Prediction of all-cause mortality in cardiovascular patients by using machine learning models

Background

Cardiovascular disease (CVD) is the leading cause of death worldwide. The prediction of all-cause death in patients with chronic cardiovascular risk will enable improved preventive strategies.

Purpose

This study aims to predict death of patients, who suffering from cardiovascular disease, within a follow-up period of 20 years. For this purpose was utilized the Ludwigshafen Risk and Cardiovascular Health (LURIC) study was utilized and employed machine learning models.

Methods

The Ludwigshafen Risk and Cardiovascular Health (LURIC) study was used in this analysis, which includes 3,316 patients (mean age 61.98 years, 69.2% male, 1722 deceased patients and 1594 alive or censored patients). After performing feature selection, 23 clinical and laboratory markers were included in our analysis. Pre-processing techniques such as SelectKbest and KNNImputer, were applied to the dataset and then it was splitted into a train and test (30%) set. The prediction of 20-year-risk of death was accomplished using the following machine learning algorithms: Logistic Regression (LR), Support Vector Machine (SVM), eXtreme Gradient Boosting (XGB) and Adaptive Boosting (AdaB). 10-fold cross validation was applied to verify the results of the models.

Results

The efficiency of each machine learning algorithm has been evaluated. XGB outperformed the others, achieving the highest accuracy (76.00%) and Area under (AUC) the Receiver Operating Characteristic Curve (ROC curve) value (0.776). The AUC value of the LR, SVM and AdaB are 0.761, 0.746 and 0.751, respectively. Figure 1 presents the ROC curve and AUC values for all models. Furthermore, the accuracy (ACC), precision, specificity, recall, F1-score and standard deviation were estimated for each classifier (Table 1).

Conclusion

We predicted all-cause mortality in patients with 20 years cardiovascular risk, using machine learning models and data of 3,316 patients. The most accurate predictive model was achieved by the eXtreme Gradient Boosting, with an accuracy equal to 76.00% and an AUC value of 0.776.

Funding Acknowledgement

Type of funding sources: Public grant(s) – EU funding. Main funding source(s): HORIZON2020

Coronary computed tomography angiography based endothelial wall shear stress in normal coronary arteries

Background

Endothelial wall shear stress (ESS) is a biomechanical force which plays a key role in the formation and evolution of atherosclerotic lesions.

Purpose

This study aims to evaluate ESS in coronary arteries without atherosclerosis, and to assess various factors affecting ESS values.

Methods

Coronary computed tomography angiography (CCTA) images from patients with suspected coronary artery disease were analyzed to identify coronary arteries without atherosclerosis. Minimal and maximal ESS values were calculated for 3-mm segments with dedicated software. Segments were categorized according to lumen diameter tertiles into small (<2.6 mm), intermediate (2.6–3.2 mm) or large (≥3.2 mm) size classes. Normal ranges of minimal and maximal ESS values were calculated per vessel and vessel size.

Results

A total of 349 normal vessels from 168 patients (mean age 59.4±9.2 years, 39% men) were included. ESS was highest in the left anterior descending artery compared to the left circumflex and the right coronary arteries (2.3 Pa vs. 1.9 Pa vs. 1.6 Pa for minimal ESS, p<0.001 and 3.7 Pa vs. 3.0 Pa vs. 2.5 Pa for maximal ESS, p<0.001). ESS values were highest in small vessel segments compared to intermediate or large segments (3.8 Pa vs. 1.7 Pa vs. 1.2 Pa for minimal ESS, p<0.001 and 6.0 Pa vs. 2.6 Pa vs. 2.0 Pa for maximal ESS, p<0.001). Weak to moderate correlation was found between ESS and the distance from the ostium (ρ values ranging from 0.22 to 0.62 for different vessels).

Conclusion

We derived ESS values from the CCTA images for visually normal coronary arteries. ESS values depend strongly on the lumen diameter of the coronary vessel. The normal ranges of minimal and maximal ESS can be used in future studies, where ESS values in stenotic lesions are compared to the normal values derived in the present analysis.

Funding Acknowledgement

Type of funding sources: Foundation. Main funding source(s): Finnish Foundation for Cardiovascular Research

Brain natriuretic peptide increases acutely and much more prominently than atrial natriuretic peptide during coronary angioplasty

BACKGROUND

Increased levels of plasma brain natriuretic peptide (BNP) are observed in patients with congestive heart failure, hypertension, left ventricular hypertrophy, and acute myocardial infarction. However, there are no data on serial changes in plasma levels of BNP in patients undergoing coronary angioplasty.

HYPOTHESIS

The study was undertaken to examine plasma concentrations of BNP together with those of atrial natriuretic peptide (ANP) in patients undergoing percutaneous transluminal coronary angioplasty (PTCA).

