Introduction

Arterial carbon dioxide partial pressure (PaCO₂) is the gold standard in monitoring ventilation during general anesthesia. End-tidal carbon dioxide partial pressure (PetCO₂) reflects PaCO₂ and becomes a standard monitoring during surgery. However, various pathologic processes of the cardio-respiratory system such as ventilation-perfusion mismatch or shunt as well as changes in patient positioning have been shown to influence the correlation between PaCO₂ and PetCO₂ [1]. One-lung ventilation (OLV) and lateral decubitus position during thoracic surgery impair ventilation-perfusion matching and, as a result, the difference between PaCO₂ and PetCO₂.

Transcutaneous carbon oxide partial pressure (PtcCO₂) monitoring provides a noninvasive and continuous estimation of PaCO₂ by sampling from arterialized capillary blood and is not influenced by ventilation-perfusion mismatch [2]. Previous studies [3–4] found that during short time OLV (≤ 1 hour) for thoracic surgery, the value of PtcCO₂ is closer to PaCO₂ than PetCO₂.

With the wide-spread use of mini-invasive thoracic surgery and the introduction of lung-protective ventilation strategy during thoracic anesthesia, prolonged hypercapnia originated from low tidal volume OLV in these patients are not uncommon [5–6]. We designed this study to evaluate the accuracy of PtcCO₂ in predicting PaCO₂ values during prolonged OLV and permissive hypercapnia during mini-invasive thoracic surgery.

Materials and Methods

The study protocol was approved by the Clinical Research Ethics Committee of Peking University First Hospital (2012[504]). Written informed consent was obtained from each patient.

Eighteen adult patients of ASA physical status I or II who were scheduled to undergo mini-invasive thoracic surgery with an expected OLV duration of two hours or more were recruited for this study. Patients with diagnosed cardiovascular disease were excluded.

No premedication was administrated. Before the induction of general anesthesia, an epidural puncture was performed between the fifth and eighth thoracic interspace and an epidural catheter was inserted. A test dose of 3 ml 1% lidocaine was administered and no other epidural medication was used during anesthesia.

Intraoperative monitoring included a non-invasive blood pressure, pulse oxygen saturation, an electrocardiogram, nasopharyngeal temperature, urine output, peak airway pressure, and direct arterial blood pressure measurement through a radial artery catheter. General anesthesia was induced with propofol (1–2 mg/kg) and remifentanil (effect site target control infusion at a target of 4–6 ng/ml). And rocuronium (0.6mg/kg) was administered to facilitate endotracheal intubation with a double-lumen tube by direct laryngoscopy. Patients were mechanically ventilated in a volume-controlled manner both in the supine and the lateral decubitus position. During two-lung ventilation (TLV), the fresh gas flow, tidal volume, respiratory rate and inspiratory/expiratory ratio were set at 1 L/min oxygen and 1 L/min air, 6–8 ml/kg, 10–12 breath/min, and 1:2, respectively. Anesthesia was maintained with sevoflurane inhalation (end-tidal anesthetic concentration of 0.8 MAC or above) and remifentanil infusion until the end of the surgery. Sufentanil was administered as a bolus when deemed necessary during surgery and before the end of surgery.

For all patients, the position of the double-lumen endobronchial tube was confirmed under direct vision with a fiberoptic bronchoscope (FOB). The patients were then turned to the lateral decubitus position. The tube position was then checked again with the FOB just before OLV, and the effectiveness of lung collapse during OLV was confirmed by direct observation in the operative field. During OLV, the fresh gas flow, tidal volume, respiratory rate and inspiratory/expiratory ratio were set at 1 L/min oxygen, 4–6 ml/kg, 10–16 breath/min, and 1:1.5, respectively, to maintain a SpO₂ of 90% or higher and a peak airway pressure lower than 25 cmH₂O. Intravenous ephedrine or phenylephrine or nicardipine was administrated to maintain blood pressure fluctuation within 30% from baseline. Additional doses of rocuronium were administered to maintain muscle relaxation.

