Authors: Gyeo Ra Lee, Eun Young Kim
Categories: Article, abdominal surgery, micro-circulation, postoperative prognosis, ScvO2, venous-to-arterial carbon dioxide partial pressure difference
Source: Biomedicines
Authors: Gyeo Ra Lee, Eun Young Kim
Purpose: The venous-to-arterial carbon dioxide partial pressure difference [P(v-a)CO2] reflects the adequacy of tissue perfusion, with elevated values suggesting impaired clearance of CO2. While its prognostic role has been investigated in septic shock and high-risk surgery, evidence in postoperative critically ill patients remains limited. This study aimed to evaluate the prognostic value of ΔP(v-a)CO2 after major abdominal surgery and its relationship with microcirculatory markers. Methods: We retrospectively analyzed 86 patients admitted to the intensive care unit (ICU) after major abdominal surgery between September 2020 and October 2023. Arterial and central venous blood gas analyses were performed immediately postoperatively and at 24 h. Patients were stratified into groups according to ΔP(v-a)CO2 (≤ 0 vs. >0). Postoperative outcomes and correlations with central venous oxygen saturation (ScvO2) were assessed. Results: In the subgroup analysis of patients with an initial P(v-a)CO2 > 6 mmHg, those in the ΔP(v-a)CO2 > 0 group required mechanical ventilation (54.5% vs. 22.2%, p = 0.033) and continuous renal replacement therapy (36.4% vs. 8.9%, p = 0.020) more frequently, with longer durations of both interventions (p = 0.011 and p = 0.016, respectively). ICU length of stay and the incidence of acute kidney injury were significantly lower in the ΔP(v-a)CO2 ≤ 0 group. In addition, a modest negative correlation was observed between ScvO2 measured at 24 h postoperatively and ΔP(v-a)CO2. Conclusions: ΔP(v-a)CO2 may serve as a useful marker for postoperative risk stratification in critically ill patients undergoing major abdominal surgery. However, given the retrospective design, small sample size, and single-center setting, these findings should be considered hypothesis-generating and require confirmation in larger, prospective multicenter studies.
The primary objective of hemodynamic resuscitation in patients admitted to the intensive care unit (ICU) post-surgery is the normalization of reduced microcirculation due to intraoperative hypoperfusion or tissue injury, to meet the oxygen and metabolic demands of major organs. Although blood flow to vital organs is typically maintained by the host’s auto-regulatory mechanisms, it becomes compromised in shock or clinically deteriorating conditions. In these cases, maintaining appropriate blood flow to major organs is challenging, thus making the accurate assessment and monitoring of blood flow status crucial. Throughout the resuscitation process, the restoration of impaired microcirculation is mainly assessed indirectly through macrohemodynamic variables such as mean arterial pressure or stroke volume, which refers to the amount of blood ejected by the left ventricle with each heartbeat and reflects the effectiveness of cardiac output. However, these variables do not adequately represent impaired microcirculation, which may result from factors such as increased blood viscosity, endothelial dysfunction, or microthrombi formation. It should also be considered that in certain cases, particularly following excessive fluid administration, hemodilution may occur, potentially offsetting the impact of viscosity-related impairment. Moreover, fluid resuscitation based on incorrect parameters or a high blood pressure target above a mean arterial pressure (MAP) of 65 mmHg could lead to an inappropriate positive fluid balance. Previous studies by Opsina et al. [1] and Pottecher et al. [2] have shown that responses to fluid administration do not directly correlate with effects on changes in systemic circulation status. Therefore, it is essential to use appropriate markers that more accurately reflect the status of microcirculation in critically ill patients, particularly in those at risk of tissue or endothelial injury due to surgical stress or sepsis-related microthrombi formation. The venous-to-arterial carbon dioxide tension difference [P(v-a)CO2] refers to the difference in the partial pressure of carbon dioxide (CO2) between venous and arterial gas analyses. P(v-a)CO2 is primarily an indicator of the adequacy of cardiac output relative to CO2 production, rather than a pure marker of microcirculation. When cardiac output is low, CO2 clearance decreases and venous PCO2 rises relative to arterial PCO2, a mechanism well explained by Fick’s principle for CO2 [3,4]. This parameter has been shown to be associated with mortality, organ dysfunction, and adverse postoperative outcomes in high-risk surgical patients. The usual pathological threshold is approximately 6 mmHg, and many studies have used values above this threshold as a sign of hypoperfusion or inadequate flow, with poor prognosis in septic shock and high-risk surgical populations [3,4,5,6]. In addition, ScvO2 measurement supplemented with lactate has been recommended to uncover residual hypoperfusion, and recent reviews [3,7] have highlighted the complementary role of CO2-derived indices in guiding resuscitation and risk assessment in critically ill patients. Importantly, in major abdominal surgery, the gap in P(v-a)CO2 has been shown to correlate with microcirculatory markers, and persistent elevation reflects inadequate perfusion despite apparently normal macrohemodynamic parameters. Several studies have demonstrated that P(v-a)CO2 is not only a physiological marker of CO2 clearance but also provides prognostic information regarding organ dysfunction and survival in surgical and critically ill populations. We have earlier reported that a P(v-a)CO2 value of 8.6 mmHg or higher, measured at 24 h post-surgery, significantly predicts 7-day mortality following major surgery in critically ill patients [8]. However, no studies have reported the relationship between changes in P(v-a)CO2 measured after surgery and postoperative patient prognosis. Thus, the present study was designed to evaluate the prognostic significance of changes in P(v-a)CO2 among critically ill patients after abdominal surgery, and to explore the relationship between ΔP(v-a)CO2 and other conventional microcirculatory markers, such as ScvO2, in order to better understand its potential role in postoperative risk stratification.
