Prognosis and Treatment of Necrotizing Soft Tissue Infections: A Prospective Cohort Study

Introduction:

Necrotizing soft-tissue infections (NSTI) are among the most serious and deadly infections known. They are characterized by rapidly progressing soft-tissue inflammation with necrosis and can quickly cause multiple organ failure and death. Mortality has been shown to be 25-35 %, with survivors coping with amputations and prolonged rehabilitation.

Currently, there is a lack of proper tools to evaluate the severity and prognosis of NSTI in individual patients. This results in necessary, yet sometimes overzealous surgical debridement, culminating in prolonged patient rehabilitation and invalidity. Hyperbaric oxygen therapy (HBOT) may be added as adjunctive therapy of NSTI. However, there is no clear understanding of the effectiveness of HBOT on NSTI. The investigators seek to remedy these two issues by examining multiple biomarkers over the course of several studies.

Methodology:

Location: Copenhagen University Hospital, Rigshospitalet, Denmark.

Design: Observational cohort study.

Cohort: All NSTI patients in Denmark since 2013.

Controls: 50-100 Patients undergoing elective, orthopedic surgery at Rigshospitalet.

Biomarkers: soluble thrombomodulin, syndecan-1, sE-selectin, VE-cadherin, protein C, suPAR.

Sample size calculations:

1: The test kits the investigators will be using to measure the primary outcome sTM (Human sCD141 ELISA kit, Nordic Biosite) have an interassay standard variation of 0.58 ng/ml. In order to be certain that measured changes in sTM concentration are not a result of interassay standard deviation, the investigators have set the mimimum relevant difference in sTM to 3 x the interassay standard variation, thus 1.75 ng/ml.

The investigators prepared a power calculation using a Wilcoxon rank sum test. Assuming an estimated standard deviation of 4.6 ng/ml and a mean of 9.9 ng/ml, the investigators will need to include a maximum of 150 NSTI patients and 50 elective surgery patients to reach a statistical power of at the very least 60 % (a very conservative estimate) and presumably closer to 85 % (more realistic estimate) at a 5 % significance level. The estimates depend on data distribution.

2: The test kits the investigators will be using to measure the primary outcome sE-selectin (Human CD62E ELISA kit, Diaclone) have an interassay standard variation of 0.37 ng/ml. In order to be certain that measured changes in sE-selectin concentration are not a result of interassay standard variation, the investigators have set the mimimum relevant difference in sE-selectin to 3 x the interassay standard variation, thus 1.1 ng/ml.

Assuming an estimated standard deviation of 209 ng/ml (septic shock) vs. 23 ng/ml (severe sepsis and sepsis) and means of 295 vs. 181 ng/ml, respectively, the investigators will need to include at least 132 NSTI patients and 50 elective surgery patients to reach a statistical power of 90 % at a 5 % significance level.

3: suPAR levels during NSTI have never previously been examined. In order to estimate sample size and since NSTI patients are also septic, the investigators are basing the sample size calculation on a previous study concerning the correlation between suPAR and sepsis. This study found statistically significant correlation between suPAR levels and mortality in 141 patients. This is also the goal of this study. Further studies have also found significant correlations between suPAR, sepsis and mortality in 132 patients. The investigators will include at least 150 NSTI patients during this study.

Statistical considerations:

To check whether the HBOT treatment has an effect on the range of biomarkers, the investigators will analyze the means and variances of the biomarkers in the NSTI group and the two control groups, the orthopaedic patients and the sepsis patients. Non-parametric data will be log-transformed and will be presented as median values with IQR. Wilcoxon rank sum tests will be used for group comparisons. Fisher's exact test will be used for categorical data. Correlation analysis will be performed using Spearman rank correlation or Pearson correlation. To assess the quality of suPAR as a predictor of health outcomes, a model selection exercise will be conducted with various types of regression models. The type of regression will vary with the type of health-outcome, with suPAR as the predictor in all cases. Receiver operating characteristic (ROC) curve analysis will be applied to determine suPARs accuracy as a marker of severity and mortality in patients with NSTI. The investigators will construct Kaplan-Meier curves for survival data. Statistically significant results are when p<0.05.

Data:

Data will be handled according to the National Data Protection Agency. All original records (including consent forms and questionnaires) will be archived at the trial site for 15 years. The National Data Protection Agency has approved the biobank (RH-2016- 199). Data checks have been programmed into the data registry to warn when input variables are outside of predefined possible clinical range. All registry data will be compared to external data sources, i.e. medical records, to ensure accuracy. Standard Operating Procedures have been implemented regarding data collection. Patients with missing data for calculating for example SAPS II scores etc. will be excluded from the study.

Ethics:

The trial will adhere to the Helsinki Declaration and Danish law. The National Ethics Committee and the Regional Ethics Committee (H-16021845) have approved this study.

Biomarker analyses, data extraction and interpretation will be performed once the recruitment of participants has ended.