Proteína TREM-1 solúvel como preditor de mortalidade em pacientes HIV/AIDS internados: uma coorte prospectiva

Abstract

Introduction: The Human Immunodeficiency Virus (HIV) is one of the main causes of morbidity and mortality in the world. Viral reservoirs are the main barrier to achieve a cure for HIV in infected individuals, even in the presence of Antiretroviral Therapy (ART). HIV infected macrophages migrate to various tissues, forming a stable viral reservoir in continuous production. TREM-1 is a myeloid cell protein-1, expressed in monocytes/macrophages and neutrophils, fundamental in the regulation and amplification of inflammation in the immune response. Objective: To assess whether the sTREM-1 protein has a predictive value for mortality in HIV/AIDS patients hospitalized for HIV-related causes of hospital admission, in a well-known and reputed hospital in the city of Manaus-AM, Brazil. Methodology: A prospective cohort study was carried out in patients hospitalized at the Fundação de Medicina Tropical Doutor Heitor Vieira Dourado (FMT – HVD), with a 28-day segment for predicting mortality. A total of 340 patients who met the eligibility criteria were identified in the study, in which 260 were included. sTREM-1 plasma was quantified by ELISA. To evaluate the correlation between sTREM-1 and mortality prediction in HIV patients, the ROC Curve, Area Under the Curve (AUC) and Poisson Multivariate Regression were calculated. Results: From 260 patients included, 81.9% survived and 18.1% died. In the death group, the mean value of sTREM-1 was higher (885.9±197.7) compared to those who survived (555.8±134.3; p = 0.0001), with a large magnitude of effect (Cohen's Q) 1.96, for a mean difference of 392.4±54.7pg/mL (Mann-Whitney test for ranks; p<0.0001). The Area AUC was 0.91 95% CI 0.85 – 0.97 (p < 0.0001). The selected cut-off value for high sTREM-1 was 683.8 pg/mL with the best correct classification of cases at 84.23%, sensitivity of 89.36% (95% CI 74.8 – 94.2 %) and specificity 83.10% (95% CI 80.1 – 89.6%). The calculated positive odds ratio (LR+) was 5.28 and negative (LR-) 0.13. Poisson Multivariate Regression was calculated by classifying the serum levels of sTREM-1 in the serum according to the cut-off value (High > 683.8pg/mL or Low <683.8pg/mL), where high sTREM-1 showed the best performance for predicting death with a high Incidence Density Ratio (IRR 16.7, p = 0.000). Conclusion: In summary, these findings show that sTREM-1 is a predictor of mortality in hospitalized HIV patients.