Mild Hypercapnia Increases Subcutaneous and Colonic Oxygen Tension in Patients Given 80% Inspired Oxygen during Abdominal Surgery

With approval of the Medical University of Vienna Institutional Review Board (Vienna, Austria) and written informed consent from the patients, we studied 30 patients with American Society of Anesthesiologists physical status I, II, or III who were undergoing elective colorectal resection for cancer of the left colon between September 2003 and August 2004. Patients were recruited from the operating room schedule on the day before surgery. Exclusion criteria included documented coronary or peripheral artery disease, diabetes mellitus, and any symptoms of infection or bowel obstruction. We also excluded patients with preoperative systolic arterial blood pressure greater than 170 mmHg or diastolic arterial pressure greater than 90 mmHg and patients taking vasoactive drugs.

Patients were given a standard mechanical bowel preparation (Klean-Prep®; Norgine, Marburg, Germany; macrogol 3350, polyethylenglycol 59 g) the night before surgery. Patients were allowed to drink until midnight and received 1,000 ml lactated Ringer’s solution overnight. Thirty minutes before skin incision, patients were given prophylactic intravenous antibiotic therapy (1.5 g metronidazole and 1.5 g cefuroxime). Anesthesia was induced with fentanyl (1–3 μg/kg), propofol (2–3 mg/kg), and rocuronium (0.6 mg/kg); anesthesia was maintained with sevoflurane adjusted to keep arterial blood pressure within 20% of the preinduction value. Fentanyl was given continuously during surgery at a rate of 3 μg · kg⁻¹· h⁻¹. No vasoactive drugs were used intraoperatively.

After induction of anesthesia and tracheal intubation, patients were assigned to normocapnia or mild hypercapnia using a set of computer-generated random numbers; assignments were kept in sealed envelopes until used. All patients received 80% oxygen in air. End-tidal carbon dioxide partial pressure was maintained near 35 mmHg (normocapnia) or 50 mmHg (mild hypercapnia). A parallel-group randomized design was used in lieu of a crossover design because there was insufficient time during surgery to test more than one end-tidal carbon dioxide partial pressure. End-tidal carbon dioxide partial pressure was chosen as the end point because it was easy to adjust a patient’s carbon dioxide concentration to these levels. The designated end-tidal concentration in the normocapnia group was maintained simply by adjusting the rate of mechanical ventilation with a 10-ml/kg tidal volume and 1–3 l/min fresh gas flow. In the mild hypercapnia group, the soda lime carbon dioxide absorbent was removed from the anesthesia machine. Ventilation was maintained with a 10-ml/kg ideal body weight tidal volume and a respiratory rate of 8 breaths/min. Fresh gas flow was then adjusted to obtain an end-tidal concentration of 50 mmHg. The attending anesthesiologists were aware of group assignment.

Lactated Ringer’s solution (500 ml) was given in the recovery room. A bolus of 10 ml/kg lactated Ringer’s solution was given before induction of anesthesia. Subsequently, the solution was infused at a rate of 10 ml · kg⁻¹· h⁻¹. Blood loss was replaced with colloid (Voluven®, HES 130/0.4, 6%; Fresenius Kabi GmbH, Graz, Austria) at a 1-2:1 ratio; supplemental fluid was given as necessary to maintain urine output of at least 1 ml · kg⁻¹· h⁻¹. Because we excluded patients with significant cardiovascular disease, erythrocytes were transfused as necessary to maintain hematocrit at 26% or greater.

Upper-body forced-air warming was used on the left arm and head to keep patients normothermic. However, local heating at the tissue oxygen measurement sites (described in the next section) was avoided.

Demographic data, American Society of Anesthesiologists physical status, and duration of surgery were recorded. We also recorded all routine anesthetic, respiratory, and hemodynamic values; fluid administration; and urinary output. Inspired oxygen fraction, oxygen saturation, end-tidal sevoflurane, and carbon dioxide concentrations were measured during anesthesia. Core temperature was measured in the distal esophagus (Mon-a-therm; Tyco-Mallinckrodt Anesthesiology Products, St. Louis, MO). Routine anesthetic measurements were recorded at 10-min intervals throughout the perioperative period.

