Silent cerebral infarction after proximal thoracic aortic surgery under deep hypothermic circulatory arrest: an exploratory prospective DW-MRI study

Abstract

Background

Deep hypothermic circulatory arrest (DHCA) is widely used in proximal thoracic aortic surgery; however, its association with subclinical cerebral ischemic changes remains poorly understood. Diffusion-weighted magnetic resonance imaging (DW-MRI) enables sensitive detection of silent cerebral infarction (SCI), which may occur in the absence of overt neurological deficits.

Methods

This prospective observational study included 35 consecutive adult patients undergoing proximal thoracic aortic surgery under DHCA at 18 °C without selective cerebral perfusion. Preoperative and postoperative DW-MRI scans were obtained in all patients. Silent cerebral infarction (SCI) was defined as the presence of at least one new ischemic lesion on postoperative DW-MRI not present preoperatively. Perioperative neuron-specific enolase (NSE) levels and intraoperative near-infrared spectroscopy (NIRS) measurements were recorded. Factors associated with SCI were explored using univariable and multivariable logistic regression analysis.

Results

Postoperative silent cerebral infarction was detected in 19 patients (54.3%), whereas no ischemic lesions were observed on preoperative DW-MRI.

In univariable analysis, increasing age (OR 1.06, 95% CI 1.01–1.13; p = 0.021), Cardiopulmonary bypass time (CPB time) (OR 1.01, 95% CI 1.00–1.03; p = 0.024), and aortic cross-clamp time (CC time) (OR 1.04, 95% CI 1.01–1.07; p = 0.006) were associated with SCI. In age-adjusted multivariable models constructed separately for operative time variables, CC time remained independently associated with SCI (OR 1.03, 95% CI 1.01–1.06; p = 0.015), whereas CPB time did not retain statistical significance (OR 1.02, 95% CI 0.99–1.03; p = 0.089). Additional collinearity assessment showed a strong correlation between CPB time and CC time, and sensitivity analyses supported the persistence of the association between CC time and SCI. DHCA time, NIRS parameters, and NSE changes were not significantly associated with SCI. Although postoperative NSE levels increased compared with preoperative values, the change in NSE did not predict the occurrence of SCI.

Conclusions

Silent cerebral infarction was frequently detected on postoperative DW-MRI following proximal thoracic aortic surgery performed under DHCA. Aortic cross-clamp time was associated with SCI development; however, this association should be interpreted cautiously, as it may reflect procedural complexity, cumulative aortic manipulation, and embolic burden rather than the isolated effect of circulatory arrest time.

Trial registration

NCT04755439. Date of trial registration: February 17, 2021.

Introduction

Deep hypothermic circulatory arrest (DHCA) remains a widely used cerebral protection strategy in complex proximal thoracic aortic surgery, particularly when a bloodless surgical field is required. Through deep systemic cooling, DHCA reduces cerebral metabolic demand and allows temporary interruption of cerebral blood flow for a limited period. Despite its long-standing clinical use, neurological injury remains one of the most feared complications of aortic surgery, ranging from overt stroke to more subtle forms of cerebral damage that may remain clinically silent.

Silent cerebral infarctions (SCIs) are small ischemic lesions detected on neuroimaging in the absence of focal neurological deficits. Although these lesions may remain clinically silent in the early postoperative period, their detection is important because they may indicate subclinical cerebral injury. Diffusion-weighted magnetic resonance imaging (DW-MRI) is a sensitive imaging method for detecting acute ischemic brain lesions [1]. It enables the identification of subclinical cerebral lesions that would otherwise remain undetected using conventional imaging techniques [1].

Neurological injury after aortic surgery is commonly evaluated through clinically evident outcomes such as stroke. However, MRI-detected subclinical cerebral injury may provide additional information, particularly in patients undergoing DHCA without adjunctive cerebral perfusion. In this specific setting, the incidence, anatomical distribution, and perioperative factors associated with SCI development remain insufficiently characterized in contemporary surgical practice.

Therefore, this exploratory prospective study aimed to assess the incidence of silent cerebral infarction detected by DW-MRI after proximal thoracic aortic surgery performed under DHCA and to explore clinical and intraoperative factors associated with these lesions.

