Imbalance between oxygen delivery and consumption during cooling and rewarming phases of cardiopulmonary bypass and association with hyperlactatemia in total aortic arch replacement: a prospective observational study

Abstract

Background

Inadequate oxygen perfusion during cardiac surgery increases risk of postoperative complications. In order to optimize perfusion strategies, the aim of this exploratory study was to investigate oxygen delivery, consumption and extraction during hypothermic circulatory arrest and their association with postoperative hyperlactatemia.

Method

A consecutive sample of 44 adult patients scheduled for total aortic arch replacement under cardiopulmonary bypass and hypothermic circulatory arrest were prospectively enrolled during the study period. Oxygen delivery index, oxygen consumption index and oxygen extraction rate were measured during cooling from 36 to 24 ℃ and rewarming from 24 to 36 ℃. The primary outcome was the association between intraoperative oxygen metabolism parameter and postoperative hyperlactatemia. Hyperlactatemia was defined as peak lactate ≥ 4 mmol/L within 24 h after surgery.

Results

A total of 41 patients were enrolled in the final analysis. Serum lactate rose slightly from 1.5 (1.3–2.2) mmol/L at baseline to 2.0 (1.5–2.6) mmol/L before hypothermic circulatory arrest and again to 4.0 (3.6–5.1) mmol/L after 29.17 ± 5.87 min of arrest, finally peaking at 5.0 (2.9–7.1) mmol/L within 24 h after surgery. The three oxygen parameters decreased progressively during cooling and increased gradually during rewarming. All three parameters at a given temperature were obviously higher during rewarming than cooling. Oxygen extraction rate at 24 ℃ during cooling correlated positively with peak lactate concentration within 24 h after surgery (Spearman’s ρ = 0.67, p = 0.001). Hyperlactatemia occurred in 24 patients (58.54%), whose oxygen extraction rate and oxygen consumption index at 24 ℃ during cooling were significantly higher than in those who did not suffer hyperlactatemia. In fact, these two parameters discriminated hyperlactatemia with respective areas under the receiver operating characteristic (AUC) curves of 0.83 and 0.80.

Conclusion

In this exploratory study, a relatively higher oxygen extraction rate at 24 °C during cooling appeared to be associated with postoperative hyperlactatemia (AUC 0.83, cut-off > 0.102). This preliminary finding warrants validation in larger prospective studies.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13019-026-04532-9.

Background

Cardiopulmonary bypass remains associated with high risk of perioperative mortality and morbidity, with 15–30% of patients suffering acute kidney injury, among whom 2–5% require dialysis (carrying a mortality of 50–80%), and with 30–70% of patients suffering delirium, cognitive dysfunction, or stroke [1–5]. Risk of postoperative complications is even higher in those whose cardiopulmonary bypass involves hypothermic circulatory arrest [6, 7].

A major contributor to this risk is imbalance between oxygen delivery and oxygen consumption during cardiopulmonary bypass [8–11], such as reduced systemic oxygen delivery, which can precipitate a shift to anaerobic metabolism and elevated harmful lactate levels [12]. For example, maintaining an index of oxygen delivery (DO₂I) around 270–280 mL/min/m² may be effective for avoiding acute kidney injury [9, 10]. We suspect that optimizing perfusion based only on DO₂I may be less effective because the efficiency of aerobic metabolism depends not only on oxygen delivery but also on its extraction, which in turn may depend on the patient’s characteristics and on whether the patient is being cooled or rewarmed during hypothermic circulatory arrest [13].

In order to develop potentially more comprehensive guidance for optimizing perfusion during cardiopulmonary bypass involving hypothermic circulatory arrest, the aim of this exploratory study was to investigate oxygen metabolism during the cooling and rewarming phases of patients undergoing total aortic arch replacement and to explore their relationship with postoperative hyperlactatemia.

Methods

Patient population

This observational study was approved by the Ethics Committee of West China Hospital, Sichuan University (protocol 20241686). Given the emergency nature of acute type A aortic dissection, a deferred consent procedure was also approved. A consecutive series of adults diagnosed with acute type A aortic dissection and scheduled for total aortic arch replacement under cardiopulmonary bypass and hypothermic circulatory arrest at West China Hospital, Sichuan University were prospectively enrolled in the study between August 1, 2024 to March 1, 2026. Informed consent or a deferred consent was obtained from either the patients or their legally authorized representative.Patients were excluded if they or their representative refused consent, or if the surgery was switched to a procedure not involving hypothermic circulatory arrest. All procedures were performed in accordance with the Helsinki declaration, and data were collected as part of routine clinical measurements during total aortic arch replacement at our hospital.

