The study was conducted in the Electrophysiology Laboratory (EP Lab), Heart Center, Stony Brook University Hospital, Stony Brook, New York, USA.

This study was approved by our university Institutional Review Board (IRB) before any patients were enrolled. All patients provided written informed consent prior to any research interventions.

We studied adult patients undergoing atrial fibrillation ablation procedures under general anesthesia between December 2015 and November 2016. Eligible patients were identified and recruited to the study during consultation with the cardiac electrophysiologist in the cardiology clinic. Written informed consent was obtained by the anesthesiology team prior to any procedures being performed.

Adults over 18 who were undergoing cardiac ablation at Stony Brook University Hospital EP Lab were included in the study. Patients were excluded if they had any contraindication for jet ventilation such as obesity (BMI ≥ 40), lung disease or respiratory disease (e.g. severe COPD), or abnormality of the skin preventing application of tcPCO₂ probe such as jaundice (which affects skin pigmentation, interfering with function of tcPCO₂ sensor).

We used the SenTec AG (Therwil, Switzerland, www.sentec.com) Digital Monitoring System (see Figure 1) which monitors tcPCO₂ levels. TcPCO₂ measurements were recorded with the V-Sign^TM Sensor 2, recording at 42.0˚C, under software version MPB-SW:V05.03.02/SMB-SW:V07.03.1. The Sentec DMS monitor also records O₂ saturation and pulse rate but we did not utilize those measurements for this study. End-tidal CO₂ was recorded using the standard inline sampling line incorporated into the GE Aestiva 5 anesthesia machine.

This was an unblinded observational study. Patients were not randomized; they were used as their own control. Respiratory support was performed in three phases. The order of phases was determined by the requirements of clinical care and did not vary.

1) Baseline―Conventional ventilation (CV) using GE Aestiva 5 anesthesia machine.

2) During Cardiac Ablation―High Frequency Jet Ventilation (JV) using Acutronic Monsoon III high frequency jet ventilator (Susquehanna Micro Inc., Windsor PA, http://www.susquemicro.com/).

3) Post-Ablation―patient was reconnected to GE Aestiva anesthesia machine for respiratory support.

A researcher in the OR manually recorded tcPCO₂ and EtCO₂ at variable intervals (10 - 15 minutes in most cases). Measurements were taken at frequent intervals during the entire procedure, including periods before, during and after jet ventilation. ABG samples were taken at the discretion of the anesthesia provider. These measurements were entered into an Excel file and graphed over time. Relevant events in the management of the patient (e.g. onset and offset of jet ventilation) were noted on the datasheet. For statistical analysis, measurements of pCO₂ (EtCO₂, tcPCO₂, PaCO₂) were averaged within-subject to give one measurement in each of the three conditions (Baseline, Jet Ventilation, Post-Jet Ventilation). Measures were compared to each other within condition by paired t-test, with a significance level of P < 0.05. Statistical analyses were performed using IBM SPSS Statistics v22.

Twenty patients were enrolled in the study. Of these, one was a screen failure (BMI over the limit of 40); one had surgery cancelled; and 3 had unusable data due to recording problems. We have evaluable data for 15 patients. Patients (9 M, 6 F) were ASA 3 and 4, and ranged in age from 26 to 82 years (median 66 years). BMI ranged from 24 to 35 m/kg² (median 28.5). No patient admitted to being a current smoker; 8 (53.3%) were former smokers. Hypertension was common, with 10 (66.7%) patients having this diagnosis. Pulmonary disease was rare in this sample, with one patient having a history of asthma and one having mild COPD. Two patients (14.3%) had been diagnosed with kidney disease; one of these had undergone nephrectomy. There were no diabetic patients in this sample.

The majority (n = 11) had paroxysmal atrial fibrillation (Afib); 3 patients had persistent Afib and 1 was described with atrial flutter. Left ventricle ejection fraction was available on 7 patients and ranged from 27 to 65 (median 57). Seven patients (46.7%) had a history of cardiac artery disease. One patient (6.7%) had had a prior MI, and another one (6.7%) had a prior TIA. Two patients (13.3%) had a pacemaker (one of these was also stented) while a third patient had a stent only. Five patients had undergone prior cardiac surgery, including 3 patients who had prior AF ablation, but none had prior CABG surgery.