Methods

Study design and patient population

This study was a single-center, prospective, observational cohort conducted at a tertiary cardiovascular surgery center from December 2020 to January 2024. The study protocol was approved by the institutional ethics committee (Decision No: 2020/KK/239) and registered on ClinicalTrials.gov (NCT04755439). Written informed consent was obtained from all participants before enrollment.

Consecutive adult patients (aged 18–85 years) undergoing proximal thoracic aortic surgery under DHCA were prospectively enrolled. Eligible surgical procedures included ascending aortic replacement, aortic root replacement (Bentall procedure), and combined valve and coronary interventions requiring DHCA. Exclusion criteria were a history of stroke or transient ischemic attack, contraindications to MRI, or lack of postoperative DW-MRI imaging.

DHCA without adjunctive cerebral perfusion was selected according to institutional practice, operative strategy, anatomical considerations, and the anticipated duration of circulatory arrest, particularly in cases requiring a bloodless field for open distal anastomosis or limited proximal arch/hemiarch reconstruction. In selected patients with fragile or diseased distal aortic tissue, an open distal anastomosis under DHCA was preferred to provide a bloodless and controlled operative field and to reduce the technical risk of distal anastomotic tearing, provided that the anticipated circulatory arrest time was expected to remain limited.

Neuroimaging protocol and outcome definition

All patients underwent preoperative DW-MRI within 1–3 days of surgery and postoperative DW-MRI within 7 days of surgery. Imaging was performed using a standardized protocol that included diffusion-weighted and apparent diffusion coefficient (ADC) sequences.

SCI was defined as the presence of at least one new ischemic lesion on postoperative DW-MRI that was not present on preoperative imaging, in the absence of focal neurological deficits. All DW-MRI images were independently evaluated by two experienced neuroradiologists who were blinded to clinical and intraoperative data. Discrepancies were resolved by consensus.

The primary outcome of the study was the presence of at least one postoperative silent cerebral infarction.

Perioperative variables and monitoring

Baseline demographic characteristics, cardiovascular risk factors, and operative details were recorded prospectively. Intraoperative parameters included CPB time, CC time, and time of DHCA.

Cerebral oxygenation was continuously monitored intraoperatively using near-infrared spectroscopy (NIRS) with bilateral frontal probes. Baseline regional cerebral oxygen saturation (rSO₂) was recorded before anesthesia induction. NIRS-derived variables included baseline values, the lowest intraoperative values during DHCA, and a time-integrated measure of intraoperative NIRS change calculated from the average of right and left hemispheric NIRS values over time.

Serum neuron-specific enolase (NSE) levels were measured preoperatively on the morning of surgery and postoperatively between 6 and 36 h after the procedure.

Statistical analysis

Statistical analyses were conducted using IBM SPSS Statistics version 29.0 (IBM Corp., Armonk, NY, USA). Continuous variables were assessed graphically and summarized as medians and interquartile ranges (IQRs), whereas categorical variables were presented as counts and percentages.

Given the single-cohort observational design, no direct group comparisons were planned. Perioperative changes in NSE levels (postoperative minus preoperative) were evaluated in the entire cohort using the Wilcoxon signed-rank test.

Associations between clinical and intraoperative variables and the occurrence of silent cerebral infarction were examined using logistic regression. Univariable logistic regression was performed for each candidate variable. Because of the limited number of outcome events, the multivariable analysis was constructed using a clinically driven limited-variable approach to reduce the risk of overfitting. Age was included as a clinically relevant baseline covariate. Since CPB time and CC time were clinically interrelated and showed potential collinearity, these variables were not entered into the same multivariable model. Therefore, two separate age-adjusted multivariable logistic regression models were constructed: Model 1 included age and CC time, and Model 2 included age and CPB time.

To further assess the relationship between key operative time variables, a correlation matrix was generated for CPB time, CC time, and DHCA time. In addition, variance inflation factors (VIFs) were calculated to evaluate potential multicollinearity among these variables. The distribution of specific surgical procedures was also summarized according to SCI status to better characterize procedural heterogeneity between SCI-positive and SCI-negative patients. To assess the robustness of the multivariable findings and to reduce potential instability related to collinearity between CPB time and CC time, additional sensitivity logistic regression models were performed. These models evaluated CC time and CPB time in separate models, adjusted for age, and also assessed a model including age, CC time, and DHCA time. Results were reported as odds ratios (OR) with 95% confidence intervals (CI). Statistical significance was defined as p < 0.05.

There were no missing data for the primary outcome or key covariates; therefore, no data imputation was required.

