GEN

ERA

LT

HO

RA

CIC

Retrograde Jejunogastric Decompression AfterEsophagectomy Is Superior to Nasogastric DrainageVarun Puri, MD,* Yinin Hu, MS,* Tracey Guthrie, RN, Traves D. Crabtree, MD,Daniel Kreisel, MD, Alexander S. Krupnick, MD, G. Alexander Patterson, MD, andBryan F. Meyers, MD

Department of Surgery, Washington University School of Medicine, St. Louis, Missouri

tsmg3pclba

ti

Background. Nasogastric tubes (NG) are commonlyused for maintaining conduit decompression after esoph-agectomy. We investigated the use of retrograde tubegastrostomy (RG) after esophagectomy.

Methods. Patients underwent either NG or RG place-ment for postoperative conduit decompression. Bothtubes were maintained on low continuous suction.

Results. Between 2000 and 2008, 306 patients under-went esophagectomy with reconstruction. One hundredninety-three patients underwent NG and 113 underwentRG placement. The 2 groups were comparable in age,gender, tumor stage, and smoking status. Patients in theNG group were more likely to have received neoadjuvanttherapy and to have a thoracotomy for esophagectomy.The incidence of respiratory complications was lower inthe retrograde group compared with the NG group:Pneumonia, 9 of 113(8.0%) vs 50 of 193 (25.9%), p < 0.001;respiratory failure requiring bronchoscopy or reintuba-

tion, 12 of 113 (10.8%) vs 46 of 193 (23.8%), p � 0.004;

Tower, Barnes Jewish Hospital, One Barnes Jewish Hospital Plaza, St.Louis, MO 63110; e-mail: puriv@wudosis.wustl.edu.

© 2011 by The Society of Thoracic SurgeonsPublished by Elsevier Inc

aspiration, 4 of 113 (3.5%) vs 20 of 193 (10.4%), p �0.045.The incidence of cardiac dysrhythmias was also lower inthe retrograde group (18 of 113 [15.9%] vs 69 of 193[35.8%], p < 0.001). The incidence of wound complica-ions, myocardial infarction, stroke, and conduit necro-is-anastomotic leak was similar between groups. In aultivariate regression model an NG tube was the stron-

est predictor for postoperative pneumonia (odds ratio.27, 95% confidence interval 1.50 to 7.12). The otherredictors were prior chest surgery, smoking, and thora-otomy incision. There were 4 minor complications re-ated to the retrograde tube (wound infection n � 1,roken tube requiring endoscopy n � 2, tube caught innastomosis detected intraoperatively n � 1).Conclusions. Retrograde gastrostomy decompression of

he conduit after esophagectomy is effective and dimin-shes complications compared with NG tube drainage.

(Ann Thorac Surg 2011;92:499–503)

© 2011 by The Society of Thoracic Surgeons

Postoperative distention of the gastric conduit afteresophagectomy can predispose the patient to signif-

icant complications. Animal models have shown thatgastric distention leads to ischemia in the gastric tube [1].Additionally, pulmonary aspiration of gastrointestinalcontents is a significant risk in the presence of a dilated,fluid-filled conduit [2]. A dilated conduit may also stressthe anastomosis mechanically, thus compromising anas-tomotic healing. The majority of surgeons use nasogastrictube drainage for gastric conduit decompression afteresophagectomy. This decompression is usually requiredfor at least 5 to 7 days.

Nasogastric tubes are uncomfortable and lead to manycomplications of their own. Inadvertent tube removal,tube migration, gastrointestinal bleeding, nasal alar ne-crosis, sinusitis, and aspiration pneumonitis are only afew of these complications [3]. The use of retrograde tube

Accepted for publication March 15, 2011.

*The first two authors contributed equally to this manuscript.

Presented at the Poster Session of the Forty-seventh Annual Meeting ofThe Society of Thoracic Surgeons, San Diego, CA, Jan 31–Feb 2, 2011.Winner of the Blue Ribbon as the top General Thoracic Poster.

