As a result of continuing improvements in surgical techniques and cardiovascular anesthesia, an increasing number of elderly patients with a similarly increasing number of comorbidities are able to safely undergo cardiac surgery. Although mortality after cardiac surgery has decreased, neurologic complications remain an important and widespread issue in an ever-aging surgical population. Delirium, reflecting acute changes in cognition and attention, is common after cardiac surgery, occurring in 9-52% of patients.1,2 In particular, recent studies have revealed that postoperative delirium may have more than transient effects, and can negatively affect long-term morbidity and mortality.1,3-5 Therefore, understanding the risks for postoperative delirium is an important consideration for surgeons and anesthesiologists.

Cardiopulmonary bypass (CPB) may result in neurological injury caused by emboli, systemic inflammation, and unpredictable intraprocedural hypoperfusion.6 In an attempt to avoid these adverse CPB sequelae, off-pump coronary artery bypass grafting (OPCAB) was introduced during the early 1990s. This technique was expected to reduce postoperative neurological complications, but recent studies largely failed to show any significant benefit.7-9

Microembolization has been considered a significant contributor to postoperative delirium.10,11 However, relatively few studies have investigated the relation between the brain lesions that would be expected to result from a significant embolic load and postoperative delirium. Indeed, studies of cerebral emboli in conventional on-pump coronary artery bypass surgery (CABG) using transcranial Doppler (TCD) have failed to show any significant association with postoperative delirium.12 In regard to OPCAB, however, there are a few reports investigating the relation between cerebral embolia and postoperative delirium. Importantly, not all of the cerebral emboli detected by TCD would be expected to induce a clinically apparent ischemic lesion. Therefore, magnetic resonance diffusion-weighted imaging (DWI) might provide more information about the more subtle ischemic lesions caused by cerebral emboli.

With respect to carotid artery lesions, we previously reported that carotid artery stenosis was significantly associated with postoperative stroke and delirium.13 Embolic signals measured using TCD were even observed in patients with asymptomatic carotid artery stenosis, resulting in silent cerebral infarction.14,15 Significant atherosclerosis, manifesting as carotid stenosis or ascending aortic plaque, was also associated with delirium after CABG.16 Therefore, evaluation of carotid artery stenosis and ascending aortic plaque to identify its relation to both postoperative delirium and new magnetic resonance imaging (MRI)-detectable ischemic lesions may provide important information on the pathophysiology of these neurological abnormalities.

The principal aim of the present study was to investigate the relation between postoperative delirium and new ischemic lesions detected by MRI-DWI after OPCAB. The impact of magnetic resonance angiographic (MRA) detectable intracranial artery stenoses that could result in cerebral hypoperfusion was also explored. Finally, as recent studies have reported that white matter hyperintensities may predict postoperative delirium with conventional on-pump CABG,17-19 we assessed whether these white matter lesions were associated with postoperative delirium after OPCAB.

Methods

Patients

The institutional review board of our hospital approved this study (September 2008). All participants provided written informed consent prior to enrolment. From February 2009 to July 2011, consecutively consenting patients undergoing elective OPCABG who were more than 20 yr old were enrolled. Exclusion criteria included patients with contraindications to MRI (e.g., claustrophobia or anxiety) or who had psychiatric disease (depression) or Kawasaki’s disease.

Brain MRI

Brain MRI was performed using two 1.5-T systems (MAGNETOM Vision or MAGNETOM Sonata; Siemens Healthcare, Tokyo, Japan). All patients underwent preoperative MRI within three days of surgery, with postoperative MRI being performed within two weeks after removal of the temporary pacing leads. Preoperative MRI was performed using T2-weighted imaging (T2WI) with TR/TE of 5400/99 msec or 5400/85 msec; T2*-weighted imaging (T2*WI) with TR/TE of 736/20 msec and a flip angle of 30°; fluid-attenuated inversion recovery (FLAIR) with TR/TE/TI of 9000/105/2400 msec; diffusion-weighted imaging (DWI) with TR/TE of 4000/100 msec and b of 1000 sec·mm−2 and MRA. A slice thickness of 4 mm with a slice interval of 2 mm was used for T2WI, T2*WI, FLAIR, and DWI. Postoperative MRI was performed using the same sequences as for preoperative MRI. MRA was not performed postoperatively.