Operative technique

All patients received standard premedication with intramuscular midazolam (0.03–0.05 mg/kg) 30 min prior to surgery. Intraoperative monitoring included electrocardiography, invasive arterial blood pressure, pulse oximetry, arterial blood gas analysis, and near-infrared spectroscopy (NIRS) (Medtronic INVOS NIRS Software, Monitoring System Analytics Tool, Minneapolis, MN) for cerebral oxygenation.

General anesthesia was induced with intravenous propofol (2–2.5 mg/kg), fentanyl (2.5–5 µg/kg), and rocuronium bromide (0.6 mg/kg), and maintained with continuous infusions of propofol, fentanyl, and rocuronium. Central venous and urinary catheters were placed in all patients.

Systemic anticoagulation was achieved with intravenous heparin (300–400 IU/kg), targeting an activated clotting time > 420 s. At the end of CPB, heparin was reversed with protamine sulfate.

Following median sternotomy, CPB was initiated, and an aortic cross-clamp was applied. All procedures were performed using a conventional roller-pump CPB system. Myocardial protection was provided using antegrade blood cardioplegia. Perfusion flow was adjusted according to body surface area, maintaining a mean arterial pressure of 60–80 mmHg. During cooling and rewarming, pump flow was adjusted according to body surface area, perfusion pressure, and the phase of hypothermia or rewarming. Alpha-stat acid–base management was used throughout cooling, DHCA, and rewarming. The oxygen fraction delivered to the membrane oxygenator was initially set at 1.0 and subsequently adjusted according to serial arterial blood gas analyses. The target hematocrit during hypothermic CPB was maintained at approximately 24–30%.

Before initiating DHCA, patients were cooled to 18 °C, with additional cranial cooling applied using ice packs at 24 °C. Nasopharyngeal temperature monitoring was used for temperature assessment. Before DHCA, approximately one-third of the patient’s blood volume (1600–1800 ml) was withdrawn into the venous reservoir.

The distal aorta-graft anastomosis was performed under DHCA using the open anastomosis technique. The arterial cannula was reinserted to the apex of the distal aorta-graft anastomosis site following the "Kaplan Technique [2]," and DHCA was terminated. In this technique, arterial cannulation is performed directly through the aneurysmal aortic tissue without involving any healthy segment of the native aortic tree.

Rewarming was performed gradually after completion of distal anastomosis. Topical ice packs were removed during the early rewarming phase, at approximately 24 °C. Active rewarming was not initiated during the first 5 min after reinstitution of CPB. Thereafter, rewarming was performed according to our institutional protocol. The temperature gradient between the perfusate and the patient’s blood was maintained at approximately 3 °C between 18 °C and 24 °C, 5 °C up to 32 °C, and 8 °C up to 37.6 °C, Following removal of the aortic cross-clamp and restoration of hemodynamic stability, CPB was terminated and patients were transferred to the intensive care unit.

Results

Patient characteristics and operative data

A total of 35 patients were included in the study. The median age was 58.0 years (IQR 24–78), and the median body surface area was 1.9 m² (IQR 1.7–2.6). Hypertension (HT) was present in 15 patients (42.9%), and diabetes mellitus (DM) in 7 patients (20.0%). Surgical procedures included ascending aortic replacement, Bentall procedures, and combined valve and coronary interventions. Operative and demographic characteristics are summarized in Table 1.

Table 1.

Distribution of Descriptive Characteristics n (%)

	Median (IQR)
Age	58,0 (24–78)
BSA m2	1,9 (1,7–2,6)
Hypertension (HT)
Yes	15 (42,9)
Diabetes Mellitus (DM)
Yes	7 (20,0)
Surgery
AsAO	5 (14,3%)
CABG + AsAO	2 (5,7%)
CABG + AVR + AsAO	4 (11,4%)
Bentall	11 (31,4%)
AVR + AsAO	10 (28,6%)
Bentall + MVR + CABG	1 (2,9%)
Bentall + CABG	1 (2,9%)
AsAO + MVR + Tricuspit Devega	1 (2,9%)
Smoking
Yes	7 (20,0)
Alcohol
Yes	5 (14,3)
Bypass time (min)	222,0 (134–332)
Aortic Cross Clamp time (min)	147,0 (84–224)
DHCA time (min)	23,0 (21–30)
Stay of Hospital (day)	11,0 (4–22)
Duration of mechanical ventilation (hour)	11,0 (5–144)
Postop awakening time (hour)	7,0 (2–120

AsAO ascending aorta replacement, AVR Aortic valve replacement, BSA Body Surface Area, CABG Coranary artery bypass grafting, DHCA Deep Hypotermic Circulatory Arrest, MVR Mitral valve replacement

The median CPB time was 222 min (IQR 134–332), the median CC time was 147 min (IQR 84–224), and the median DHCA time was 23 min (IQR 21–30).