From September 2020 to October 2023, an observational study was conducted in the 22-bed ICU of a single tertiary hospital, and the data were retrospectively reviewed. All patients aged 18 years or older, admitted to the ICU after abdominal surgery under general anesthesia, were included, regardless of the surgical method (including laparotomy, laparoscopic surgery, or robotic surgery). We retrospectively included consecutive patients meeting these criteria, without additional exclusion related to the indication for catheterization, since both arterial and central venous catheters are inserted routinely as part of standard postoperative management in our ICU. Of the 86 patients included, 55 underwent elective or scheduled surgeries, and 31 underwent emergency surgeries. For the emergency surgery group, preoperative management was performed according to the Surviving Sepsis Campaign guidelines, including adequate hemodynamic resuscitation and timely administration of broad-spectrum antibiotics, in order to minimize the risk of insufficient preoperative preparation. In our ICU, invasive arterial and central venous catheter monitoring is routinely performed in all patients admitted after major abdominal surgery, regardless of intraoperative course or risk profile. Accordingly, all patients in this study had both a central venous catheter and an arterial line in place, followed by venous blood gas analysis (VBGA) and arterial blood gas analysis (ABGA) within 1 h (T0) and 24 h (T1) after ICU admission, respectively. Only patients with complete paired ABGA and VBGA results at both T0 and T1 were included in the final analysis to ensure internal consistency. As a result, patients without complete gas analysis data were excluded, which may introduce a risk of selection bias. However, since invasive arterial and central venous catheter monitoring is a routine practice in our ICU, we believe that this selection reflects standard clinical practice rather than selective catheter placement for research purposes. We retrospectively reviewed and analyzed data from patients who had complete ABGA and VBGA results at both T0 and T1. Patients were excluded if they met any of the following (1) under 18 years of age; (2) pregnancy; (3) death within 48 h after ICU admission, which made it impossible to collect data at both T0 and T1; (4) receipt of extracorporeal membrane oxygenation support therapy; (5) lack of an arterial or venous catheter owing to vessel stricture or thrombosis; (6) lack of venous or arterial gas analysis results at either T0 or T1; (7) preoperative ventilator care; (8) severe pulmonary dysfunction making general anesthesia unfeasible. Participants were divided into ΔP(v-a)CO2 ≤ 0 group and ΔP(v-a)CO2 > 0 group based on ΔP(v-a)CO2 variation, and demographic characteristics, fluid balance, hemodynamic values, and clinical outcomes were compared. A reference value of ≤6 mmHg for P(v-a)CO2 was adopted based on prior studies [3,4]. Although ROC analysis was conducted using P(v-a)CO2 values at T0 to assess the predictive value for postoperative ventilator care, the analysis did not yield a statistically significant cutoff for broader clinical outcomes such as mortality or complications, likely due to the limited sample size. However, the derived curve demonstrated a trend consistent with previous studies, including our prior work, thereby supporting the clinical relevance of the commonly referenced 6 mmHg threshold.
Accordingly, we performed subgroup analyses using this widely accepted cutoff value to ensure comparability with the existing literature and to explore potential associations between P(v-a)CO2 and clinical outcomes within our cohort. Consequently, participants were categorized into two groups based on their initial P(v-a)CO2 low P(v-a)CO2 (≤6 mmHg) and high P(v-a)CO2 (>6 mmHg), and a subgroup analysis was conducted on these groups. In addition, for practical interpretation in line with our primary objective, patients were stratified into two groups (Δ ≤ 0 vs. Δ > 0), which allowed direct assessment of outcome differences according to the direction of change. This study received approval and monitoring from the Institutional Review Board of The Catholic University of Korea, Seoul St. Mary’s Hospital (No. IRB; KC24RISI0743) and adhered to the Declaration of Helsinki and its amendments.
