Pentraxin 3 in patients with severe sepsis or shock: the ALBIOS trial

Abstract

Background: The long pentraxin PTX3 is a key component of the humoral arm of innate immunity related to sepsis severity and mortality. We evaluated the clinical and prognostic significance of circulating PTX3 in the largest cohort ever reported of patients with severe sepsis or septic shock.

Materials and methods: Plasma PTX3 was measured on days 1, 2 and 7 after randomization of 958 patients to albumin or crystalloids for fluid resuscitation in the multicentre Albumin Italian Outcome Sepsis (ALBIOS) trial. We tested the association of PTX3 and its changes over time with clinical severity, prevalent and incident organ dysfunctions, 90-day mortality and treatment.

Results: PTX3 was high at baseline (72 [33-186] ng/mL) and rose with the severity and number of organ dysfunctions (P < 0·001) and the incidence of subsequent new failures. The PTX3 concentration dropped from day 1 to 7, but this decrease was less pronounced in patients with septic shock (P = 0·0004). Higher concentrations of PTX3 on day 1 predicted incident organ dysfunctions. Albumin supplementation was associated with lower levels of PTX3 in patients with septic shock (P = 0·005) but not in those without shock. In a fully adjusted multivariable model, PTX3 on day 7 predicted 90-day mortality. Smaller drops in PTX3 predicted higher 90-day mortality.

Conclusions: In severe sepsis and septic shock, early high PTX3 predict subsequent new organ failures, while a smaller drop in circulating PTX3 over time predicts an increased risk of death. Patients with septic shock show lower levels of PTX3 when assigned to albumin than to crystalloids.

Keywords: Albumin; pentraxin 3; prognosis; septic shock; severe sepsis.

Figure 1. PTX3 concentration and incident organ failures

Relations between tertiles of PTX3 concentration on day 1 and incident organ failures. New organ failures were defined as a change in each component of the SOFA score during the study to 3 or 4 from 0, 1 or 2 at baseline. Number of patients without organ-specific failure at study entry: respiration (525), coagulation (862), liver (896), cardiovascular (392), renal (715). P across tertiles by chi-square test.

Figure 2. Time-course of PTX3 concentration by shock and randomized treatment

Plasma PTX3 concentrations on days 1, 2 and 7 are reported in patients with the three time-points available, i.e. 311 with severe sepsis without shock (151 randomized to albumin, 160 to crystalloids) and 377 with septic shock (190 randomized to albumin, 187 to crystalloids). Data are shown as median [Q1-Q3] after stratification for sepsis severity alone (a) or sepsis severity and randomized treatment (b,c). Repeated measures analysis of variance were done on log-transformed concentrations. In all patients: time p<0.0001, shock p<0.0001, interaction time by shock p=0.0004; in patients with severe sepsis without shock: time p<0.0001, treatment p=0.82, interaction time by treatment p=0.90; in patients with septic shock: time p<0.0001, treatment p=0.005, interaction time by treatment p=0.55.

Figure 3. Kaplan-Meier survival curves for 90-day mortality by tertiles of PTX3 on days 1, 2 and 7

The number of patients at risk and events are shown below the curves. Tertiles of PTX3: day 1 (lower < 41.1, mid 41.1 to 137.5, upper > 137.5 ng/mL), day 2 (lower < 25.3, mid 25.3 to 54.6, upper >54.6 ng/mL), day 3 (lower < 11.8, mid 11.8 to 26.9, upper >26.9 ng/mL). P for log-rank test <0.0001 for all.

Figure 4. Modified Poisson regression models for 90-day mortality by changes in PTX3 concentrations over time

The relation between continuous changes in PTX3 concentrations between days 1, 2 or 7 was estimated with simple or adjusted modified Poisson regression models. Multivariable model 1 included the log-transformed concentration of PTX3 on day 1, age, sex, body-mass index, reason for ICU admission, SAPS II and SOFA scores, pre-existing conditions (liver disease, chronic obstructive pulmonary disease, chronic renal failure, immunodeficiency, congestive or ischemic heart disease), mean arterial pressure, fraction of inspired oxygen FiO₂, diuresis, serum concentrations of lactate, albumin, bilirubin and creatinine, blood platelets, septic shock at randomization, mechanical ventilation, antibiotics at randomization, positive blood culture. Model 2 = model 1 plus plasma concentrations of presepsin and high-sensitive cardiac troponin T on day 1.