Forearm-minus-fingertip skin temperature gradients were measured on the right arm as an index of arteriovenous shunt vasoconstriction. Skin temperature gradients correlate well with volume plethysmography.16Values exceeding 0°C indicate vasoconstriction and correlate with onset of the core temperature plateau.17 

After induction of anesthesia, a 20-gauge cannula was inserted into a radial artery. Arterial pressure was measured continuously from this catheter throughout surgery; arterial blood was sampled for gas analysis when bowel tissue oxygen tension was measured (details below).

After induction of anesthesia, a silastic tonometer was inserted in the subcutaneous tissue of the right upper arm to measure tissue oxygen tension and temperature. The tonometer consisted of 15 cm of tubing filled with hypoxic saline; 10 cm of the tube was tunneled in subcutaneously. A Clarke-type oxygen sensor and thermistor (Licox; Gesellschaft für Medizinische Sondensysteme, GmBH, Kiel, Germany) were placed in the subcutaneous portion of the silastic tube. In vitro  accuracy of these oxygen sensors is ±3 mmHg for the range from 0 to 100 mmHg and ±5% for 100 to 360 mmHg (in a water bath at 37°C).18The oxygen sensor calibration remains stable (within 8% of baseline value for room air) in vivo  for at least 8 h. Temperature sensitivity of the sensors is 0.255/°C. However, thermistors are incorporated into the probes, and temperature compensation was included in the tissue oxygen tension (Psqo²) calculations. We calibrated individual oxygen sensors by exposing them to room air (ambient partial pressure of oxygen = 154 mmHg); measurements in air were within 10% of 154 mmHg in all cases. After each investigation, the sensors were again exposed to ambient air to exclude a significant drift of the oxygen sensor; none drifted more than 10% from baseline.

To measure intramural intestinal oxygen partial pressure (Pimo²), the same surgeon inserted oxygen sensors through 20-gauge cannula into a section of descending colon (approximately 20 cm away from the tumor) that was subsequently to be resected. The probes were inserted into the tissue plane between the serosa and mucosa after mobilization of the left colon. In addition, needle temperature probes were inserted adjacent to the oxygen probes (Tyco-Mallinckrodt Anesthesiology Products). To obtain temperature corrected measurements, we adjusted the Licox monitor according to the measured temperature. Care was taken to minimize handling of the intestine and to return the bowel to a neutral position.

All intraoperative measurements except Pimo²were recorded at 10-min intervals, after an initial 30-min equilibration period. Pimo²was measured every 60 s for 15 min, starting approximately 1 h after the beginning of surgery. Surgery was discontinued during Pimo²measurements to enhance reading stability. Upon completion of the measurement period, the measurement sites were inspected for edema or hematoma.

The number of patients required for this trial was estimated as follows: Previous studies suggest that a difference in Psqo²of approximately 15 mmHg is clinically important. We have previously shown that hypercapnia increases subcutaneous tissue oxygenation by approximately the same amount.11,13The standard deviations of Psqo²values in similar studies range from 10 to 20 mmHg, with an average of 15 mmHg. Assuming a difference of at least 15 mmHg between our two treatment groups, 15 patients in each group provide an 80% power to detect a significant difference between the two groups at an α level of 0.05.

Because intraindividual values were stable during our 15-min measurement period, values repeated over time were first averaged within each patient over the 15-min period of intestinal tissue oxygen measurements and then averaged across each group. Potential confounding factors and outcomes of the study were analyzed with chi-square, unpaired two-sided t , or Mann–Whitney rank sum tests. Data that were normally distributed are presented as mean ± SD; others are presented as median (interquartile range). P < 0.05 was considered statistically significant.