The venous partial pressure of CO2 (PCO2) can be measured by obtaining mixed venous blood via a pulmonary artery catheter or central venous blood through a central venous catheter. However, pulmonary artery catheters are rarely utilized in our ICU, and previous studies [5,6] have demonstrated consistent results between the central venous-arterial PCO2 differences and the mixed venous-arterial PCO2 differences. Consequently, P(v-a)CO2 was calculated using PCO2 from the central venous blood in the current study. In our ICU, invasive arterial and central venous monitoring is performed as a routine practice for all patients requiring postoperative intensive care, according to institutional protocols. All participants were equipped with a central venous catheter and an arterial line either before or immediately upon admission to the ICU, and blood gas analysis was performed within 1 h (T0) and 24 h (T1) after admission. The central venous catheter was inserted via the Seldinger technique [7] after the internal jugular vein was visualized by ultrasound. Following the catheter insertion, a physician verified that the tip was positioned in the superior vena cava or upper right atrium by X-ray. The arterial lines were placed using a 22-gauge catheter (BD Angiocath Plus^®^, Becton Dickison Medical, Franklin Lakes, NJ, USA) in either the dorsalis pedis or radial arteries using aseptic techniques. Should securing an arterial line in these locations prove impossible, a 4-French catheter was placed in the femoral artery using the Seldinger technique. All ABGA and VBGA results were acquired using a point-of-care gas analyzer (ABL800 FLEX^®^, Radiometer, Copenhagen, Denmark) throughout the entire study period, thereby avoiding inter-device variability. To minimize bias affecting the results, the interval between the VBGA and ABGA was kept under 5 min, in accordance with our institutional protocol. All syringes used for blood sampling were pre-heparinized with the same balanced heparin solution, and blood gas analysis was performed immediately after collection to minimize temperature- and anticoagulant-related variability. P(v-a)CO2 was defined as follows and was calculated.P(v-a)CO2 = pCO2 of central venous blood − pCO2 of arterial blood
Also, ΔP(v-a)CO2 was defined and calculated as follows.ΔP(v-a)CO2 = P(v-a)CO2 measured at T1 − P(v-a)CO2 measured at T0
Clinical data were prospectively recorded in the electronic medical records and retrospectively reviewed for this study. The acquired information included demographics, fluid balances, underlying disease, and disease severity using APACHE II scores at ICU admission. Organ dysfunction was evaluated daily with SOFA scores starting from the time of ICU admission. Daily fluid balance, defined as the difference between total daily intake (including intravenous fluids, enteral fluids, medications, and blood products) and total output (encompassing losses through gastrointestinal, urinary, and other drains), ranged from 6 a.m. the previous day to 6 a.m. on the day of measurement. Nurses consistently recorded and calculated daily fluid balances using a specifically designed document. After conducting VBGA and ABGA, retrospective collection and recording of hemodynamic values such as lactate, central venous oxygen saturation (ScvO2), pH, bicarbonate, and P(v-a)CO2 were performed. The point in time to calculate the delta value is T1, with the delta value being defined as the difference between the values at T1 and T0.
Morbidity during postoperative hospitalization was monitored and recorded to determine clinical outcomes. Morbidity was classified according to the Clavien–Dindo classification, including only complications of grade III or higher in the analysis. Grade III complications were postoperative conditions that required endoscopic, radiological, or surgical intervention, whereas grade IV complications were life-threatening conditions that necessitated ICU management. Furthermore, grade V referred to a patient’s death following surgery. The primary outcome of the study was to determine prognosis differences based on changes in P(v-a)CO2 levels in critically ill patients experiencing an immediate post-abdominal surgery increase in P(v-a)CO2. The secondary outcome assessed the correlation between changes in P(v-a)CO2 and other conventional markers of microcirculation status in critically ill patients.
Acute kidney injury (AKI) was characterized by any of the following criteria based on the Kidney Disease Improving Global Outcomes (1) a urinary output of <0.5 mL/kg/hr for at least 6 h; (2) an increase in serum creatinine of ≥0.3 mg/dL(≥26.5 μmol/L) within 48 h; or (3) an increase in serum creatinine to ≥1.5 times the baseline within the previous 7 days [8].
Defects in the integrity of the anastomosis wall, including suture and staple lines, can lead to communication with extra-luminal compartments. An abscess near the anastomosis site was classified as anastomotic leakage [9].
Bile leakage was defined as a bilirubin concentration in the drain fluid that is at least 3 times higher than the simultaneously measured serum bilirubin concentration on or after postoperative day 3, or by the need for radiologic or surgical intervention due to biliary collections or bile peritonitis [10].
Intra-abdominal fluid collection encompasses both non-abscess abdominal fluid collection and abdominal abscess. A non-abscess intra-abdominal fluid collection is defined as a fluid accumulation measuring ≥3 cm in its largest dimension (length, width, or depth), as identified by computed tomography or ultrasonography, without signs of infection. An abdominal abscess is identified by a fluid collection detected through computed tomography or ultrasonography, with positive cultures from percutaneous drainage or reoperation [11].