The distribution of specific surgical procedures according to SCI status is summarized in Table 2. Bentall procedure was performed in 11 patients, including 7 SCI-positive and 4 SCI-negative patients. Ascending aortic replacement with aortic valve replacement (AVR) was performed in 10 patients, including 3 SCI-positive and 7 SCI-negative patients. Isolated ascending aortic replacement was performed in 5 patients, all of whom were SCI-negative. More complex combined procedures, including ascending aortic replacement with AVR and coronary artery bypass grafting (CABG), ascending aortic replacement with CABG, Bentall with CABG, Bentall with mitral valve replacement (MVR) and CABG, and ascending aortic replacement with MVR and tricuspid DeVega repair, were observed predominantly in the SCI-positive group.

Table 2.

Distribution of surgical procedures according to SCI status

Surgical procedure	Total cohort (N = 35)	SCI-negative (n = 16)	SCI-positive (n = 19)
Bentall	11 (31.4%)	4 (25.0%)	7 (36.8%)
AsAO + AVR	10 (28.6%)	7 (43.8%)	3 (15.8%)
AsAO	5 (14.3%)	5 (31.2%)	0 (0.0%)
AsAO + AVR + CABG	4 (11.4%)	0 (0.0%)	4 (21.1%)
AsAO + CABG	2 (5.7%)	0 (0.0%)	2 (10.5%)
AsAO + MVR + Tricuspid DeVega	1 (2.9%)	0 (0.0%)	1 (5.3%)
Bentall + CABG	1 (2.9%)	0 (0.0%)	1 (5.3%)
Bentall + MVR + CABG	1 (2.9%)	0 (0.0%)	1 (5.3%)

Data are presented as n (%). Percentages for the total cohort were calculated using N = 35, whereas percentages for SCI-negative and SCI-positive groups were calculated within each respective group

SCI silent cerebral infarction, AsAO ascending aortic replacement, AVR aortic valve replacement, CABG coronary artery bypass grafting, MVR mitral valve replacement

Neuroimaging findings

Preoperative DW-MRI showed no ischemic lesions in any patient. Postoperative DW-MRI detected at least one new SCI in 19 patients (54.3%). Figure 1 demonstrates the development of lesions using preoperative and postoperative DW-MRI and ADC mapping. The median number of lesions per patient was 1.0 (IQR 0–11). The anatomical distribution of lesions is presented in Table 3, with the frontal lobe the most frequently affected region, followed by the cerebellum and the occipital lobe (Fig. 2).

Fig. 1.

DWI and ADC maps from the preoperative (A, B) and postoperative (C, D) periods

Table 3.

Distribution and Anatomical Localization of Silent Cerebral Infarctions Detected on Postoperative DW-MRI n (%)

Median (IQR)
Preoperative DW-MRI
Ischemic lesions detected	0 (0,0%)
Postoperative DW-MRI
Silent cerebral infarction present (≥ 1 lesion)	19 (54,3%)
Lesion burden
Number of lesions per patient,	1,0 (0–11)
Anatomical distribution of lesions
Frontal lobe	9 (25,7%)
Parietal lobe	5 (14,3%)
Temporal lobe	2 (5,7%)
Occipital lobe	6 (17,1%)
Cerebellum	7 (20,0%)
Brainstem	2 (5,7%)
Basal ganglia	2 (5,7%)
Thalamus	2 (5,7%

DW-MRI Diffusion-weighted magnetic resonance imaging

Fig. 2.

Anatomical distribution of silent cerebral infarctions detected on postoperative diffusion-weighted magnetic resonance imaging

No patient developed a clinically evident stroke or focal neurological deficit during the hospital stay.