PtcCO₂ was measured with a TCM3 transcutaneous CO₂/oxygen device (Radiometer, Copenhagen, Denmark). The monitoring technique was standardized by applying the probe to the upper part of the patient’s dependent arm in the lateral decubitus position. Before each study, the device was calibrated by using a two-point self-calibration (5% and 10% CO₂) and the working temperature of the electrode was maintained at 42°C to “arterialize” the skin capillary blood flow according to the manufacturer’s recommendations. The monitor used an internal adjustment to compensate for the effects of the heated probe on CO₂ tension. It took appropriately 20 minutes for initial stabilization after probe attachment. The end-tidal concentrations of the anesthetics and CO₂ were measured with a AS/5 monitor (Datex-Ohmeda, Helsinki, Finland) which was calibrated in 5% CO₂ and 20.9% oxygen gas before the study. Continuous sampling was obtained from a connector attached to the proximal end of heat moisture exchanger in the respiratory circuit. Arterial blood samples were obtained during TLV, just before the initiation of OLV, and every 30 minutes during OLV until 120 minutes. Arterial blood gas analysis was performed using a GEM premier 3000 analyzer (Instrumentation Laboratory, USA). Data of PtcCO₂, PetCO₂, and PaCO₂ monitoring results were collected simultaneously. The heart rate, mean arterial pressure, pulse oxygen saturation, and nasopharyngeal temperature were also recorded at same time-points.

Quantitative data were presented as means ± standard deviation (SD). Bland-Altman method was used to analyze the agreement between PaCO₂ and PetCO₂ or between PaCO₂ and PtcCO₂. The bias (the mean difference between the values) and the precision (the SD of the bias) were calculated. Student’s unpaired t-tests were also used to compare the differences between PaCO₂ and PetCO₂ or between PaCO₂ and PtcCO₂. A p value of less than 0.05 was regarded as statistically significant. Statistical analysis was conducted using SPSS version 14.0 (Chicago, IL, USA).

Results

All eighteen patients completed the study protocol. The demographic data were shown in Table 1. Surgical procedures included lobectomy or pneumonectomy for lung cancer, and thymectomy for thymoma. A total of 90 data sets consisting of the simultaneous measurements of PtcCO₂, PetCO₂ and PaCO₂ at five time points were obtained (Table 2). The heart rate and mean arterial pressure did not significantly change from the preoperative values during the study period. The body temperature remained constant between 35.5 and 36.5°C.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic data (n = 18).

https://doi.org/10.1371/journal.pone.0138912.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Carbon dioxide level in five time points.

https://doi.org/10.1371/journal.pone.0138912.t002

When PtcCO₂ and PaCO₂ were compared, the bias and precision were 4.1 ± 6.5 mmHg during TLV and 2.9 ± 6.1 mmHg during OLV, respectively (Fig 1). When PetCO₂ and PaCO₂ were compared, the bias and precision were -11.8 ± 6.4 mmHg during TLV and -11.8 ± 4.9 mmHg during OLV, respectively (Fig 2). The values of bias and precision were stable and the difference between PtcCO₂ and PaCO₂ was significantly lower than that between PetCO₂ and PaCO₂ throughout the 2-hour period of OLV (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Agreement between transcutaneous CO₂ (PtcCO₂) and arterial CO₂ (PaCO₂).

Bland-Altman analysis of PtcCO₂ versus PaCO₂ during two-lung ventilation (TLV) and one-lung ventilation (OLV). Bias was labeled. The 95% limits of agreement of the average PtcCO₂ –PaCO₂ difference during TLV and OLV were 4.1 ± 6.5 mmHg and 2.9 ± 6.1 mmHg (mean ± 1.96 standard deviation), respectively.

https://doi.org/10.1371/journal.pone.0138912.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Agreement between end-tidal CO₂ (PetCO₂) and arterial CO₂ (PaCO₂).

Bland-Altman analysis of PetCO₂ versus PaCO₂ during two-lung ventilation (TLV) and one-lung ventilation (OLV). Bias was labeled. The 95% limits of agreement of the average PetCO₂ –PaCO₂ difference during TLV and OLV were -11.8 ± 6.4 mmHg and -11.8 ± 4.9 mmHg (mean ± 1.96 standard deviation), respectively.

https://doi.org/10.1371/journal.pone.0138912.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Bland-Altman analysis at different time points.

https://doi.org/10.1371/journal.pone.0138912.t003

Discussion

This study demonstrates that PtcCO₂ monitoring is a more accurate estimation of PaCO₂ than PetCO₂ during OLV of two hours or more with low tidal volume for thoracic surgery. Our results are consistent with previous ones on the accuracy of PtcCO₂ monitoring during thoracic surgery of shorter duration [3–4]. Moreover, we find that the PtcCO₂ is accurate of in estimating the PaCO₂ in the clinical condition of permissive hypercapnia which is not observed in previous studies [3–4].

PtcCO₂ monitoring used in the present study is a non-invasive method for continuous measurement of transcutaneous CO₂ partial pressure. It uses skin electrodes to quantify the amount of CO₂ that diffuses to the electrode on the surface of the skin. Local heating is required in order to increase the local blood circulation in the capillary bed below the sensor. The accuracy of PtcCO₂ monitoring is influenced by several factors including methodological limitations (e.g., stabilization time, reaction time, periodic repositioning of the sensor, the need for membrane restoration, and baseline calibration), technical mistakes (e.g., improper application of the sensor, trapped air bubbles in the electrolyte solution, damage to the sensor membrane, improper calibration) and hypoperfusion (e.g., vasoconstriction, hypothermia, shock, low cardiac output or local edema) [7].

