Development and validation of a multivariable prediction rule for detecting a severe acquired ADAMTS13 activity deficiency in patients with thrombotic microangiopathies

Thrombotic microangiopathies (TMAs) are a group of diseases that have different aetiologies and treatments, but a clinical differential diagnosis remains difficult. Among TMAs, thrombotic thrombocytopenic purpura (TTP) is characterised by a severe ADAMTS13 functional deficiency. However, assays exploring ADAMTS13 activity are limited to some specialised laboratories. Our objective was to develop and validate a diagnostic method for TTP in adult patients with TMA. We generated a multivariable model (four predictors) on a cohort of 174 TMA patients in order to predict an ADAMTS13 activity deficiency (AUC of 0.927). The multivariable model was simplified into a binary rule to facilitate the interpretation of the predictions. There were two scenarios for a patient: (1) Predicted ADAMTS13 deficiency; if the patient met four conditions simultaneously (platelets ≤44×109/L, creatinine ≤2 mg/dL (≤176.84 µmol/L) for males or ≤1.9 mg/dL (≤168 µmol/L) for females, age ≤68 years and no history of haematopoietic stem cell transplant [HSCT]); or (2) Predicted "normal" activity; if any of the above conditions are not met. This rule was validated on a second cohort of 86 patients and performed with sensitivity of 87.7% and specificity of 92.7%. This could lead to the earlier confirmation or rapid exclusion of TTP when ADAMTS13 testing is not avalilable, facilitating a more suitable therapy based on the aetiology of the TMA.