ABSTRACT
BACKGROUND & AIMS: Accurate diagnosis of sarcopenia requires evaluation of muscle quality, which refers to the amount of fat infiltration in muscle tissue. In this study, we aim to investigate whether we can independently predict mortality risk in transcatheter aortic valve implantation (TAVI) patients, using automatic deep learning algorithms to assess muscle quality on procedural computed tomography (CT) scans. METHODS: This study included 1199 patients with severe aortic stenosis who underwent transcatheter aortic valve implantation (TAVI) between January 2010 and January 2020. A procedural CT scan was performed as part of the preprocedural-TAVI evaluation, and the scans were analyzed using deep-learning-based software to automatically determine skeletal muscle density (SMD) and intermuscular adipose tissue (IMAT). The association of SMD and IMAT with all-cause mortality was analyzed using a Cox regression model, adjusted for other known mortality predictors, including muscle mass. RESULTS: The mean age of the participants was 80 ± 7 years, 53% were female. The median observation time was 1084 days, and the overall mortality rate was 39%. We found that the lowest tertile of muscle quality, as determined by SMD, was associated with an increased risk of mortality (HR 1.40 [95%CI: 1.15-1.70], p < 0.01). Similarly, low muscle quality as defined by high IMAT in the lowest tertile was also associated with increased mortality risk (HR 1.24 [95%CI: 1.01-1.52], p = 0.04). CONCLUSIONS: Our findings suggest that deep learning-assessed low muscle quality, as indicated by fat infiltration in muscle tissue, is a practical, useful and independent predictor of mortality after TAVI.
ABSTRACT
The incidence of aortic valve stenosis (AoS) increases with age, and once diagnosed, symptomatic severe AoS has a yearly mortality rate of 25%. AoS is diagnosed with transthoracic echocardiography (TTE), however, this gold standard is time consuming and operator and acoustic window dependent. As AoS affects the arterial blood pressure waveform, AoS-specific waveform features might serve as a diagnostic tool. Aim of the present study was to develop a novel, non-invasive, AoS detection model based on blood pressures waveforms. This cross-sectional study included patients with AoS undergoing elective transcatheter or surgical aortic valve replacement. AoS was determined using TTE, and patients with no or mild AoS were labelled as patients without AoS, while patients with moderate or severe AoS were labelled as patients with AoS. Non-invasive blood pressure measurements were performed in awake patients. Ten minutes of consecutive data was collected. Several blood pressure-based features were derived, and the median, interquartile range, variance, and the 1st and 9th decile of the change of these features were calculated. The primary outcome was the development of a machine-learning model for AoS detection, investigating multiple classifiers and training on the area under the receiver-operating curve (AUROC). In total, 101 patients with AoS and 48 patients without AoS were included. Patients with AoS showed an increase in left ventricular ejection time (0.02 s, p = 0.001), a delayed maximum upstroke in the systolic phase (0.015 s, p < 0.001), and a delayed maximal systolic pressure (0.03 s, p < 0.001) compared to patients without AoS. With the logistic regression model, a sensitivity of 0.81, specificity of 0.67, and AUROC of 0.79 were found. The majority of the population without AoS was male (85%), whereas in the population with AoS this was evenly distributed (54% males). Age was significantly (5 years, p < 0.001) higher in the population with AoS. In the present study, we developed a novel model able to distinguish no to mild AoS from moderate to severe AoS, based on blood pressure features with high accuracy. Clinical registration number: The study entailing patients with TAVR treatment was registered at ClinicalTrials.gov (NCT03088787, https://clinicaltrials.gov/ct2/show/NCT03088787 ). The study with elective cardiac surgery patients was registered with the Netherland Trial Register (NL7810, https://trialsearch.who.int/Trial2.aspx?TrialID=NL7810 ).