trials has suggested that complication rates may be reduced when plasmatic albumin concentration is maintained above 30 g/l. To clarify these controversial findings, a multicenter double-blinded randomized prospective study has been performed in about 7000 critically ill patients, comparing the effects of 4% albumin vs. saline infusion for volume replacement. Although no difference in the overall mortality has been observed, a predefined subgroup analysis has shown a trend of improvement in survival in patients treated with albumin affected by severe sepsis (p=0.09). In contrast, a trend towards a worse outcome in patients treated with albumin has been observed in the subgroup of patients with trauma (p=0.06), likely due to the higher number of patients with trauma associated to brain injury who did not survive in the albumin-treated group. These findings, therefore, highlight the importance of characterizing different types of critically ill patients who may benefit or not from albumin administration.

More recently, a further prospective, randomized controlled study has shown, in a series of 100 patients admitted to an intensive care unit (ICU), a significant reduction of organ dysfunction, as detected by the Sequential Organ Failure Assessment (SOFA) score, after correction of hypoalbuminemia with albumin administration.

In parallel to the investigations on the role of albumin in critically ill patients, an increasing interest has been focused on the timing and the hemodynamic target of volume replacement in septic patients. In particular, a recent randomized controlled study performed in patients with sepsis has shown a better survival in the group of patients in which fluid replacement has been obtained as early as possible according to predefined hemodynamic targets ("early goal-directed therapy"), as compared to the control group. Focusing our attention on septic patients, we can, therefore, conclude that: a - the early-goal directed volume replacement improves survival, and, b - the "type" of volume replacement may have a further effect on survival. In fact, volume replacement with the use of saline requires a greater amount of fluid and may lead to metabolic acidosis. On the other hand, volume replacement with the use of hydroxyethyl starch is potentially harmful. Volume replacement with the use of albumin does not involve any known risk and may also be beneficial in septic patients.

The current study aims to verify whether volume replacement with the use of albumin and its maintenance within plasmatic physiologic ranges may have beneficial effects in terms of mortality, morbidity and length of stay in patients with severe sepsis or septic shock, as compared to a standard volume replacement with the use of crystalloids. For this purpose, and to overcome the possible biases explained above, the study design will include two different and important aspects: 1 - for both arms of the study population, i.e., patients treated with albumin or with crystalloids, volume replacement will be performed according to the "early-goal directed therapy"; 2 - during volume replacement, and for the following days of treatment until the 28th day of admission in ICU (or until the day of ICU discharge, whichever comes first), serum albumin level will be monitored and kept equal or above a level of 30 g/l only in the albumin treated group. Such a study design will have several advantages. In particular, the introduction of the "early-goal directed therapy" both in patients treated with albumin or crystalloids, with the use of pre-defined hemodynamic targets, will standardize and optimize volume replacement for all the septic patients according to the standard care at the moment suggested worldwide. Moreover, it will allow us to specifically observe the direct effects of albumin administration per se and the maintenance of its serum level within normal range. In fact, besides its oncotic properties, it is conceivable that the physiological characteristics of albumin potentially important in septic patients (such as NO modulation, free oxygen radical scavenging, and acid-base homeostasis) may have a possible benefit on survival, especially after the early resuscitation phase.

Under this perspective, in the attempt to elucidate the possible mechanisms by which the use of albumin for volume replacement may be beneficial in patients with sepsis, we will plan to create a centralized blood bank including blood samples sequentially withdrawn from patients included in the study population to evaluate the effects of albumin on biological markers. Albumin, in fact, may have important effects on reducing the overall systemic inflammation ongoing in septic patients, thereby explaining a possible direct benefit of its administration. For this purpose, we plan to involve a sub-group of 50 participating ICUs enrolling about 700 patients, in which three serum samples will be withdrawn (at day 1, day 2 and day 7 after the enrollment) and stored in the bank.

PRIMARY OBJECTIVE To verify the hypothesis that volume replacement with albumin and its maintenance within predefined plasmatic physiologic range (equal to or greater than 30 g/l) improves survival of patients with severe sepsis or septic shock, as compared to a volume replacement with the use of crystalloids. Survival will be measured until the 28th and 90th day after enrollment.

SECONDARY OBJECTIVES

To verify the hypothesis that volume replacement with albumin and its maintenance within predefined plasmatic physiologic range (equal to or greater than 30 g/l) reduces:

1. The numbers and the severity of organ dysfunction, as detected by the SOFA score
2. The hospital length of stay
3. ICU length of stay

SUB-STUDY To understand the possible mechanisms by which albumin administration and the maintenance of its serum level equal or above 30 g/l might be potentially beneficial, as compared to crystalloids, in patients with severe sepsis of septic shock, with a particular regard to the inflammatory process characterizing this category of patients. For this purpose, by creating a blood bank of sequential plasmatic samples of patients included in the study, we aim to detect any possible difference between the plasmatic levels of biologic markers of systemic inflammation.

