Related Experiment Video
Updated: Jul 13, 2026
A Preclinical Controlled Cortical Impact Model for Traumatic Hemorrhage Contusion and Neuroinflammation
Published on: June 10, 2020
Novel Hematologic Inflammatory Biomarker Ratios Derived From Routinely Ordered Labs Upon Admission: Enhancement of
Kyle M Rei1, Javed Siddiqi2,3,4,5
1Neurology, University of Colorado Anschutz Medical Campus, Denver, USA.
Abstract:
Introduction Many novel hematologic inflammatory biomarker ratios derived from routinely ordered labs upon admission have demonstrated a significant predictive ability of six-month traumatic brain injury (TBI) outcomes. However, these biomarkers have not been evaluated in the context of existing validated TBI prognostic models. Methods This retrospective cohort study compared the predictive ability of 36 biomarkers measured upon admission to the International Mission for Prognosis and Analysis of Clinical Trials (IMPACT) prognostic model for six-month unfavorable outcomes and mortality of 199 neurosurgery patients between 2020 and 2024 in the United States. Model enhancement was evaluated via logistic regression comparing the IMPACT lab model to an extended model including one additional biomarker using likelihood ratio (LR) and Nagelkerke R². Results Of the 199 moderate-to-severe TBI patients included, 116 (58.3%) had six-month unfavorable outcomes, and 83 (41.7%) suffered six-month mortality. The IMPACT lab model showed excellent discrimination for both unfavorable outcomes (area under the receiver operating characteristic curve (AUROC) 0.887, p<.001) and mortality (AUROC 0.880, p<.001). Among 36 biomarkers, 19 showed significant discrimination on univariable analysis via AUROC. However, only 11 biomarkers remained significant after adjusting for the IMPACT lab model. Four biomarkers improved the IMPACT lab model for six-month unfavorable outcomes (R²=0.632): red cell distribution width (RDW) to platelet ratio (RPR) (LR=6.138, p=0.013, R²=0.654), lactate to platelet ratio (LPR) (LR=7.494, p=0.006, R²=0.672), prothrombin time (PT) (LR=5.636, p=0.018, R²=0.650), and lactate (LR=6.794, p=0.009, R²=0.670). Nine biomarkers improved the IMPACT lab model for six-month mortality (R²=0.659): RPR (LR=5.476, p=0.019, R²=0.678), LPR (LR=11.012, p<0.001, R²=0.694), lactate to albumin ratio (LAR) (LR=3.957, p=0.047, R²=0.737), platelets (LR=3.886, p=0.049, R²=0.672), lactate (LR=8.759, p=0.003, R²=0.686), albumin (LR=4.639, p=0.031, R²=0.739), PT (LR=12.764, p<0.001, R²=0.699), international normalized ratio (INR) (LR=12.02, p<0.001, R²=0.697), and activated partial thromboplastin time (aPTT) (LR=9.572, p=0.002, R²=0.688). Conclusions We demonstrated that many biomarkers showing significant predictive ability on univariable analysis lacked robustness when adjusting for a validated prognostic model, and only nine biomarkers enhanced the IMPACT lab model: RPR, LPR, LAR, platelets, PT, INR, aPTT, lactate, and albumin. RPR provided the IMPACT lab model with superior predictive ability for six-month unfavorable outcomes as compared to its individual constituent biomarkers, RDW and platelets, which demonstrates the utility of novel biomarker ratios. LPR and LAR did not provide the IMPACT lab model with superior predictive ability as compared to their constituent biomarkers, which may be preferred due to superior parsimony. Additional research is needed to validate the addition of these biomarkers to the IMPACT lab model. Future studies should evaluate TBI biomarkers in the context of existing validated models such as IMPACT or Corticosteroid Randomization After Significant Head Injury (CRASH).
