Comparison of Machine Learning Models Including Preoperative, Intraoperative, and Postoperative Data and Mortality After Cardiac Surgery.

Importance: A variety of perioperative risk factors are associated with postoperative mortality risk. However, the relative contribution of routinely collected intraoperative clinical parameters to short-term and long-term mortality remains understudied. Objective: To examine the performance of multiple machine learning models with data from different perioperative periods to predict 30-day, 1-year, and 5-year mortality and investigate factors that contribute to these predictions. Design, Setting, and Participants: In this prognostic study using prospectively collected data, risk prediction models were developed for short-term and long-term mortality after cardiac surgery. Included participants were adult patients undergoing a first-time valve operation, coronary artery bypass grafting, or a combination of both between 1997 and 2017 in a single center, the University Medical Centre Groningen in the Netherlands. Mortality data were obtained in November 2017. Data analysis took place between February 2020 and August 2021. Exposure: Cardiac surgery. Main Outcomes and Measures: Postoperative mortality rates at 30 days, 1 year, and 5 years were the primary outcomes. The area under the receiver operating characteristic curve (AUROC) was used to assess discrimination. The contribution of all preoperative, intraoperative hemodynamic and temperature, and postoperative factors to mortality was investigated using Shapley additive explanations (SHAP) values. Results: Data from 9415 patients who underwent cardiac surgery (median [IQR] age, 68 [60-74] years; 2554 [27.1%] women) were included. Overall mortality rates at 30 days, 1 year, and 5 years were 268 patients (2.8%), 420 patients (4.5%), and 612 patients (6.5%), respectively. Models including preoperative, intraoperative, and postoperative data achieved AUROC values of 0.82 (95% CI, 0.78-0.86), 0.81 (95% CI, 0.77-0.85), and 0.80 (95% CI, 0.75-0.84) for 30-day, 1-year, and 5-year mortality, respectively. Models including only postoperative data performed similarly (30 days: 0.78 [95% CI, 0.73-0.82]; 1 year: 0.79 [95% CI, 0.74-0.83]; 5 years: 0.77 [95% CI, 0.73-0.82]). However, models based on all perioperative data provided less clinically usable predictions, with lower detection rates; for example, postoperative models identified a high-risk group with a 2.8-fold increase in risk for 5-year mortality (4.1 [95% CI, 3.3-5.1]) vs an increase of 11.3 (95% CI, 6.8-18.7) for the high-risk group identified by the full perioperative model. Postoperative markers associated with metabolic dysfunction and decreased kidney function were the main factors contributing to mortality risk. Conclusions and Relevance: This study found that the addition of continuous intraoperative hemodynamic and temperature data to postoperative data was not associated with improved machine learning-based identification of patients at increased risk of short-term and long-term mortality after cardiac operations.

Ensemble machine learning prediction and variable importance analysis of 5-year mortality after cardiac valve and CABG operations.

Despite having a similar post-operative complication profile, cardiac valve operations are associated with a higher mortality rate compared to coronary artery bypass grafting (CABG) operations. For long-term mortality, few predictors are known. In this study, we applied an ensemble machine learning (ML) algorithm to 88 routinely collected peri-operative variables to predict 5-year mortality after different types of cardiac operations. The Super Learner algorithm was trained using prospectively collected peri-operative data from 8241 patients who underwent cardiac valve, CABG and combined operations. Model performance and calibration were determined for all models, and variable importance analysis was conducted for all peri-operative parameters. Results showed that the predictive accuracy was the highest for solitary mitral (0.846 [95% CI 0.812-0.880]) and solitary aortic (0.838 [0.813-0.864]) valve operations, confirming that ensemble ML using routine data collected perioperatively can predict 5-year mortality after cardiac operations with high accuracy. Additionally, post-operative urea was identified as a novel and strong predictor of mortality for several types of operation, having a seemingly additive effect to better known risk factors such as age and postoperative creatinine.

Acute Kidney Injury Classification Underestimates Long-Term Mortality After Cardiac Valve Operations.