STUDY POPULATION Every patient admitted to ICU throughout the study period will be screened for eligibility. International standard criteria will be applied to identify patients with severe sepsis or septic shock. According to the "early-goal directed therapy", after the diagnosis, patients will receive arterial and central venous catheterization, if not yet performed. In addition to volume replacement and albumin administration scheduled for the study, patients will be clinically treated according to the criteria of the treating physician. We will strongly encourage the participating Units to follow the guidelines of treating sepsis published by the Surviving Sepsis Campaign. In addition to study fluids, physicians will be allowed to administer blood products, and either enteral or parenteral nutrition, when needed. Screening for patient eligibility will begin either at ICU entry or at the emergency room, wherever the diagnosis of severe sepsis or septic shock will be made. After assessing for eligibility, each Unit will notify the Coordinating Center to be assigned to either the treatment or the control arm. The entire study will then be conducted in ICU until the 28th day of admission or until ICU discharge, whichever comes first.

Inclusion criteria:

Patients with severe sepsis or septic shock, if each one of the following criteria is satisfied:

1. Proved or suspected infection in at least one site:

   1. lung
   2. abdomen
   3. genito-urinary tract
   4. other (blood, skin and soft tissue, central nervous system, bones and joints, cardiac system, catheter-related infection, other)

2. Two or more of the following:

   1. a core temperature ≥ 38° C o ≤ 36° C;
   2. a heart rate ≥ 90 beats/min;
   3. a respiratory rate ≥ 20 breaths/min or PaCO2 ≤ 32 mmHg or use of mechanical ventilation for an acute process;
   4. a white blood cell count ≥ 12000/ml or ≤ 4000/ml or immature neutrophils > 10%.

3. Presence of at least a severe organ dysfunction, as measured by the modified Sequential Organ Failure Assessment (SOFA) score:

   1. respiratory score > 1;
   2. hematologic score > 1;
   3. hepatic score > 1;
   4. cardiovascular score equal to 1, 3 or 4;
   5. renal score > 1.

Exclusion criteria:

1. Age below 18 years
2. Terminal state
3. Known adverse reaction to albumin administration
4. Severe sepsis or septic shock in patients after proved or suspected head injury, clinically active
5. Congestive heart failure (NYHA score III and IV)
6. Pathological conditions in which albumin administration is clinically indicated (hepatic cirrhosis with ascites, intestinal malabsorption syndrome, nephritic syndrome, burns)
7. More than 24 hours since inclusion criteria were met
8. Religious objection to the administration of human blood products
9. Inclusion in other experimental study

INTERVENTION/EXPOSURE After its approval by the Ethical Committee, the study will be activated as follows: 1- screening for eligibility, 2- informed consent, 3- inclusion in the study via centralized randomization.

Volume replacement will be performed in both the treated and the control group according to the "early-goal directed therapy". In the treated group, after randomization and simultaneously to volume replacement, 300 ml of 20% of albumin solution (total amount of 60 gr) will be infused over a period of 3 hours. From day 2 to day 28 (or until ICU discharge, whichever comes first), fluid will be administered as follows:

1. treated group: albumin will be infused on a daily basis, aimed to maintain its serum concentration equal or above 30 g/l (8). In particular, after the daily determination of its serum level:

   1. if lower than 25 g/l, 300 ml of 20% of albumin solution (total amount of 60 gr) will be infused;
   2. if equal or higher than 25 g/l and below 30 g/l, 200 ml of 20% of albumin solution (total amount of 40 gr) will be infused;
   3. if higher than or equal to 30 g/l, no albumin will be infused. Albumin solutions will be infused over a period of 3 hours. Further infusion of crystalloids will be allowed, when necessary, according to the clinical judgment. No infusion of colloids, other than albumin, will be admitted.

2. control group: crystalloids infusion will be allowed whenever necessary on a clinical basis. Albumin administration will be restricted to emergency use, as clinically judged and documented according to the standard criteria of each participating unit. No other colloids will be allowed.

OUTCOMES

Primary outcomes:

1. Absolute risk reduction of overall mortality of 7.5% at the 28th, with a further control at 90th day, after randomization

Secondary outcomes:

1. Reduction of the number and the severity of organ dysfunction (as recorded by the SOFA score)
2. Reduction of the length of stay in ICU
3. Reduction of the hospital length of stay

INFORMATION RETRIEVAL Blood samples for the biohumoral sub-study will be obtained on day 1, 2 and 7 following randomization, to be stored in the central blood bank for subsequent analysis.

