Study design and participants
This prospective observational study was approved by the Research Ethics Committee of The Second Hospital of Hebei Medical University (No.2022-R707), registered in the Chinese Clinical Trial Registry (No. ChiCTR2200063774), and conducted at The Second Hospital of Hebei Medical University from September 2022 to June 2023. Written informed consent for study participation was obtained from all patients or their legal representatives.
After obtaining written informed consent, patients aged ≥ 65 years, with an American Society of Anesthesiologists (ASA) physical status of Ⅲ–Ⅳ, and scheduled for OPCABG were recruited from October 2022 to June 2023 (Fig. 1 for STROBE diagram). The exclusion criteria were any of the following: (1) education < 6 years; (2) Montreal Cognitive Assessment Scale-Basic (MoCA-B) < 18; (3) pre-existing psychological and/or mental illness; (4) Parkinson's disease or Epileptic disease; (5) history of previous brain surgery; (6) taking sedatives or antidepressants in the last year; (7) severe hepatic insufficiency, acute kidney injury or renal insufficiency (eGFR < 30 ml/min/1.73m2), aortic arch atherosclerosis thicker than 4 mm; (8) alcoholism or drug abuse; (9) audition, vision or language troubles impeding communication; (10) situations unsuitable for an MRI scan (claustrophobia); (11) contraindications to gadolinium contrast agents. The eliminate criteria were: (1) emergency intraoperative extracorporeal circulation; (2) secondary postoperative thoracotomy; or (3) postoperative stroke.
Anesthesia and perioperative management
All patients underwent standardized anesthetic management. All preoperative cardiac medications were continued until the morning of surgery. In the operating room, all patients were monitored using electrocardiography, pulse oximetry, end-tidal carbon dioxide, bispectral index (BIS), body temperature, invasive blood pressure, and central venous pressure and received general anesthesia, with or without a parasternal nerve block at the discretion of the attending anesthesiologist. Anesthesia was induced with midazolam (0.03–0.05 mg/kg), etomidate (0.2–0.3 mg/kg), sufentanil (0.4–0.6 µg/kg) and rocuronium (0.9 mg/kg). Anesthesia was maintained with continuous intravenous infusion of ciprofol (0.5-1 mg·kg-1·h-1) and sufentanil (0.5-1 µg·kg-1·h-1), inhaled of sevoflurane at a minimum end-tidal concentration of 0.5 to 1 minimal alveolar concentration (MAC), and an intravenous injection of rocuronium according to the procedure of the operation and the intraoperative conditions of the patients. The MAP was maintained at 65 mmHg (± 30% of the baseline value). BIS value was maintained at 40–60. End-tidal carbon dioxide partial pressure was maintained at 35–45 mmHg. The nasopharyngeal temperature was maintained at 36–37°C. The surgical incision performed a median sternotomy for gaining access to the heart. Heparin was given before surgical manipulation of the coronary arteries was started. Based on the condition of coronary artery stenosis, the number and location of the surgery was determined. The hemodynamic was maintained stability during surgery, and gave vasoactive drugs when necessary. After vascular anastomosis, protamine was given to counteract heparin. All patients received intraoperative salvage autologous blood transfusion. Red blood cell transfusions should be considered if the hematocrit level less than 30% during surgery.
After surgery, patients were transferred to the intensive care unit (ICU) where they received necessary sedation and analgesia until qualified for tracheal extubation. ciprofol (0.4-1mg·kg-1·h-1) was used for postoperative sedation. Postoperative pain was managed using patient-controlled intravenous of sufentanil (2µg/kg in 200 ml of 0.9% normal saline) at a background infusion rate of 2 ml/h with 1ml boluses available and a locking time interval of 15 minutes. The goal of analgesia was to maintain the numerical rating scale (NRS) score at 0–3 at rest.
Assessment of POD
The primary outcome was the occurrence of POD. Assessment of POD was performed twice daily (between 06:00–08:00, and 18:00– 20:00) for postoperative days 1 through 5 by trained POD assessors, using the 3-Minute Diagnostic Confusion Assessment Method (3D-CAM) in non-intubated patients[25] or the CAM for the Intensive Care Unit (CAM-ICU) in intubated patients[26]. The POD was defined and assessed based on four features: (1) acute onset of mental status changes or a fluctuating course, (2) inattention, (3) disorganized thinking, and (4) altered level of consciousness. POD was diagnosed based on the patient displayed both features 1 and 2, with either 3 or 4 during the assessment period. Applied tests correspond with the DSM-V and has been validated with acceptable sensitivity and specificity[25, 26]. Secondary outcome was the severity of POD. The severity of delirium was assessed using the long form of the Chinese version of CAM-Severity (CAM-S)[27], with scores ranging from 0 (no delirium features) to 19 (most severe). The researcher responsible for the delirium assessment participates in a training session, at the end of which a quiz is performed and all participants must answer all quiz questions correctly. Delirium assessors were blinded to data collected during surgery and MRI.
