Endoscopy 2012; 44(06): 565-571
DOI: 10.1055/s-0031-1291664
Original article
© Georg Thieme Verlag KG Stuttgart · New York

A pilot study to assess mediastinal emphysema after esophageal endoscopic submucosal dissection with carbon dioxide insufflation

Y. Maeda
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
D. Hirasawa
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
N. Fujita
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
T. Obana
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
T. Sugawara
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
T. Ohira
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
Y. Harada
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
T. Yamagata
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
K. Suzuki
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
Y. Koike
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
Y. Yamamoto
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
Z. Kusaka
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
,
Y. Noda
Department of Gastroenterology, Sendai City Medical Center, Sendai, Miyagi, Japan
› Author Affiliations
Further Information

Corresponding author

Y. Maeda, MD
Department of Gastroenterology
Sendai City Medical Center
5-22-1, Tsurugaya
Miyagino-ku
Sendai 983-0824
Miyagi
Japan   
Fax: +81-22-2529431   

Publication History

submitted 26 June 2011

accepted after revision 28 December 2011

Publication Date:
09 March 2012 (online)

 

Background and aims: Mediastinal emphysema sometimes develops following esophageal endoscopic submucosal dissection (ESD) without perforation because the esophagus has no serosa. Carbon dioxide (CO2) insufflation during esophageal ESD may reduce the incidence of mediastinal emphysema. The aim of the present study was to compare the incidence and severity of post-ESD mediastinal emphysema in patients receiving CO2 insufflation vs. standard air insufflation during esophageal ESD.

Patients and methods: A total of 27 patients who had undergone esophageal ESD with insufflation of CO2 between July 2009 and March 2010 were enrolled in this study (CO2 group). Another 105 patients who had undergone esophageal ESD with air insufflation between March 2004 and May 2009 were included as historical controls (air group). Multi-detector row computed tomography (MDCT) was carried out immediately after ESD. A conventional chest radiograph was taken the next day. Mediastinal emphysema findings on MDCT and radiography were compared between the groups.

Results: Mediastinal emphysema detected by chest radiography was 0 % in the CO2 group vs. 6.6 % in the air group (n.s.). Mediastinal emphysema on MDCT was significantly less frequent in the CO2 group compared with the air group (30 % vs. 63 %; P = 0.002). The severity of mediastinal emphysema also tended to be lower in the CO2 group.

Conclusions: Whereas mediastinal emphysema detected by radiography is not so common, MDCT immediately after ESD revealed a certain prevalence of post-ESD mediastinal emphysema. Insufflation of CO2 rather than air during esophageal ESD significantly reduced postprocedural mediastinal emphysema. CO2 can be considered as insufflating gas for esophageal ESD.


Introduction

Insufflation is required during esophageal endoscopic submucosal dissection (ESD) and also during standard gastrointestinal endoscopy to achieve adequate visualization of the bowel lumen. Regardless of the type of endoscope used, it is standard practice to use ambient atmospheric air (also referred to as room air) to insufflate the lumen. After insufflation, air is not absorbed and remains in the gastrointestinal tract, which can lead to bowel distension and abdominal discomfort. In several studies, CO2 insufflation has been shown to be effective in reducing procedure-related pain and discomfort [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]. This is due to the rapid absorption of CO2 from the gastrointestinal tract into the bloodstream and to subsequent pulmonary excretion, unless some type of pulmonary dysfunction exists. The safety of CO2 insufflation for ESD has also been shown in several studies [2] [12] [13].

It is known that mediastinal emphysema can develop even if no perforation is recognized during or at the end of esophageal ESD because the esophagus has no serosa [14] [15] [16] [17] [18] [19] [20]. Since 2004, we have prospectively evaluated mediastinal emphysema after esophageal ESD with multi-detector row computed tomography (MDCT) and reported its prevalence. Most mild mediastinal emphysema observed only by MDCT and not by simple radiography was asymptomatic. However, there were some cases of extensive mediastinal emphysema in which radiography failed to demonstrate the condition. Severe inflammatory changes tend to develop in patients with extensive mediastinal emphysema and such patients have a longer febrile period. MDCT evaluation is desirable in patients who have suspected mediastinal emphysema [21].