Perioperative monitoring

Patients were routinely monitored through five-lead electrocardiography, capnography and measurement of pulse oxygen saturation (SpO₂), invasive blood pressures at both radial arteries and left dorsal pedis artery, as well as nasopharyngeal and bladder temperatures. In addition, regional cerebral oxygen saturation (rScO₂) was monitored through self-adhesive transcutaneous oximetry sensors (EGOS-600 A, Suzhou Engine Bio-medical Electronics, Suzhou, China) on the right and left sides of the forehead.

Anesthesia

General anesthesia was induced with midazolam (0.04–0.1 mg/kg), sufentanil (1–2 µg/kg), and rocuronium (0.5–1.2 mg/kg), then maintained with sevoflurane inhalation (1–2%) and intermittent administration of sufentanil and rocuronium. After tracheal intubation, patients were placed on pressure-controlled mechanical ventilation to keep end-tidal carbon dioxide in the normal range (35–45 mmHg). A central venous catheter was inserted in the right jugular vein, and transesophageal echocardiography (iE33; Phillips Medical System, Andover, MA, USA) was routinely performed before surgery.

Surgery and cardiopulmonary bypass

The perfusionists and surgical team were unaware of the study hypothesis. All perfusion management, including pump flow rate adjustments, target hematocrit, and temperature was performed according to our institutional standardized clinical protocol, without any intervention or guidance from the study. All patients underwent total aortic arch replacement via median sternotomy while supine. Cardiopulmonary bypass was established using a Stockert-5 roller pump (Sorin, Munich, Germany) with a disposable hollow-fiber membrane adult oxygenator (Affinity NT; Medtronic, Minneapolis, MN, USA). The patient was perfused systemically after cannulation of the aorta, right axillary artery, or femoral artery; systemic venous return was achieved through cannulation of the vena cava or femoral vein.

Once systemic heparinization had achieved an activated clotting time > 480 s, cardiopulmonary bypass was initiated. During cooling before circulatory arrest, pump flow rate was adjusted according to our institutional protocol, which aimed at maintaining a mean arterial blood pressure between 50 and 80 mmHg and a mixed venous oxygen saturation (SvO₂) > 80%. Consequently, pump flow rate was typically gradually reduced from approximately 3.2 to 2.2 L/min/m² as temperature decreased. The target hematocrit was > 21% during bypass. Cardiac arrest was induced with cold blood cardioplegia [crystalloid: blood ratio, 1:4] at a dose of 20–30 mL/kg. Cardioplegia was repeated at a dose of 10–15 mL/kg every 30–40 min during surgery.

When nasopharyngeal temperature reached 24 °C and bladder temperature < 28 °C, circulatory arrest was established. Selective antegrade cerebral perfusion was performed via cannulation of the innominate artery or axillary artery at a flow rate between 6 and 12 mL/min/kg in order to maintain cerebral perfusion pressure between 40 and 50 mmHg.

Systemic perfusion was restored after anastomosis of the artificial graft to the descending aorta and left subclavian artery. After 5 min of perfusion, rewarming was initiated at approximately 0.5 °C per min. During cooling and rewarming, water and nasopharyngeal temperatures were maintained within 10 ℃ of each other.

Alpha–stat management was used during the cooling and rewarming phases, while pH-stat was applied during hypothermic circulatory arrest and selective antegrade cerebral perfusion. After surgery, patients were transferred to the intensive care unit for respiratory and circulatory support.

Data collection

Pre- and intraoperative data were collected from the electronic medical record system. Arterial and mixed venous blood samples were collected intermittently at predefined time points: (1) at each target nasopharyngeal temperature during cooling (36, 32, 28, 24 ℃) and rewarming (24, 28, 32, 36 ℃), and (2) every 6 h for the first 24 postoperative hours in the intensive care unit. Blood gas analysis was performed immediately using a cobas b 123 < 4> POC system (Roche Diagnostics GmbH, Mannheim, Germany), which provided measurements of hemoglobin, arterial oxygen saturation (SaO₂), SvO₂, arterial oxygen partial pressure (PaO₂), mixed venous oxygen partial pressure (PvO₂), and lactate concentration. At the same time points, data on pump flow, bladder temperature and rScO₂ were recorded.