The median duration of anesthesia was 6 hours 22 minutes (range 2:43 to 9:28). Jet ventilation was employed for a median of 2 hours 56 minutes (range 1:41 to 4:45 hours). The anesthesia provider was free to choose the anesthesia technique and medications provided to the patient. All patients were treated with midazolam premedication. During baseline, 8 patients received inhalational anesthesia using sevoflurane or desflurane, and 7 were given propofol infusion. During the JV phase, all patients were given total intravenous anesthesia (TIVA) using propofol infusion and narcotics. Fentanyl was used in 13 cases (86.7%) and remifentanil in 5 cases (33.3%); some patients received both agents. Muscle relaxant (rocuronium) was used in all patients with reversal by either neostigmine (n = 6) or sugammadex (n = 9). Fluid hydration was maintained with normal saline. All patients received low dose (0.1 - 0.5 mcg/kg/min) phenylephrine infusion during the ablation phase.

14 patients (93.3%) were extubated at the conclusion of the procedure in the EP Lab suite; only 1 patient remained intubated on transfer to the recovery area. There were no reports of serious adverse events or anesthesia complications in this sample. Patients remained overnight for observation as per standard procedure, and were discharged to home the next day.

The mean values of measured tcPCO₂, EtCO₂ and PaCO₂ are given in Table 1. Both tcPCO₂ and EtCO₂ slightly (but statistically significantly) underestimated PaCO₂ in all conditions (except tcPCO₂ in baseline). However, tcPCO₂ values were consistently higher than were EtCO₂ values in baseline (P = 0.009) and post-jet ventilation (P = 0.009) by paired t-test. Results from a typical patient are shown in Figure 1.

Correlations between the different measures of CO₂ are shown in Table 2. During baseline, correlations between arterial blood and both EtCO₂ and tcPCO₂ were low and not statistically significant. The higher variability of measurements during baseline is reflected in the low correlations from this period. During jet ventilation, the correlation between arterial blood CO₂ and tcPCO₂ rose to a highly significant value (r = 0.843, P < 0.001). End-tidal measure of CO₂ was not available during this period. Post-jet ventilation, the high correlation between tcPCO₂ and PaCO₂ was maintained (r = 0.874, P < 0.001) and was matched by a similarly high correlation between EtCO₂ and PaCO₂ (r = 0.741, P = 0.002). Correlations between the two measures of tcPCO₂ and EtCO₂ were non-significant

Results of paired t-tests are indicated as follows: TcPCO₂ versus PaCO₂: *P < 0.05 ***P = 0.001. EtCO₂ versus PaCO₂: ^##P < 0.01 ^###P < 0.001. TcPCO₂ versus EtCO₂: ⁺⁺P < 0.01.

at baseline (r = −0.423, P < 0.12) but quite high during the post-jet period (r = 0.696, P = 0.006).

Figure 2 shows tcPCO₂ compared to the gold standard, arterial blood gas measurement (PaCO₂) in baseline, JV and post-JV conditions, using the Bland-Altman technique [17] . This technique plots the difference between measures versus their mean, giving 95% limits of agreement (i.e. 95% confidence limits). Comparing PaCO₂ to tcPCO₂, these plots show a small but consistent bias where the mean difference between these two measures is 0.94 + 6.7 at baseline, and 3.15 + 2.94 during JV (see Table 1). Under JV conditions, variability of the measurements was reduced, as shown by the narrower 95% confidence interval. Reduced variability of the difference scores appears to persist into the post-JV period (see Figure 2 and Table 3). During and post-JV, the 95% confidence interval for the mean (shown by the dashed orange lines) does not include zero, indicating that the mean difference of tcPCO₂ is significantly (P < 0.05) different from zero.

Comparison of tcPCO₂ to traditional EtCO₂ before and post-JV is shown in Figure 3, using the same graphical technique. In both cases, the mean difference between the two measures is approximately 3 - 5 mmHg, with standard deviation of 4 - 6 mmHg (see Table 3). This is well within the range of variability that is accepted in clinical practice. Variability in the 95% limits of agreement (as

shown by the dashed red lines) is slightly decreased after JV. However, for both baseline and post-JV, the 95% confidence interval for the mean (as shown by the dashed orange lines) does not include zero, indicating a significant difference (P < 0.05) between the end-tidal and the transcutaneous measures.

To our knowledge, this study is the first to evaluate use of tcPCO₂ to manage respiratory state during high frequency jet ventilation with a closed airway. It is also one of very few in the literature to directly compare tcPCO₂ and EtCO₂ during conventional ventilation. We found a correlation between tcPCO₂ and EtCO₂ ranging from −0.423 at baseline to r = +0.696 post-jet ventilation. The latter value is very similar to the value of r = 0.707 reported by Zasa et al. [18] for these two measures during conventional ventilation. In our patient sample, both EtCO₂ and tcPCO₂ estimates of PCO₂ were statistically lower than the measured PaCO₂ by paired t-tests. However these differences were small in magnitude and are not clinically relevant. It should be noted however, that in the baseline condition with conventional ventilation, there was no significant difference between tcPCO₂ and arterial PaCO₂.