Perioperative NSE and NIRS measurements

The median preoperative NSE level was 16.2 ng/mL (IQR 8–78), and postoperative NSE increased to a median of 31.4 ng/mL (IQR 15–93). The median perioperative NSE change was 10.7 ng/mL (IQR − 53.2 to 70.0). The increase in NSE from preoperative to postoperative measurements across the entire cohort was statistically significant (Wilcoxon signed-rank test, p < 0.001). The distribution of NSE values and time-integrated intraoperative NIRS change is shown in Table 4.

Table 4.

Distribution of Perioperative NSE Levels and NIRS Measurements

	Median (IQR)
Preoperative NSE	16,2 (8–78)
Postoperative NSE	31,4 (15–93)
NSE change	10,7 (−53,2–70,0)
Time-integrated intraoperative NIRS change	506,5 (56–3804)

^*Data are presented as median (interquartile range). NSE change represents postoperative minus preoperative values. No group comparisons were performed

^*The time-integrated intraoperative NIRS change represents a summary measure derived from the average of right and left hemispheric NIRS values over time and should not be interpreted as a diagnostic AUROC value

NSE neuron specific enolaz

The median time-integrated intraoperative NIRS change was 506.5 (IQR, 56–3804). Neither the baseline nor the lowest intraoperative NIRS value was significantly associated with SCI.

Factors associated with silent cerebral infarction

To further evaluate the relationship between operative time variables, a correlation matrix was generated for CPB time, CC time, and DHCA time. A strong positive correlation was observed between CPB time and CC time (r = 0.875), whereas DHCA time showed only weak correlations with both CPB time (r = 0.090) and CC time (r = 0.106). VIF analysis showed moderate collinearity for CPB time and CC time, with VIF values of 4.28 and 4.30, respectively. The VIF for DHCA time was 1.01, indicating no relevant collinearity for this variable.

Univariable logistic regression analysis identified increasing age (OR 1.06, 95% CI 1.01–1.13; p = 0.021), CPB time (OR 1.01, 95% CI 1.00–1.03; p = 0.024), and CC time (OR 1.04, 95% CI 1.01–1.07; p = 0.006) as variables associated with higher odds of SCI. DHCA time and NSE change were not significantly associated with SCI (Table 5).

Table 5.

Univariable and multivariable logistic regression analysis of factors associated with Silent Cerebral Infarction (SCI)

Variable	Univariable OR (95% CI)	p-value	Multivariable Model 1* OR (95% CI)	p-value	Multivariable Model 2** OR (95% CI)	p-value
Age (per year)	1.06 (1.01–1.13)	0.021	1.05 (0.99–1.13)	0.102	1.06 (0.99–1.13)	0.084
Aortic cross clamp time (per min)	1.04 (1.01–1.07)	0.006	1.03 (1.01–1.06)	0.015	—	—
Cardiopulmonary bypass time (per min)	1.01 (1.00–1.03)	0.024	—	—	1.02 (0.99–1.03)	0.089
DHCA time (per min)	0.97 (0.73–1.30)	0.874	—	—	—	—
NSE change (ng/mL)	1.04 (0.98–1.11)	0.125	—	—	—	—
Hypertension (yes vs no)	2.44 (0.61–9.80)	0.207	—	—	—	—
Diabetes mellitus (yes vs no)	2.50 (0.41–15.11)	0.318	—	—	—	—
Smoking (yes vs no)	1.15 (0.21–6.14)	0.865	—	—	—	—
			−2 Log likelihood:33.82, R2 = 0,452		−2 Log likelihood:38.54, R2 = 0,324

DHCA Deep Hypothermic Circulatory arrest, NSE neuron-specific enolase

Model 1*: Age + Aortic cross clamp time, Model 2**: Age + Cardiopulmonary bypass time, Multivariable models were constructed based on clinical relevance (a priori selection) rather than solely univariable p-values. Due to potential collinearity between aortic cross clamp time and cardiopulmonary bypass time, these variables were not included in the same model. Given the limited number of events (n = 19), the number of variables in multivariable models was restricted to avoid overfitting

In the age-adjusted multivariable models constructed separately for operative time variables, CC time remained independently associated with SCI in Model 1 (OR 1.03, 95% CI 1.01–1.06; p = 0.015), whereas age did not retain statistical significance. In Model 2, which included age and CPB time, CPB time did not retain statistical significance (OR 1.02, 95% CI 0.99–1.03; p = 0.089). Model fit statistics favored Model 1 compared with Model 2, with a lower − 2 log-likelihood value and higher pseudo-R² (33.82 and 0.452 vs. 38.54 and 0.324, respectively).