DO₂I during bypass, in the units of ml/min/m², was computed with the formula [14].

1

while the index of oxygen consumption (VO₂I), with the same units, was calculated with the formula.

2

Finally, their ratio was calculated to yield oxygen extraction rate (O₂ER):

3

These parameters were calculated off-line, and they were not used to guide intraoperative perfusion management (i.e., no goal-direct perfusion protocol was applied). Data were recorded only during aortic cross-clamping, because the calculation of DO₂I and VO₂I [Eqs. (1) and (2)] assumes that total systemic flow equals the pump flow, an assumption that is valid only when the aorta is cross-clamped (the heart does not contribute to forward flow). The standardized clinical protocols for anesthesia, perfusion, and surgical management at our institution can be available upon request. This observational study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Study outcomes

The primary outcome was the association between intraoperative oxygen metabolism parameters and postoperative hyperlactatemia, and the analyses were exploratory, aiming to identify potential critical physiological phases rather than solely testing a single pre-specified primary hypothesis. Hyperlactatemia was defined as peak serum lactate concentration ≥ 4 mmol/L within 24 after surgery, as this threshold has been shown to be independently associated with increased morbidity and mortality in cardiac surgery [15, 16]. As an exploratory analysis, the discriminative ability of variables for hyperlactatemia was also assessed using the area under the receiver operating characteristic curve (AUC).

Secondary outcomes were acute kidney injury, defined as ≥ 50% increased (or an absolute increase of ≥ 26.4 µmol/L) in creatinine level in serum from baseline during the first 48 h after surgery [17]; duration of mechanical ventilation, stay in the intensive care unit, and overall hospitalization; all-cause mortality during hospitalization or within 30 days after surgery; cardiac dysfunction, defined as an postoperative ejection fraction < 50% during hospitalization; stroke, defined as new-onset global or focal neurological deficits confirmed by cerebral computed tomography or magnetic resonance imaging; delirium, diagnosed using the conventional Confusion Assessment Method or the same method adapted to the intensive care unit; and pulmonary complications, which were defined as described and included respiratory infection, or failure, bronchospasm, atelectasis, pleural effusion, pneumothorax, and aspiration pneumonitis [18].

Statistical analysis

Statistical analyses were performed using SPSS version 17.0 (IBM, Chicago, IL, USA), GraphPad Prism 7.0 (GraphPad Software, USA), and PASS 15.0 software. Continuous data were presented as mean ± standard deviation (SD) if they showed a normal distribution based on Kolmogorov-Smirnov test, or as median (interquartile range, IQR) otherwise. Associations of variables with hyperlactatemia were explored using Pearson’s or Spearman’s correlation after assessing the linearity and normality of the data distributions. Scatterplots were visually inspected for influential outliers. Inter-group differences in continuous variables were assessed for significance using Student’s t test or Welch’s unequal variances t test for normally distributed data; otherwise, the Mann-Whitney U test was used. Categorical data were reported as n (%), and inter-group were assessed for significance using the χ² or Fisher’s exact test. As all inter-group comparisons across multiple temperature points or multiple parameters were exploratory and unadjusted, aiming to identify potential critical physiological phases not only to test a single pre-specified primary hypothesis, the observed differences should be interpreted as preliminary associations rather than independent or causal effects. For the key correlation analysis between peak lactate and O₂ER at different temperatures, we additionally applied false discovery rate (FDR) correction using the Benjamini-Hochberg method to control for type I error. Additionally, the ability of variables to identify hyperlactatemia was also assessed in terms of the area under the receiver operating characteristic curve (AUC). Differences that were associated with p < 0.05 were considered statistically significant. However, given the exploratory nature of the study, p-values should be interpreted descriptively, and all findings are considered hypothesis-generating.

In this exploratory observational study, sample size was informed by a pilot data (9 patients, 5 with postoperative hyperlactatemia; O₂ER difference at 24 °C during cooling: 0.03 ± 0.024 [pooled SD], Cohen’s d = 1.25). For a two-group comparison with 80% power at a α = 0.05 and 10% attribution, we aimed for 11 per group. Given an expected hyperlactatemia incidence of ~ 50%-60%, this required a total of approximately 22 patients. To further enhance statistical stability for exploratory analyses across temperatures, we ultimately enrolled 44 consecutive patients.