METHODS

Plasma concentrations of BNP and ANP were examined in 13 patients with stable angina pectoris and good left ventricular function undergoing PTCA. Blood samples were taken from the femoral vein at baseline, after the first balloon inflation, after the end of the procedure, and 4 h thereafter.

RESULTS

Plasma BNP levels were 14 +/- 4 at baseline, 22 +/- 10 after the first balloon inflation, 28 +/- 12 at the end of the procedure, and 15 +/- 4 pgr/ml 4 h thereafter (F = 13.05, p < 0.00001). Plasma ANP levels were 80 +/- 15, 86 +/- 14, 90 +/- 24, and 75 +/- 6 fmol/l (F = 5.95, p = 0.002), respectively. The increase of BNP at the end of the procedure was related to the increase of ANP (r = 0.78, p = 0.002).

CONCLUSION

Plasma BNP levels increase acutely and much more prominently than those of plasma ANP during coronary angioplasty; however, plasma BNP levels return to baseline values shortly after the end of the procedure.

Patient and personnel dosimetry in endovascular radiotherapy with 90Sr/90Y sources

Endovascular brachytherapy (EVBT) is an established treatment to reduce the probability of restenosis after a percutaneous coronary intervention. The purpose of this study was to assess (1) the manufacturer's stated dosimetric data for (90)Sr/(90)Y source trains to be used in EVBT and (2) the procedure-related radiation burden. The radiation fields in water around six (90)Sr/(90)Y source trains were studied using phantoms made of 'solid water' and MD55-2 radiochromic films. The water equivalence of the phantom material was tested by applying quantitative computed tomography. Thermoluminescence dosemeters were used to assess personal radiation burden and crosscheck the dose distribution along the source trains. Technical failure was observed in one source train and this train was excluded from analysis. The measured dose rate in water at 2 mm radial distance was on average 8% higher than the manufacture's stated value (range of measured to stated values 1.05--1.15). The dose rate decreased exponentially with radial distance between 2 and 6 mm. The dose rate in contact with the source viewing window of the delivery devices ranged between 0.5 and 7.5 mGy h(-1). Low-energy photons were the main contributors to personal dose.

Effects of delayed (24 h postintervention) β-irradiation therapy after coronary angioplasty and stenting in de novo native coronary artery lesions

BACKGROUND

The effects of brachytherapy performed 24 h postintervention in de novo native coronary artery lesions.

METHODS AND RESULTS

Thirty-nine patients with 39 de novo coronary artery lesions were randomised to prompt (immediately after intervention, n=21) or delayed (24 h later, n=18) beta brachytherapy ((90)Sr/(90)Y) after been successfully revascularized with stenting. Clinical follow-up data up to 21 months (median time) were compared. After irradiation and at 6-month follow-up, paired volumetric analysis of the stented segment and the 5-mm proximal and distal reference segments was performed; this included measurements of the external elastic membrane, lumen, plaque, and media (external elastic membrane minus lumen), stent and intima hyperplasia (stent minus lumen). Baseline clinical and angiographic characteristics were similar in the two groups. Postintervention measurements of stent, lumen, and intima hyperplasia volumes as well as postintervention minimum lumen cross-sectional areas were not different. In the stented segments and in the segments 5 mm proximal and distal to the stent, similar changes of all IVUS measurements were measured in the two patient groups, but in the lumen volume at the proximal stent edge of patients irradiated 24 h postinjury. At 21 months median follow-up time, target lesion revascularization (TLR) was performed in 8 patients (38%) in the prompt irradiation group compared with 3 (17%) in the delayed (P=.17).

CONCLUSION

Beta irradiation is similarly effective whether performed immediately after percutaneous coronary intervention or 24 h later.

Platelet-associated and secreted PAF-acetylhydrolase activity in patients with stable angina: sequential changes of the enzyme activity after angioplasty

BACKGROUND

Platelet-activating factor (PAF), the lipid mediator of inflammation and potent platelet agonist, can be hydrolysed and inactivated by PAF-acetylhydrolase (PAF-AH). We investigated the PAF-AH activity in relation to PAF formation in platelets from patients with stable angina undergoing elective percutaneous transluminal coronary angioplasty (PTCA).

DESIGN

Twenty-seven patients with stable angina, undergoing PTCA, and 30 age- and sex-matched controls were studied. The platelet-associated and secreted PAF-AH activity was measured, before PTCA, as well as at 4 h, 48 h and 6 months afterwards. PAF formation by thrombin-stimulated platelets and the platelet aggregation responses to PAF and ADP were also determined.

