Strategy to Reduce Hypercapnia in Robot-Assisted Radical Prostatectomy Using Transcutaneous Carbon Dioxide Monitoring: A Prospective Observational Study

Introduction

Estimation of the arterial carbon dioxide partial pressure (PaCO₂) by direct analysis of arterial blood gases is the gold standard for monitoring carbon dioxide during general anesthesia. However, because the direct measurement of PaCO₂ is invasive and intermittent, end-tidal carbon dioxide partial pressure (P_ETCO₂) monitoring is preferred for the continuous monitoring of carbon dioxide.¹ An important limitation of P_ETCO₂ monitoring is its reduced accuracy due to factors such as ventilation-perfusion mismatch, shunt, and the surgical position of the patient, including the Trendelenburg or lateral decubitus position.²

Transcutaneous carbon dioxide partial pressure (P_TCCO₂) monitoring offers an alternative, noninvasive method for the continuous measurement of PaCO₂ from arterialized capillary blood in tissues. Unlike P_ETCO₂ monitoring, it is not influenced by ventilation–perfusion mismatch. The accuracy of P_TCCO₂ has been proven in pediatric patients, thoracic anesthesia, and laparoscopic surgery.^3–5

Robot-assisted radical prostatectomy (RARP) for the treatment of prostate cancer requires that the patient should be placed in the steep Trendelenburg position, followed by the inflation of carbon dioxide gas into the peritoneal cavity to improve the surgical view and reduce bleeding. As a result, the organs in the abdominal cavity are pushed towards the diaphragm, thereby reducing both the functional residual volume and lung compliance. In addition, in robot-assisted surgery, the angle of the surgical table is steeper than the angle used in other laparoscopic surgeries, which may worsen ventilation-perfusion mismatch and cause pronounced intraoperative hypercapnia.^6,7 RARP is mainly performed in the elderly, in whom a higher risk for intraoperative hypercapnia has been attributed to the age-related decline in lung function.⁸ In addition, subcutaneous emphysema is more common in this population due to weakened tissue, which in turn also contributes to the development of hypercapnia, acidosis and therefore sympathetic excitation, tachycardia, hypertension, hyperkalemia, and other complications.^9,10 As these complications may be lethal in older patients with preexisting cardiac and pulmonary disease, it is important to maintain normocapnia during RARP by accurately monitoring PaCO_2.

We hypothesized that, compared to P_ETCO₂ monitoring, P_TCCO₂ monitoring provides a more accurate approximation of PaCO₂ and contributes more to ventilator adjustment during RARP performed in patients in a steep Trendelenburg position and under prolonged capnoperitoneum. Therefore, this prospective study evaluated the accuracy of two noninvasive monitoring systems, P_TCCO₂ and P_ETCO₂, and the predictive power of each in detecting hypercapnia.

Materials and Methods

Participants

This study prospectively included 46 patients classified as American Society of Anesthesiologists physical status I–III scheduled for RARP from January 2020 to April 2021. Patients with a history of severe cardiovascular or respiratory disease, neuromuscular disease, a body mass index > 35 kg/m², or who required a vasoconstrictor during surgery were excluded.

Anesthesia

General anesthesia was induced with propofol (1–2 mg/kg), fentanyl (1–2 µg/kg), and rocuronium (0.6 mg/kg). Then the trachea was intubated, and the lungs were mechanically ventilated under pressure-controlled ventilation volume-guaranteed with a tidal volume of 8 mL per predicted body weight, and an I:E ratio of 1:2 with a positive end-expiratory pressure of 5 cmH₂O and a respiratory rate of 10 respirations per minute. The tidal volume was reduced by 25 mL to maintain a peak inspiratory pressure < 35 cmH₂O. At P_ETCO₂ > 40 mmHg or a P_TCCO₂ > 45 mmHg, the respiratory rate was increased appropriately and arterial blood gases (ABG) were simultaneously analyzed. Anesthesia was maintained using 1–1.5 minimum alveolar concentration of desflurane using 50% oxygen in air. A bolus of fentanyl or remifentanil infusion was administered as needed to ensure a bispectral index value between 40 and 60 and a systolic blood pressure within 20% of baseline. Patients who required a vasoconstrictor to increase blood pressure during surgery were removed from the study.

