Plasma soluble vascular endothelial growth factor receptor-1 levels predict outcomes of pneumonia-related septic shock patients: a prospective observational study

This prospective, observational study was conducted at Taipei Veterans General Hospital, a tertiary medical center in Taiwan. From January 2006 to February 2008, all patients who were at least 18 years of age and who were admitted to the medical intensive care unit (ICU) and respiratory ICU were screened. Patients who had a diagnosis of pneumonia and who fulfilled the American College of Chest Physicians/Society of Critical Care Medicine criteria for septic shock [1], which is defined as refractory hypotension despite adequate fluid supplement and the requirement of vasopressors to maintain the mean arterial blood pressure of at least 65 mm Hg, were eligible for enrollment. The diagnosis of pneumonia was confirmed on the basis of typical clinical presentations, fever, leukocytosis, and new infiltrates on chest radiographs [22, 23]. Patients with underlying malignancies, autoimmune disorders, active thromboembolic disease and those who failed to give informed consent were excluded. During the same period, we also enrolled 20 patients who were from the general medical ward and who had a diagnosis of pneumonia without sepsis or organ dysfunction as a control group. All patients were treated in accordance with the current treatment guidelines, and broad-spectrum antibiotics were administered within 1 hour of the onset of septic shock, as is routine in our ICUs. The antibiotics were also adjusted on the basis of the clinical response and the susceptibility profile of bacterial cultures. All patients with septic shock received a physiological dose of corticosteroids (hydrocortisone, 200 mg/day, divided in four doses). The study protocol was approved by the Taipei Veterans General Hospital Institutional Review Board, and the study was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants or their authorized representatives before enrollment.

Demographic characteristics, underlying comorbidities, disease severity, and organ dysfunction were determined on the day of enrollment. Organ dysfunction was defined as in a previous study [24]; the details of the definition are provided as supplemental material (Additional file 1). Disease severity was evaluated by the Acute Physiology and Chronic Health Evaluation II (APACHE II) score on the day of enrollment. Multi-organ dysfunction was defined as septic shock plus one or more organ dysfunctions, and the severity of multi-organ dysfunction was evaluated by the Sequential Organ Failure Assessment (SOFA) score. Survival status at 28 days and beyond was determined.

Blood samples were collected from peripheral vessels within 24 hours of shock onset and study entry. Plasma was separated from whole blood and stored at -70°C until analysis. Total forms of plasma VEGF and sVEGFR1 concentrations were measured in duplicate with commercial enzyme-linked immunosorbent assay kits (Quantikine; R&D Systems, Inc., Minneapolis, MN, USA). Plasma uPA activity was evaluated with a uPA Activity Assay kit (Chemicon International, Temecula, CA, USA).

Statistical analysis was performed with SPSS 14.0 software (SPSS, Inc., Chicago, IL, USA). Continuous variables between subgroups were compared with the Mann-Whitney U test or independent t test, and categorical variables were compared using Pearson's chi-square test. Binary logistic regression analysis was performed to determine the independent variables for multi-organ dysfunction. To determine the predictive accuracy of biomarker levels for survival and organ dysfunction, receiver operating characteristic (ROC) curves were constructed, and the areas under the curves (AUCs) were calculated.

For survival analysis, patients were stratified into subgroups according to plasma biomarker levels. Survival time was estimated by the Kaplan-Meier method, and the log-rank test was used to compare mortality between patients with quartile biomarker levels. Censored analysis was used since observation time was limited by discharge from the hospital. A multivariate Cox proportional hazards regression model with forward stepwise selection procedures was used to identify the risk factors for 28-day mortality. A P value of less than 0.1 in the univariate analysis was required for a variable to enter the multivariate model. A P value of less than 0.05 was considered statistically significant for all tests.

The study profile showing the number of cases and reasons for exclusion is shown in Figure 1. Ultimately, 81 patients with pneumonia complicated with septic shock were enrolled in the study. We also enrolled 20 patients with pneumonia without organ failure for comparison; all of these patients survived and were successfully discharged from the hospital. Among the 81 pneumonia patients with septic shock, 43 died within 28 days (28-day mortality of 53.1%), 11 died during their hospital stay after day 28 (hospital mortality of 66.7%), and 27 (33.3%) survived and were discharged from the hospital. The demographic characteristics of all pneumonia patients are shown in Table 1. The mean age of these patients was 79.3 ± 10.9 years, and the majority were male (87/101, 86.1%). There were no differences in age, gender, and underlying comorbidities between the patients with pneumonia without organ dysfunction, the septic shock survivors, and the non-survivors at day 28. The non-survivors of pneumonia-related septic shock had a significantly higher APACHE II score (P < 0.001), lower PaO₂/FiO₂ (arterial partial pressure of oxygen/fraction of inspired oxygen) ratio (P < 0.001), higher SOFA score (P < 0.001), and more organ dysfunctions as compared with the survivors. Pathogens were identified more frequently in patients with pneumonia-related septic shock as compared with those without organ dysfunction (P < 0.001).