Baseline data foreseen for the inclusion procedure, including source of ICU admission, by definition will be collected and validated centrally in coincidence with the randomization. As soon as each case is concluded, the original copy of the self-copying Clinical Research Form (CRF) is sent to the Coordinating Center, for quality control of clinical data, which is the prerequisite for the formal checks and input into the study database. A copy of the original paper CRF will remain at the Clinical Center, to comply with all legal and regulatory requirements foreseen in the International Conference on Harmonization-Good Clinical Practice (ICH-GCP) rules.

The following clinical variables, usually included in the standard care of septic patients, will be collected: hemodynamics (HR, MAP, CVP, urinary output; type and dosage of vasoactive agents), blood gas analysis values, body temperature, laboratory tests (serum level of urea, creatinine, potassium, sodium, bilirubin, albumin; hemoglobin level, hematocrit, WBC count, glycemia, platelet count, prothrombin and activated partial thromboplastin time), Glasgow Coma Scale, SOFA score, infection and antibiotic treatment, amount and type of fluids administered within the previous 24 hours (or prior randomization), net positive fluid balance, ventilatory setting, adverse or specific clinical event. Variables will be retrieved:

1. At baseline (within 6 hours from randomization)
2. On a daily basis, from day 1 to day 28 after randomization or until ICU discharge (whichever comes first)

We will apply the following indicators:

1. Simplified Acute Physiologic Score (SAPS II), to assess the severity of systemic disease at study entry.
2. Sequential Organ Failure Assessment (SOFA) score, to monitor organ failure at study entry and during its course.

The study will not be conducted in a blinded fashion, as in that case it would need a specifically designed setting for fluid administration, excessively expensive, and, de facto, unpractical (in terms of workload and organization). The use of the "early-goal directed therapy" for hemodynamic treatment, with the use of predefined hemodynamic targets, should assure the approach to hemodynamic optimization independently of the type of fluid employed.

Data will be collected at 24 hour interval (once daily), usually during the "morning" shift of each day of the study (between 7:00 AM and 11:00 AM, using the values closest to 9:00 AM). Data will be collected from the 1st day until the 28th day of ICU admission, or until ICU discharge, whichever comes first. Survival data will be collected until the 90th day after randomization.

Due to the characteristics of the study population, "lost to follow-up" patients are expected to be an exceptionally rare event, not influencing the main analyses on survival, based on the standard "intention to treat" criteria.

MONITORING OF THE STUDY

The strategy adopted to assure a monitoring as closely as possibly compliant to the GCP requirements responds to two characteristics of the study:

1. The protocol coincides with what normally required for the care of septic patients;
2. It is assumed that the criteria for the management of critically ill patients are substantially respected across the large spectrum of the participating units.

The standard GCP requirements related to point a, will be assured through the monitoring role of the clinical pharmacists of each center, coordinated by the Study Center of the Italian Society of Hospital Pharmacists (Centro Studi SIFO), which has already a recognized tradition in this area (www.sifoweb.it). The type of treatment which is the object of the study is particularly suited for this choice, as by definition albumin distribution must be already monitored on a per patient basis, and all the requirements related to the control of compliance with the study procedures is easily integrated with the clinical monitoring described above.

1. To assure and monitor the compliance with point b, the following procedures are foreseen: the large network of participating centers will be divided into "subgroups", each including 15-20 clinical units, under the responsibility of a monitor selected by the study organization with formal competences in the management of the clinical conditions under study.
2. The standard GCP "initiation visits" will be transformed into training seminars for each group, with the participation of at least 2 persons per center, including the Principal Investigator (PI) of the unit (and/or his/her delegate) and a nurse in charge of the operational procedures.
3. The clinical monitor will assure, specifically over the first year, at least 1 visit for the centers of his/her subgroup, to review with the clinical staff how inclusion criteria have been adhered to, and the targets of the background treatment strategies could be respected. The clinical monitor will also assure a permanent "on call" availability, to guarantee a correct and more efficient application of study criteria.

The coordination and management of the specific strategy illustrated above is under the joint responsibility of Consorzio Mario Negri Sud and of the Centro Studi SIFO, which are not involved in any of the activities related to patient selection and treatment, and have a long tradition in the area of organizing and monitoring large-scale clinical trials, conducted by and with not-for-profit collaborative groups.

SAMPLE SIZE ESTIMATES In Italy, ICU mortality for severe sepsis can be assumed to be about 45%. The objective of a 15% relative reduction of mortality, corresponding to an absolute mortality reduction of 7.5%, has been chosen as a clinically relevant objective, based on what observed at day 28 in the subgroup analysis of the SAFE study. The randomization of ~1350 patients allows reaching this objective, with a power of 0.80, and a two-sided level of significance of 0.05. As the number of participating centers appears already compatible with the possibility of including a larger cohort, it is anticipated that at the occasion of the second interim analysis, a specific request could be made to the Data and Safety Monitoring Board (DSMB) to advise on the opportunity of targeting a sample size of ~1800 patients, which is suitable to reliably assess also a lower (but still clinically relevant) absolute reduction of 6.5%.