Abnormal lesions were diagnosed as follows. Preoperative cerebral infarction was detected using preoperative FLAIR and DWI. A significant intracranial arterial stenosis was defined as > 50% stenosis on MRA in one or more of the intracranial carotid, vertebral, basilar, first and second portions of the anterior cerebral, and/or the horizontal portion of the middle cerebral arteries. Using DWI and FLAIR, new postoperative ischemic lesions were defined as cerebral lesions > 2 mm that were not present before surgery. Minor intracranial bleeding was diagnosed using T2*WI. White matter lesions were defined as periventricular hyperintensity (PVH) and deep subcortical white matter hyperintensity (DSWMH). They were graded (grades 0-3) according to the Fazekas scale: 0, absence; 1, cap of pencil-thin PVH or punctate focal DSWMH; 2, smooth halo PVH or early confluence of focal DSWMH; 3, irregular PVH extending into the deep white matter or large confluent areas of DSWMH.20 An experienced neuroradiologist blinded to the patients’ preoperative and postoperative clinical status read the MR images. Preoperative carotid artery duplex scanning was performed to assess the severity of the carotid stenosis, which was quantified as present (stenosis > 50%) or absent (stenosis < 50%).21 Transesophageal echocardiography was performed in all patients to evaluate the severity of aortic atherosclerosis (using the Katz score).22

Delirium assessment

All patients were assessed for delirium six to 24 hr after extubation by a nurse trained to use the delirium rating scale R98 (DRS-R98). The DRS-R98 is composed of a 16-item clinician-rated scale. It include three severity items and three diagnostic items, for a maximum total score of 46 points.23 The DRS-R98 divides delirium into three categories based on severity: no delirium (total points 0), subclinical delirium (total points 1-7), and delirium (total points ≥ 8).24

Operative procedure

All patients underwent standard OPCAB using an aortic “no-touch” technique, avoiding aortic side clamping and manipulation of the ascending artery to minimize the risk of stroke. Vein grafts were anastomosed with an assist device where indicated, using either the Enclose 2 (Novare Surgical Systems, Inc. Cupertino, CA, USA) or the PAS PORT system (Cardiac Inc., Redwood City, CA, USA). Surgical procedures, operative times, and use of anesthetic and narcotics were obtained from the operative record. We selected potential risk factors of postoperative delirium from a review of the literature. Candidate variables included age, body weight, history of smoking, hypertension, hyperlipidemia, heart failure, diabetes, perioperative atrial fibrillation, myocardial infarction, cerebral infarction, chronic obstructive pulmonary disease, chronic kidney disease, duration of the operation and anesthesia, intubation time, and MRI findings.

Statistical analysis

Our initially planned sample size was based on an expected delirium incidence of 12% in patients without carotid stenosis. Using a power of 80%, a 5% two-sided significance level, and a 2.5 ratio of patients with carotid artery stenosis to patients without carotid artery stenosis, with a 2.0 relative risk of delirium in those with carotid stenosis,13 our final sample size was estimated to be 370.

Continuous variables are presented as the mean (SD), and discrete variables are presented as the frequency and percentages. Univariate analysis with the proportional odds model was used to evaluate risk factors for postoperative delirium. The proportional odds ratio (OR) was assumed to be equal between the two groups’ comparisons (i.e., DRS score between 0 and 1-7; as well as between 1-7 and ≥ 8) in this model. In addition, we conducted a multivariate analysis that included risk factors with a univariate P < 0.3. The model was selected by the Akaike’s information criterion (AIC),25 with low AIC values being preferred. ORs and 95% confidence intervals (CI) were also determined. Postoperative outcome, length of stay (LOS) in the intensive care unit (ICU), intubation, and hospital LOS were tested between patients with or without delirium using the non-parametric Jonckheere-Terpstr test. P values < 0.05 (two-sided) were considered to indicate significance. Statistical analyses were performed with SAS version 9.3 (SAS Institute, Cary, NC, USA).