The mean VEGF level of the septic shock non-survivors was 386.5 ± 524.1 pg/mL, which was significantly lower than that of the patients with pneumonia without organ dysfunction (688.9 ± 616.9 pg/mL, P = 0.005) but comparable to that of the septic shock survivors (219.9 ± 232.1 pg/mL, P = 0.455) (Figure 2). The mean sVEGFR1 level of the septic shock non-survivors was 659.3 ± 1,022.8 pg/mL, which was significantly higher than that of the septic shock survivors (221.1 ± 268.9 pg/mL, P < 0.001) and those with pneumonia without organ dysfunction (199.3 ± 81.6 pg/mL, P = 0.006). The mean plasma uPA activity of the septic shock non-survivors was 47.2 ± 40.6 units, which was significantly higher than that of the septic shock survivors (27.6 ± 17.2 units, P = 0.001) but comparable to that of the patients with pneumonia without organ dysfunction (44.3 ± 17.9 units, P = 0.243).

The ROC curves of day-1 plasma levels of VEGF and sVEGFR1, uPA activities, and APACHE II scores in predicting 28-day mortality of septic shock patients are shown in Figure 3. APACHE II score, sVEGFR1 level, and uPA activity were good predictors of 28-day mortality, and the AUCs were 0.861 (95% confidence interval [CI] 0.765 to 0.929, P < 0.001), 0.756 (95% CI 0.647 to 0.846, P < 0.001), and 0.716 (95% CI 0.646 to 0.812, P < 0.001), respectively. The AUC of plasma VEGF level for 28-day mortality was 0.55 (95% CI 0.434 to 0.662, P = 0.46).

The Kaplan-Meier analyses of survival according to quartile biomarker levels in septic shock patients are shown in Figure 4. Patients with higher plasma sVEGFR1 levels and uPA activities had significantly higher mortality (sVEGFR1, P < 0.001; uPA, P = 0.031) as compared with those with lower levels. The absolute difference in survival between the subgroups as represented by these biomarkers was evident from the early stage of septic shock. The survival curves overlapped between patients with different plasma VEGF levels (P = 0.58). Based on the optimal cutoff point of plasma sVEGFR1 level determined from the ROC curve (Figure 2), the Kaplan-Meier analysis of survival of the subgroups of patients divided by APACHE II scores and sVEGFR1 levels is shown in Figure 4d. In patients with an APACHE II score of at least 25, those with higher sVEGFR1 levels (>224 pg/mL) had a significantly higher mortality rate as compared with those with lower sVEGFR1 levels (P < 0.001). Similarly, in patients with an APACHE II score of less than 25, a trend toward higher mortality was found in those with higher sVEGFR1 levels (>224 pg/mL), but statistical significance was not achieved (P = 0.11).

Cox regression analysis models by quartile plasma biomarker levels, APACHE II scores, and comorbidities for the hazard ratio of mortality in septic shock patients are shown in Table 2. Both APACHE II score and increased plasma sVEGFR1 level were significant risk factors of 28-day mortality. The hazard ratios were 1.55 (P = 0.029) for every escalation of the sVEGFR1 quartile and 2.30 (P < 0.001) for that of APACHE II scores.

The occurrence of organ dysfunction, including acute respiratory distress syndrome, renal dysfunction, metabolic acidosis, and hematologic dysfunction, based on biomarker levels is shown in Figure 5. When the day-1 plasma biomarker levels were stratified according to the optimal cutoff point determined from the ROC curve (Figure 3), patients with higher sVEGFR1 levels and uPA activities had significantly higher incidences of organ dysfunction. The mean plasma sVEGFR1 levels and uPA activities of patients with various numbers of organ dysfunctions are also shown in Figure 6. Patients with more organ dysfunctions had significantly higher plasma sVEGFR1 levels and uPA activities compared with those with only one organ dysfunction.

Multivariate analysis by quartile plasma biomarker levels, APACHE II scores, and comorbidities for the hazard ratio of multi-organ dysfunction are shown in Table 3. Increased plasma sVEGFR1 levels and uPA activities were independent risk factors for multi-organ dysfunction. The hazard ratios were 2.82 (P = 0.021) for every escalation of the sVEGFR1 quartile and 2.75 (P = 0.037) for that of the uPA quartile.