Imaging
All magnetic resonance images were acquired using an MRI scanner (GE SIGNA Architect, 3.0T) at the Second Hospital of Hebei Medical University. MRI was performed 1–5 days before surgery. During the MRI data acquisition, the participants had to remain relaxed and still. Main scanning sequence and parameters: 3D T1-BRAVO and 3D Osag T2-Flair CUBE images displayed anatomic brain structures, diffusion-weighted imaging (DWI) with apparent diffusion coefficient mapping, whereas DCE Whole-Brain sequence was used to measure BBB permeability. DCE scanning: axial, TR 4.4ms, TE 1.6ms, slice thickness 2.0 mm with 0.6mm gap, FOV 26.9cm×24cm. The Flip Angle was 12°. The scanning times for T1-BRAVO, T2-FLAIR, and DCE were 224, 367, and 264 seconds, respectively. At the beginning of the third dynamic scan, 0.1mmol/kg gadolinium contrast agent was injected through the elbow vein using a high-pressure syringe 20 ml at an injection rate of 4.5ml/s, and the scan lasted for 40 phases.
At the end of scanning, the images obtained by DCE were processed by GE AW4.7 workstation GEN IQ module (GE Medical Systems). The quantitative analysis was based on modified Tofts two-compartment pharmacokinetic model, with an important physiological parameter Ktrans[6, 16, 17, 22]. The software has a specialised template for the head, thereby facilitating convenient and accurate operations. Furthermore, it incorporates 3D motion correction technology, which prevents involuntary motion from affecting quantitative accuracy. Before calculating Krans, acquisition of the vascular inflow function was performed in Auto Mode to minimise errors due to flow near vessel boundaries and T1 correction due to the relationship between gadolinium concentration and signal intensity depending on baseline T1. In the process of calculating Ktrans from the DCE-MRI, the region of interest (ROI) was defined by utilizing the 3D approach and 3D Osag T2-Flair CUBE images. The hippocampus, thalamus, frontal lobe, amygdala, cingulate cortex, precuneus, temporal lobe and parietal lobe were analysed as ROIs. The software was programmed to automatically calculate the Ktrans value when the ROI was manually labelled. The Ktrans value was averaged from the each layer.
AW Volumeshare 7 software module for GE was used for segmentation and volume estimation, based on 3D Osag T2-Flair CUBE images. There are sagittal, horizontal, coronal, and reconstructed 3D images. Adjust the window width and level based on T1 images to achieve a significant contrast between gray and white matter. The boundaries of the hippocampus, thalamus, and whole brain were delineated according to established criteria from 3D maximum intensity projection. Reconstruct the coronal plane and manually outline the boundary of the ROI. The software will automatically define the edges between adjacent layers and calculate the absolute volume of the ROI. Then hippocampal and thalamus volume were standardized. Standardized volume were calculated as follows: (measured two sides hippocampal volume and thalamus volume /whole brain volume) ×1000[28]. MRI was evaluated and processed by two qualified radiologists who were blinded to the clinical and surgical variables.
Statistical analysis
No formal sample size analysis was performed, because no relevant references exist for the estimation of postoperative outcomes through preoperative BBB status. Accordingly, the sample size was based on two pieces of evidence: first, a study by Chagnot et al.[29] reporting that the Ktrans range of low permeability of the BBB is between 10− 4 and 10-3min-1; and second, combining Nation et al.[6] research and preliminary experiments, the BBB in the hippocampus of delirium patients may be within 5×10-3min-1. In our study, the sample size calculation was based on the Ktrans of patients who have not experienced delirium is approximately half of those who have POD. It is expected expected 40% incidence of POD after cardiac surgery[30, 31]. Considering a significance level of 0.05, SD 3, and a power of 0.8 (β = 0.2), a sample size of fifty patients was required. To compensate for missing data and dropouts, the sample size was increased to a total of 60 patients.
Statistical calculations were conducted using SPSS 27.0 software program. Patients were categorized into two groups: POD and Non-POD (NPOD), based on the delirium assessment results. Continuous variables were presented as means (standard deviation), if normally distributed, and median (quartile 1– quartile 3) if not. Group comparisons were performed using the independent sample t-test for normally distributed variables and the Mann-Whitney test for non-normally distributed variables. Categorical data were presented as frequencies and percentages, and analyzed using 2-tailed χ2 tests or the Fisher exact test. Univariable logistic regression analysis was used to evaluate the relationship between hippocampus, thalamus, frontal lobe, amygdala, cingulate cortex, precuneus, temporal lobe and parietal lobe Ktrans values and Brain, Hippocampus, Thalamus volume and POD. Multivariable analysis was conducted by including Hippocampus, Thalamus Ktrans values and controlling for 2 well-established POD risk factors. Mediation analysis was performed to evaluate whether hippocampus Ktrans values mediates the association between MoCA-B and POD. We also examined whether the results differed by interleukin-6(IL-6) concentrations, using the P-value of the interaction term in regression models to assess significance. Correlation analyses were conducted using Graphpad Prism 9.5 software program, Bivariate correlations between POD severity and Ktrans values and volumes of the hippocampus and thalamus using the nonparametric Spearman correlation coefficient. All statistical tests were two-sided, and a P value less than 0.05 was considered statistically significant.