Novel biomarkers for detection of early acute kidney injury, renal recovery and bacterial infections in critically ill patients

Abstract: Diagnosis of infection in the intensive care unit (ICU) is challenging because the signs and symptoms normally attributed to infection are quite common also in ICU patients without infection. This is a problem as delayed antibiotic therapy may increase the risk of organ failure and ultimately, death. One example of organ failure is acute kidney injury (AKI), which affects more than 1/3 of ICU patients. Diagnosis of AKI and decision to initiate supportive treatment (e.g. renal replacement therapy, RRT) is largely based on markers of kidney dysfunction - rather than markers of kidney damage. Moreover, markers to predict successful discontinuation of RRT are currently lacking. It is possible that we in a foreseeable future will be able to detect and treat both infection and AKI in the ICU earlier than we can today. A method that has been suggested is the use of biomarkers - biological markers that we can measure in the patient's blood or urine. The aim of this thesis is to study a number of potential biomarkers to predict AKI development and renal recovery (studies I and IV) and to detect infection (studies II and III) in ICU patients. Study I examined if endostatin – a potential marker of renal epithelial and endothelial damage – could predict the development of AKI within 72 hours after ICU admission. Of the 93 studied patients, 21 developed AKI within 72 hours. We also created a clinical risk prediction model based on age, APACHE II score and early oliguria. The statistical model predicted outcome with fair accuracy. Adding endostatin to the model increased prediction accuracy. In study II we measured daily plasma calprotectin levels in 110 ICU patients in order to assess calprotectin as an early marker of infection in the ICU. Altogether, 58 patients developed infection. The study showed that, in ICU patients, plasma calprotectin was as good as C-reactive protein (CRP) in predicting infection and better than white blood cell count (WBC) and procalcitonin. In study III we examined dimeric neutrophil-gelatinase associated lipocalin (dNGAL), a protein released from activated neutrophils, as an early marker of infection in the ICU and its response to antibiotic therapy. Daily plasma dNGAL was measured in 198 ICU patients. We found that infection, but not AKI, was independently associated with greater dimeric NGAL levels. However, its value as an early marker of bacterial infection was limited. Following initiation of appropriate antibiotic therapy, dNGAL decreased more rapidly than the traditional biomarkers CRP and PCT. In study IV we studied 135 ICU patients with AKI requiring RRT. We assessed if biomarker concentrations in plasma and urine (NGAL, endostatin, cystatin C, creatinine, urea), before and during RRT, alone and together with a clinical prediction model, could improve prediction of renal recovery within 60 days of ICU admission (alive and without need for RRT). By day 60, renal recovery was found in 98 of the 135 patients. The individual biomarkers in plasma or urine were poor predictors of renal recovery. The clinical prediction model, based on patient age and daily urine output, predicted renal recovery with reasonable accuracy.