Address correspondence to Dr Puri, Cardiothoracic Surgery, 3108 Queeny

gastrostomy after esophagectomy has been described inthe past [4]. We faced many of the aforementionedcomplications with nasogastric tubes and thus institutedthe use of retrograde gastrostomy decompression tubesafter esophagectomy in our patients several years ago. Tostudy the impact of retrograde gastrostomy drainage, thisretrospective review was performed.

Patients and Methods

We performed a review of all patients undergoing esoph-agectomy at our institution between January 2000 andDecember 2008. Patient data were abstracted from aprospectively maintained institutional database and in-dividual chart review carried out for missing data. Alloperative records were individually reviewed. An Insti-tutional Review Board exemption was obtained for thisretrospective study.

All patients undergoing esophagectomy received ei-ther nasogastric (NG) or retrograde gastrostomy (RG)tubes for postoperative conduit decompression. The NGtubes were 16-18 Fr in size and were passed into theconduit during creation of the anastomosis. They weresecured in place with adhesive tape to the nasal ala. The

RG tubes were18 Fr silastic catheters (Kendall Argyle

0003-4975/$36.00doi:10.1016/j.athoracsur.2011.03.082

500 PURI ET AL Ann Thorac SurgCONDUIT DECOMPRESSION AFTER ESOPHAGECTOMY 2011;92:499–503

GEN

ERA

LT

HO

RA

CIC

Silicone Salem Sump tube; Covidien, Mansfield, MA)with multiple holes at the end and had a sump portsimilar to NG tubes. The RG tube was threaded into theconduit in retrograde fashion by a double purse-stringsuture on the antimesenteric wall of a proximal jejunalloop approximately 15 to 20 cm from the ligament ofTreitz (Figs 1–3). A jejunal feeding tube was placed intothe same loop about 5 cm distal to the RG tube. Tubeposition was confirmed on postoperative X-rays (Fig 4).

Fig 1. Relation of retrograde gastrotomy and feeding jejunostomy.

Fig 2. Proximal jejunum is tacked to the anterior abdominal wall after

placement of retrograde gastrostomy and feeding jejunostomy tubes.

Both tubes were brought out through the anteriorabdominal wall and secured in place. Our technique issimilar to that described by Urschel and colleagues [4].Nasogastric or RG tubes were maintained on lowcontinuous suction. When decompression was no lon-ger needed, NG tubes were removed and RG tubeswere capped and removed prior to discharge.

Fig 3. Final position of retrograde gastrostomy and feeding jejunostomytubes.

Fig 4. X-ray verification of tube position and conduit decompression.

The inset shows the radio-opaque tip of the tube.

s

501Ann Thorac Surg PURI ET AL2011;92:499–503 CONDUIT DECOMPRESSION AFTER ESOPHAGECTOMY

GEN

ERA

LT

HO

RA

CIC

Statistical AnalysisContinuous data were summarized by the mean �standard deviation or median (interquartile range),and comparisons between the 2 groups were madewith unpaired Student t tests for means of normallydistributed continuous variables and Wilcoxon ranksum tests for skewed data. Categoric variables weresummarized by frequency with percentage, and �2 orFisher exact tests were used to compare categoric databetween groups. To determine the predictors of post-operative pneumonia and dysrhythmia, binary logisticregression models were utilized for each dependentvariable of interest. Multivariate models were con-structed by purposeful selection of covariates to in-clude variables with p values less than 0.20 in thebivariate analysis. All data analyses were done usingSPSS (SPSS 13.0 for Windows; SPSS Inc, Chicago, IL).A p value less than 0.05 was considered statisticallyignificant.

Results

Between January 2000 and December 2008, 306 patientsunderwent esophagectomy with reconstruction. Of these,54 [17.6%] were female and the mean age of patients was60.8 (range 26 to 88) years. The majority of operationswere performed for cancer. One hundred forty nine of306 (48.7%) patients underwent transhiatal esophagec-tomy while 157 (51.3%) patients underwent Ivor-Lewis orMcKeown three-stage esophagectomy. One hundredninety three of 306 (63%) patients underwent NG tubeplacement while 113 of 306 (37%) had RG insertion. Thegroups were similar in age, gender, history of priorthoracic surgery, smoking status, and indication for sur-gery. Patients in the NG group were more likely to havereceived neoadjuvant chemotherapy or neoadjuvant ra-diation therapy (Table 1). Additionally, patients in theNG group underwent thoracotomy more frequently thanthose in the RG group.