Unlike air, CO2 is rapidly cleared from the gastrointestinal tract by passive absorption, CO2 insufflation during esophageal ESD may reduce the incidence of mediastinal emphysema compared with air. The aim of the current pilot study was to estimate the impact of CO2 insufflation on the incidence and severity of post-ESD mediastinal emphysema in preparation for a randomized controlled trial.


Patients and methods

Participants

This study was approved by the institutional review board of the Sendai City Medical Center, and all included patients gave written informed consent before esophageal ESD.

Between July 2009 and March 2010, 30 consecutive patients with 31 esophageal epithelial neoplasms were treated by ESD at the Department of Gastroenterology, Sendai City Medical Center. Two patients were excluded due to severe chronic obstructive pulmonary disease (COPD) and 28 patients with 29 lesions underwent ESD with insufflation of CO2. One further patient was excluded due to synchronous multiple lesions, leaving 27 patients with 27 lesions enrolled in the study (CO2 group).

Between March 2004 and May 2009, 110 consecutive patients with 127 esophageal epithelial neoplasms were treated by ESD with air insufflation at the department. After exclusion of those who had synchronous multiple lesions or those who did not give informed consent for MDCT, 105 patients with 105 lesions were included as historical controls (air group).

Patients who were treated in June 2010 were excluded from this study because insufflation gas was selected at the discretion of the endoscopist during that period.


ESD procedure

ESD was performed using a GIF Q260 J or GIF Q260 endoscope (Olympus Optical Co., Ltd., Tokyo, Japan) with a transparent hood (D-2201 – 11304; Olympus) attached to the tip of the scope. A water jet function (GIF-Q260J; Olympus) or hand-made external water channel (GIF-Q260; Olympus) was utilized for flushing during ESD [22].

After the lesion had been stained with iodine, it was marked and underwent local injection into the submucosa of 10 % glycerin with 0.007 % epinephrine. ESD was performed, as described by Oyama et al. [15], using a hook knife and an electrocautery unit (ICC200; ERBE, Tübingen, Germany). The modes of electric power used were the 50 W auto-cut mode and the 50 W spray-coagulation mode [21]. The ESD procedure and periprocedural management were identical in both groups except for the type of insufflation gas used ([Fig. 1]).

Zoom
Fig. 1 The endoscopic submucosal dissection (ESD) procedure. a White light imaging. b Narrow band imaging (NBI). c Iodine staining imaging. d Marking following iodine staining. e Dissection following submucosal injection. f Ulcer after ESD. g Resected specimen with iodine staining. h Hook knife. i CO2 regulator.

The ESD procedures in the study were performed by five endoscopists who each had at least 5 years’ experience in endoscopy and experience of 20 cases of gastric ESD. The procedure was performed on an inpatient basis.


CO2 insufflation and monitoring system

In the CO2 group, CO2 was administered using a commercially available CO2 regulation unit (OLYMPUS UCR; Olympus Medical System Corp., Tokyo, Japan), which was connected to a CO2 bottle ([Fig. 1i]). A CO2 nasal sampling set with O2 tubing (CapnoLine H O2; Oridion Medical 1987 Ltd., Jerusalem, Israel) was used to monitor endotidal CO2 pressure.


Sedation

The sedation protocol was standardized for all patients. No premedication was given. Standard monitors including electrocardiography, an oscillometric blood pressure cuff, and a pulse oximeter were employed. First, 20 – 50 mg of propofol was administered slowly, followed by constant intravenous infusion (1 – 5 mg/kg/hour) using a syringe pump to control the depth of sedation. When a patient moved, 5 – 10 mg of propofol was administered as a bolus injection and the infusion rate of propofol was increased. The rate of extracellular fluid infusion and the volume of oxygen inhalation were increased against cardiopulmonary suppression, and the rate of propofol infusion was decreased.

An analgesic (pentazocine, 7.5 – 15 mg) was given intravenously at the beginning of sedation in both groups and further injection was administered depending on the patient’s condition. The level of sedation was evaluated according to the American Society of Anesthesiologists classification and maintained at a moderate-to-deep level [23] [24].


Periprocedural patient management

On the day of ESD and the day after, the patient was kept fasted. An antibiotic (cefazoline 1 g twice daily) was administered intravenously for 3 days after the procedure. When a patient suffered from a fever of over 38 °C and/or from poststernal pain, fasting and administration of antibiotics were prolonged until improvement of the symptoms.