ORGANIZATIONAL CHARACTERISTICS Up to 150 ICUs (which have already a tradition of cooperative studies) are expected to be able to randomize patients into the trial. A further expansion of the network is foreseen, as the Italian Society of Anesthesia and Intensive Care (S.I.A.A.R.T.I.), of whom the study PI is the President for the study period (2007-2010) has explicitly endorsed the protocol. All ICUs of the Society could adhere to the study, provided they are ready to document their logistic capability, and to be assessed and trained for the specific requirements of the study.

The study is coordinated by a Steering Committee which includes the scientific representatives of the Institutions assuming all the competences and organizational know-how required for the trial:

1. Istituto di Anestesia e Rianimazione, Fondazione IRCCS - "Ospedale Maggiore Policlinico, Mangiagalli, Regina Elena", Università degli Studi di Milano [Coordinating Center]: L. Gattinoni, P. Caironi
2. Dipartimento di Medicina Perioperatoria e Terapia Intensiva, Azienda Ospedaliera San Gerardo di Monza, Università degli Studi Milano-Bicocca: A. Pesenti, R. Fumagalli
3. Consorzio Mario Negri Sud, S. Maria Imbaro, Chieti: G. Tognoni, M. Romero
4. Istituto di Ricerche Farmacologiche Mario Negri, Milano: R. Latini, S. Masson

The DSMB includes two internationally well-known ICU clinicians (P.M. Suter, J.L.Vincent), an independent statistician, specifically competent on survival analyses of complex patients-populations (M.G. Valsecchi), one of the leading Italian figures in the area of bioethical and legal requirements in medical research (A. Santosuosso).

TIMING

The calendar below provides a reasonably realistic timetable for the activities (Time 0 is assumed to be the notification of the study approval):

1. 0-6 months: protocol presentation and approval by the Ethical Committees; final centers' selection and investigators' training
2. 7-28 months: patient randomization
3. First interim analysis (after the first ~600 patients)
4. Second interim analysis (after 1000 patients)
5. Closure of the study (after 28 months)
6. Presentation of the final report (4 months after the database locking)

STATISTICAL ANALYSIS All the efficacy analyses will be based on the intent to treat population (ITT), consisting of all randomized patients. Background and relevant baseline information will be summarized for the ITT population by treatment and presented using descriptive statistics. Incidence event rate will be estimated using Kaplan-Meier survival curves that will be compared using logrank analysis. Additionally, treatment efficacy will be assessed by multivariate analyses using Cox's regression model. Other secondary analyses will include the evaluation of efficacy on all secondary end-points. The Mantel-Haenszel procedure will be applied to test for the linearity of effects among subgroups; the chi-square test will be applied to test for heterogeneity of effects among the subgroups. Finally, recursive partitioning techniques will be applied to identify homogeneous subgroups of patients showing a higher likelihood to benefit from the study intervention. Two interim analyses are planned for the double scope of monitoring safety and of verifying the correctness of the assumptions made for sample size estimation regarding the primary end-point event rate in relation to the anticipated survival benefit. No stopping rules are foreseen, for utility or for efficacy.

Inclusion Criteria:

• Patients with severe sepsis or septic shock, if each one of the following criteria is satisfied:

  • Proved or suspected infection in at least one site:

    • lung
    • abdomen
    • genito-urinary tract
    • other (blood, skin and soft tissue, central nervous system, bones and joints, cardiac system, catheter-related infection, other)

  • Two or more of the following:

    • a core temperature ≥ 38° C o ≤ 36° C
    • a heart rate ≥ 90 beats/min
    • a respiratory rate ≥ 20 breaths/min or PaCO2 ≤ 32 mmHg or use of mechanical ventilation for an acute process
    • a white blood cell count ≥ 12000/ml or ≤ 4000/ml or immature neutrophils > 10%

  • Presence of at least a severe organ dysfunction, as measured by the modified Sequential Organ Failure Assessment (SOFA) score:

    • respiratory score > 1
    • hematologic score > 1
    • hepatic score > 1
    • cardiovascular score equal to 1, 3 or 4
    • renal score > 1

Exclusion Criteria:

• Age below 18 years
• Terminal state
• Known adverse reaction to albumin administration
• Severe sepsis or septic shock in patients after proved or suspected head injury, clinically active
• Congestive heart failure (NYHA score III and IV)
• Pathological conditions in which albumin administration is clinically indicated (hepatic cirrhosis with ascites, intestinal malabsorption syndrome, nephritic syndrome, burns)
• More than 24 hours since inclusion criteria were met
• Religious objection to the administration of human blood products
• Inclusion in other experimental study