To address potential instability related to collinearity between CPB time and CC time, additional sensitivity models were performed (Table 6). In a model including age and CC time but excluding CPB time, CC time remained significantly associated with SCI (OR 1.04, 95% CI 1.01–1.07; p = 0.015). Similarly, in a model including age, CC time, and DHCA time, CC time remained associated with SCI (OR 1.04, 95% CI 1.01–1.07; p = 0.017), whereas DHCA time was not significant. In contrast, when CC time was excluded and CPB time was included with age, CPB time showed only borderline significance (OR 1.02, 95% CI 1.00–1.03; p = 0.089).

Table 6.

Sensitivity logistic regression models for silent cerebral infarction

Model	Variable	OR	95% CI	p value
Primary model	Age	1.06	0.99–1.14	0.089
	CPB duration	0.98	0.95–1.02	0.252
	Aortic cross-clamp time	1.07	1.00–1.14	0.039
Sensitivity model 1	Age	1.06	0.99–1.13	0.100
	Aortic cross-clamp time	1.04	1.01–1.07	0.015
Sensitivity model 2	Age	1.06	0.99–1.12	0.084
	CPB duration	1.02	1.00–1.03	0.089
Sensitivity model 3	Age	1.07	0.99–1.14	0.073
	Aortic cross-clamp time	1.04	1.01–1.07	0.017
	DHCA duration	0.83	0.54–1.27	0.386

Abbreviations: CI confidence interval, CPB cardiopulmonary bypass, DHCA deep hypothermic circulatory arrest, OR odds ratio, SCI silent cerebral infarction

Primary model included age, CPB duration, and aortic cross-clamp time. Sensitivity model 1 included age and aortic cross-clamp time. Sensitivity model 2 included age and CPB duration. Sensitivity model 3 included age, aortic cross-clamp time, and DHCA duration

Discussion

This exploratory prospective study demonstrates that SCI detected by DW-MRI is a frequent radiological finding after proximal thoracic aortic surgery performed under DHCA. A substantial proportion of patients developed at least one new ischemic lesion, despite the absence of clinically evident stroke or focal neurological deficit. These findings highlight the discrepancy between radiological and clinical neurological outcomes and emphasize the importance of imaging-based endpoints for detecting subclinical cerebral injury [1, 3–6].

A key finding of this study is that procedural duration parameters, particularly, aortic cross-clamp were more strongly associated with SCI than the time of circulatory arrest itself. In multivariable analysis, CC time remained independently associated with SCI, whereas DHCA time did not. Sensitivity analyses performed to address potential instability related to collinearity between CPB time and CC time showed that the association between CC time and SCI persisted when CPB time was excluded from the model and when DHCA time was included. However, given the limited number of outcome events and the interrelationship between operative time variables, this finding should still be interpreted as exploratory rather than definitive. This observation supports the concept that cerebral injury in complex cardiac surgery is multifactorial and may be more closely related to overall procedural complexity and systemic ischemic stress than to isolated periods of circulatory arrest [7–10]. Similar associations between longer operative times and neurological injury have been reported in previous studies of CPB–related cerebral dysfunction [6, 11]. However, this finding should be interpreted cautiously because operative time variables are clinically interrelated. In the additional collinearity assessment, CPB time and CC time showed a strong positive correlation, whereas DHCA time showed only weak correlations with both variables. VIF analysis also demonstrated moderate collinearity between CPB time and CC time. Therefore, CC time should not be considered an isolated ischemic exposure. Rather, it may serve as a surrogate marker of overall procedural complexity, cumulative aortic manipulation, cannulation, clamping and declamping, handling of diseased or calcified aortic tissue, and potential embolic burden.

This interpretation is further supported by the distribution of surgical procedures according to SCI status. More complex combined procedures were observed predominantly in the SCI-positive group, whereas isolated ascending aortic replacement was observed only in SCI-negative patients. Although the small sample size precludes definitive subgroup conclusions, these findings suggest that SCI development may be more closely related to the cumulative burden of complex aortic surgery than to DHCA time alone.