Outcome measurement

Chest MDCT was carried out immediately after ESD without administration of contrast medium. Two types of MDCT scanner were used, as available (Aquilion 4 and Aquilion 64 TSX-101A; Toshiba Medical Systems Co., Tokyo, Japan).

Mediastinal emphysema was graded into four categories: Grade 0, no mediastinal emphysema; Grade I, bubbles around the esophagus; Grade II, mediastinal emphysema around the thoracic aorta; Grade III, mediastinal emphysema extending around the heart and/or beyond the mediastinum into the neck; and Grade IV, mediastinal emphysema with pneumothorax and/or subcutaneous emphysema ([Fig. 2]) [21]. The following day, a conventional chest radiograph was taken and presence of mediastinal emphysema was evaluated.

Zoom
Fig. 2 Grade of mediastinal emphysema on multi-detector row computed tomography. Reproduced from Maeda et al. Dig Endosc 2011; 23: 221 – 226, with permission from John Wiley and Sons.

Mediastinal emphysema findings on MDCT and radiography were compared between the groups. Laboratory tests were carried out before ESD (baseline), and on day 1 and day 3 post-ESD. Fever and white blood cell count (WBC) the day after ESD were compared between the two groups. Elevation of serum C-reactive protein (CRP) level between baseline and 3 days after ESD was also compared between the two groups.


Statistical analysis

A two-sided P value of < 0.05 was considered statistically significant for all tests. Continuous variables (e. g. procedure time) were compared by using the two-sample t test, and categorical variables (e. g. incidence of mediastinal emphysema) were compared by using the χ2 test.



Results

Patients

[Tab.1] shows the patient characteristics. The two groups were similar with regard to sex, location of the tumor, histology, and histological depth of invasion of the tumor. The mean age was higher in the CO2 group than in the air group.

Table 1

Characteristics of patients undergoing esophageal endoscopic submucosal dissection with either air or CO2 insufflation.

CO2 group

Air group

P value

Total no. of patients

27

105

Sex, male/female, n

23/4

99/6

0.236

Age, mean ± SD, years

73.4 ± 8.8

68.8 ± 8.0

0.011

Location in esophagus[1]

0.943

Cervical

 0

  1

Upper thoracic

 1

  6

Middle thoracic

20

 74

Lower thoracic

 5

 22

Abdominal

 1

  2

Histology[*]

0.778

Squamous cell carcinoma

26

100

Barrett’s adenocarcinoma

 1

  5

Histological depth[1]

0.120

EP

 9

 46

LPM

12

 29

MM

 2

 21

SM1

 0

  3

SM2~

 4

  6

Tumor size, mean ± SD, mm

21.3 ± 25.3

26.5 ± 31.3

0.365

Resection size, mean ± SD, mm

33.7 ± 35.9

38.6 ± 41.5

0.581

En bloc resection, n (%)

25 (93)

104 (99)

0.199

Local recurrence, n (%)

0 (0)

1 (1)

(0.462)

Procedure time, mean ± SD, minutes

57.6 ± 64.8

62.9 ± 73.1

0.730

EP, carcinoama in situ; LPM, tumor invades lamina propria mucosa; MM, tumor invades muscularis mucosa; SM1, tumor invades upper third of the submucosal layer; SM2~, tumor invades middle third of the submucosal layer or deeper.

* Based on the Japanese Classification of Esophageal Cancer [25].


Although most patients in both groups had squamous cell carcinoma, one patient in the CO2 group and five patients in the air group had Barrett’s adenocarcinoma.

The average procedure time was 57.6 minutes in the CO2 group and 62.9 minutes in the air group (n.s.). The average size of the resected specimen was 33.7 mm vs. 38.6 mm, respectively (n.s.). The rate of en bloc resection was 93 % vs. 99 %, respectively (n.s.). Two patients in the CO2 group and one patient in the air group underwent piecemeal resection. One patient each in both groups underwent piecemeal resection because they had recurrent superficial esophageal cancer that was difficult to treat by ESD. The other patient in the CO2 group had a lesion in the lower thoracic esophagus after Nissen operation, and it was hard to keep adequate visualization during the procedure due to the deformity of the esophagus. One patient in the air group developed local recurrence and was treated by repeat ESD. No perforation or postprocedural bleeding requiring transfusion or endoscopic treatment was encountered in either of the groups.