Intraoperative Monitoring

Intraoperative monitoring included an electrocardiogram, pulse oxygen saturation, noninvasive blood pressure, arterial blood pressure, peak airway pressure, and oropharyngeal body temperature. The patient’s body temperature was kept at 36–37°C. P_ETCO₂ was monitored using a side-stream infrared CO₂ analyzer (Avance CS², GE Healthcare, Madison, WI, USA). P_TCCO₂ was measured using a TCM4 device (Radiometer, Copenhagen, Denmark). Before its placement, the electrode was cleaned, a new membrane was applied, and the device was calibrated according to the manufacturer’s recommendation. The electrode was placed on the upper left chest and set at a working temperature of 42°C. The skin where the electrode was placed was swabbed with alcohol to facilitate adhesion of the disc to the skin. Capnoperitoneum was established and the intra-abdominal pressure (IAP) was maintained at 15–20 mmHg at the surgeon’s discretion.

Outcome Assessment

The primary outcome was the agreement of noninvasive monitoring (P_TCCO₂ and P_ETCO₂) with PaCO₂. To assess of this, ABG were sampled and noninvasive monitoring measurements were recorded in the pre-capnoperitoneum state in a supine position, 30 minutes after CO₂ insufflation in the Trendelenburg position and every hour thereafter or whenever the P_TCCO₂ was > 45 mmHg or the P_ETCO₂ was > 40 mmHg, and 20 minutes after CO₂ deflation on resumption of the supine position. The first ABG sampling was conducted when the patient’s blood pressure and heart rate had stabilized and the respiratory rate was constant for at least 5 minutes after tracheal intubation. The TCM4 device was calibrated in vivo based on the results of the first ABG analysis. We calculated the differences and absolute differences between the noninvasive monitoring values (P_TCCO₂ and P_ETCO₂) and PaCO₂. The absolute differences were determined because negative numbers would artificially lower the mathematical mean of the difference. An absolute difference of 5 mmHg was defined as within the clinically acceptable range indicative of the interchangeability of the two methods.^10,11 The secondary outcomes were the sensitivity and predictive power for detecting hypercapnia, defined as PaCO₂ > 45 mmHg, which would be the basis for deciding to adjust the ventilator settings. To assess these, P_TCCO₂ and P_ETCO₂ were dichotomized at P_TCCO₂ > 45 mmHg and P_ETCO₂ > 40 mmHg. No data were recorded within 30 minutes after the occurrence of an already-recorded hypercapnia event to avoid bias due to over-representation of any single hypercapnia event. P_TCCO_2, PaCO_2, P_ETCO₂, arterial blood pressure, heart rate, oropharyngeal temperature, and IAP were recorded simultaneously.

Statistical Analysis

The sample size was calculated using G*Power (ver. 3.1.4). This indicated that 48 subjects were required to achieve a 90% power to detect a 7.5 mmHg difference (with α = 0.05) between the two methods with an estimated standard deviation of 15 mmHg using a paired t-test and a 10% dropout rate. This difference was based on a previous study.¹²

The statistical analyses were performed using SPSS ver. 26.0 (SPSS, USA). Quantitative data are presented as the mean ± standard deviation (SD) or median (interquartile range [IQR]) depending on the normality of the distribution. Correlation analysis with Pearson’s correlation coefficient (r) was used to establish the relationship between the two noninvasive measurements (P_TCCO₂ and P_ETCO₂) and PaCO₂. To evaluate their agreement with PaCO₂, the bias (mean difference between the noninvasive monitoring values and PaCO₂) and precision (SD of the bias) were evaluated, using Bland–Altman analysis. The reliability of noninvasive monitoring was compared using the mean absolute difference, using a two-tailed Wilcoxon signed-rank test for pairs after assessing normality. Chi-square analysis or Fisher’s exact test was used to compare the number of data with an absolute difference exceeding 5 mmHg. The sensitivity of the two noninvasive CO₂ monitoring systems for detecting hypercapnia was compared using Yates corrected chi-square method. The predictive power for detecting hypercapnia was compared by constructing a receiver operating characteristic curve and calculating the area under the curve (AUC). A P-value < 0.05 was considered statistically significant.

