An Evaluation of Transcutaneous Carbon Dioxide Partial Pressure Monitoring during Apnea Testing in Brain-dead Patients

The study was approved by our local ethics committee (Comité de Protection des Personnes se Prîetant à la Recherche Biomédicale, Groupe Hospitalier Pitié-Salpîetrière, Paris, France). Thirty-two patients clinically suspected of being brain dead were investigated prospectively during a 7-month period (December 2004 to June 2005). All of them had been admitted to the intensive care unit (ICU) for severe coma resulting mainly from spontaneous intracranial hemorrhage, head injury, or cerebral anoxia. The cause of coma was established for every patient, and reversible abnormalities (drug and metabolic intoxications, hypothermia < 35°C, and shock) were excluded. Because of the severity of their cerebral lesions at the time of admission into the ICU, all of these patients were potentially expected to develop brain death. Care of the patients conformed to standard procedures in our ICU for severely comatose patients. The patients were monitored with an arterial pressure catheter, enabling samples to be taken for arterial blood gas measurements. Ptcco²was continuously measured with a heated transcutaneous ear sensor (V-Sign® Sensor, Sentec Digital Monitoring System; SENTEC-AG, Therwil, Switzerland), which also combines a pulse oximetry sensor. The Ptcco²measurement by the V-Sign® Sensor is based on a Severinghaus-type electrochemical carbon dioxide tension sensor. The sensor temperature is warmed up to 42°C to achieve local arterialization of the skin at the monitoring site for the Ptcco²measurement. According to the manufacturer's recommendations, the sensor was automatically calibrated in vitro  in its integrated calibration unit (“docking station”) before each apnea test. After calibration, the sensor was applied to the patient's ear lobe using a single-use ear clip and a thin layer of sensor gel, and a 20-min equilibration time was allowed before measurements.

At the time of investigation, all patients were in a deep, unresponsive coma. They lacked all bulbar reflexes (pupillary [light], corneal, oculocardiac, and oropharyngeal [gag and cough] reflexes), had no spontaneous breathing movements, and usually showed vasoplegia and diabetes insipidus. All these findings strongly indicate brain death.19Because the apnea test has been shown to be deleterious in some patients and may therefore limit organ procurement for transplantation,9,20–22we have decided in our ICU to perform the apnea test only after brain death has been confirmed by electrocortical silence on one electroencephalogram with maximal amplification. On the other hand, when the electroencephalogram is unhelpful in confirming brain death (mainly because of hypothermia < 35°C or because of a significant residual blood concentration of sedative drugs), we usually require the absence of intracerebral blood flow on four-vessel cerebral angiography. However, angiography is not only potentially deleterious, but also risky because of transportation of the patient to the radiology department, especially when there is major hemodynamic instability.23,24Therefore, when four-vessel cerebral angiography is mandatory, we usually perform the apnea test before angiography. In such cases, before the apnea test, we always verify the absence of intracerebral blood flow by transcranial Doppler ultrasonography.3 

The apnea test was performed after a 20-min preoxygenation period with an inspired oxygen fraction of 100%. After the ventilator was disconnected, a 9-l/min oxygen flow was delivered through the endotracheal tube via  an oxygen cannula (12-French catheter). The patient was then closely observed for respiratory efforts. If spontaneously respiratory efforts or complications (major hemodynamic instability despite increase in the dose of catecholamine and/or severe hypoxemia) occurred, the apnea was discontinued and the patient was immediately reconnected to the ventilator. Otherwise, the apnea was continued, and afterward, the patient was reconnected to the ventilator at the end of the test. The apnea test was considered positive if there was no respiratory effort and if Paco²reached at the end of the test was 60 mmHg or higher.3 

In the first 20 brain-dead patients (20-min apnea test group), the apnea test was performed according to our standard guideline, i.e. , over a 20-min fixed period. Using the receiver operating characteristic (ROC) curve, this enabled us to calculate the best Ptcco²target value which estimates that the Paco²threshold of 60 mmHg has been reached. Thereafter, for the following 12 brain-dead patients (Ptcco²targeted apnea test group), the apnea test was performed until this previously calculated Ptcco²target had been reached.