BACKGROUND: Perioperative acute kidney injury (AKI) is an important predictor of long-term all-cause mortality after coronary artery bypass (CABG). However, the effect of AKI on long-term mortality after cardiac valve operations is hitherto undocumented. METHODS: Perioperative renal injury and long-term all-cause mortality after valve operations were studied in a prospective cohort of patients undergoing solitary valve operations (n = 2,806) or valve operations combined with CABG (n = 1,260) with up to 18 years of follow-up. Postoperative serum creatinine increase was classified according to AKI staging 0 to 3. Patients undergoing solitary CABG (n = 4,938) with cardiopulmonary bypass served as reference. RESULTS: In both valve and valve+CABG operations, postoperative renal injury of AKI stage 1 or higher was progressively associated with an increase in long-term mortality (hazard ratio [HR], 2.27, p < 0.05 for valve; HR, 1.65, p < 0.05 for valve+CABG; HR, 1.56, p < 0.05 for CABG). Notably, the mortality risk increased already substantially at serum creatinine increases of 10% to 25%-that is, far below the threshold for AKI stage 1 after valve operations (HR, 1.39, p < 0.05), but not after valve operations combined with CABG or CABG only. CONCLUSIONS: An increase in serum creatinine by more than 10% during the first week after valve operation is associated with an increased risk for long-term mortality after cardiac valve operation. Thus, AKI classification clearly underestimates long-term mortality risk in patients undergoing valve operations.

Dopamine treatment attenuates acute kidney injury in a rat model of deep hypothermia and rewarming - The role of renal H2S-producing enzymes.

Hypothermia and rewarming produces organ injury through the production of reactive oxygen species. We previously found that dopamine prevents hypothermia and rewarming-induced apoptosis in cultured cells through increased expression of the H2S-producing enzyme cystathionine ß-Synthase (CBS). Here, we investigate whether dopamine protects the kidney in deep body cooling and explore the role of H2S-producing enzymes in an in vivo rat model of deep hypothermia and rewarming. In anesthetized Wistar rats, body temperature was decreased to 15°C for 3h, followed by rewarming for 1h. Rats (n≥5 per group) were treated throughout the procedure with vehicle or dopamine infusion, and in the presence or absence of a non-specific inhibitor of H2S-producing enzymes, amino-oxyacetic acid (AOAA). Kidney damage and renal expression of three H2S-producing enzymes (CBS, CSE and 3-MST) was quantified and serum H2S level measured. Hypothermia and rewarming induced renal damage, evidenced by increased serum creatinine, renal reactive oxygen species production, KIM-1 expression and influx of immune cells, which was accompanied by substantially lowered renal expression of CBS, CSE, and 3-MST and lowered serum H2S levels. Infusion of dopamine fully attenuated renal damage and maintained expression of H2S-producing enzymes, while normalizing serum H2S. AOAA further decreased the expression of H2S-producing enzymes and serum H2S level, and aggravated renal damage. Hence, dopamine preserves renal integrity during deep hypothermia and rewarming likely by maintaining the expression of renal H2S-producing enzymes and serum H2S.

S1P1 receptor modulation preserves vascular function in mesenteric and coronary arteries after CPB in the rat independent of depletion of lymphocytes.

BACKGROUND: Cardiopulmonary bypass (CPB) may induce systemic inflammation and vascular dysfunction. Sphingosine 1-phosphate (S1P) modulates various vascular and immune responses. Here we explored whether agonists of the S1P receptors, FTY720 and SEW2871 improve vascular reactivity after CPB in the rat. METHODS: Experiments were done in male Wistar rats (total n = 127). Anesthesia was induced by isoflurane (2.5-3%) and maintained by fentanyl and midazolam during CPB. After catheterization of the left femoral artery, carotid artery and the right atrium, normothermic extracorporeal circulation was instituted for 60 minutes. In the first part of the study animals were euthanized after either 1 hour, 1 day, 2 or 5 days of the recovery period. In second part of the study animals were euthanized after 1 day of postoperative period. We evaluated the contractile response to phenylephrine (mesenteric arteries) or to serotonin (coronary artery) and vasodilatory response to acethylcholine (both arteries). RESULTS: Contractile responses to phenylephrine were reduced at 1 day recovery after CPB and Sham as compared to healthy control animals (Emax, mN: 7.9 ± 1.9, 6.5 ± 1.5, and 11.3 ± 1.3, respectively). Mainly FTY720, but not SEW2871, caused lymphopenia in both Sham and CPB groups. In coronary and mesenteric arteries, both FTY720 and SEW2871 normalized serotonin and phenylephrine-mediated vascular reactivity after CPB (p<0.05) and FTY720 increased relaxation to acetylcholine as compared with untreated rats that underwent CPB. CONCLUSION: Pretreatment with FTY720 or SEW2871 preserves vascular function in mesenteric and coronary artery after CPB. Therefore, pharmacological activation of S1P1 receptors may provide a promising therapeutic intervention to prevent CPB-related vascular dysfunction in patients.