Table 1. Comparison of Patient Demographics in theNasogastric and Retrograde Gastrostomy Groups

Patient DemographicsNasogastric

(n � 193)

RetrogradeGastrostomy

(n � 113) p Value

PreoperativeMale 156 (80.8%) 96 (85.0%) 0.438Age 60.4 � 12.1 61.4 � 12.3 0.471Prior thoracic surgery 24 (12.4%) 13 (11.5%) 0.858History of tobacco use 155 (80.3%) 84 (74.3%) 0.252Malignant indication 188 (97.4%) 108 (95.6%) 0.507Neoadjuvant chemo 94 (48.7%) 11 (9.7%) �0.001Neoadjuvant radiation 95 (49.2%) 11 (9.7%) �0.001

OperativeThoracotomy incision

(Ivor-Lewis,McKeown)

119 (61.7%) 39 (34.5%) �0.001

Pyloroplasty- 99 (51.3%) 88 (77.9%) �0.001

pyloromyotomy

The incidence of respiratory complications was sig-nificantly lower among patients receiving RG for con-duit decompression. Retrograde gastrostomy was as-sociated with lower risk of pneumonia (8% vs 25.9%,p � 0.001), respiratory failure requiring bronchoscopyor reintubation (10.6% vs 23.8%, p � 0.004), and aspi-ration (3.5% vs 10.4%, p � 0.05). Additionally, theincidence of cardiac dysrhythmias was significantlylower in the RG group (15.9% vs 35.8%, p � 0.001)(Table 2). Atrial fibrillation and supraventricular tachy-cardia comprised the majority of dysrhythmias. Theincidence of stroke, chylothorax, conduit necrosis, em-pyema, hemorrhage, pulmonary embolism, wound de-hiscence, wound infection, and myocardial infarctionwere similar between the 2 groups (Table 2). Therewere 4 minor and no major complications related to RGtube placement. One patient developed a skin-levelwound infection requiring bedside debridement andoral antibiotics. Two patients presented with brokentubes at the level of the skin or the jejunum andunderwent endoscopy for tube retrieval. In 1 patient,the RG tube was inadvertently caught in the end-to-end stapled anastomosis between the esophageal rem-nant and the stomach conduit. This required partialopening of the stapled anastomosis and closure withhand-suturing. Anecdotally, several patients who un-derwent NG tube placement inadvertently pulled outtheir tubes and required repositioning with or withoutfluoroscopic-endoscopic guidance. The exact incidenceof this complication was not recorded.

A logistic regression multivariate analysis was per-formed using postoperative pneumonia and dysrhyth-mias as dependent variables. Presence of an NG tube,prior thoracic surgery, history of tobacco use, and useof a thoracotomy incision were predictors of postoper-ative pneumonia. Among these, the presence of an NG

Table 2. Univariate Analysis of Postoperative Morbidity inthe Two Groups

ComplicationsNasogastric

(n � 193)

RetrogradeGastrostomy

(n � 113) p Value

Aspiration 20 (10.4%) 4 (3.5%) 0.045Atelectasis 27 (14.0%) 4 (3.5%) 0.003Cerebrovascular

accident2 (1.0%) 1 (0.9%) 1.000

Chylothorax 19 (9.8%) 10 (8.8%) 0.842Conduit necrosis 7 (3.6%) 3 (2.7%) 0.750Dysrhythmia 69 (35.8%) 18 (15.9%) �0.001Empyema 5 (2.6%) 1 (0.9%) 0.419Hemorrhage 12 (6.2%) 5 (4.4%) 0.611Pulmonary embolism 8 (4.1%) 2 (1.8%) 0.333Pneumonia 50 (25.9%) 9 (8.0%) �0.001Respiratory failure 46 (23.8%) 12 (10.6%) 0.004Wound dehiscence 6 (3.1%) 4 (3.5%) 1.000Wound infection 16 (8.3%) 9 (8.0%) 1.000Anastomotic leak 3 (1.6%) 2 (1.8%) 1.000

tube was the strongest predictor (odds ratio 3.274; 95%

avtTnotegb

ecptpnaftt

Pi

waatlrald

aTtnfcp

NPHT

502 PURI ET AL Ann Thorac SurgCONDUIT DECOMPRESSION AFTER ESOPHAGECTOMY 2011;92:499–503

GEN

ERA

LT

HO

RA

CIC

confidence interval 1.5 to 7.16; p � 0.001) (Table 3). Age,NG tube, and history of smoking were predictors ofpostoperative dysrhythmia (Table 4).