Results

Subject Characteristics

Of the 67 patients assessed for eligibility, 19 were excluded and 48 patients were enrolled in the study. Two patients were removed because they required vasoconstrictors during surgery to treat low blood pressure. Ultimately, the final analysis included 219 datasets from 46 patients (Figure 1).

Figure 1 Flow chart of participant recruitment.

All patients were male, with a mean age of 67.93 ± 9.30 years. Table 1 summarizes the patients’ demographic factors and surgical characteristics. All patients underwent RARP performed by either of two surgeons. The angle of the operation table was 28°–30°, which was the maximum angle of the two tables used in our hospital. Thirty-two patients were placed in 28° of Trendelenburg and 14 patients at 30°. The IAP was maintained between 15 and 20 mmHg according to the surgeon’s preference. The mean duration of capnoperitoneum was 175.93 ± 41.26 minutes and the mean body temperature of the patients during surgery was 36.10 ± 0.32°C. No complications related to prolonged P_TCCO₂ monitoring occurred.

Table 1 Clinical and Surgical Characteristics of the Patients (n = 46)

Primary Outcome

Table 2 gives the mean values of PaCO₂, P_ETCO₂, and P_TCCO₂ before, during, and after capnoperitoneum. Correlation analysis at each time-point showed correlations between P_TCCO₂ and PaCO₂ (r = 0.99, 0.84, 0.67, respectively; P < 0.05), and between P_ETCO₂ and PaCO₂ (r = 0.70, 0.70, 0.57, respectively; P < 0.05). Bland–Altman analysis indicated better agreement with lower bias and higher precision of P_TCCO₂ for PaCO₂ compared with P_ETCO₂ (Figure 2). As shown in Table 3, the mean absolute difference between PaCO₂ and P_TCCO₂ was smaller than the difference between P_ETCO₂ and PaCO₂ at all time points, indicating better reliability. The P_ETCO₂ continued to differ from PaCO₂ by more than 5 mmHg after 1 h of CO₂ insufflation. Of the 219 datasets, an absolute difference ≥ 5 mmHg was observed in 100 P_ETCO₂ datasets (45.7%) and in 28 P_TCCO₂ datasets (12.79%).

Table 2 Correlation Analysis of PETCO2 and PaCO2, PTCCO2, and PaCO2 at Three Time Points

Table 3 Bias and Mean Absolute Difference Between PaCO2 and Noninvasive Measurements

Figure 2 Agreement of two noninvasive monitoring systems (PTCCO2 and PETCO2) and PaCO2 by Bland–Altman analysis. Black circles indicate the difference in PaCO2 and PTCCO2 and white circles indicate the difference in PaCO2 and PETCO2. (A) During the pre-capnoperitoneum period in the supine position, the difference between PaCO2 and PTCCO2 converged on zero, while the difference between PaCO2 and PETCO2 was greater than zero and the range of agreement was wider. (B) During the period of capnoperitoneum in a steep Trendelenburg position, the difference between PaCO2 and PTCCO2 was closer to zero than the difference between PaCO2 and PETCO2. The triangles indicate the data for the patient with subcutaneous emphysema: black triangles are the difference between PaCO2 and PTCCO2 and white triangles the difference between PaCO2 and PETCO2. (C) During the post-capnoperitoneum period in the supine position, the bias in the PETCO2 remained higher, even after CO2 deflation. The black triangle indicates the difference in PaCO2 and PTCCO2 for the patient with subcutaneous emphysema.

Secondary Outcome

During surgery, the respiratory rate was adjusted 51 times in 31 patients, according to the CO₂ management protocol. The increment in the respiratory rate was made based on the P_TCCO₂ in 23 times (45.10%), on the P_ETCO₂ in 9 times (17.65%), and on both parameters in 19 times (37.25%). There were 37 hypercapnia events in 19 patients; 28 during capnoperitoneum and nine after CO₂ deflation. Of the 28 events during capnoperitoneum, a P_TCCO₂ > 45 mmHg occurred in 23 events and a P_ETCO₂ > 40 mmHg in 14 events. Using the predefined cut-off values of P_TCCO₂ > 45 mmHg and P_ETCO₂ > 40 mmHg, the sensitivity of two monitoring methods was 82.14% and 50.0%, respectively (P = 0.024). Of the nine events after CO₂ deflation, P_TCCO₂ > 45 mmHg occurred in seven events, and P_ETCO₂ > 40 mmHg in two. The sensitivity of P_TCCO₂ and P_ETCO₂ monitoring was 77.78% and 22.22%, respectively (P = 0.059).