Microarray analysis of gene expression profiles in the rat kidney demonstrates a local inflammatory response induced by cardiopulmonary bypass.

CONTEXT: Cardiopulmonary bypass (CPB) is a commonly used technique in cardiac surgery but is associated with acute, transient, renal dysfunction that has a negative impact on long-term survival. OBJECTIVE: To unravel the molecular pathogenesis of renal injury following CPB. DESIGN: To obtain insight into the pathogenesis of renal dysfunction following CPB, we performed a microarray analysis of renal gene expression in the rat. SETTING: University Medical Centre Groningen. INTERVENTION: Rats underwent CPB or a sham procedure for 60 min and were sacrificed at 60 min, 1 and 5 days after the procedure. MAIN OUTCOME MEASURES: Renal gene expression profile as determined by microarray analysis. RESULTS: Expression of 420 genes was significantly altered in CPB compared to the sham procedure, and in 407 genes, this was evident in the acute phase (60 min) following CPB. Gene ontology analysis revealed 28 of these genes were involved in inflammatory responses, with high expression of genes downstream of mitogen-activated protein-kinase (MAP-kinase) signalling pathways. Potent inducers identified are from the interleukin-6 cytokine family that consists of interleukin-6 and oncostatin M (OSM), which signal through the gp130-cytokine receptor complex. The plasma concentration of interleukin-6 was hugely increased by CPB as measured by ELISA. Expression of genes downstream of these signalling pathways that lead to production of chemokines, adhesion molecules and molecules involved in coagulative pathways, was upregulated. CONCLUSION: CPB induces an acute and local inflammatory response in the kidney, which might contribute to renal injury. The signalling pathways involved identified by gene expression analysis may represent pharmacological targets to limit renal injury following CPB.

Troubleshooting the rat model of cardiopulmonary bypass: effects of avoiding blood transfusion on long-term survival, inflammation and organ damage.

INTRODUCTION: Rat models of cardiopulmonary bypass (CPB) have been used to examine the mechanisms of associated organ damage and to test intervention strategies. However, these models only partly mimic the clinical situation, because of the use of blood transfusion and arterial inflow via the tail artery. Thus a model using arterial inflow in the aorta and a miniaturized CPB circuit without need of transfusion was validated by examining intra-procedure characteristics, mortality and the effects of CPB on biomarkers of inflammation and cerebral injury during 5days follow-up. METHODS: Male Wistar rats (n=95) were anesthetized with isoflurane (2.5%) and fentanyl/midazolam during CPB. Animals were assigned to Control (n=6), Sham (n=40) or normothermic CPB (n=49) groups. Both Sham and CPB were cannulated in the aorta via the left carotid artery and in the right common jugular vein for access into the right heart. Extracorporeal circulation (ECC) was instituted for 60min only in CPB at a flow rate of 120mLkg(-1)min(-1) employing a CPB circuit of 15ml primed with 6% hydroxyethyl starch 60mgml(-1) solution. Rats were sacrificed at either 1h or 1, 2 or 5days after Sham or weaning from CPB. Plasma IL-6 and s100Beta levels were measured and blood cell counts were performed. RESULTS: Mortality in CPB animals (3 out of 49) and Sham (4 out of 40) did not differ (chi-square=0.46, dF=1, P>0.5). Compared to baseline (1.87±0.46∗10^9cells/L), Sham procedure (cannulation and anesthesia) significantly increased blood neutrophil count at the end of the period matching ECC (6.34±2.36∗10^9cells/L, P<0.05). CPB induced neutrophilia which persisted during 24h recovery. Also, CPB caused a rapid and prominent increase in plasma IL-6 from the first hour of the postoperative period (~1200pg/ml) with continuation (50-90pg/ml) up to 5th day of recovery. S100Beta levels were above detection level only in 3 out of 42 samples from CPB animals. DISCUSSION: Our rat model of CPB without homologous blood transfusion produces a reproducible and reliable systemic inflammatory response, with low mortality rates on long term follow up. The model more closely mimics the human situation in respect to arterial inflow site and avoidance of blood transfusion. Thus, our CPB model is suitable to study its influence on systemic inflammation, ischemia-reperfusion injury, microcirculation and vascular dysfunction in vivo, and to evaluate potential therapeutic interventions.