Pleural effusion was defined by the presence of interlobar fluid or blunting of the phrenico-costal angle on chest X-ray, based on a clear comparison between preoperative and postoperative images [12]. Chest X-rays were performed with the patient positioned in a semi-upright posture as much as possible. In addition, pleural effusion was routinely assessed using bedside ultrasound, which was performed as part of the institutional protocol for evaluating volume status.
Postoperative pneumonia is characterized by the development of new consolidation, infiltration, or cavitation on postoperative chest X-ray, along with at least one of the (1) fever (>38 °C) without another identified cause, (2) leukocytosis (white blood cell count > 12 × 10^9^ L) or neutropenia (white blood cell count < 4 × 10^9^ L), and at least two of the (1) new or worsening cough, dyspnea, or tachypnea, (2) rales or bronchial breath sounds, (3) increased respiratory secretion, change in sputum character, new onset of purulent sputum, increased suction demand, or (4) gas exchange deterioration [13].
The treatment status and duration of treatment for patients undergoing mechanical ventilation or continuous renal replacement therapy were documented. The length of ICU stay, postoperative stay, and hospital stay were also recorded. Deaths during hospitalization were reviewed, and mortality rates at 7 days, 28 days, and during hospital stay post-surgery were analyzed.
All statistical analyses were conducted using the SPSS statistical package software (version 24.0 for Windows; SPSS, Inc., Chicago, IL, USA). Based on the normal reference range of P(v-a)CO2, participants were divided into low PvaCO2 (≤6 mmHg) and high PvaCO2 (>6 mmHg) groups. We compared demographic characteristics, fluid balance, hemodynamic values, and clinical outcomes between the two groups. Continuous data are presented as mean ± standard deviation, and overall differences were tested using Student’s t-test or analysis of variance. The normal distribution of variables was assessed using the Kolmogorov–Smirnov test. Variables that were not normally distributed were analyzed using the Mann–Whitney test. Categorical variables were calculated using Fisher’s exact test or the Chi-squared (χ^2^) test. The relationship between the ΔP(v-a)CO2 and various parameters measured at T1 and T0 was assessed using partial correlation, linear regression analysis, and Bland–Altman plots. Differences were considered statistically significant for p-values of <0.05. Multivariate logistic regression analysis was performed to identify predisposing factors for 28-day mortality and postoperative mechanical ventilator support using significant variables from the univariate analysis.
During the study period, a total of 1275 patients were admitted to the surgical ICU post-surgery and enrolled in the study. Due to the retrospective design of the study, VBGA was not routinely performed until a certain time point. After its implementation as part of routine clinical practice, most patients from the earlier period were excluded because of missing ABGA/VBGA data. So, 1190 were excluded due to lack of ABGA or VBGA records at T0 or T1. Consequently, 86 patients were analyzed, as shown in Figure 1. The primary outcome of this study was the need for postoperative mechanical ventilation. Secondary outcomes included CRRT use, acute kidney injury (AKI), other postoperative morbidities, ICU/hospital stay, and mortality.
Participants were divided into ΔP(v-a)CO2 ≤ 0 group (n = 49, 57%) and ΔP(v-a)CO2 > 0 group (n = 37, 43%), and their demographic characteristics, fluid balance, hemodynamic values, and clinical outcomes were compared (Table 1).
No significant differences were found in demographics such as age, sex, body mass index (BMI), and underlying diseases between the ΔP(v-a)CO2 ≤ 0 group and the ΔP(v-a)CO2 > 0 group. Disease severity, assessed by the total SOFA score at T0 (3.7 ± 2.9 vs. 2.9 ± 2.8, p = 0.243), and septic shock status at T0 (n = 8, 16.3% vs. n = 2, 5.4%, p = 0.117), were also comparable between the two groups. Similarly, the delta value of the total SOFA score did not differ. Regarding fluid balance, no significant differences were noted in total intake and output amounts on both the day of surgery and the following day. In terms of hemodynamic parameters, there were no significant differences between the two groups in lactate levels at T0, Δlactate at T1, ScvO2 at T0, and ScvO2 at T1. The value of P(v-a)CO2 at T0 was significantly higher in the ΔP(v-a)CO2 ≤ 0 group (10.6 ± 3.8 vs. 4.5 ± 3.3, p < 0.001). However, P(v-a)CO2 at T1 (5.9 ± 2.8 vs. 9.5 ± 4.3, p < 0.001) and ΔP(v-a)CO2 at T1 (−4.6 ± 3.8 vs. 5 ± 4.3, p < 0.001) were significantly lower in the ΔP(v-a)CO2 ≤ 0 group than in the ΔP(v-a)CO2 > 0 group. Regarding primary outcome, in the overall study population of 86 patients who were eligible for analysis, the need for postoperative mechanical ventilation occurred in 19/86 patients (22%), with no significant difference between the two groups (n = 12, 24.5% vs. n = 7/37, 18.9%, p = 0.607). For secondary outcomes, no significant differences were found in CRRT use, AKI, ICU stay, or mortality between the two groups. During the study period, mortality occurred in 5 patients (5.8%): 7-day (n = 1, 1.2%), 28-day (n = 4, 4.7%), and in-hospital (n = 5, 5.8%), without significant group differences.