Results

A total of 44 consecutive patients were initially enrolled. Three patient were excluded (two did not undergo hypothermic circulatory arrest, one had missing primary outcome data), leaving 41 patients in the final analysis (Supplementary Fig. 1). The 41 patients (33 men) had a mean age of 51.85 ± 10.60 year and underwent cooling for 56.33 ± 14.04 min and rewarming for 112.19 ± 49.99 min. A total of 24 patients (58.54%) experienced hyperlactatemia. No significant differences were observed between the groups with respect to preoperative characteristics, extent of dissection, surgical duration, cardiopulmonary bypass time, circulatory arrest time, cooling time, rewarming time, nadir nasopharyngeal/bladder temperatures, perfusion strategy, transfusion requirement, or other perioperative variables (Table 1, Supplementary Table 1). After surgery, 11 patients (26.83%) experienced acute kidney injury, 16 (39.02%) experienced delirium and 24 (58.54%) experienced pulmonary complications. Three patient died two weeks after surgery. No significant difference in secondary outcomes was detected between groups (Table 2).

Table 1.

Clinicodemographic and intraoperative characteristics of enrolled patients

Characteristic	All patients (N = 41)	No hyperlactatemia (n = 17)	Hyperlactatemia (n = 24)	P
Age, yr	51.85 ± 10.60	51.59 ± 9.26	52.33 ± 11.62	0.714
Female	8 (19.51)	4 (23.53)	4 (16.67)	0.698
Body weight, kg	71.87 ± 13.73	71.08 ± 14.90	72.43 ± 13.14	0.761
Height, cm	169.95 ± 9.45	168.70 ± 10.42	170.83 ± 8.86	0.484
Body surface area, m2	1.80 ± 0.21	1.78 ± 0.23	1.81 ± 0.20	0.661
Smoking	16 (39.02)	6 (35.29)	10 (41.67)	0.764
Drinking	14 (34.14)	6 (35.29)	8 (33.33)	1.000
Hypertension	39 (95.12)	17 (100)	22 (91.67)	0.502
Diabetes	25 (60.97)	11 (64.70)	14 (58.33)	0.806
Congestive heart failure	6 (14.63)	2 (11.76)	4 (16.67)	1.000
Arrhythmia	7 (17.07)	3 (17.64)	4 (16.67)	1.000
COPD	0	0	0	1.000
Pleural effusion	5 (12.19)	2 (11.76)	3 (12.5)	1.000
Ischemic stroke	4 (9.76)	2 (11.76)	2 (8.33)	1.000
Hemorrhagic stroke	0	0	0	1.000
Renal failure	6 (14.63)	2 (11.76)	4 (16.67)	0.685
EuroScore II, %	6.24 ± 2.83	5.33 ± 3.42	6.92 ± 2.19	0.212
Ejection fraction < 50%	7 (17.07)	3 (17.64)	4 (16.67)	1.000
Cardiac tamponade	6 (14.63)	2 (11.76)	4 (16.67)	0.685
Malperfusion	25 (60.97)	10 (58.82)	15 (62.5)	0.768
Hemoglobin, mg/dL	129.86 ± 20.23	129.78 ± 25.42	129.92 ± 16.55	0.988
Glucose, mg/L	8.02 ± 2.39	8.11 ± 2.56	7.95 ± 2.37	0.881
BNP, pg/dL	249 (135–653)	240 (132–620)	306 (154–1169)	0.477
Lactate at baseline, mmol/L	1.15 (1–1.23)	1 (1–1.1)	1.2 (1.18–1.43)	0.508
Preoperative medication
β blocker	14 (34.14)	4 (23.52)	10 (41.67)	0.321
Calcium channel blocker	3 (14.29)	2 (22.22)	1 (8.33)	1.000
ACEI	0	0	0	1.000
Type of systemic perfusion
Trans-femoral	18 (43.90)	7 (41.18)	11 (45.3)	0.769
Trans-aorta	9 (21.95)	4 (23.53)	5 (20.83)	0.723
Trans-axillary	14 (34.14)	6 (35.29)	8 (33.33)	1.000
Duration (min) of
Cardiopulmonary bypass	205.00 ± 55.67	187.56 ± 53.45	218.08 ± 55.88	0.222
Cross-clamping	149.14 ± 48.54	130.33 ± 50.63	163.25 ± 43.73	0.127
Cooling	56.33 ± 14.04	54.00 ± 9.64	58.08 ± 16.82	0.523
Rewarming	112.19 ± 49.99	111.89 ± 57.03	112.42 ± 46.67	0.981
ACP	28.14 ± 5.42	26.78 ± 4.74	29.17 ± 5.87	0.331
Hypothermic circulatory arrest	28.14 ± 5.42	26.78 ± 4.74	29.17 ± 5.87	0.331
Surgery	7.11 ± 1.54	6.92 ± 1.82	7.25 ± 1.36	0.640
Nadir temperatures, ℃
Nasopharyngeal	23.44 ± 1.05	23.13 ± 1.38	23.68 ± 0.69	0.252
Bladder	25.71 ± 0.94	25.47 ± 0.98	25.89 ± 0.903	0.317
Red blood cell infusion	21 (51.21)	7 (41.17)	14 (58.33)	0.350
Urine output, ml	1626.67 ± 1095.39	1906.67 ± 1433.14	1416.67 ± 758.69	0.322