The AUC [95% confidence interval (CI)] of P_TCCO₂ at detecting PaCO₂ > 45 mmHg during and after capnoperitoneum was 0.88 (0.82–0.95) and 0.90 (0.78–1.00), respectively. The corresponding AUCs (95% CI) of the P_ETCO₂ were 0.81 (0.71–0.90) and 0.80 (0.62–0.98), respectively. The overall AUC value of P_TCCO₂ was significantly higher than that of P_ETCO₂ (0.916 vs 0.826; P = 0.044).

A Case Report

One case merits further discussion (Table 4). In an 82-year-old patient with moderate aortic regurgitation and mild chronic obstructive pulmonary disease (COPD), P_TCCO₂ increased suddenly after 2 minutes of CO₂ gas insufflation, from 37 to 46 mmHg, whereas the P_ETCO₂ increased slowly to 40 mmHg. Subcutaneous emphysema was detected by palpating the upper chest. After the surgeon was notified, the respiratory rate was adjusted from 10 to 12 respirations per minute. During the subsequent 30 minutes of capnoperitoneum, CO₂ retention continued, resulting in PaCO₂, P_ETCO₂, and P_TCCO₂ values of 65.7, 55, and 78 mmHg, respectively, despite increasing the respiratory rate to 20 respirations per minute and lowering the IAP from 20 to 12 mmHg. The ABG analysis showed a pH of 7.186, indicative of acidosis. Approximately 1 h later, there was little change in his PaCO₂ (66.1 mmHg); his P_ETCO₂ decreased to 49 mmHg but his P_TCCO₂ decreased only slightly, to 72 mmHg. The highest P_ETCO₂ value after the detection of subcutaneous emphysema was 55 mmHg, whereas the P_TCCO₂ peaked at 89 mmHg. Twenty minutes after CO₂ deflation, the patient’s PaCO₂ was 40.8 mmHg and his P_ETCO₂ 36 mmHg, but his P_TCCO₂ was still elevated at 64 mmHg. The patient completed RARP without conversion to open surgery.

Table 4 PaCO2, PETCO2 and PTCCO2 in a Patient with Subcutaneous Emphysema

Discussion

This study demonstrated the superior accuracy of transcutaneous carbon dioxide measurement over P_ETCO₂ monitoring to estimate PaCO₂ during RARP. P_TCCO₂ is more sensitive in detecting hypercapnia during prolonged capnoperitoneum, which can otherwise lead to respiratory acidosis. The combined use of P_TCCO₂ and P_ETCO₂ monitoring may enable anesthesiologists to provide more meticulous ventilator management.

A previous study reported the inconsistent correlation between P_ETCO₂ and PaCO₂ after CO₂ inflation in laparoscopic surgeries with prolonged capnoperitoneum.¹⁰ In morbidly obese patients undergoing laparoscopic bariatric surgery, capnoperitoneum may exacerbate a reduction in the functional residual capacity and increase ventilation–perfusion mismatch, thereby diminishing the accuracy of P_ETCO₂.¹¹ In our study, the correlation of P_TCCO₂ and P_ETCO₂ with PaCO₂ decreased as capnoperitoneum was prolonged. The mean absolute difference between PaCO₂ and P_ETCO₂ exceeded 5 mmHg after 1 h of capnoperitoneum.^10,13,14 By contrast, the mean absolute difference between PaCO₂ and P_TCCO₂ was < 5 mmHg throughout the period of capnoperitoneum. The scatter diagram of P_ETCO₂ and PaCO₂ on a Bland–Altman plot also showed many data points outside the acceptable range, indicating poor agreement between PaCO₂ and P_ETCO₂.