Results

Because patient recruitment was slower than expected, we ceased enrolment after a convenience sample of 105 eligible patients (Figure). Before the baseline data collection, seven patients were excluded for claustrophobia, anxiety, or refusing informed consent. Ten patients were further excluded from analysis because of either the MRI or carotid artery ultrasonography not being completed. We thus analyzed 88 patients. The patients’ preoperative characteristics and intraoperative parameters are shown in Table 1. MRI findings are shown in Table 2. Preoperative MRA revealed intracranial artery lesions in 24 (27%) of the 88 patients. PVH was found in 71 (81%) patients and DSWMH in 76 (86%) patients. Postoperative DWI or FLAIR MRI detected new ischemic lesions in seven (7.9%) patients. Overall, 25 (32%), 48 (60%), and seven (8%) patients had total DRS scores of 0, 1-7, and ≥ 8, respectively.

Figure
figure 1

Patient enrolment and outcomes. MRI = magnetic resonance imaging

Full size image
Table 1 Patients’ characteristics and intraoperative parameters (n = 88)
Full size table
Table 2 Findings of brain magnetic resonance imaging (n = 88)
Full size table

Table 3 shows the univariate analysis of factors correlated with postoperative delirium. Older age (≥ 70 yr), preoperative myocardial infarction, new ischemic lesions on MRI, new white matter disease (WMD; i.e., DSWMH or PVH) had a significant association with postoperative delirium. Multivariate analysis demonstrated a significant association with new ischemic lesions (OR 11.07, 95% CI 1.53 to 80.03; P = 0.017), carotid artery stenosis (OR 7.06, 95% CI 1.59 to 31.31; P = 0.010), myocardial infarction (OR 3.78, 95% CI 1.05 to 13.65; P = 0.043), and WMD (OR 3.04, 95% CI 1.14 to 8.12; P = 0.027) (Table 4). As the DRS advanced from 1-7 to ≥ 8, the ICU LOS was significantly greater in patients with postoperative delirium than in those without delirium (Table 5).

Table 3 Univariate analysis of risk factors of postoperative delirium
Full size table
Table 4 Multivariate analysis of risk factors of postoperative delirium
Full size table
Table 5 Relationship between delirium and duration of ICU and hospital stay
Full size table

There was no significant association between new ischemic lesions detected by MRI and carotid artery stenosis, cerebral artery stenosis, or the degree (determined by the Katz score) of aortic atherosclerosis.22

Discussion

This prospective observational study examined whether brain MRI findings and previously reported risk factors for postoperative delirium were associated with delirium following OPCABG surgery. Our main finding was that new ischemic lesions seen on MRI, carotid artery stenosis > 50%, and DSWMH were independently associated with postoperative delirium. We also demonstrated a relation between delirium and myocardial infarction.

The incidence of new ischemic lesions detected by brain MRI in our study was 7.9%. This contrasts with on-bypass cardiac surgery studies in which a high number of postoperative lesions were detected by MRI: 47% after valvular surgery26 and 30% after on-pump CABG.27 In a prior study of OPCAB, although the patient number was small (n = 13), no postoperative embolic ischemic events were observed.28 Thus, in comparison to reports of on-pump cardiac surgery, the incidence of injuries related to brain embolism following OPCAB was relatively low. The difference in the rates of DWI-defined ischemic lesions between the on-bypass and off-bypass cases might be explained by the numbers of cerebral emboli that occurred. For example, Liu et al. also found that avoiding CPB during CABG decreased the number of postoperative cerebral microemboli measured with bilateral TCD ultrasonography.29