Frequency and severity of mediastinal emphysema

Mediastinal emphysema was visualized on MDCT significantly less often in the CO2 group compared with the air group (30 % vs. 63 %; P = 0.002) ([Fig. 3]). Chest radiography could not detect any mediastinal emphysema in the CO2 group, whereas 6.6 % of the mediastinal emphysema cases in the air group were detected on radiography (P = 0.370) ([Fig. 4]).

Zoom
Fig. 3 Detection rate of mediastinal emphysema by multi-detector row computed tomography.
Zoom
Fig. 4 Detection rate of mediastinal emphysema by chest radiography.

The severity of mediastinal emphysema on MDCT was Grade I, 26 % in the CO2 group vs. 47 % in the air group; Grade II, 3.7 % vs. 11 %; Grade III, 0 % vs. 5.7 %; and Grade IV, 0 % in both groups ([Tab. 2]). The CO2 group showed a lower grade of mediastinal emphysema on MDCT (P = 0.031).

Table 2

Incidence and degree of mediastinal emphysema on multi-detector row computed tomography.

Severity of mediastinal emphysema

CO2 group (n = 27), %

Air group (n = 105), %

P value[*]

Grade 0

70.4

37.1

0.002

Grade I

25.9

46.7

0.051

Grade II

 3.7

10.5

0.275

Grade III

 0

 5.7

0.204

Grade IV

 0

 0

 – 

* P values for differences between each grade compared with the other grades.



Clinical course

The mean WBC count on the day after the procedure was 8995 /μL in the CO2 group vs. 9387 /μL in the air group (n.s.). The mean serum CRP level elevation on day 3 after the procedure from baseline was significantly lower in the CO2 group (2.2 mg/dL vs. 3.3 mg/dL; P = 0.021) ([Fig. 5]). The incidence of fever of over 38 °C was 19 % vs. 11 %, respectively (n.s.). The mean duration of fever over 38 °C was 1.2 days vs. 1.5 days, respectively (n.s.). The mean duration of fasting was the same in both groups ([Table 3]). All patients recovered with conservative treatment, such as prolonged fasting and administration of antibiotics. No perforation or post-procedural hemorrhage occurred in either group. The incidence of adverse events was low and did not differ between the two groups. No serious cardiopulmonary complications occurred ([Table 4]).

Table 3

Clinical course of patients with mediastinal emphysema after esophageal endoscopic submucosal dissection.

CO2 group (n = 27)

Air group (n = 105)

P value

Fever ≥ 38 °C, %

19

11

0.32

Duration of fever ≥ 38 °C, days

 1.2

 1.5

0.32

Duration of fasting, days

 2.4

 2.4

0.33

Table 4

Complications.

CO2 group (n = 27), n (%)

Air group (n = 105), n (%)

P value

Perforation

0 (0)

 0 (0)

 – 

Post-procedure hemorrhage

0 (0)

 0 (0)

 – 

Esophageal stricture with dysphagia

1 (3.7)

14 (13)

0.286

Sinusitis

0 (0)

 1 (1)

0.462

Pneumonia

0 (0)

 1 (1)

0.462

Liver abscess

1 (3.7)

 0 (0)

0.462

Zoom
Fig. 5 C-reactive protein (CRP) elevation and white blood cell count (WBC) after endoscopic submucosal dissection.


Discussion

CO2 is rapidly absorbed from the gastrointestinal tract into the bloodstream and subsequently excreted through expiration. The usefulness and safety of CO2 as an alternative to air has been demonstrated in several randomized controlled studies in patients who underwent diagnostic colonoscopy, endoscopic retrograde cholangiopancreatography (ERCP), and double-balloon enteroscopy in conscious or intravenously sedated patients [3] [4] [5] [6] [7] [8] [9] [10] [26]. No pulmonary complications or CO2 retention from CO2 insufflation was reported in patients who did not have any pulmonary dysfunction. As in previous studies, the present study also excluded patients with severe COPD.

