Artificial intelligence-based body composition predicts postoperative complications and mortality in advanced heart failure patients receiving long-term mechanical circulatory support

Abstract

Abstract Background Obesity is an independent cardiovascular risk factor and is associated with higher postoperative infection rates in patients after cardiac surgery. However, in heart failure, obesity has been described as beneficial in terms of survival (“obesity paradox”); whereas sarcopenia is associated with poor prognosis. An increasing number of advanced heart failure patients requires surgical treatment in terms of left ventricular assist device (LVAD) implantation. The postoperative in-hospital survival has improved in recent years but is still relatively high (11.1%). The impact of body composition on outcome in this population remains unclear. Purpose The aim of this investigation was to examine the preoperative computed tomography (CT) body composition based on artificial intelligence (AI) analysis as a predictor of the postoperative outcome in advanced heart failure patients, who underwent LVAD implantation. Methods Preoperative CT scans of 137 patients who received LVAD implantations between 2015 and 2020 were retrospectively analyzed using an AI-powered automated software tool based on a convolutional neural network, U-net, developed for image segmentation (Visage version 7.1., Visage Imaging GmbH, Berlin, Germany; Figure 1). Assessment of body composition included visceral and subcutaneous adipose tissue areas (VAT, SAT), psoas and total abdominal muscle areas (PMA, TAMA) and sarcopenia (defined as lumbal skeletal muscle index ≤38.5 cm2/m2 in women and ≤54.4 cm2/m2 in men). The body composition parameters were correlated with postoperative major complication rates and survival. Results The mean age of patients was 58.21±11.9 years; 122 (89.1%) were male. Most patients had severe heart failure requiring inotropes (INTERMACS profile I-III, 71.9%) secondary to coronary artery diseases or dilated cardiomyopathy (96.4%). 44 (32.1%) patients were obese (BMI ≥30 kg/m2), 96 (70.1%) were sarcopene and 19 (13.9%) were sarcopene obese. Adipose tissue was associated with a significantly higher risk of overall postoperative complications (SAT 158.22 cm2 [63.09, 256.78 cm2] vs. 124.04 cm2 [42.86, 187.68 cm2], p=0.049) and in-hospital mortality (VAT 168.11 cm2 [134.19, 285.27 cm2] vs. 135.42 cm2 [49.44, 227.91 cm2], p=0.033; SAT 227.28 cm2 [139.38, 304.35 cm2] vs. 173.81 cm2 [97.65, 254.16 cm2], p=0.009). Sarcopenia did not influence the postoperative outcome and survival within 1 year post LVAD implantation. Conclusions Preoperative AI-based CT body composition identifies patients with poor prognosis after LVAD implantation. Adipose tissue areas predict in-hospital mortality, thus, obesity as a perioperative risk factor outweighs the protective effect known in heart failure patients when undergoing LVAD implantation. Funding Acknowledgement Type of funding sources: None.