For subgroup analysis, the analysis was selectively re-performed only in participants with high P(v-a)CO2(>6 mmHg) at T0 (Table 2).
The subjects of analysis were the ΔP(v-a)CO2 ≤ 0 group (n = 45, 80.4%) and ΔP(v-a)CO2 > 0 group (n = 11, 19.6%). There were no significant differences in demographics or disease severity, as represented by the total SOFA score, between the two groups. In terms of hemodynamic parameters, no significant differences were observed in lactate and ScvO2 levels at T0 between the two groups. Furthermore, the proportion of patients who underwent laparoscopic or robotic surgery did not differ significantly between the two groups (12 of 45 patients [26.7%] in the ΔP(v-a)CO2 ≤ 0 group vs. 3 of 11 patients [27.3%] in the ΔP(v-a)CO2 > 0 group). Among these patients, P(v-a)CO2 values at T0 (10.6 ± 3.6 vs. 10.4 ± 2.7 mmHg, p = 0.845) and T1 (7.6 ± 3.9 vs. 7.0 ± 5.0 mmHg, p = 0.652), as well as ΔP(v-a)CO2 (−3.0 ± 5.3 vs. −3.4 ± 5.8 mmHg, p = 0.818), showed no significant differences between groups. However, lactate levels at T1 (2.5 ± 1.6 vs. 4.8 ± 5.2, p = 0.011) and Δlactate at T1(−0.9 ± 2.4 vs. 1.5 ± 2.8, p = 0.006) were significantly lower in the ΔP(v-a)CO2 ≤ 0 group than in the ΔP(v-a)CO2 > 0 group. The initial P(v-a)CO2 value at T0 was significantly higher in the ΔP(v-a)CO2 ≤ 0 group (11 ± 3.5 vs. 8.6 ± 1.8, p < 0.036). However, P(v-a)CO2 at T1(6.2 ± 2.8 vs. 12.6 ± 5.3, p < 0.003) and ΔP(v-a)CO2 at T1(−4.8 ± 3.8 vs. 3.9 ± 5.2, p < 0.001) were significantly lower in the ΔP(v-a)CO2 ≤ 0 group than in the ΔP(v-a)CO2 > 0 group. In contrast to the overall study population, patients with P(v-a)CO2 > 6 mmHg at T0 showed that the incidence of mechanical ventilation was significantly lower in the ΔP(v-a)CO2 ≤ 0 group (n = 10, 22.2%) than in the ΔP(v-a)CO2 > 0 group (n = 7, 54.5%, p = 0.033). The duration of mechanical ventilation was also shorter (0.9 ± 2.2 vs. 3.6 ± 5.6 days, p = 0.011).
Regarding secondary outcomes, CRRT was required less frequently in the ΔP(v-a)CO2 ≤ 0 group (n = 4, 8.9%) compared to the ΔP(v-a)CO2 > 0 group (n = 4, 36.4%, p = 0.020). The duration of CRRT was also shorter (0.5 ± 1.6 vs. 2.8 ± 5.6 days, p = 0.016). The incidence of AKI was significantly lower (n = 5, 11.1% vs. n = 4, 36.4%, p = 0.041). ICU stay was shorter in the ΔP(v-a)CO2 ≤ 0 group, though hospital stay did not differ (Table 3).
During the study period, mortality occurred in 4 cases (7.1%) among all participants, with 7-day, 28-day, and in-hospital mortality rates of 1 (1.8%), 4 (7.1%), and 4 patients (7.1%), respectively. Among these, only the 7-day mortality rate was significantly lower in the ΔP(v-a)CO2 ≤ 0 group (n = 0, 0% vs. n = 1, 9.1%, p = 0.041).
Partial correlation analysis was conducted to assess the association between ΔP(v-a)CO2 and various hemodynamic parameters measured at T1 and T0, with adjustments for age and sex (Table 4).
Among the parameters, ScvO2 measured at T1 exhibited a negative correlation with ΔP(v-a)CO2 (partial correlation −0.273, p = 0.045) and was also negatively correlated with ΔP(v-a)CO2 (β ± SE: −0.522 ± −0.221, correlation −0.305, p = 0.022) in the linear regression analysis (Table 5).
Figure 2 further demonstrates the negative correlation between ScvO2 measured at T1 and ΔP(v-a)CO2.
Table 6 displays the outcomes of the logistic regression analysis concerning postoperative 28-day mortality.