Clinical characteristics, etiology of brain death, and hemodynamic variables (heart rate, systolic arterial blood pressure, and oxygen peripheral saturation) were recorded. Ptcco²was continuously monitored before, during, and after the apnea test, and data were stored on a computer for off-line analysis. Complications during the apnea test were recorded as hypotension (defined as a decrease in systolic arterial blood pressure of more than 20% of baseline value and/or the need for an increase in the dose of catecholamine administered), hypertension (defined as an increase in arterial blood pressure of more than 20% of baseline value and/or the need for a decrease in the dose of catecholamine administered), and severe hypoxemia (defined as a decrease of oxygen peripheral saturation < 90%). For Paco²measurements (Blood Gas Analyzer ABL725; Radiometer, Copenhagen, Denmark), arterial blood gases were obtained through an indwelling radial arterial line and were compared with the simultaneous Ptcco². In the 20-min apnea test group, arterial blood gases were sampled before the apnea test (baseline) and thereafter at 5, 10, 15, and 20 min of apnea. In the Ptcco²targeted apnea test group, arterial blood gases were sampled before the apnea test (baseline) and thereafter at the end of the apnea test when the Ptcco²had reached the calculated target value. In addition, after reconnection of the patient to the ventilator at the end of the apnea test, arterial blood gases were sampled at 5-min intervals for 30 min in 10 patients of the 20-min apnea test group.

Data are expressed as mean ± SD. Comparison of two means was performed using the Student t  test. The ROC curve was used to determine the best threshold value for Ptcco²to predict that Paco²has reached the mandatory threshold of 60 mmHg. The area under the ROC curve was also calculated. Sensitivity, specificity, positive and negative predictive values, accuracy (defined as the sum of concordant cells divided by the sum of all cells in the two-by-two table), and their 95% confidence intervals were calculated. The best threshold value was defined as the one that simultaneously minimizes the distance to the ideal point (sensitivity = specificity = 1) and that provides a positive predictive value as close as possible to 1. All P  values were two tailed, and a P  value of less than 0.05 was considered significant. The NCSS 2001 statistical program (Statistical Solutions Ltd., Cork, Ireland) was used for all statistical analyses.

The apnea test was performed 32 times in 32 patients, 24 men and 8 women (mean age, 48 ± 14 yr). Causes of brain death were cerebral hemorrhage (n = 20), blunt head trauma (n = 5), cerebral anoxia (n = 4), and cerebral gunshot injury (n = 3). At the time of investigation, 31 patients required catecholamine administration, and 23 patients exhibited diabetes insipidus. The confirmatory test of brain death was electroencephalogram in 19 patients, whereas the 13 other patients required cerebral angiography because of significant residual blood concentration of sedative drugs. The apnea test was completely performed in the 20 patients of the 20-min apnea test group, and none of them showed any spontaneous respiratory movement during the 20-min apnea period. Similarly, in the 12 patients of the Ptcco²targeted apnea test group, the apnea test was performed until Ptcco²had reached the calculated target value, and none of them showed any spontaneous respiratory movement during apnea. Fourteen patients in the 20-min apnea test group and 4 patients in the Ptcco²targeted apnea test group showed a significant hypotension requiring an increase in the dose of catecholamine administered (P < 0.05). Finally, whatever the apnea test group, none of the 32 investigated patients showed significant hypoxemia during the apnea test.

The mean Paco²–Ptcco²gradient was 0.7 ± 3.6 mmHg for baseline measurement before the apnea test in the 32 investigated patients. Figure 1presents the typical recording of Paco²and Ptcco²during the apnea test in one patient of the 20-min apnea test group. The increases in Paco²and Ptcco²during the apnea test in the 20-min apnea test group are shown on figure 2. During the apnea test, the mean Paco²–Ptcco²gradient was fairly stable, approximately 8.7 ± 7.1. The box plot representation of the Paco²–Ptcco²gradient clearly shows that the gradient observed during the apnea test could not be analyzed in the same manner as the baseline gradient (fig. 3A). The Bland-Altman analysis for comparison of Ptcco²versus  Paco²during the 20-min apnea test (i.e. , excluding baseline measurements) revealed a mean bias of 8.6 mmHg, with limits of agreement (± 1.96 • SD) of 22.4 and −5.3 mmHg (fig. 3B).25There was no significant correlation between the Paco²–Ptcco²gradient and the rate of increase in Paco²during the apnea (R = 0.19). At the end of the 20-min apnea test, we observed major hypercapnia, acidosis, and decrease in Pao²as compared with baseline measurement (table 1). During the first 30 min after reconnection of the patient to the ventilator, the mean Paco²–Ptcco²gradient progressively increased from −1.8 ± 11.5 mmHg to 8.5 ± 6.0 mmHg (fig. 2). The capacity of Ptcco²to predict that the targeted Paco²60 mmHg had been reached was assessed with a ROC curve analysis (fig. 4). The area under the ROC curve was 0.983 ± 0.013 (P < 0.001). The best calculated threshold value of Ptcco²was 60 mmHg, with a sensitivity of 0.80 [0.68–0.88], a specificity of 1.00 [0.89–1.00], a positive predictive value of 1.00 [0.93–1.00], a negative predictive value of 0.72 [0.57–0.83], and an accuracy of 0.87 [0.78–0.92] (table 2).