From a practical perspective, the high incidence of MRI-detected SCI observed in the present cohort suggests that optimization of the entire operative pathway may be important, even in the absence of clinically apparent stroke. Prolonged CC and CPB times may serve not only as time-dependent risk factors, but also as surrogate markers of procedural complexity, aortic manipulation, embolic burden, and perioperative hemodynamic instability. Therefore, strategies aimed at reducing SCI may include meticulous preoperative planning, minimizing CC and CPB time whenever feasible, limiting unnecessary aortic manipulation, maintaining adequate perfusion pressure and oxygen delivery during CPB, avoiding excessive temperature gradients during rewarming, and individualizing cerebral protection strategies according to patient- and procedure-related risk. In more complex or high-risk cases, adjunctive cerebral perfusion strategies, particularly selective antegrade cerebral perfusion, should be considered within an individualized cerebral protection strategy.

The anatomical distribution of SCI may also provide mechanistic insight. In the present cohort, lesions were most frequently observed in the frontal lobe, followed by the cerebellum and occipital lobe. This distribution, together with the focal nature of the lesions, appears more compatible with a scattered embolic pattern than with a purely global hypoperfusion mechanism. Hypoperfusion-related injury would be expected to involve watershed territories or broader, less well-demarcated ischemic areas. Therefore, the association between CC time and SCI may reflect the cumulative embolic risk associated with aortic manipulation, cannulation, clamping, and declamping rather than global ischemia duration alone. Postoperative atrial fibrillation (AF) may also represent a potential embolic contributor to cerebral ischemic lesions after cardiac Surgery [12]; however, no new-onset AF or clinically relevant rhythm disturbance was documented before postoperative DW-MRI in the present cohort.

The high incidence of MRI-detected SCI observed in this cohort aligns with prior reports showing frequent subclinical ischemic lesions after cardiac surgery when sensitive imaging modalities are used [3, 5, 6, 13, 14]. In a prospective DHCA cohort, Zhuang et al. similarly reported that DW-MRI reveals a substantial burden of cerebral parenchymal injury even in the absence of overt neurological deficits, supporting the role of MRI-based endpoints in this setting [14]. However, it remains unclear how these lesions translate into long-term neurocognitive impairment. Although population-based studies have linked SCI to increased risk of dementia and cognitive decline [4], the clinical relevance of perioperative MRI-detected lesions remains debated, particularly in the absence of formal neuropsychological testing. The lack of overt neurological deficits in our cohort suggests that high SCI rates may partly reflect the sensitivity of DW-MRI rather than clinically meaningful neurological injury.

Our findings also raise important questions about commonly used intraoperative and biochemical monitoring tools. Although perioperative NSE levels increased significantly after surgery, the change in NSE was not associated with SCI. This finding aligns with previous studies indicating that NSE reflects global cerebral stress rather than focal ischemic injury and may therefore have limited utility as a surrogate marker for subclinical cerebral damage [15]. Similarly, intraoperative NIRS showed poor discriminative ability for SCI. Although NIRS provides continuous monitoring of cortical oxygenation, its ability to detect deep, microembolic, or watershed-type ischemic lesions is inherently limited [16]. These results suggest that both NSE and NIRS may be insufficient as standalone tools for predicting MRI-detected cerebral injury in this setting.

In addition to neuroimaging and intraoperative NIRS monitoring, circulating biomarkers may have a potential role in the perioperative assessment of cerebral injury. Although NSE was assessed in the present study, the lack of a standardized postoperative sampling time limited the interpretation of its perioperative changes. Several circulating biomarkers have been proposed as potential indicators of neuronal, neuroaxonal, and glial injury after cardiac surgery [17]. Other biomarkers, including S100B, neurofilament light chain, and glial fibrillary acidic protein, may provide complementary information regarding neuronal and glial injury and should be considered in future studies investigating cerebral injury after DHCA. Larger prospective studies with standardized biomarker sampling, systematic neurocognitive assessment, and appropriate control groups are needed to clarify the clinical significance of MRI-detected SCI and to identify modifiable strategies for neuroprotection.

Several studies have suggested that adjunctive cerebral perfusion strategies, particularly selective antegrade cerebral perfusion, may reduce the incidence of neurological injury during aortic arch surgery [10, 18–20]. In our institution, DHCA has been used selectively in patients requiring a bloodless operative field for open distal anastomosis, distal ascending aortic or proximal arch involvement, hemiarch replacement, or complex aortic reconstruction, particularly when the anticipated circulatory arrest time is limited. In contrast, our study focused exclusively on DHCA without adjunctive perfusion, reflecting institutional practice and anatomical considerations. Although large observational series have reported acceptable clinical outcomes with DHCA when arrest times are kept within reasonable limits [7, 9, 21, 22], our data indicate that subclinical cerebral injury remains common, even with relatively short circulatory arrest time. These findings underscore the complexity of cerebral protection during aortic surgery and suggest that optimizing overall procedural efficiency may be as important as selecting a cerebral protection strategy.