The reason for the consistently higher values of tcPCO₂ compared to EtCO₂ is due to the effect of alveolar dead space. End-tidal PCO₂ at the end of expiration during tidal breathing is assumed to represent the alveolar gas. However it is lower than ‘ideal’ alveolar P_ACO₂, because the almost CO₂-free gas from alveolar dead space dilutes and lowers the end-tidal PCO₂ reading. Transcutaneous CO₂ monitoring directly samples the PaCO₂ from the blood vessels, and is therefore expected to be closer to the true PaCO₂ level.

There is no data in the current medical literature on efficacy or accuracy of tcPCO₂ monitoring during closed airway jet ventilation for cardiac ablation. The available studies on surgical patients under general anesthesia support the use of transcutaneous capnography compared to end-tidal measurement. Mizushima et al. (2003) described an excellent correlation (r = +0.96) between the ABG measure of PaCO₂ and the transcutaneous CO₂ monitor in 15 patients undergoing microlaryngosurgery with high frequency jet ventilation [19] . In thoracotomy patients receiving one lung ventilation, end tidal CO₂ monitoring may not be accurate to estimate PaCO₂ due to mismatch between perfusion and ventilation. Choi et al. (2008) found that tcPCO₂ was more accurate and useful for assessing CO₂ levels during one lung ventilation [9] .

The manufacturer recommends sensor placement on the cheeks, forehead, deltoid muscle of upper arm or pectoralis muscle on the chest for maximal efficiency in measuring CO₂, SpO₂ and pulse rate. However, these sites were not accessible in our patients due to surgical drapes. We used the left pectoralis or the deltoid muscle, which gives acceptable sensitivity in measurement of tcPCO₂ [20] . There are 3 “outlier” data points shown in Figure 2 and Figure 3. All three of these patients had the sensor probe placed on the left pectoralis. None of the “outlier” patients were smokers and none had COPD. We did not find any immediate explanation for their “outlier” status. But it should be noted that by definition, 95% confidence limits will exclude 5% of the data points; i.e., one in 20. Thus the number of outliers is consistent with statistical expectations.

This pilot study is limited by the small sample size. The study design would have been improved by collecting a prescribed number of ABG samples at standardized intervals. In order to follow standard clinical practice as closely as possible, we chose to allow the anesthesiologist to use their clinical judgment in the number and timing of ABG samples. We did not perform extensive follow-up of the patients. They were monitored in a cardiac recovery unit overnight as per our standard hospital procedure, and discharged the next day. None of the patients experienced adverse sequelae. To correct the tcPCO₂ readings for change over time, we applied the “drift correction” algorithm supplied by the device manufacturer. In most cases, visual inspection showed the corrected and uncorrected values were not substantially different (1 - 2 mmHg) over a period of 4 - 6 hours of recording. The tcPCO₂ data reported in this paper are the uncorrected values recorded manually from the monitor during anesthesia.

Withdrawing arterial blood from the arterial line does carry a minimal degree of risk of introducing potential infection or an air bubble; this risk is increased when done multiple times during the procedure. In our EP Lab setting, sending an ABG sample would also require a dedicated person to carry the actual sample to our respiratory laboratory. In practice this person has to leave the floor for a period of 5 - 10 minutes, followed by an additional period of time for the ABG results to be presented via the computerized system. To that one should add over two hundred dollars for hospital charges per sample.

If validated for this purpose, tcPCO₂ monitoring may increase patient safety during JV for cardiac ablation procedures. It will also eliminate the need for frequent samples of arterial blood gas, and decrease the time and expense associated with STAT laboratory analysis of blood gas samples.

Accurate monitoring of respiratory function is a vital component of good anesthesia care. During much of the cardiac ablation procedure, the anesthesiologist is “flying blind” without real-time feedback on this vital function. This pilot study suggests that transcutaneous monitoring of CO₂ is accurate and reliable under these conditions with general anesthesia. Utilization of tcPCO₂ has the potential to replace the need for periodic sampling of ABGs during jet ventilation. This will increase patient safety by providing more information to the anesthesiologist and allowing better-informed respiratory care, while reducing costs of analysis of the blood samples.