Following univariate analysis, significant risk factors identified included total SOFA scores at T0 and T1, Δbicarbonate at T1, lactate levels at T1, Δlactate at T1, and ScvO2 levels at T1. However, no significant risk factors emerged in predicting postoperative 28-day mortality following multivariate analysis. Regarding postoperative mechanical ventilator treatment, significant variables identified after univariate analysis included changes in ΔP(v-a)CO2 status, increases in ΔSOFA scores at T1, and decreases in ScvO2 levels at T1, as shown in Table 7.
Among these, increased ΔSOFA scores at T1 and decreased ScvO2 levels at T1 were confirmed as significant risk factors for postoperative mechanical ventilator treatment [[2], and (OR = 0.898, 95% CI: 0.819–0.985, p = 0.022), respectively].
Our results indicated that ΔP(v-a)CO2 values measured at T1, 24 h after ICU admission, were significantly correlated with ScvO2 levels at T1, suggesting an association with tissue perfusion status. Among patients in the high-risk group with P(v-a)CO2 values exceeding 6 mmHg at T0, those with ΔP(v-a)CO2 values of 0 mmHg or lower at T1 had significantly better clinical outcomes, including reduced duration of mechanical ventilator and CRRT use, shorter ICU stays, lower incidence of postoperative AKI, and reduced 7-day mortality rate.
Maintaining adequate tissue perfusion and oxygenation in patients who have undergone major surgery is essential to preserve organ function and surgical site integrity. ScvO2 has been used as one of the various tissue perfusion parameters in clinical settings. It is known to reliably reflect changes in oxygen delivery and tissue oxygen consumption [14], and a previous study reported that close monitoring of tissue O2 extraction using ScvO2 measurements can reduce postoperative organ dysfunction through the early detection and correction of abnormal tissue oxygenation [15]. Unfortunately, in this study, the ScvO2 level measured at T0 did not show a clear correlation with clinical outcomes. However, the reduced ScvO2 level at T1 was a significant risk factor for postoperative mechanical ventilation after multiple logistic regression analysis. A decrease in ScvO2 generally indicates increased metabolic demands, especially when oxygen extraction may be abnormally increased for oxygen uptake in cases of extensive tissue injury such as major surgery or multiple trauma. Additionally, an abnormally decreased ScvO2 can indicate hypovolemia or anemia due to intraoperative hemorrhage or insufficient oxygen supply from decreased cardiac output. Therefore, patients with reduced ScvO2 levels would require early correction of the imbalance between oxygen delivery and consumption, and they may require more aggressive ventilator therapy to improve oxygen supply, in addition to general conservative oxygen support [16]. In particular, patients who underwent major abdominal surgery, the focus of our study, often experienced intraoperative bleeding or early hemodynamic instability when admitted to the ICU after surgery. Additionally, decreased hemoglobin or cardiac output can result in inadequate oxygen supply to tissues and vital organs, leading to vital organ impairment. Moreover, respiratory muscle movement is often restricted by postoperative pain and residual anesthetic effects, leading to frequent asynchrony immediately after surgery. This asynchrony promotes atelectasis and increased dead space in the immediate postoperative period, ultimately leading to increased ventilation/perfusion mismatch. Furthermore, O2 extraction often increases in response to the elevated oxygen demand for recovery from damage caused by extensive tissue manipulation during surgery or for healing at the anastomosis site. Since the subjects of this study were limited to postoperative patients with high risks such as dynamic imbalance of respiratory muscles or increased O2 extraction from tissues, differences in clinical outcomes might be observed based on ScvO2 values. We anticipate that the clinical application of ScvO2 could be more beneficial in monitoring the condition of postoperative patients and improving clinical outcomes compared to other patient groups. Furthermore, physicians should be aware that the ScvO2 value measured 24 h post-surgery in patients who have undergone major abdominal operations could indicate a risk factor for requiring postoperative mechanical ventilation. If the ScvO2 value decreases, the patient’s condition should be closely monitored, including cardiac output, accurate metabolic demands, and volume status; physicians should also promptly provide adequate fluid therapy and administer vasopressors.