Clinical effectiveness of centrifugal pump to produce pulsatile flow during cardiopulmonary bypass in patients undergoing cardiac surgery.

Although the centrifugal pump has been widely used as a nonpulsatile pump for cardiopulmonary bypass (CPB), little is known about its performance as a pulsatile pump for CPB, especially on its efficacy in producing hemodynamic energy and its clinical effectiveness. We performed a study to evaluate whether the Rotaflow centrifugal pump produces effective pulsatile flow during CPB and whether the pulsatile flow in this setting is clinically effective in adult patients undergoing cardiac surgery. Thirty-two patients undergoing CPB for elective coronary artery bypass grafting were randomly allocated to a pulsatile perfusion group (n = 16) or a nonpulsatile perfusion group (n = 16). All patients were perfused with the Rotaflow centrifugal pump. In the pulsatile group, the centrifugal pump was adjusted to the pulsatile mode (60 cycles/min) during aortic cross-clamping, whereas in the nonpulsatile group, the pump was kept in its nonpulsatile mode during the same period of time. Compared with the nonpulsatile group, the pulsatile group had a higher pulse pressure (P < 0.01) and a fraction higher energy equivalent pressure (EEP, P = 0.058). The net gain of pulsatile flow, represented by the surplus hemodynamic energy (SHE), was found much higher in the CPB circuit than in patients (P < 0.01). Clinically, there was no difference between the pulsatile and nonpulsatile groups with regard to postoperative acute kidney injury, endothelial activation, or inflammatory response. Postoperative organ function and the duration of hospital stay were similar in the two patient groups. In conclusion, pulsatile CPB with the Rotaflow centrifugal pump is associated with a small gain of EEP and SHE, which does not seem to be clinically effective in adult cardiac surgical patients.

Dexamethasone: benefit and prejudice for patients undergoing on-pump coronary artery bypass grafting: a study on myocardial, pulmonary, renal, intestinal, and hepatic injury.

STUDY OBJECTIVES: Cardiac surgery with cardiopulmonary bypass (CPB) results in perioperative organ damage caused by the systemic inflammatory response syndrome (SIRS) and ischemia/reperfusion injury. Administration of corticosteroids before CPB has been demonstrated to inhibit the activation of the systemic inflammatory response. However, the clinical benefits of corticosteroid therapy are controversial. This study was designed to document the effects of dexamethasone on cytokine release and perioperative myocardial, pulmonary, renal, intestinal, and hepatic damage, as assessed by specific and sensitive biomarkers. DESIGN AND PATIENTS: A prospective, double-blind, placebo-controlled, randomized trial for dexamethasone was conducted in 20 patients receiving either dexamethasone (1 mg/kg before anesthesia induction and 0.5 mg/kg after 8 h; n = 10) or placebo (n = 10). Different markers were used to assess the SIRS: interleukin (IL)-6, IL-8, IL-10, C-reactive protein (CRP), and tryptase; and organ damage: heart (plasma heart-type fatty acid binding protein, cardiac troponin I [cTnI], creatine kinase-MB), kidneys (N-acetyl-glucosaminidase [NAG], microalbuminuria), intestine (intestinal-type fatty acid binding protein [I-FABP]/liver-type fatty acid binding protein [L-FABP]), and liver (alpha-glutathione S-transferase). RESULTS: Dexamethasone modulated the SIRS with lower proinflammatory (IL-6, IL-8) and higher antiinflammatory (IL-10) IL levels. CRP and tryptase were lower in the dexamethasone group. cTnI values were lower in the dexamethasone group at 6 h in the ICU (p = 0.009). Patients in the dexamethasone group had a longer time to tracheal extubation (18.86 +/- 1.13 h vs 15.01 +/- 0.99 h, p = 0.02 [mean +/- SEM]), with a lower oxygenation index at that time: Pa(O2)/fraction of inspired oxygen ratio, 37.17 +/- 1.8 kPa vs 29.95 +/- 2.1 kPa (p = 0.009). The postoperative glucose level (10.7 +/- 0.6 mmol/L vs 7.4 +/- 0.5 mmol/L, p = 0.005) was higher in the dexamethasone group. Serum glucose was independently associated with intestinal injury (urine I-FABP peak, R2 = 42.5%, beta = 114.4 +/- 31.4, significant at p = 0.002; urine L-FABP peak, R2 = 47.3%, beta = 7,714.1 +/- 1,920.9, significant at p = 0.001) and renal injury (urine NAG, R2 = 32.1%, beta = 0.21 +/- 0.07, significant at p = 0.009). Tryptase peaks correlated negatively with peaks of intestinal and renal injury biomarkers. CONCLUSION: Even while inhibiting SIRS, dexamethasone treatment offered no protection against transient, subclinical, perioperative abdominal organ damage. Tryptase release could have a preconditioning effect, offering protection against perioperative intestinal and renal damage. Dexamethasone treatment resulted in more pronounced postoperative pulmonary dysfunction, prolonged time to tracheal extubation, and initiated postoperative hyperglycemia in patients undergoing elective on-pump coronary artery bypass graft surgery.