Therefore, the 12 patients of the Ptcco²targeted apnea test group were investigated using this calculated Ptcco²target value of 60 mmHg. The mean duration of the Ptcco²targeted apnea test was significantly reduced: 11 ± 4 min (P < 0.001 vs.  20 ± 0 min). None of the 12 patients exhibited a Paco²lower than 60 mmHg at the end of the apnea test (extreme values: 64.1–79.4 mmHg). The mean Paco²–Ptcco²gradient at the end of the apnea test in the Ptcco²targeted apnea test group was 12.0 ± 4.6 mmHg (table 1). Figure 5presents the typical recording of Paco²and Ptcco²during the apnea test in one patient of the Ptcco²targeted apnea test group. As expected, table 1shows that hypercapnia, acidosis, and decrease in Pao²at the end of the apnea test were significantly reduced in the Ptcco²targeted apnea test as compared with the 20-min apnea test group.

In this study, we have shown that Ptcco²monitoring during the apnea test in brain-dead patients (1) exhibits a mean Paco²–Ptcco²gradient of 8.7 ± 7.1 mmHg that is fairly stable during a 20-min apnea test; and (2) could accurately predict that the target Paco²of 60 mmHg had been reached, therefore enabling us to shorten the apnea test and thus significantly reduce hypercapnia, acidosis, and decrease in Pao²at the end of the apnea test.

The mean Paco²–Ptcco²gradient was 0.7 ± 3.6 mmHg for the baseline measurement before the apnea test in the 32 brain-dead patients. This close agreement between Paco²and Ptcco²had been previously reported both in volunteers and in anesthetized patients.11–16However, most of these studies investigated Ptcco²monitoring at equilibrium during stable ventilatory and circulatory conditions. Conversely, during the apnea test, we observed that the mean Paco²–Ptcco²gradient was fairly stable, approximately 8.7 ± 7.1 mmHg in the 20-min apnea test group, and 12.0 ± 4.6 mmHg in the Ptcco²targeted apnea test group. This increased gradient in such dynamic conditions is fully understandable because of the continuous increase in Paco²during the apnea. Indeed, the “delay” of Ptcco²monitoring as compared with the simultaneous Paco²measured by blood gas analysis could be linked to the time for diffusion of carbon dioxide from the ear lobe capillary to the sensor, and eventually the time of calculation of Ptcco²by the monitor. At last, the slightly higher mean Paco²–Ptcco²gradient in the Ptcco²targeted apnea test group as compared with the 20-min apnea test group could probably be explained by the decline in the increase rate in Paco²throughout the duration of the apnea test.6–8 

The mean rate of increase in Paco²during the 20-min apnea test was 2.9 ± 0.8 mmHg/min in our study, close to those reported by Ropper et al.  4(2.6 ± 0.8 mmHg/min) and Orliaguet et al.  7(2.7 mmHg/min). We did not find a significant correlation between the Paco²–Ptcco²gradient and the rate of increase in Paco²during the apnea. On one hand, the Paco²increase is highly variable among brain-dead patients during the apnea test because of carbon dioxide washout, atelectasis, cardiac-induced ventilations, and other potentially unknown factors.6On the other hand, the instantaneous Paco²–Ptcco²gradient during such dynamic conditions probably depends on many factors, such as the rate of increase in Paco², but also hemodynamic, temperature, ear lobe vascularization, and skin condition.26At last, 5 min after reconnection of the patient to the ventilator, Ptcco²was very close to Paco², likely because the rapid decrease in Paco²overtook the delay of Ptcco²variations (fig. 2). Afterward, the late increase in the mean Paco²–Ptcco²gradient is unexpected and could be due to a drift of the transcutaneous sensor because of the extreme and rapid changes in Paco²during the apnea test and reconnection to the ventilator.

