The SOPRANI network has brought together the different relevant backgrounds of investigators and PhD students and holds biannual meetings that focus on education as well as on scientific progress and discussion. In addition, the network has established access for its collaborators to large amounts of patient data in the field of traumatic brain injury by data sharing agreements with most of the existing data repositories originating from studies and trials, as well as a joint controller agreement within the network. Thereby, the network has created the largest dataset ever, operated on through a federate analysis set up. Five PhD students are applying advanced modeling using these repositories. A Python library was created that enables the visualization of insult burden curves relating secondary event parameters to outcome, and was applied to body temperature, intracranial pressure and cerebral perfusion pressure. This work was extended for the first time to spinal cord injury by establishing insult burden curves for intraspinal pressure and spinal cord perfusion pressure. Safety and feasibility of intraspinal monitoring was demonstrated using a multicenter dataset. While patients are further recruited, intraspinal microdialysis output is being correlated with intraspinal pressure and neurological recovery data. Physiological monitoring signatures of ICU intervention annotations in brain injury are being investigated for automated classification and integration in effect prediction models. Proof of concept in this matter has been delivered for osmotherapy interventions, and more interventions are in the pipeline. Further, inherent instability of clustering-based patient groupings was demonstrated, showing how conformal prediction can provide meaningful uncertainty estimates for outcome prediction, and methods are being prepared that translate these findings into clinically relevant guidance. Older outcome prediction and risk assessment models are being updated using recent patient cohorts, introducing time-dynamic biomarker data and dynamic imputation of missing values in continuous vital-sign data. Four students are fine tuning pathological event understanding and detection in animal models in order to develop novel monitoring tools. Two are working on cerebral blood flow autoregulation, unraveling existence of hysteresis patterns and interference of vasoactive drugs. It was found in this set up that functional ultrasound-derived signals reliably capture cerebrovascular responses, supporting its potential for continuous assessment of autoregulation in patients. One student investigates how the protective blood brain barrier (dys)functions in a rat model under varying conditions that mimic circumstances experienced by patients in the neuro-ICU, and was able to establish a robust and reproducible MRI acquisition and analysis pipeline. Another student has developed information-theoretic metrics to better characterize pathological brain hemodynamics from currently available bedside monitoring signals. The latter student is working on a physical model of brain hemodynamics to better capture pressure-flow relationships underlying cerebral autoregulation, in close collaboration with the first 2 colleagues. Finally, one student works on data synthesis preparing decision support platforms, has completed a systematic review on microdialysis biomarkers and their association with outcome and currently performs an individual participant data meta-analysis on intracranial pressure related variables and outcomes.
Since TBI is a sex dependent pathology, controlling for sex difference is essential. The proportion of male/female patients in the Center-TBI database is 67/33%, and a fairly similar distribution is found in Track-TBI, Brain-IT and other patient data repositories the students have access to. The higher female representation than in older literature is explained by the increasing proportion of TBI in older adults, that have more or less equal male/female distributions. In the outcome prediction model updates (Utrecht students), sex is inherently included. The secondary insult burden curves (KUL) are being refined as access to larger databases is available now, and this work includes the investigation of differences of tolerance curves for intracranial pressure and other insults for the 2 sexes (which was not available until now). The piglet model experiments on effects of vasoactive drugs on autoregulation and of hysteresis are being performed in male subjects first and findings will be separately investigated/confirmed in female animals. For the rat blood brain barrier experiments, at present the experiments are also done exclusively on male rats. There is a stringent reason for this: spontaneously hypertensive rats (SHR) used in these experiments show clear sex differences that would affect the experimental pipeline beyond TBI-pathophysiology related sex differences: females have lower blood pressure, greater NO bioavailability and better renal outcomes than age-matched males. This is a limitation, and findings are therefore contextualized within existing data on sex-specific (hormonal, vascular etc) sex-specific mechanisms.