The bias in the PaCO₂ and P_TCCO₂ in our study was much smaller than reported in previous studies.¹⁵ This was likely to be because none of our patients experienced severe hypercapnia and acidosis, which were avoided by the meticulous adjustments that were made based on both P_TCCO₂ and P_ETCO₂ monitoring, with the exception of a single case of subcutaneous emphysema. The respiratory rate was adjusted in 45% of the cases based on the P_TCCO₂ values compared to 19% based on P_ETCO₂ values, highlighting that the additional use of P_TCCO₂ monitoring can compensate for the deficiencies of P_ETCO₂ monitoring. Considering previous reports of an increasing difference between PaCO₂ and P_TCCO₂ along with an increase of PaCO₂ levels,^16,17 the relatively small change in the PaCO₂ levels of our patients may account for the discrepancy. In addition, the stable cardiovascular function of the patients, the exclusion of those with severe lung diseases, and the maintenance of a constant body temperature during surgery may have reduced the bias in the PaCO₂ and P_TCCO₂ values in this study.

Previous studies raised concerns about the limitations of P_TCCO₂ measurements, such as the relatively slow response time, difficulty maintaining good contact between the patient’s skin and the sensor, and the long warm-up time needed for the sensor to reach its final operating temperature. In addition, the variability in skin thickness at the sensor attachment site may affect the accuracy of the P_TCCO₂ measurements. Nishiyama et al^18,19 reported that the PaCO₂ was more precisely measured by attaching the electrode to the chest rather than to the upper arm, forearm, or earlobe. In our patients, the electrode was attached to the upper left area of the chest, where adherence of the disc to the skin could be checked by the anesthesiologist even when the patient was in the steep Trendelenburg position during RARP. Skin tissue perfusion is one of the most important factors determining the accuracy and precision of P_TCCO₂. A low environmental temperature causes vascular contraction in the skin, reducing blood flow. Bladder irrigation, which is frequently performed during urological surgery, may also cause a drop in the patient’s body temperature. In our patients, efforts were therefore made to maintain their body temperature, such as by using heated breathing circuits, an intravenous fluid warmer, and a forced-air warming system and by maintaining the temperature of the operating room above 23°C. The accuracy of P_TCCO₂ monitoring in patients administered vasoconstrictors is a matter of debate. Rodriguez et al²⁰ reported that catecholamine support did not affect the accuracy P_TCCO₂ monitoring, but other studies came to the opposite conclusion.^12,14,16,21 In our study, two patients who needed a vasoconstrictor to increase their intraoperative blood pressure were excluded from the final analysis, to eliminate a possible confounder.

In the aforementioned patient with subcutaneous emphysema, P_TCCO₂ increased more rapidly than P_ETCO₂ because the former was measured in an area of subcutaneous emphysema. Likewise, the P_TCCO₂ decreased slowly because of the accumulated CO₂ at that site. Because the development of subcutaneous emphysema in the chest area is common during robotic surgery with the patient in the Trendelenburg position, the attachment of a sensor to the chest area would allow the rapid detection of subcutaneous emphysema. However, once the P_TCCO₂ increases, it would overestimate the PaCO₂ due to the accumulation of CO₂ in the subcutaneous layer.

We initially attempted to compare the decreasing trends in P_ETCO₂ and P_TCCO₂ with the decrease in PaCO₂ during CO₂ elimination. However, the brief time (< 30 minutes) from the end of capnoperitoneum to extubation was insufficient for the cutaneous capillary and systemic PaCO₂ to reach equilibrium, thus ruling out a comparison. Nonetheless, our results showed that the superiority of P_TCCO₂ over P_ETCO₂ was maintained even during the CO₂ elimination period after CO₂ insufflation was ended. Further studies are needed to determine whether the good correlation between P_TCCO₂ and PaCO₂ is retained after CO₂ elimination.

The limitations to our study include the following. First, only healthy male patients without significant co-morbidities were enrolled, such that our findings may not be generalizable to patients with severe coexisting conditions. In addition, patients treated with vasoactive drugs were excluded from our study, although transcutaneous monitoring might be advantageous for sedated patients in the intensive care unit who often need vasoactive drugs.

Conclusion

In conclusion, P_TCCO₂ monitoring was more accurate than P_ETCO₂ monitoring for measuring PaCO₂ and was better able to detect hypercapnia in surgery requiring prolonged capnoperitoneum and the steep Trendelenburg position. Because capnography enables the detection of acute life-threatening events, such as apnea, airway obstruction, ventilator disconnection, or esophageal intubation, P_TCCO₂ cannot substitute for P_ETCO_2. Nonetheless, it is valuable as an adjunct method in situations in which ventilation–perfusion mismatch interferes with the gradient between P_ETCO₂ and PaCO₂.