Values are n (%), mean ± SD, or median (interquartile range)

ACEI, angiotensin-converting enzyme inhibitor; ACP, antegrade cerebral perfusion; COPD, chronic obstructive pulmonary disease; BNP, B-type natriuretic peptide

Table 2.

Comparison of secondary outcomes between patients who experienced hyperlactatemia or not

Outcome	All patients (N = 41)	No hyperlactatemia (n = 17)	Hyperlactatemia (n = 24)	P
Acute kidney injury	11 (26.83)	3 (17.65)	8 (33.33)	0.324
Arrhythmia	3 (7.32)	1 (5.88)	2 (8.33)	1.000
Cardiac dysfunction	16 (39.02)	6 (35.29)	10 (41.67)	0.764
Delirium	16 (39.02)	6 (35.29)	10 (41.67)	0.764
Duration of
Mechanical ventilation, h	22.38 (13.05–51.33)	13.25 (12.08–37.72)	33.28 (14.68–61.83)	0.155
Intensive care unit, d	4 (2–6)	4 (2–4)	5 (3–7)	0.587
Hospitalization, d	11 (8–14)	9 (8–15)	11 (8–13)	0.616
Hemorrhagic stroke	0	0	0	1.000
Ischemic stroke	6 (14.63)	2 (11.76)	4 (16.67)	1.000
Mortality
In-hospital	0	0	0	1.000
30-day	3 (7.32)	1 (5.88)	2 (8.33)	1.000
Paraplegia	0	0	0	1.000
PPC	24 (58.54)	8 (47.06)	16 (66.67)	0.335
Reoperation	6 (14.63)	2 (11.76)	4 (16.67)	1.000

Values are n (%) or median (IQR). PPC, postoperative pulmonary complication

Among all patients, the serum lactate level rose progressively throughout the perioperative period. From 1.5 (1.3–2.2) mmol/L at 36 ℃, the level increased slightly to 2.0 (1.5–2.6) mmol/L immediately before hypothermic circulatory arrest. After 29.17 ± 5.87 min of arrest, the level surged to 4.0 (3.6–5.1) mmol/L after restoration of lower body circulation. The level remained stable during the subsequent rewarming and peaked at 5.0 (2.9–7.1) mmol/L within the first 24 h after surgery (Fig. 1). The levels at the various temperatures during rewarming were significantly higher than those at the same temperatures during cooling.

Fig. 1.

Levels of lactate in serum, oxygen delivery index (DO₂I), oxygen consumption index (VO₂I) and oxygen extraction rate (O₂ER) during cooling and rewarming phases of cardiopulmonary bypass involving hypothermic circulatory arrest (HCA). * p < 0.05

DO₂I, VO₂I, and O₂ER decline progressively during cooling and increased gradually during rewarming. All three parameters were significantly higher during rewarming than cooling at the same target temperature, though this was not true for DO₂I at 28 ℃ or O₂ER at both 28 ℃ and 36 ℃. DO₂I and VO₂I correlated positively with each other during cooling phase and during rewarming at 28, 32 and 36 ℃, with correlation coefficients ranging from 0.40 to 0.61 (all p < 0.05; Fig. 2).

Fig. 2.

Analysis of linear correlations between oxygen delivery index (DO₂I) and oxygen consumption index (VO₂I) at the indicated temperatures during the cooling or rewarming phases of cardiopulmonary bypass involving hypothermic circulatory arrest

Among the exploratory correlations examined across all temperature points and phases, a moderate positive association was observed between peak lactate level within 24 h after surgery and O₂ER at 24 ℃ during cooling (Spearman’s ρ = 0.67, p = 0.001; Fig. 3), and a moderate negatively association was observed with SvO₂ at the same time point (ρ = -0.68, p < 0.001; Fig. 4). After Benjamini-Hochberg FDR correction for the correlations examined across temperatures, these associations remained significant at q < 0.01. In addition, the duration of hypothermic circulatory arrest and the peak lactate level showed no significant linear association (ρ = 0.316, p = 0.163).

