High levels of endogenous estrogens are associated with death in the critically injured adult

Abstract

Background: Sex hormones exhibit predictable changes in their physiologic patterns during critical illness. Endogenous estrogens are elevated in both genders as a result of the peripheral conversion of androgens to estrogens by the aromatase enzyme. Elevated endogenous estrogens have been associated with death in medical and mixed surgical intensive care unit (ICU) patients. Our objective was to determine the relationship between endogenous estrogens and outcomes in critically injured patients.

Methods: A prospective cohort of injured patients remaining in the ICU for at least 48 hours at two trauma centers was enrolled. Sex hormones (estradiol, progesterone, testosterone, prolactin, and dehydroepiandrosterone-sulfate) were assayed and mortality was assessed. A logistic regression model was used to determine the association between estradiol and death. The area under the receiver operating characteristic (AUROC) curve was used to estimate the accuracy of estradiol in predicting death.

Results: Nine hundred ninety-one patients were enrolled with a 13.4% mortality rate. Despite no detectable difference in mortality among genders, estradiol was significantly elevated in nonsurvivors (16 pg/mL vs. 35 pg/mL, p < 0.001). Estradiol was a marker for injury severity with the most severely injured patients exhibiting the highest levels. The ability of estradiol to predict death (AUROC = 0.65) was comparable with Trauma and Injury Severity Score (AUROC = 0.65) and superior to Injury Severity Score (AUROC = 0.54) in this cohort.

Conclusions: Serum estradiol is a marker of injury severity and a predictor of death in the critically injured patient, regardless of gender. Whether or not estradiol plays a causal role in outcomes is unclear, but estrogen modulation represents a potential therapy for improving outcomes in critically ill trauma patients.

Trial registration: ClinicalTrials.gov NCT00170560.

Figures

Fig. 1

Relationship between APACHE II and serum estradiol. Regression model using restricted cubic spline covariates demonstrating predicted estradiol with increasing APACHE II.

Fig. 2

Relationship between mortality and increasing serum estradiol. Regression model using restricted cubic spline covariates demonstrating probability of death with increasing serum estradiol levels (pg/mL). Although estradiol seems to be protective within the physiologic range (

Fig. 3

The receiver operating characteristic (ROC) curve for estradiol compared with other outcome predictors. The ROC curves for estradiol, ISS, and TRISS for predicting mortality are displayed. The curve for estradiol is comparable with TRISS and the area under the ROC curve (AUROC) is the same for these two variables.