Mediastinal emphysema can develop after esophageal ESD even in the absence of perforation because the esophagus has no serosa. However, whereas mediastinal emphysema detected by radiography is not so common, MDCT immediately after ESD revealed a certain prevalence of post-ESD mediastinal emphysema. Previous studies have reported that the incidence of mediastinal emphysema after esophageal ESD with air insufflation as assessed by radiography was about 7 % [15] [16] [17] [18]. The present study showed a similar incidence of 6.6 % in the air group. However, the incidence of mediastinal emphysema detected by radiography in the CO2 group was 0 %, showing a tendency of reduction. Mediastinal emphysema recognized by radiography was Grade II or more on MDCT. Mild mediastinal emphysema on MDCT could not be detected by radiography. In the CO2 group, the grade of mediastinal emphysema on MDCT also tended to be lower than that in the air group, and radiography could not detect any mediastinal emphysema in the CO2 group. MDCT is 10 times more sensitive than radiography in detecting mediastinal emphysema. CT was performed immediately after the procedure and a radiograph was taken the next morning. The discrepancy between the frequency of depiction of mediastinal emphysema in each modality may be influenced by the difference in the timing of imaging, sensitivity of each modality to gas, and spontaneous absorption.

The incidence of fever and its duration showed no difference between the groups. The mean CRP level on day 3 after ESD was lower in the CO2 group, though the mean WBC on the day after ESD was not different. Fever or laboratory data can be affected not only by mediastinal emphysema but also by another febrile complication such as pneumonia. Furthermore, mild mediastinal emphysema such as CT Grade I (bubbles around the esophagus) is asymptomatic. Although patients showing severe mediastinal emphysema tended to develop severe inflammatory changes and have a longer febrile period, demonstration of the difference of clinical course resulting from mediastinal emphysema would require a very large number of patients because the number of patients who had severe mediastinal emphysema was small [21].

During ESD, it is mandatory to maintain an adequate endoscopic view with insufflation of gas in order to perform a safe procedure. In cases with exposure of the muscular layer, leakage of the insufflated gas into the mediastinum via the gaps between the muscle fibers is considered to be the mechanism of development of mediastinal emphysema. However, mediastinal emphysema can develop even in cases without exposure of the muscular layer [21]. Preservation of the submucosa is not a perfect barrier against air leakage. Insufflation of CO2 compared with air during esophageal ESD significantly reduced postprocedural mediastinal emphysema. CO2 insufflation may limit the increase in inner pressure of the esophagus as a result of rapid absorption into the bloodstream. Mediastinal emphysema resulting from CO2 insufflation may also rapidly disappear because leaking CO2 in the mediastinum is also quickly absorbed into the bloodstream.

The present study has some limitations, such as the expertise in ESD of endoscopists, which may have affected the incidence of mediastinal emphysema because the CO2 group procedures were performed at a later date than the air group ones. Additional work is needed to demonstrate the usefulness of CO2 insufflation during esophageal ESD in a vigorous way, such as a randomized controlled trial.

Conclusion

Insufflation of CO2 during esophageal ESD significantly reduced postprocedural mediastinal emphysema compared with air insufflation. CO2 insufflation can be considered for esophageal ESD.




Competing interests: None

Acknowledgments

Part of this paper was presented at Digestive Disease Week, 7 – 11 May 2011, Chicago, Illinois, USA [27].


Corresponding author

Y. Maeda, MD
Department of Gastroenterology
Sendai City Medical Center
5-22-1, Tsurugaya
Miyagino-ku
Sendai 983-0824
Miyagi
Japan   
Fax: +81-22-2529431   


Zoom
Fig. 1 The endoscopic submucosal dissection (ESD) procedure. a White light imaging. b Narrow band imaging (NBI). c Iodine staining imaging. d Marking following iodine staining. e Dissection following submucosal injection. f Ulcer after ESD. g Resected specimen with iodine staining. h Hook knife. i CO2 regulator.
Zoom
Fig. 2 Grade of mediastinal emphysema on multi-detector row computed tomography. Reproduced from Maeda et al. Dig Endosc 2011; 23: 221 – 226, with permission from John Wiley and Sons.
Zoom
Fig. 3 Detection rate of mediastinal emphysema by multi-detector row computed tomography.
Zoom
Fig. 4 Detection rate of mediastinal emphysema by chest radiography.
Zoom
Fig. 5 C-reactive protein (CRP) elevation and white blood cell count (WBC) after endoscopic submucosal dissection.