Limitations

This study has several limitations. Its relatively small sample size and single-center observational design limit the generalizability of the findings and preclude causal inference. Although collinearity assessment, variance inflation factor analysis, sensitivity models, and procedure-specific analyses were performed, CC time, CPB time, and overall surgical complexity remain closely interrelated; therefore, the association between CC time and SCI should be interpreted as hypothesis-generating rather than causal. The limited number of SCI events also restricted the complexity and stability of the multivariable model, and residual confounding related to procedural complexity, cumulative aortic manipulation, embolic burden, perioperative hemodynamics, and aortic atherosclerotic plaque burden cannot be excluded. In addition, the absence of a control group undergoing comparable CPB surgery without DHCA or with selective antegrade cerebral perfusion limits the ability to attribute MRI-detected lesions specifically to DHCA. Formal neurocognitive testing and standardized neurological scoring were not routinely performed, so subtle functional or cognitive changes may have been missed. Furthermore, postoperative NSE sampling was performed within a broad 6–36-h window, which may have influenced its interpretation. Finally, although no significant carotid pathology or new-onset postoperative AF before postoperative DW-MRI was observed, these factors could not be fully explored in the regression analysis. These findings should therefore be validated in larger, multicenter prospective studies with appropriate control groups, standardized neurological and neurocognitive assessment, uniform biomarker sampling, detailed evaluation of aortic atherosclerotic burden, and individualized cerebral protection strategies.

Conclusion

In this exploratory prospective study, silent cerebral infarction was frequently detected on postoperative DW-MRI after proximal thoracic aortic surgery performed under DHCA, despite the absence of clinically evident stroke or focal neurological deficit. Aortic cross- clamp time was associated with SCI development; however, this relationship should be interpreted cautiously, as it may reflect overall procedural complexity, cumulative aortic manipulation, and embolic burden rather than the isolated effect of circulatory arrest time. These findings highlight the value of imaging-based endpoints for detecting subclinical cerebral injury and support the need for larger controlled studies incorporating standardized neurocognitive assessment, uniform biomarker sampling, and individualized cerebral protection strategies.

Abbreviations

ADC

Apparent Diffusion Coefficient

AF

Atrial Fibrillation

AVR

Aortic Valve Replacement

BSA

Body Surface Area

CABG

Coronary Artery Bypass Grafting

CC time

Aortic cross-clamp time

CI

Confidence Interval

CPB

Cardiopulmonary Bypass

CTA

Computed Tomography Angiography

DHCA

Deep Hypothermic Circulatory Arrest

DM

Diabetes Mellitus

DW-MRI

Diffusion-Weighted Magnetic Resonance Imaging

HT

Hypertension

IQR

Interquartile Range

MRI

Magnetic Resonance Imaging

MVR

Mitral Valve Replacement

NIRS

Near-Infrared Spectroscopy

NSE

Neuron-Specific Enolase

OR

Odds Ratio

rSO₂

Regional Cerebral Oxygen Saturation

SCI

Silent Cerebral Infarction

Authors’ contributions

- Burak Bozkurt: Conceptualization, study design, manuscript writing. - Okan Akıncı: MRI data acquisition and interpretation, literature support. - Mukan Kağan Kuş: MRI interpretation, drafting methods section. - Ceyda Ören: MRI data acquisition and interpretation, literature support. - Zeynep Yıldız: NIRS data interpretation, results drafting. - İnanç Yeşilkaya: Radiological supervision, critical review of imaging data. - Ozan Ertürk: Clinical neurological analysis, manuscript revision. - Mehmet Kaplan: Surgical planning and oversight, final manuscript approval.

Funding

No Funding.

Data availability

The datasets used and/or analysed during the current study are not publicly available due to patient privacy and ethical restrictions but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Regional ethics comitee approved our study. Decision No: 2020/KK/239, Date:23/11/2020) and was registered on ClinicalTrials.gov under the identifier NCT04755439. The study was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants before enrollment.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.