Trauma Sub-Endotypes Identified Using Latent Class Analysis Have a Differential Response to Blood Product Transfusion Ratios. A secondary analysis of the PROPPR Randomized Trial

Abstract

Background: The American Red Cross declared the first ever national blood shortage crisis amid the COVID epidemic in 2022. In severe traumatic injury, large volumes of blood products are often required to resuscitate patients with hemorrhagic shock. A 1:1:1 transfusion ratio (plasma:platelets:red blood cells) compared to a 1:1:2 ratio improves time to bleeding cessation in severely injured patients with hemorrhagic shock, but has not been associated with reduced mortality in randomized trials. It remains unclear whether certain patients within the trauma population benefit from a 1:1:1 or 1:1:2 resuscitation strategy. My objective was to derive trauma sub-endotypes using latent class analysis (LCA) of molecular data and determine whether these subgroups were associated with mortality and differential treatment response to 1:1:1 vs. 1:1:2 resuscitation strategies. Methods: A secondary analysis of the PROPPR trial, a randomized trial published in JAMA in 2015 that demonstrated no difference in 30-day mortality between the two blood transfusion resuscitation groups. LCA was performed on a panel of twelve plasma biomarkers in 478 severely injured patients (out of a total of 680 patients in the original PROPPR Trial) who had a complete panel of biomarkers drawn at the time of presentation and before blood product resuscitation. I tested for an association between trauma endotypes and 30-day mortality using multivariable relative risk (RR) regression adjusting for age, sex, trauma center, mechanism of injury, and injury severity score (ISS). I tested for a differential treatment response to transfusion strategy using a RR regression model for 30-day mortality by incorporating an interaction term for the product of endotype group and treatment group adjusting for age, sex, trauma center, mechanism of injury, and ISS. Results: LCA identified two endotype groups as the optimal model within the population. Trauma endotype-1 (TE-1, n = 248) was associated with significantly higher risk for 30-day mortality compared with trauma endotype-2 (TE-2, n = 230). Mortality in TE-1 was 28.6% with 1:1:2 treatment vs. 32.6% with 1:1:1 treatment, whereas mortality in TE-2 was 24.5% with 1:1:2 treatment vs. 7.3% with 1:1:1 treatment. There was a significant interaction between treatment arm and TE for 30-day mortality (p-value for interaction = 0.02). Conclusions: Endotypes derived from plasma biomarkers in trauma patients at hospital arrival were associated with a differential response to 1:1:1 vs. 1:1:2 resuscitation strategies in severely injured trauma patients. These findings support the concept of molecular heterogeneity in critically ill trauma populations and have implications for tailoring therapy for patients at high risk for adverse outcomes.