Immediate postoperative renal function deterioration in cardiac surgical patients predicts in-hospital mortality and long-term survival.

Postoperative renal function deterioration is a serious complication after cardiac surgery with cardiopulmonary bypass and is associated with increased in-hospital mortality. However, the long-term prognosis of patients with postoperative renal deterioration is not fully determined yet. Therefore, both in-hospital mortality and long-term survival were studied in patients with postoperative renal function deterioration. Included were 843 patients who underwent cardiac surgery with cardiopulmonary bypass in 1991. Postoperative renal function deterioration (increase in serum creatinine in the first postoperative week of at least 25%) occurred in 145 (17.2%) patients. In these patients, in-hospital mortality was 14.5%, versus 1.1% in patients without renal function deterioration (P < 0.001). Multivariate analysis significantly associated in-hospital mortality with postoperative renal function deterioration, re-exploration, postoperative cerebral stroke, duration of operation, age, and diabetes. In patients who were discharged alive, during long-term follow-up (100 mo), mortality was significantly increased in the patients with renal function deterioration (n = 124) as compared with those without renal function deterioration (hazard ratio 1.83; 95% confidence interval 1.38 to 3.20). Also after adjustment for other independently associated factors, the risk for mortality in patients with postoperative renal function deterioration remained elevated (hazard ratio 1.63; 95% confidence interval 1.15 to 2.32). The elevated risk for long-term mortality was independent of whether renal function had recovered at discharge from hospital. It is concluded that postoperative renal function deterioration in cardiac surgical patients not only results in increased in-hospital mortality but also adversely affects long-term survival.

Off-pump coronary revascularization attenuates transient renal damage compared with on-pump coronary revascularization.

STUDY OBJECTIVES: Cardiopulmonary bypass (CPB) represents a specific risk factor for renal damage during coronary revascularization. The purpose of this study was to compare the perioperative renal damage in patients undergoing on-pump and off-pump coronary surgery. DESIGN AND PATIENTS: The progress and extent of renal damage was prospectively studied in two groups of patients undergoing cardiac surgery without concomitant morbidity, undergoing elective coronary revascularization with (n = 12) and without (n = 10) CPB. Markers of glomerular function (creatinine clearance) and damage (microalbuminuria), and markers of tubular function (fractional excretion of sodium [FENa] and free water clearance) and damage (N-acetyl-beta-D glucosaminidase [NAG]) were evaluated. Measuring plasma concentrations of free hemoglobin assessed hemolysis. Plasma and urinary specimens were obtained at the following points: (1) baseline; (2) heparinization; (3) the end of CPB or completing graft for off-pump surgery; (4) skin closure; (5) the sixth hour in the ICU; and (6) the second postoperative day. Free water and creatinine clearances, FENa, and the urinary excretion of microalbumin and NAG were calculated for the corresponding time intervals. SETTING: University hospital. RESULTS: We found that off-pump coronary revascularization induced significantly less changes in microalbuminuria, FENa, free water clearance, NAG, and free hemoglobin as compared with operations with CPB. Markers returned to baseline within 2 days after the operation, and there was no clinical or laboratory evidence of overt renal dysfunction in both groups. CONCLUSION: Off-pump coronary surgery attenuates transient renal injury compared with traditional on-pump coronary artery bypass grafting.