Dados do Trabalho

Título

A MODEL TO CLASSIFY THROMBOTIC MICROANGIOPATHY (TMA) PHENOTYPE IN HOSPITALIZED PATIENTS (TMA-ETI SCORE)

Introdução

Thrombotic Microangiopathy (TMA) is a syndrome characterized by the presence of anemia and thrombocytopenia as a consequence of microthrombi formation, resulting in organ damage by ischemic injury. The primary aim is to develop an algorithm to classify thrombotic microangiopathy (TMA) phenotype in hospitalized patients (TMA-ETI score).

Material e Método

This is a single-center retrospective cohort study including hospitalized patients with TMA at Hospital das Clínicas of Botucatu, UNESP. We included all consecutive patients diagnosed with TMA between Jan 1, 2012 and Dec 31, 2021. We defined TMA as a presence of anemia (Hb<10g/dL) and thrombocytopenia (platelet count<150,000/µL) associated with signs of hemolysis (increase in lactate dehydrogenase (LDH) more than 1.5 times the upper reference, or consumed haptoglobin, or the presence of schistocytes).
We classified patients in 8 categories: Infections; Malignant Hypertension; Transplant; Malignancy; Pregnancy; Thrombotic Thrombocytopenic Purpura (PTT); Shiga Toxin-mediated hemolytic uremic syndrome (STEC-SHU) and Complement Mediated TMA (CM-TMA or aHUS). We fitted a model to classify patients using clinical characteristics, biochemical exams and mean arterial pressure (MAP) at presentation. The associated conditions and comorbidities were extracted by the evolution field of electronic medical records. All laboratory exams were retrieved within the course of hospitalization.

Resultados

We retrospectively retrieve a large number of TMA phenotypes using automatic strategies in electronic health records in almost 10 years (n=2407). Secondary TMA was found in 97.5% and Primary TMA was found in 2.47% of the patients (TTP/aHUS). The best model was LightGBM with accuracy of 0.979, and multiclass ROC-AUC of 0.966. The predictions were very accurate in infectious related, malignancy, malignant hypertension, pregnancy related, transplant, and TTP. The predictions had lower confidence in aHUS and STEC-HUS cases.The ten strongest predictors were: presence of infection, presence of hypertensive emergency, active malignancy, active pregnancy, minimum platelets, age, active transplant and maximum total bilirubin.

Discussão e Conclusões

Secondary conditions were the most common etiologies of TMA in hospitalized patients. We retrieved comorbidities, associated conditions, and MAP to fit a model to predict TMA and also define TMA phenotypic characteristics. This is the first multiclass model to predict TMA in an eight class category including primary and secondary conditions.

Palavras Chave

Thrombotic Microangiopathy; aHUS; Hemolysis; Algorithm

Área

Doenças do glomérulo