Estimating coagulopathy in an ovine acute lung injury model of sepsis using a disease progression model

Introduction

Acute lung injury (ALI) caused by smoke inhalation with or without bacterial pneumonia remains a significant cause of morbidity and mortality among burn patients. Bacterial pneumonia in an ALI patient is particularly worrisome because it often leads to sepsis. Although much of the literature surrounding ALI and pneumonia-induced sepsis has rightfully focused on pulmonary and endothelial changes, a major consequence of ALI and an area of continued research and drug development is coagulopathy. The objective, therefore, of our study was to determine whether coagulopathy differs between types of ALI and whether the dysregulation can be estimated using a disease progression model.

Methods

Nineteen sheep with acute lung injury were incorporated into this pneumonia-sepsis model. Pneumonia was induced by inoculating the airway with ~2.5 × 10¹¹ colony-forming units (CFUs) methicillin-resistant Staphylococcus aureus (MRSA), while smoke injury was created through inhalation of cotton smoke. The injury groups studied were as follows; MRSA and smoke inhalation (M+S), MRSA untreated (M), MRSA treated (M+T), and smoke inhalation only (S). Data were modeled over 24 hours. First, all the sheep were modeled together to determine a rank-order of the injury groups. After rank-ordering the groups, the groups became model inputs and in conjunction with other clinical and laboratory variables were used to estimate the output parameter, prothrombin time (PT). In order to minimize overparameterization of a small patient population, the model was allowed to estimate PT using only two parameters.

Results

The number of sheep in each group was as follows; seven M+S, three M, three M+T, and six S. The rank-order of injury from least to greatest severity was M+T, S, M, M+S. The two highest-ranking parameters in estimating PT were calcium and injury. When using calcium and injury alone, the model estimate agreement with measured PT was r² = 0.70 and r² = 0.36, respectively. Allowing the model to combine the inputs did not improve the model estimate (r² = 0.70) compared with when calcium was used alone.

Conclusions

The progression model allowed all individual sheep to be characterized as to the severity of resulting coagulopathy and identified some important co-factors. Acute lung injury can lead to systemic coagulopathy even without MRSA infection, but the extent and severity is greater with infection.