Fig. 3.

Analysis of linear correlations between oxygen extraction rate (O₂ER) at the indicated temperatures during the cooling or rewarming phases and peak lactate level in serum within 24 h after surgery

Fig. 4.

Analysis of linear correlations between mixed venous oxygen saturation (SvO₂) at the indicated temperatures during the cooling and rewarming phases and peak lactate levels in serum within 24 h after surgery

Compared with the non-hyperlactatemia group, the hyperlactatemia group had significantly higher VO₂I and O₂ER at 24 ℃ during cooling, and significantly lower SvO₂ at the same time point (Table 3). Blood glucose concentration, bladder temperature, and rScO₂ at a given target temperature during cooling or rewarming did not differ significantly between patients who experienced hyperlactatemia or not (Supplementary Table 2).

Table 3.

Comparison of parameters of oxygen consumption and extraction between patients who experienced hyperlactatemia or not

Temp, ℃	Phase during bypass	No hyperlactatemia (n = 17)	Hyperlactatemia (n = 24)	P
Oxygen consumption index (VO 2 I), ml/min/m 2
36	Cooling	78.86 ± 14.90	84.00 ± 17.27	0.483
	Rewarming	88.14 ± 20.84	109.32 ± 18.06	0.022*
32	Cooling	57.92 ± 13.02	56.23 ± 12.39	0.765
	Rewarming	75.79 ± 10.19	85.80 ± 18.44	0.160
28	Cooling	51.89 ± 31.82	45.40 ± 10.35	0.514
	Rewarming	57.70 ± 6.75	63.37 ± 14.00	0.278
24	Cooling	29.20 ± 6.12	36.97 ± 5.98	0.009*
	Rewarming	48.21 ± 12.74	53.55 ± 16.91	0.439
Oxygen extract rate (O 2 ER)
36	Cooling	0.24 ± 0.05	0.24 ± 0.04	0.914
	Rewarming	0.25 ± 0.06	0.28 ± 0.04	0.114
32	Cooling	0.18 ± 0.06	0.17 ± 0.03	0.793
	Rewarming	0.23 ± 0.03	0.24 ± 0.04	0.441
28	Cooling	0.18 ± 0.13	0.15 ± 0.03	0.462
	Rewarming	0.21 ± 0.03	0.20 ± 0.03	0.449
24	Cooling	0.10 ± 0.02	0.13 ± 0.02	0.006*
	Rewarming	0.21 ± 0.09	0.20 ± 0.09	0.913
Mixed venous oxygen saturation (SvO2), %
36	Cooling	79.94 ± 4.82	79.84 ± 5.03	0.963
	Rewarming	80.23 ± 5.20	77.73 ± 3.92	0.223
32	Cooling	86.36 ± 7.06	84.69 ± 3.61	0.489
	Rewarming	79.94 ± 3.42	80.11 ± 4.60	0.929
28	Cooling	85.01 ± 14.28	86.72 ± 4.15	0.697
	Rewarming	81.34 ± 4.08	82.98 ± 4.48	0.401
24	Cooling	93.17 ± 1.92	89.18 ± 3.07	0.003*
	Rewarming	78.54 ± 12.02	79.89 ± 11.47	0.797

Values are mean ± SD, * p < 0.05

In exploratory ROC analyses, O₂ER at 24℃ during cooling discriminated between patients with and without postoperative hyperlactatemia with an AUC of 0.83 (95% confidence interval 0.69–0.97). A cut-off O₂ER > 0.102 yielded a sensitivity of 91.7% and specificity of 82.4%. Similarly, VO₂I at 24 ℃ during cooling showed an AUC of 0.80 (95% confidence interval 0.64–0.96); a cut-off > 27.98 ml/min/m² gave a sensitivity of 87.5% and specificity of 71.6% (Fig. 5).

Fig. 5.