Although ScvO2 is often used as an indicator of global oxygen balance, its role as a direct marker of microcirculatory function remains limited. In certain clinical conditions, such as sepsis or septic shock, elevated ScvO2 levels may reflect impaired oxygen extraction due to microcirculatory shunting or mitochondrial dysfunction, which can limit its reliability in accurately reflecting tissue perfusion. As noted in a multicenter study by Pope et al., the interpretation of ScvO2 should consider the clinical context, as overly high ScvO2 values may reflect impaired oxygen utilization rather than adequate perfusion [17]. Nonetheless, compared to lactate, which typically reflects delayed metabolic consequences of hypoperfusion, ScvO2 and ΔP(v-a)CO2 may offer more immediate insights into circulatory adequacy and oxygen dynamics. While each parameter has its own limitations, the combined interpretation of ScvO2 and ΔP(v-a)CO2 may enhance the early detection of perfusion mismatch and allow more timely intervention. In postoperative patients with high metabolic demands and fluctuating volume status, this can be particularly beneficial. Thus, our findings suggest that complementary use of these markers may provide additional insights alongside conventional parameters such as lactate. Impaired microcirculation disorders can result from increased blood viscosity, endothelial dysfunction, or glycocalyx degradation associated with severe tissue injury or sepsis [18], or from microthrombi formation associated with disseminated intravascular coagulation. In our study population—comprising patients who underwent major abdominal surgery—large-volume fluid administration is commonly required, which may lead to hemodilution. This hemodilution can, in some cases, attenuate the viscosity-related component of microcirculatory impairment. However, despite potentially preserved macro-hemodynamic parameters such as blood pressure, perfusion to peripheral organs may still be compromised. Therefore, macro-circulation variables may not reliably reflect the underlying status of the microcirculation in such clinical settings. In such instances, although blood pressure might be maintained, perfusion to major organs in the periphery could be compromised, hence macro-circulation variables may not accurately reflect the real microcirculation status. Moreover, the glycocalyx, which possesses a net negative charge, functions as an endothelial barrier by forming a wall for negatively charged proteins in the plasma, thereby helping to maintain osmotic pressure toward the blood vessel’s lumen. When plasma protein leaks into the interstitium due to glycocalyx degradation, tissue perfusion does not improve as expected after sufficient fluid resuscitation because oncotic pressure cannot be maintained [18,19]. Therefore, additional biological markers that can more accurately represent tissue perfusion are needed in these cases. CO2 serves as an indicator of tissue perfusion because it is about 20 times more soluble in water than oxygen. In ischemic tissues with poor perfusion, oxygen deficiency triggers anaerobic respiration for energy production, which also increases CO2 production as a by-product. Consequently, the increased CO2 is likely to diffuse into the venous effluent due to its higher solubility in water, making CO2 in venous blood an effective indicator of tissue hypoperfusion. The results of this study reveal that in the high-risk group, a P(v-a)CO2 measured at T1 greater than zero with an increase over 6 mmHg post-admission correlates with increased durations of mechanical ventilation and CRRT usage, extended ICU stays, and elevated incidences of postoperative AKI and 7-day mortality rate. Gustavo et al. [20] reported that persistently elevated P(v-a)CO2 values in patients newly diagnosed with septic shock correlate with severe multi-organ dysfunction and 28-day mortality. In our previous study published in 2024 [21], we established that a P(v-a)CO2 value of 8.6 mmHg or higher, measured 24 h post-admission, served as a significant prognostic indicator for 7-day mortality among ICU patients after abdominal surgery. Consequently, based on prior research and our findings, in patients entering the ICU post-major surgery, a P(v-a)CO2 level exceeding 6 mmHg immediately post-admission should alert physicians to possible decreases in tissue perfusion, serial measurements of P(v-a)CO2 may be helpful in monitoring perfusion changes, though prospective studies are needed. Furthermore, the authors anticipate that whether timely and aggressive intervention based on these findings improves outcomes requires confirmation in future prospective trials.
Interestingly, our findings showed that the ScvO2 value measured at T1, indicating tissue perfusion, demonstrated a negative correlation with ΔP(v-a)CO2. Typically, an increase in ScvO2 suggests enhanced oxygen delivery, whereas a decrease in P(v-a)CO2 signals improved tissue perfusion and CO2 clearance. Consequently, ScvO2 and P(v-a)CO2 are interrelated and often assessed together to evaluate tissue perfusion in critically ill patients. Ospina et al. [1] observed that low ScvO2 (<70%) and high P(v-a)CO2 (>6 mmHg) correlate with adverse outcomes in septic shock patients. Moreover, several studies show that ScvO2 and P(v-a)CO2 are inversely proportional. Another study [5] also found that increased P(v-a)CO2 in septic shock patients corresponded with reduced ScvO2 levels and poor tissue perfusion. Additionally, successful resuscitation in septic shock patients led to a significant rise in ScvO2 and a decline in P(v-a)CO2 values. These findings are consistent with our study results, which demonstrated a negative correlation between ScvO2 levels and the change in P(v-a)CO2 values at T1. These findings suggest that a combined interpretation of ScvO2 and P(v-a)CO2 may help physicians gain a more nuanced understanding of hemodynamic status in patients with compromised tissue perfusion, although further prospective studies are warranted to determine its role in guiding interventions.