Intramural intestinal oxygen partial pressure was recorded from all 15 patients in each group; however, data of one patient had to be excluded because of technical problems with the oxygen probe. Psqo²was only recorded from 12 hypercapnia patients and 14 normocapnia patients. Age, weight, height, and American Society of Anesthesiologists physical status classification were similar in the two groups (table 1). The groups did not differ with regard to duration of surgery, end-tidal sevoflurane concentration, or amount of intraoperative fluid or drugs administered. Core temperatures and urinary output were similar, as well (table 2). Heart rate was significantly higher in the hypercapnia group (76 ± 12 vs.  68 ± 9; P = 0.049). No edema or hematoma was observed at the intestinal oxygen and temperature measurement sites.

During the entire surgery, arterial carbon dioxide pressure was 54 ± 6 mmHg in the hypercapnia group and 39 ± 3 mmHg in the normocapnia group (P < 0.001); arterial oxygen pressure was similar in the two groups. pH was slightly, but significantly, less in the hypercapnia group than in the normocapnia group (7.27 ± 0.05 vs.  7.38 ± 0.03; P < 0.001). Psqo²was significantly greater in the hypercapnia group (116 ± 29 mmHg) than in the normocapnia group (84 ± 25 mmHg).

During the measurement period of Pimo², arterial carbon dioxide pressure was 55 ± 5 mmHg in the hypercapnia group and 40 ± 3 mmHg in the normocapnia group (P < 0.001); arterial oxygen pressure was similar in the groups (table 3). Psqo²was significantly greater in the hypercapnia group (116 ± 29 mmHg) than in the normocapnia group (84 ± 25 mmHg; P = 0.006; fig. 1). Pimo²was significantly greater in patients assigned to hypercapnia (107 [81–129] mmHg) than normocapnia (53 [41–104] mmHg; P = 0.020; fig. 2). Subcutaneous and colon temperatures were similar in the groups during these measurements, as were sevoflurane concentration, core temperature, arterial oxygen pressure, and mean arterial pressure (table 3).

There was weak correlation between tissue oxygenation in the arm and in the colon (r = 0.46, P = 0.022; fig. 3).

Wound infections are established perioperatively, a period that was identified as “decisive” nearly 50 yr ago.19This decisive period explains why perioperative factors influence the incidence of infection even though infections are typically detected days after surgery. Wound infections are still considered a surgical complication, although it is well established that infection risk is related to preexisting diseases and type and duration of surgery, and can be reduced by providing appropriate and timely antibiotics.20 

Another strong predictor of surgical site infection is subcutaneous oxygenation and perfusion.7For example, subcutaneous tissue oxygen values less than 60 mmHg are associated with a wound infection rate exceeding 20%.7Furthermore, the leakage rate for anastomoses ranges from 3 to 10%2and correlates with tissue oxygenation in the colon. For example, the leakage rate progressively increased to 100% as bowel oxygen tension was reduced to less than 20 mmHg in a rabbit model.8Another study in humans showed that perianastomotic oxygen tensions less than 20 mmHg are associated with an extremely high leakage rate.9 

Our current study extends previous results from animals11and confirms that mild hypercapnia substantially improves subcutaneous and colonic oxygenation, even in patients whose tissue oxygenation is already high because they are given 80% inspired oxygen. Specifically, subcutaneous tissue oxygen values were roughly 84 ± 25 mmHg in normocapnic patients given supplemental oxygen. With mild hypercapnia in addition to supplemental oxygen, tissue oxygen tension was 116 ± 29 mmHg, a statistically significant increase. Intramural colon tissue oxygen tension in the hypercapnia group was nearly twice that of the normocapnia group (approximately 53 [41–104] mmHg with normocapnia and 107 [81–129] mmHg with hypercapnia)—an even more important result because the colon is probably the critical tissue determining risk of both infection and wound dehiscence.