Discrimination of hyperlactatemia in our sample based on oxygen extraction rate (O₂ER) or oxygen consumption index (VO₂I) at 24 ℃ during the cooling phase of cardiopulmonary bypass involving hypothermic circulatory arrest. Receiver operating characteristic curves are shown, together with their exploratory diagnostic performance

To explore whether oxygen metabolism parameters were associated with secondary clinical outcomes, we compared the most relevant parameter, O₂ER at 24 ℃ during cooling, between patients who did and did not develop each secondary outcome. O₂ER at 24 ℃ during cooling was numerically lower in patients who developed acute kidney injury (0.11 ± 0.02 vs. 0.13 ± 0.02, p = 0.210), delirium (0.11 ± 0.02 vs. 0.12 ± 0.02, p = 0.287), and pulmonary complications (0.11 ± 0.02 vs. 0.12 ± 0.02, p = 0.455), but these differences did not reach statistical significance. Given the small number of events, these findings should be interpreted cautiously, and larger studies are needed to asses the relationship between oxygen metabolism parameters and clinical outcomes.

Discussion

Here we comprehensively characterized clinical parameters of oxygen metabolism during the cooling and rewarming phases of hypothermic circulatory arrest in patients undergoing total aortic arch replacement. We detected a path or history dependence of oxygen metabolism: DO₂I, VO₂I, and O₂ER at a given temperature were higher during rewarming than during cooling. And we detected a potentially important association of hyperlactatemia risk with O₂ER and VO₂I at 24 ℃ during cooling.

Our study provided the first preliminary indication that optimizing oxygen metabolism during cardiopulmonary bypass should take into account not only current temperature but also the direction of temperature change [19]. Previous studies reported desirable cut-off values for DO₂I of 340 mL/min/m² for normothermic bypass or 262–300 mL/min/m² for mild hypothermic bypass, but these studies did not examine cooling or rewarming phases separately [20–22]. Our work suggests that an oxygen delivery-consumption imbalance may occur during cooling phase, followed by compensatory increases in metabolic demand during rewarming. As a result, VO₂I, O₂ER and DO₂I are significantly higher during rewarming than cooling at the same temperature. Thus, perfusion strategies might benefit from being tailored separately for the cooling and rewarming phases.

Such optimization may need to pay special attention to oxygen metabolism at 24 ℃ during cooling involving hypothermic circulatory arrest. At this point, O₂ER correlated positively (ρ = 0.67, p = 0.001) and SvO₂ correlated negatively (ρ = -0.68, p < 0.001) with peak lactate level within 24 h after surgery. And O₂ER and VO₂I were observed to be significantly higher, in the hyperlactatemia group compared to the no-hyperlactatemia group. These may reflect an ineffective extraction process at 24 ℃ during cooling, despite adequate oxygen delivery [23]. Several speculative mechanisms could potentially explain why this temperature emerged as relevant in our exploratory analysis, but these require direct investigation in future studies. One possible explanation is that the activity of mitochondrial cytochrome c oxidase (Complex IV), which catalyzes > 90% of cellular oxygen consumption, falls by > 60% below 28℃ and reaches its minimum at 22–25 ℃ [24–26]. A second hypothesized contributor is that hypothermia-induced leftward shift of the oxygen-hemoglobin dissociation curve may further compromises tissue oxygen releasing [27]. Third, it is plausible that hypothermia and systemic inflammatory response may lead to microcirculatory disturbance, resulting in a state of relatively low SvO₂ but high oxygen extraction while tissue hypoxia persists [28, 29]. However, due to no significant differences were observed in perioperative confounders associated with postoperative lactate levels between groups, we did not perform rigorous multivariable adjustment for these factors. Therefore, based on our exploratory findings, we hypothesize that 24 °C during cooling may represent a temperature at which the balance between oxygen delivery and cellular utilization becomes particularly vulnerable, and this interpretation remains to be validated in larger, hypothesis-driven cohorts. Direct measurements of mitochondrial function, microcirculatory imaging, and oxygen-hemoglobin dissociation kinetics are needed to evaluate these hypotheses. Additionally, our findings suggest the potential value of continuous O₂ER monitoring. However, any derived cut-off values (e.g., an O₂ER of 0.102) should not be intended to guide clinical decisions until validated in large prospective studies.

In the study, we used alpha-stat management during cooling/rewarming and pH-stat during hypothermic circulatory arrest. Alpha-stat measures blood gases at 37 °C without temperature correction, which might overestimate in vivo oxygen availability during hypothermia and could affect absolute DO₂I/VO₂I values. Therefore, caution is warranted when directly comparing our absolute values with studies that used pH-stat. Nevertheless, it is possible that because the same alpha-stat convention was applied consistently across all time points, the relative phase differences and hyperlactatemia associations may be robust. Future work is warranted to explore the impact of alpha-stat versus pH-stat on oxygen metabolism parameters.