There are several mechanisms by which ischemic lesions could result in delirium. For example, in an animal study, cholesterol cerebral emboli altered the permeability of the blood-brain barrier, activated microglia, and caused cognitive dysfunction in rats.30 After microglia returned to their resting state, the rats recovered from the cognitive decline, suggesting that inflammation due to emboli may have influenced cognitive function. Neuroinflammation may also be associated with delirium. Proinflammatory cytokines have been shown to lead to synaptic and neuronal dysfunction and to subsequent neurobehavioural and cognitive symptoms that are characteristic of delirium.31 Although the incidence of embolic events was low, it is possible that emboli contributed to the postoperative delirium given that the theories regarding the pathogenesis of delirium – neuronal aging, oxidative stress, neurotransmitter and neuroendocrine dysfunction, network disconnectivity – are complementary rather than competing.32

The presence of carotid artery stenosis > 50% implies the presence of chronic cerebral hypoperfusion and systemic arteriosclerosis. Symptomatic carotid artery stenosis also plays an important role in other cerebrovascular diseases, such as stroke, cognitive impairment, and dementia.33,34 Carotid stenosis is also associated with white matter lesions.35 A recent study indicated that asymptomatic internal carotid atherosclerosis is associated with large white matter hyperintensities, low total brain volume, and poor neuropsychological performance.36 The patients in our study who had carotid stenosis > 50% were most likely asymptomatic because at our institute carotid endarterectomy and carotid artery stenting are prioritized over CABG in patients with symptomatic carotid stenosis. Our finding that both carotid artery stenosis and white matter lesions had a significant association with postoperative delirium is compatible with the findings of previous studies.

White matter lesions appear as hyperintensities on T2-weighted MRI images. In the general population, the prevalence of white matter hyperintensities ranges from 11 to 21% in adults around 64 yr of age to 94% at age 82 yr.37,38 Previous studies reported a significant association between white matter lesions and postoperative delirium.17,18 Hatano et al. retrospectively investigated whether white matter hyperintensities predicted delirium after cardiac surgery.17 Although the incidence of postoperative delirium might have been underestimated because of the nature of their retrospective chart review using the Diagnostic and Statistical Manual of Mental Disorder, Fourth Edition, severe DSWMH had a significant OR (OR 3.9, 95% CI: 1.2 to 12.5; P = 0.02) in the multivariate logistic regression analysis. The population of the above study17 included patients with valve replacement or repair and with OPCAB, making it difficult to determine whether the effects of CPB were significant. In contrast, our study enrolled only OPCAB patients, and delirium was prospectively assessed with the DRS-98, likely resulting in more accurate findings.

Although we used the DRS-98 in the ICU, the Confusion Assessment Method-ICU (CAM-ICU) is arguably used more commonly for screening and diagnosis of delirium. However, the CAM-ICU is a categorical assessment, and thus the incidence of postoperative delirium might be underestimated by missing patients with less severe manifestations of delirium. Indeed, we examined specific score ranges to ensure that symptoms of postoperative delirium were not overlooked.

This study has several limitations. First, the number of patients studied was smaller than that of our initially calculated sample size, thus weakening the study’s statistical power. However, we obtained significant results that were compatible with those of previous studies. Second, the delirium assessment was performed only once, between six and 24 hr after extubation. Because of its fluctuating nature, we might have missed some patients who manifested delirium after the one-time assessment. In our institute, most patients are transferred to the ward of the cardiac surgical division the day after surgery. As a result, we were able to evaluate delirium status only in the ICU. Thus, delirium may have been underestimated if it occurred after patients moved to the ward. The duration of the ICU stay, however, had a significant association with postoperative delirium, meaning that the one-time assessment may have had clinical significance. Third, there was uncertainty as to what happens with these white matter lesions over the long term, although white matter lesions could indirectly predict an increased risk of stroke, dementia, and mortality.19 Fourth, we made an assumption that emboli are the key etiologic factors in all the lesions described, but we do not have any objective measurement of emboli, such as with TCD ultrasonography. Thus, we cannot rule out hypoperfusion or inflammation playing a significant role in the lesions that we described.

In conclusion, in patients undergoing OPCAB surgery, carotid artery stenosis of > 50% and the presence of white matter lesions were significantly associated with postoperative delirium in the ICU. Although MRI detected a lower incidence of new ischemic lesions than were seen in prior on-pump CABG studies, they were significantly associated with postoperative delirium.