The area under the ROC curve was 0.983 ± 0.013, indicating a very high accuracy of Ptcco²in predicting that the target Paco²60 mmHg had been reached (fig. 4). The best threshold value on ROC curve analysis was determined as a Ptcco²of 60 mmHg, providing a sensitivity of 0.80, a specificity of 1.00, and a positive predictive value of 1.00 (table 2). Indeed, none of the 12 patients of the Ptcco²targeted apnea test group investigated using this threshold of 60 mmHg exhibited a Paco²lower than 60 mmHg at the end of the apnea test. In a previous study investigating Ptcco²monitoring for apnea testing in brain-dead patients, Lang17showed that a Ptcco²of 66 mmHg had a predictive value of 82% for Paco²of 60 mmHg or greater and empirically recommended a Ptcco²of 60–66 mmHg for the confirmatory arterial blood gas check. This discrepancy as compared with our result may be fully explained, because in their analysis of the Paco²–Ptcco²gradient, they overlooked the dynamic component of Paco²increase during the apnea. Finally, the apnea test procedures performed in their study were either hypoventilation or artificial carbon dioxide insufflation, with a real apnea time of only 0.5–1 min, i.e. , far from the apnea test procedure commonly recommended throughout the world.1,3 

For the 12 patients of the Ptcco²targeted apnea test group, the mean duration of the apnea test, hypercapnia, acidosis, and decrease in Pao²at the end of the apnea test were significantly reduced as compared with the 20-min apnea test group. This result is important, because performing an apnea test in brain-dead patients may lead to complications, such as hypotension and hypoxia, especially in patients with hemodynamic instability. In the worst possible situation, the apnea test may exceptionally induce a sudden and irreversible cardiac arrest, which prevents any organ donation.9,27,28Limitation of the duration of the apnea test reduces the importance of hypercapnia, acidosis, and decrease in Pao²at the end of the apnea and therefore probably reduces the occurrence of complications related to the test. Indeed, we observed significantly less hypotension in the Ptcco²targeted apnea test group than in the 20-min apnea test group. On the other hand, whatever the apnea test group, none of our 32 patients exhibited severe hypoxemia during the apnea test. Indeed, the 20-min preoxygenation period with an inspired oxygen fraction of 100% and the 9-l/min oxygen insufflation inside the endotracheal tube during the apnea test, both performed in our study, probably limited the occurrence of significant hypoxemia during the apnea test.1,9,29Nevertheless, further studies are mandatory to determine whether reduction of the duration of the apnea test in brain-dead patients may lead to a significant improvement in the prognosis of transplanted organs.

Finally, one could argue that the 20-min apnea test period we have chosen for the first 20 investigated patients was dramatically too long, because the mean Paco²at 5 min of the apnea test (63.8 ± 10.1 mmHg, fig. 2) in the 20-min apnea test group was already higher than the threshold of 60 mmHg and because the mean duration of the apnea test in the Ptcco²targeted apnea test group was reduced to 11 ± 4 min. However, the Paco²increase during the apnea test in brain-dead patients is unpredictable from one patient to another, and shorter apnea test times, such as 10 min, have been previously reported as insufficient in some patients to reach the Paco²threshold value of 60 mmHg.6,9,30Similarly, estimation of the required apnea test duration to reach the threshold of 60 mmHg has also been reported as inefficient because of the unpredictability of Paco²increase during the apnea test.4,10This explains why some investigators, who had reported an apnea test lasting from 1 min to more than 1 h, strongly discourage a time-locked approach for the apnea test and conversely insist on arterial blood gas determinations.31On the other hand, keeping in mind the dynamic Paco²–Ptcco²gradient during the apnea test, Ptcco²monitoring offers an on-line estimation of Paco², whereas Paco²analysis from a blood gas sample requires a minimum delay of several minutes to get the result, and eventually a longer time depending on the distance between the laboratory and the ICU.32Nevertheless, one should keep in mind that the high predictive value of Ptcco²reported in our study with the V-Sign® Sensor may not be found with other transcutaneous carbon dioxide sensors, according to technology and time–response differences between devices.

In conclusion, Ptcco²monitoring during the apnea test in brain-dead patients may permit a significant reduction in the duration of the apnea test. We found that a Ptcco²of 60 mmHg has a predictive positive value of 100% in predicting that the target threshold of Paco²of 60 mmHg has been reached. Reducing the duration of the apnea test may limit hypercapnia, acidosis, and decrease in Pao²at the end of the apnea test and eventually occurrence of complications such as hypoxemia and hypotension.

The authors thank David Baker, D.M., F.R.C.A. (Department of Anesthesiology and Critical Care, Hîopital Necker-Enfants Malades, Paris, France), and Alain Mallet, Ph.D. (Department of Biostatistics, Centre Hospitalier Universitaire Pitié-Salpîetrière, Paris, France), for reviewing the manuscript.