Given the exploratory nature of our findings, any clinical implications remain preliminary. Nevertheless, our results generate the hypothesis that cooling beyond 28 °C might increase the risk of inefficiency in oxygen utilization, and that moderately hypothermic circulatory arrest (e.g., 26–28 °C) with adjunctive cerebral perfusion could be worthy of investigation as a strategy to balance cerebral protection and systemic aerobic metabolism [30, 31]. Similarly, techniques such as lower-body retrograde perfusion or efforts to minimize the duration of hypothermic circulatory arrest warrant evaluation in future physiological studies or randomized trials [32–34]. Additionally, our findings suggest the potential value of continuous O₂ER monitoring during cooling, but any derived cut-off values (e.g., an O₂ER of 0.102) should be interpreted strictly. At present, no changes to clinical practice can be recommended based solely on this exploratory study until validated in large prospective studies.

Hyperlactatemia is a reliable sensitive indicator of imbalance between oxygen delivery and consumption, and it is a useful warning sign of organ hypoperfusion and postoperative complications [35, 36]. Hyperlactatemia need not always translate into postoperative complications and poor prognosis. Indeed, we did not detect significant differences in mechanical ventilation duration, morbidity or mortality between our patients who developed hyperlactatemia or not. Likewise, we observed no significant relationship between oxygen metabolism parameters and clinical outcomes in our exploratory analyses. However, the small sample size limited our ability to perform adjusted multivariable regression analyses for potential confounders (e.g. the extent of the aortic dissection, cardiopulmonary bypass time, or preoperative comorbidities) between the groups. Future work should verify its prognosis value through comprehensive analysis of postoperative outcomes in larger samples with longer follow-up.

This study has several limitations. First, the multiple exploratory comparisons without adjustment in this small sample size may increase the possibility of Type I errors. Second, as a single-center study employing standardized perfusion protocol in a physiologically complex and unstable acute aortic dissection cohort, our findings may lack generalizability to institutions utilizing alternative strategies or to more stable populations, thereby constraining extent validity. Third, the exploratory observational design precludes causal inference and cannot fully account for unadjusted confounders. Fourth, lactate measurement was restricted to peak values, without assessment of clearance kinetics, which is a key dynamic prognostic marker. Our analysis was also limited to the single threshold ≥ 4 mmol/L for defining hyperlactatemia. Also, central venous lactate provides an integrative measure of systemic metabolism and may be insensitive to incipient regional organ ischemia. Fifth, all mechanistic interpretations regarding mitochondrial dysfunction, microcirculatory alterations, and oxygen-hemoglobin dissociation are speculative, as we did not perform direct measurements of these processes. Finally, the exploratory nature of the analyses means that the proposed cut-off values for O₂ER or VO₂I should not be used to guide clinical decisions until validated in independent prospective studies.

In conclusion, in this exploratory study, a relatively higher O₂ER at 24 °C during cooling appeared to be associated with postoperative hyperlactatemia, and that at a cut-off value > 0.102, O₂ER discriminated for hyperlactatemia with an AUC of 0.83 (95% CI: 0.69–0.97), a sensitivity of 91.7%, and a specificity of 82.4%. Oxygen metabolism may differ during the cooling from rewarming phases of cardiopulmonary bypass involving hypothermic circulatory arrest. These preliminary findings warrant validation in larger prospective studies.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

DO₂I

Index of oxygen delivery

SpO₂

Pulse oxygen saturation

rScO₂

Regional cerebral oxygen saturation

SvO₂

Mixed venous oxygen saturation

SaO₂

Arterial oxygen saturation

PaO₂

Arterial oxygen partial pressure

PvO₂

Mixed venous oxygen partial pressure

VO₂I

Index of oxygen consumption

O₂ER

Oxygen extraction rate

IQR

Interquartile range

AUC

Area under the curve

ATP

Adenosine triphosphate

Author contributions

LP, WW, and DL have given substantial contributions to theconception or the design of the manuscript, JYX to acquisition, analysis and interpretation of the data. All authors participated to draft the manuscript, LP and DL revised it critically. All authors read and approved the manuscript.

Funding

This study was conducted without specific funding from public or private sources.

Data availability

The datasets used and analyzed during this study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of West China Hospital, Sichuan University (protocol 20241686). Written informed consent was obtained. All procedures involving human participants were conducted in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.