Biomarkers for early diagnosis of Takotsubo syndrome: a model development and validation study
V Schweiger, F A Wenzl, M A Smolle, D Zech, A Akhmedov, Y Ilina, K Santos, V L Cammann, P Wang, M Roffi, L Raeber, J R Ghadri, D Di Vece, T F Luescher, C TemplinAbstract
Background
Takotsubo syndrome (TTS) frequently mimics acute coronary syndrome (ACS), and its diagnosis currently relies on clinical history, cardiac imaging, and exclusion of obstructive coronary artery disease, often resulting in delayed diagnosis and unnecessary invasive coronary angiography.
Purpose
To develop and externally validate a biomarker-based score for early differentiation of TTS and ACS.
Methods
In 3,615 consecutive patients with ACS (n=3,469, SPUM-ACS) or TTS (n=146, InterTAK) enrolled between 07/2011 and 07/2023, established and emerging cardiovascular biomarkers were measured before invasive coronary angiography. We applied machine learning-based recursive feature elimination to assess the importance of variables for diagnosing TTS. Top-ranking variables were included in a biomarker-based version of the InterTAK diagnostic score (short BioTAK) for early diagnosis of TTS. The score was developed in 1,823 patients presenting with TTS or ACS at a university hospital between 07/2011 and 01/2018 and externally validated in 1,792 patients presenting with TTS or ACS at a different time and/or study centre. Diagnostic performance of the score was assessed according to a range of performance metrics.
Results
Based on diagnostic variable importance and clinical considerations, biomarkers of congestion (N-terminal pro-B-type natriuretic peptide, peptidylglycine α-amidating monooxygenase), atherosclerotic plaque instability (soluble lectin-like oxidized low-density lipoprotein receptor-1), and lipid metabolism (low-density lipoprotein cholesterol), as well as female sex were incorporated into the score. BioTAK achieved excellent discrimination and good calibration (development cohort: AUC 0.96, [95%CI 0.94–0.98], calibration slope 1.00 [0.81–1.19], calibration-in-the-large 0.00 [-0.43–0.43]) and maintained similarly high performance on external validation (AUC 0.93 [0.90–0.96], calibration slope 0.95 [0.77–1.14], calibration-in-the-large 0.00 [-0.37–0.38]). Score values ≥35 points ruled in TTS with 98.6% specificity (PPV 60.7%) and ≤28 points ruled out TTS with 91.3% sensitivity (NPV 99.6), classifying 89.7% of patients into rule-in or rule-out categories prior to angiography.
Conclusions
A simple biomarker-based score enables accurate early rule-in and rule-out of TTS before coronary angiography. BioTAK has the potential to streamline diagnostic pathways and reduce unnecessary diagnostic procedures in patients presenting with suspected ACS.Figure 1:Variable importance diagnosing TTS.Figure 2:BioTAK Diagnostic performance.