RESULTS

The PAF-AH activity secreted by thrombin-stimulated platelets before PTCA (in pmol/10(9) cells/h) was significantly higher compared to controls (892 +/- 222 vs. 624 +/- 144, P < 0.001). The enzyme activity was not altered at 4 h after PTCA, but was significantly increased at 48 h (1284 +/- 312, P < 0.005) to return to the levels observed in the control group 6 months afterwards. Detectable levels of PAF in thrombin-stimulated platelets were found only at 6 months after PTCA. Furthermore, the cell-associated enzyme activity in resting platelets before PTCA was significantly lower compared with controls. Unlike in controls, the platelet-associated enzyme activity in the patient group was not increased after stimulation with thrombin and it was associated by a platelet hyperaggregability to PAF. Both the intact cell-associated activity and the platelet hyper-reactivity to PAF were restored at 6 months after PTCA.

CONCLUSIONS

Alterations in the platelet PAF-AH activity, which affect the PAF formation in thrombin-stimulated platelets and are associated by an increased aggregatory response to PAF, are observed in patients with stable angina and are completely restored after PTCA.

Platelet aggregatory response to platelet activating factor (PAF), ex vivo, and PAF-acetylhydrolase activity in patients with unstable angina: effect of c7E3 Fab (abciximab) therapy

OBJECTIVE

Platelet activation and aggregation is a dominant feature in the pathophysiology of unstable angina. The final step of platelet aggregation is mediated through the platelet integrin glycoprotein IIb/IIIa (GP IIb/IIIa), while abciximab (ReoPro) is one of the most potent inhibitors of this receptor. Platelet-activating factor (PAF) is a potent platelet agonist which is degraded and inactivated by PAF-acetylhydrolase (PAF-AH). The plasma form of PAF-AH is associated with lipoproteins. We studied the platelet response to the aggregatory effect of PAF, ex vivo, in relation to the plasma PAF-AH activity in 32 patients with unstable angina, as well as the effect of abciximab therapy on the above parameters.

METHODS

Thirty two patients with unstable angina and 25 sex- and age-matched healthy controls participated in the study. On the day of admission (day 1) 17 patients received a bolus of abciximab (0.25 mg/kg) followed by a 12-h infusion (10 micrograms/min). Platelet aggregation to both PAF and ADP, in platelet rich plasma, was successively studied in both patients receiving abciximab or remaining untreated. The plasma and HDL-associated PAF-AH activity was also determined at the same times.

RESULTS

In the untreated patients, the PAF EC50 values were significantly lower on the day of admission, whereas the maximal percentage of aggregation was significantly higher compared to controls (p < 0.01 for both comparisons). Similar behaviour of the platelets was observed in the aggregatory effect of ADP. This aggregatory response was not significantly altered 4 days, 7 days or 1 month afterwards. In the 17 patients who received abciximab, platelet aggregation to both PAF and ADP was inhibited by 90 +/- 5 and 96 +/- 3%, respectively, 1 h after bolus. At 2 and 3 days after treatment, platelet aggregation to both agonists was significantly recovered being similar to controls. However, it was fully restored 6 days after bolus, still being significantly higher compared to controls (p < 0.01 for PAF and p < 0.003 for ADP). The total plasma PAF-AH activity in both patient groups was not different from that of controls, whereas the HDL-associated PAF-AH activity was significantly lower. The total plasma or HDL-associated enzyme activity was not altered at any time interval studied, and it was not influenced by abciximab.

CONCLUSIONS

The increased aggregatory response of platelets to PAF and the low plasma levels of HDL-cholesterol and HDL-associated PAF-AH activity in patients with unstable angina may contribute to the severe atherosclerosis and to acute thrombosis found in these patients. Abciximab therapy may protect platelets from PAF action in vivo the first days after drug administration, but it fails to permanently restore the enhanced aggregatory response observed.

Effect of rotational coronary atherectomy on peripheral endothelin-1, atrial natriuretic peptide, and cyclic adenosine monophosphate plasma levels

We investigated the effects of coronary rotational atherectomy (PTCRA) on plasma levels of endothelin-1 (ET-1), atrial natriuretic peptide (ANP), and cyclic adenosine monophosphate (cAMP). We studied 14 patients undergoing PTCRA and compared them with 14 patients undergoing plain balloon angioplasty. Blood samples were taken from the femoral vein at baseline, after the end of the atherectomy, after the first balloon inflation, after the end of the procedure, and 4 hours later. ET-1 increased in the angioplasty group from 6.3 +/- 3.2 pmol/L at baseline to 8.5 +/- 3.9 pmol/L at the end of the procedure (F = 3.83, P = .02), whereas it did not change in the PTCRA group. ANP increased in the PTCRA group from 78.1 +/- 15.7 pmol/L at baseline to 89.7 +/- 24.0 pmol/L at the end of the procedure (F = 6.75, P = .0001), whereas it did not change in the angioplasty group. cAMP decreased in the PTCRA group, whereas it did not change in the angioplasty group. In conclusion, ET-1 increases less, ANP increases more, and cAMP decreases more during atherectomy than during plain balloon angioplasty.