However, the issue is not simply the extent to which hypercapnia improves subcutaneous oxygenation, but also the absolute oxygenation range. Hopf et al.  7showed that subcutaneous tissue oxygen levels greater than 80 mmHg in the postoperative period essentially eliminate surgical wound infections, whereas values less than 40 mmHg are associated with a nearly 50% infection rate in colon surgical patients. Our absolute values of tissue oxygen tension were already relatively high in the normocapnic patients (because of supplemental oxygen administration). Whether an increase beyond 80 mmHg further reduces wound infection is not known, especially in our fairly healthy colon surgical patient population.

In contrast, tissue oxygenation in certain high-risk patient populations remains low even with supplemental oxygen administration. Populations at high risk for wound complications include obese patients,21smokers,22diabetics,23and patients with peripheral vascular disease.24Therefore, any treatment or treatment combination that increases tissue oxygenation might be beneficial in these patient populations. We recognize that only outcome studies in various populations will determine whether the observed increases in subcutaneous and colonic oxygenation actually reduce infection risk. However, our results suggest that such studies would be worth conducting.

Subcutaneous oxygen partial pressure was measured from a needle-induced surrogate wound in the arm, which is the classic method of evaluating perioperative tissue oxygen tension and has been used in numerous previous studies.4,7,25The primary benefit of this location is its convenience and the fact that measurements can be easily conducted during surgery. Microscopic tissue injury to subcutaneous tissues at the sensor insertion site in the upper arm is not only inevitable, but a deliberate feature of the model that is designed to mimic surgical trauma.26,18It seems likely this feature contributes to the ability of upper arm oxygenation to predict infection risk.

Placement of the oxygen probes in the intestinal tissue was performed approximately 1 h after incision, when the left colon was mobilized. The probes were inserted into the descending colon (approximately 20 cm away from the tumor) that was subsequently to be resected. One single surgeon, who took special care handling the intestinal tissue, inserted all of the probes. Although we did not observe edema or hematomas at the insertion sites, microscopic tissue injury at the intestinal sensor site is inevitable. Again, this injury should not be considered a limitation of the technique, but a feature that mimics surgical trauma and, thus, more realistically reflects oxygenation that might be observed near an intestinal anastomosis.

We measured colon tissue oxygenation in the descending uninjured colon rather than at the anastomosis directly. We have shown that tissue oxygenation at the anastomosis is considerably lower in pigs.27However, supplemental oxygen administration increased tissue oxygenation significantly at the anastomosis, suggesting that interventions that improve intramural oxygenation in normal intestine are likely to have similar effects at the site of a surgical incision.

Although there was little correlation between tissue oxygenation in the arm and in the colon, both tissues reacted comparably to hypercapnia. This result mirrors those in pigs that were exposed to hypercapnia with and without supplemental oxygen.11However, we can probably not make the general assumption that the subcutaneous tissue in the upper arm reflects intestinal tissue and, thus, can be used as a surrogate measurement site. This might especially be the case during certain interventions, e.g. , supplemental fluid administration, which affect oxygenation in intestinal tissue, subcutaneous wound tissue, and subcutaneous tissue in the upper arm differently.28 

A limitation of our study was that we recorded tissue oxygen tension in the left colon only during a brief intraoperative period. It would be of substantial interest to measure tissue oxygenation in the colon and at the anastomosis during the postoperative period. However, it is not currently possible to measure colon tissue oxygenation after the abdomen is closed in humans. Although there was a statistically significant difference in blood loss between normocapnic and hypercapnic patients, it is unlikely that a difference of 150 ml would affect tissue oxygen tension.

In summary, mild intraoperative hypercapnia in patients receiving supplemental oxygen increased tissue oxygen tension in both the arm and the descending colon by amounts that are likely to produce clinically important reductions in infection risk. Therefore, results of the current study suggest that mild hypercapnia may be another inexpensive, low-risk intervention to reduce the incidence of serious and expensive complications—although this proposed pathomechanism awaits confirmation in an outcome trial.

The authors thank Gilbert Haugh, M.S., for the statistical analysis and Nancy Alsip, Ph.D., for editing the manuscript (both of the University of Louisville, Louisville, Kentucky).