Despite these interesting results, the findings of the current study should be interpreted with caution due to various limitations. First, since it uses a retrospective design, selection bias cannot be eliminated, and decisions regarding interventions and treatments were not randomized. However, all patients admitted to the ICU were managed by a single intensivist, and no significant changes in critical management principles were observed. Thus, we anticipate that this may reduce potential biases in treatment strategies. In addition, clinically important perioperative and intensive care variables such as transfusion, use of vasopressors/inotropes, and anesthesia duration were not consistently available in this retrospective dataset, which may have influenced both ScvO2 and CO2 clearance. Nevertheless, in our cohort, fluid balance, total intake, IV intake, and output showed no significant differences between the groups, supporting the validity of our analysis of ΔP(v-a)CO2 and ΔScvO2 changes between T0 and T1. Second, we analyzed only ScvO2 and P(v-a)CO2 values measured immediately and within 24 h after ICU admission. Although serial ScvO2 and P(v-a)CO2 measurements were performed several times, there were insufficient participant numbers for analysis at other time points. These values are vital not only at specific times post-major surgery but also as tools to evaluate responses. In the future, the clinical significance of ScvO2 and P(v-a)CO2 values at other times should be confirmed. Thirdly, a 24 h interval may be too long to detect trends and critical time points in P(v-a)CO2 changes. In our study, we performed timely checks of P(v-a)CO2 when the patient’s condition was unstable post-surgery, but the analysis was unsuccessful due to insufficient participant numbers. In the future, the clinical significance of short-term changes in P(v-a)CO2 should be verified. Fourthly, among the patients whose gas analysis was performed post-surgery, those who underwent both laparoscopic and robotic surgeries were included. Therefore, it is important to consider the transient rise in PaCO2 caused by CO2 pneumoperitoneum during these procedures. However, several studies have reported that in patients without preexisting pulmonary disease, acid–base imbalances and elevated PaCO2 levels caused by CO2 pneumoperitoneum generally normalize within a few hours after extubation and surgery. In our study, since the majority of patients underwent major abdominal surgery under general anesthesia, individuals with severe preoperative pulmonary dysfunction were excluded, and thus the influence of pneumoperitoneum on postoperative PaCO2 levels is presumed to be minimal [22]. Since our study focused on the P(v-a)CO2 value—an indirect indicator of tissue perfusion—and this was measured 24 h after ICU admission, the potential influence of intraoperative PaCO2 elevation is expected to be minimal. Fifthly, due to the retrospective design of the study, the exact temporal relationship between ΔP(v-a)CO2 measurements and clinical interventions such as mechanical ventilation or CRRT initiation could not be consistently confirmed. As a result, it is unclear whether changes in P(v-a)CO2 preceded or followed clinical deterioration or therapeutic actions. Nevertheless, since the majority of patients were already on mechanical ventilation at T0 (n: 18, 94.7%), we believe that the observed ΔP(v-a)CO2 values still reflect meaningful physiological changes rather than being solely post-intervention effects. Sixthly, this study has several limitations. Being a retrospective, single-center study with a small sample size—especially in the ΔP(v-a)CO2 > 0 group (n = 11)—it may have limited statistical power to detect significant differences in key clinical outcomes such as mechanical ventilation, AKI, CRRT use, and 7-day mortality. Post hoc power analysis using G*Power (version 3.1.9.7) revealed that the achieved power for most outcomes was below 0.8, suggesting insufficient power to draw definitive conclusions. Additionally, the retrospective nature of the study and the relatively small sample size—composed mainly of patients undergoing elective surgeries—likely contributed to the low incidence of AKI and CRRT use, further limiting the statistical power and generalizability of our findings. Therefore, larger prospective studies are warranted to validate our results and further clarify the prognostic value of ΔP(v-a)CO2 in this population. Finally, interpreting P(v-a)CO2 values requires a deeper analysis of various factors that affect the relationship between pCO2 and CO2 content in the blood. However, environmental conditions in our ICU, such as humidity between 35 and 40% and a temperature of 24 °C, were kept consistent to reduce these errors. Gas analysis was conducted using a precision-managed, point-of-care gas analyzer in the ICU. Nevertheless, this study offers distinct implications from previous studies. Our study demonstrated a significant correlation between ScvO2 levels and changes in P(v-a)CO2 values. These findings offer deeper insights into tissue perfusion. The relevance of this study is underscored by its focus on patients post-major surgery who are at a high risk of fatal complications from inadequate tissue perfusion. However, we were unable to demonstrate a statistical difference in postoperative morbidity, as the sample size was insufficient. In the near future, we aim to conduct a well-designed prospective randomized controlled study with a larger cohort, particularly focusing on patients with more severe conditions.
In conclusion, fluctuations in P(v-a)CO2 levels demonstrated a modest inverse association with ScvO2 and may serve as supportive markers for postoperative risk stratification in critically ill patients following major abdominal surgery. However, given the retrospective design, relatively small sample size, and the inclusion of only patients with invasive monitoring, these findings should be considered hypothesis-generating rather than definitive. Therefore, further validation in well-designed, prospective multicenter studies with larger cohorts is required to establish the prognostic utility of ΔP(v-a)CO2 in postoperative critical care.