Since its introduction into clinical practice, PtcCO₂ monitoring has received great attention in neonates [8–9]. Later on, with the improvement of technology and increasing knowledge of this monitoring modality, its use has been increasing in pediatric patient [10–11], in adult patients undergoing thoracic [3–4] and laparoscopic surgery [12–15] and in patients after surgery [16–17]. The accuracy of PtcCO₂ monitoring has been confirmed by these studies. However, some others reported that PtcCO₂ monitoring cannot precisely predict PaCO₂ in preterm infants during in their first 24 hours [18] and in some patients receiving artificial ventilation during general anesthesia [19], possibly because of the aforementioned reasons. Considering the relative high price and the failure possibility, appropriate indication should be considered in chosing this monitoring method. One suitable clinical condition for PtcCO₂ monitoring is OLV during thoracic surgery.

OLV is essential for the success of mini-invasive thoracic surgery. Advances in this technique have enabled more complex intrathoracic procedures being performed. Protective ventilation strategy [5–6] is now widely used during OLV and consists of small tidal volumes, low inspired oxygen fraction, low airway pressures, permissive hypercapnia, and positive end expiratory pressure. Our results demonstrated the applicability of PtcCO₂ during a 2-hour period of OLV with hypercapnia for thoracic surgery. The study of Oshibuchi et al. [3] found that PtcCO₂ monitoring provides a more accurate estimation of PaCO₂ than PetCO₂ during an 1-hour period of OLV with normocapnia (PaCO₂ in the range of 30–50 mmHg) for thoracic surgery. In another study, Kelly et al. [20] reported that the agreement between PtcCO₂ and PaCO₂ deteriorated at high PaCO₂ levels (>60 mmHg). Our results extrapolate these ranges, i.e., a 2-hour period of OLV with hypercapnia (PaCO₂ in the range of 35–70 mmHg) during thoracic surgery. Long time PtcCO₂ monitoring (longer than 2 hours) was only observed in nonsurgical patients in previous studies [21–22].

A difference of 5 mmHg or less between PaCO₂ and other carbon dioxide measurement was a clinically acceptable discrepancy [4,13]. Our results found that the bias between PtcCO₂ and PaCO₂ was less than 5 mmHg during either TLV or OLV, whereas the bias between PetCO₂ and PaCO₂ was lower than -11 mmHg during either TLV or OLV. In the scatter diagram of PetCO₂ and PaCO₂, 3 of 90 points were outside of the limits of agreement during OLV and all 3 points were beyond 22 mmHg. Whereas in the scatter diagram of PtcCO₂ and PaCO₂, 4 of 90 points were outside of the limits but only 1 point was beyond 22 mmHg. These also indicated the superiority of PtcCO₂ in predicting PaCO₂.

PetCO₂ monitoring is still the most convenient method in CO₂ measurement and has a unique role in judging the position of artificial airway and ventilation status. The difference between PaCO₂ and PetCO₂ increases with age, pulmonary disorders, pulmonary embolism, reduced cardiac output, hypervolemia, anesthesia, and other conditions that increase the ventilation-perfusion mismatch. In the present study, the mean difference between PetCO₂ and PaCO₂ during either TLV or OLV was higher than the previously reported ones in similar patients [3–4]. This is perhaps because the tidal volume settings during TLV (6–8 ml/kg) and OLV (4–6ml/kg) were lower in our study than in previous one (10 ml/kg throughout the operation) [3]. According to respiratory physiology, the ratio of dead space to tidal volume determines the gradient between PaCO₂ and PetCO₂ assuming that the PaCO₂ remains constant. The higher difference between PetCO₂ and PaCO₂ in our study supports the use of PtcCO₂ monitoring during prolonged OLV with lung-protective strategy of low tidal volume ventilation.

Conclusions

In conclusion, our study demonstrated that PtcCO₂ is more accurate than PetCO₂ in estimating PaCO₂ during prolonged OLV with low tidal volume ventilation for thoracic surgery.

Supporting Information

Acknowledgments

We thank Dr. Chen-Xiao Zhu (Department of Anesthesiology, Peking University First Hospital) for her help in conducting this study. We thank Ms. Sai-Nan Zhu (Department of Biostatistics, Peking University First Hospital) for her help in statistical analysis. We also thank Dr. Hong Liu (Department of Anesthesiology and Pain Medicine, University of California Davis Health System) for his critical comments during the manuscript preparation stage.