Influence of platelet size before coronary angioplasty on subsequent restenosis

Platelet size has been shown to reflect platelet activity. We prospectively measured the mean platelet volume (MPV) in 47 patients undergoing single vessel angioplasty for symptomatic angina. The patients underwent repeat angiography 4-8 months later irrespective of symptomatic status. Restenosis was assessed quantitatively by hand held calliper measurements of the lesion and qualitatively by a return of angina, ST segment changes on an exercise test and visual analysis of the lesion severity by two experienced angiographers. Twenty-four patients developed recurrent angina during the follow-up period, the MPV in the group with chest pain was 8.54 +/- 0.60 fl compared to 8.1 +/- 0.69 fl in the asymptomatic group (P = 0.04). Twenty two patients had significant ST segment changes at exercise. In this group the MPV was 8.6 +/- 0.56 fl compared to 8.0 +/- 0.70 fl for the group with a negative test (P = 0.002). Similarly visually assessed angiographic stenosis showed a significant increase in the restenotic group (8.6 +/- 0.56 vs. 8.0 +/- 0.61 fl, P = 0.001). The relative odds for developing clinically defined restenosis were 10.2 times greater if the pre-procedural MPV lay in the upper compared to the lowest quartile. There was a positive correlation between MPV and change in minimal luminal diameter between post angioplasty and follow-up angiography, assessed quantitatively, r = +0.56, P = 0.016. There was no association between clinical or angiographic definitions of restenosis and haemoglobin, red cell count, mean corpuscular volume, white cell count or platelet count. Platelet size may influence the development of restenosis after successful coronary angioplasty.

An extravascular hydraulic system to control blood pressure by a feedback regulation of the venous return

Since increases in blood pressure may be effectively controlled by hydraulic feedback by reducing the venous return through the inferior vena cava in proportion to the blood pressure this principle was applied using a totally implantable extravascular system consisting of a periaortic blood pressure sensor and a pericaval cuff around the inferior vena cava. The two cuffs were supported externally by hard skeletons, filled with water, and connected directly to each other. The two devices were tested separately and together in 10 anaesthetised mongrel dogs with normal and high blood pressure induced by a metaraminol infusion. With the periaortic sensor an increase in blood pressure of 100 mmHg caused a mean(SD) isotonic volume displacement of 0.41(0.11) ml, and an increase in the pericaval balloon energy content of 100 ml.mmHg-1 caused a fall in blood pressure of 37.8(18.3)%. The whole system prevented an excessive rise in blood pressure when metaraminol was infused, with a feedback gain of about 2.8. It is concluded that blood pressure can be maintained at acceptable levels despite strong hypertensive stimuli, by this system, without additional treatment with drugs or an external energy source. The system is totally implantable and is applied extravascularly so that no anticoagulation is needed.

Acute changes in blood pressure as a cause of cardiac arrhythmias

The effect of an acute change in blood pressure (BP) on ventricular ectopic activity and the influence of antiarrhythmic agents on this effect were examined in 24 patients. In 11 patients with premature ventricular complexes (PVCs), the BP was temporarily reduced by a sodium nitroprusside drip. In all of them the incidence of PVCs was reduced (or annihilated) by the induced hypotension. In 13 patients without ventricular ectopic activity, a metaraminol drip was given until either a PVC appeared or the systolic BP reached 200 mmHg, or symptoms appeared. In 12 cases at least one PVC appeared and in 8 of them the total number of PVCs was 13 or more, usually in the form of bigeminy. The repetition of the test following quinidine administration (serum quinidine level 1.7 +/- 0.5 ng ml-1) in 6 cases did not change this pattern, with one exception. It prevented the appearance of idioventricular accelerated rhythm in one case in whom this rhythm had been induced by the hypertension provocative test before the quinidine administration. All cases, in whom the test failed to induce more than 3 PVCs, had no cardiac problem at all. Six of the 8 cases in whom the test induced 13 or more PVCs had organic cardiac disease or palpitation. Other arrhythmias observed on BP elevation, were supraventricular extra beats, nodal escape rhythms and atrioventricular block. In one case with cardiomyopathy, the BP elevation was associated with early signs of heart failure that subsided quickly. In conclusion, acute elevation on BP may be associated with the generation of PVCs and its reduction with their reduction or disappearance.