On the prognostic value of Heparin-binding protein in Sepsis

Sammanfattning: Sepsis causes major morbidity and mortality worldwide. There is wide consensus that biomarkers have an important role in sepsis research and patient care. Heparin-binding protein (HBP) is a neutrophil-derived, proinflammatory and vascular permeability-inducing protein and a promising novel sepsis biomarker. In this thesis, we investigate some aspects of the prognostic accuracy and added value of single and repeated HBP measurements in plasma at the emergency department (ED) and the intensive care unit (ICU) regarding infection-related organ dysfunction (OD) and sepsis survival. In Paper I, we conducted a prospective, observational, convenience sample study (NCT02366650) and recruited patients with affected vital signs regardless of infection suspicion at the ED of four centres in three countries in 2015 to 2016. Among 524 included patients, we found that plasma HBP on ED admission had an area under the receiver operating characteristics curve (AUC) of 0.73 (95% CI: 0.68-0.78) to discriminate the pre-specified primary endpoint of infection-induced OD within 72 hours compared to an AUC of 0.82 (0.78-0.86) for C-reactive protein and 0.69 (0.64-0.74) for procalcitonin. In Paper II, we did a post hoc study including patients with severe sepsis or septic shock and available plasma samples on ICU admission from the prospective, observational FINNAKI study conducted in 17 Finnish ICUs in 2011 to 2012. In a total of 511 patients, addition of plasma HBP to a prediction model including age, simplified acute physiology score II, and creatinine 48 hours pre-ICU increased the AUC from 0.78 (0.73–0.84) to 0.82 (95% CI: 0.77–0.87) regarding the primary endpoint of acute kidney injury (AKI) stage 2-3 from 12 hours up to 5 days. In Paper III, we conducted an observational, convenience sample study recruiting patients with suspected septic shock at two general mixed ICUs in Skåne, Sweden, and sampled patients for plasma from ICU admission and every 4 hours for 3 days. Among 24 included patients, we found that plasma HBP is highly variable in concentration between 4-hour measurements and that every 100 ng/mL increase in HBP (range 0 to 932 ng/mL) corresponded to 1.4 mmHg decrease in mean arterial pressure in a linear mixed-effects model adjusted for time, noradrenaline dose and vasopressin use (95% CI: -1 to -2.3 mmHg, p=0.04). In Paper IV, we performed another post hoc investigation of the FINNAKI study, this time including longitudinal sampling for plasma HBP up to seven times during the first five days (hour 0, 12, 24, 36, 48 and day 3 and 5). We pre-published a statistical analysis plan (ISRCTN15560762). In a total of 652 patients, we found that longitudinal HBP adds a small but statistically significant prognostic value to a prediction model (including age, sex, functional performance pre-ICU, sequential organ failure assessment score, lactate and pre-existing chronic health conditions) regarding the primary endpoint of 90-day survival in a complete case analysis (HR 1.06, 95% CI: 1.01 to 1.12, n=576, p=0.019) and in a post-hoc analysis using multiple imputation and nonlinear HBP over time (HR 1.26, 95% CI: 1.11 to 1.43, p<0.001) but not in the pre-specified analysis using multiple imputation and linear HBP over time (HR 1.11, 95% CI: 0.93 to 1.31, p=0.245).In summary, the findings from this thesis indicate that plasma HBP does not outperform current biomarkers to prognosticate infection-induced OD within 72 hours in patients in the ED with and without suspected infection, that plasma HBP can add prognostic value to known risk factors regarding development of sepsis-related AKI in the ICU, that repeatedly measured plasma HBP concentrations can be highly variable over time during sepsis and that longitundinal HBP adds little prognostic value to known risk factors regarding 90-day survival in ICU patients with sepsis.