Comment

Reconstruction after esophagectomy commonly utilizes agastric tube or the colon as conduit. In the early postop-erative period, it is imperative to maintain effectivedecompression of the conduit. Preventing conduit dis-tention helps avoid aspiration of gastrointestinal contentsfrom a dilated, fluid-filled conduit. Conduit distentioncan lead to venous congestion and eventual ischemia,while aspiration can result in postoperative respiratorycomplications.

Nasogastric tubes (16-18 Fr) are generally placed intra-operatively and maintained for varying periods of time inthe postoperative period. Commonly, the tubes are keptin position from 3 to 8 days depending upon surgeonpreference and postoperative radiologic evaluation ofconduit emptying and anastomotic integrity. Nasogastrictubes predispose to respiratory complications includingaspiration, pneumonia, sinusitis, pharyngitis, and laryn-gitis, particularly in patients who have recently under-gone a major operation [5]. Other complications associ-

ted with NG tubes include nasal alar necrosis, hollowiscus perforation, gastrointestinal bleeding, inadvertentracheal intubation, and accidental removal of the tube.his last complication, after an esophagectomy, oftenecessitates replacement of the NG tube, either blindlyr under fluoroscopic or endoscopic guidance. Althoughhe incidence of unintentional tube removal is hard tostablish, occurrence is common enough that some sur-eons routinely use nasal stitches or nasal cartilageridles for NG tube retention.An alternative mode of conduit decompression after

sophagectomy has been suggested by Schuchert andolleagues [6], whereby a pharyngostomy is created toass a tube into the gastric or colonic conduit. However,

his procedure puts the hypoglossal nerve, the glosso-haryngeal nerve, and the lowest branch of the facialerve at risk. Other complications associated with thispproach include tract infection, neck cellulitis, inef-ective drainage, and bleeding. Additionally, antegradeube positioning is more effective with use of suction,

Table 3. Multivariate Analysis of Predictors of PostoperativePneumonia (n � 59)

Variable Number OR 95% CI p Value

asogastric tube 193 3.274 1.498–7.157 0.001rior thoracic surgery 37 2.951 1.329–6.555 0.009istory of tobacco use 239 3.021 1.133–8.053 0.014horacotomy incision 158 2 1.048–3.814 0.032

Significant predictors are shown. Nonsignificant variables include age,gender, neoadjuvant chemotherapy, neoadjuvant radiation, and pyloro-plasty-pyloromyotomy.

CI � confidence interval; OR � odds ratio.

herefore both NG and pharyngostomy tubes are sig-

nificantly more effective with wall suction than withgravity drainage [2].

We investigated the use of retrograde gastrostomytubes for conduit decompression after esophagectomy.Similar decompressive techniques have been used bysurgeons in the past. The RG tube is a soft silastic tube122 cm in length and has a sump port built in, similarto a routine NG tube. The tube is secured to theabdominal wall with simple stitches like routine ab-dominal drains. Not only does the tube drain bygravity and (or) suction, it also stents the pylorus, thusencouraging antegrade flow of gastric secretions. Ad-ditionally, spontaneous gastric emptying can be eval-uated by simply capping the tube.

Our data show a clear advantage for the use of retro-grade gastrostomy over conventional NG drainage afteresophagectomy. The presence of an NG tube was thestrongest of 4 significant risk factors for postoperativerespiratory complications. Nasogastric tubes have beenimplicated in colonization of the upper airways withgastrointestinal flora which secondarily cause pneumo-nia [7]. Also, the presence of an NG tube hamperseffective coughing and compromises pulmonary toilet [8].

ooled upper aerodigestive secretions can easily contam-nate the tracheobronchial tree.

We also noted a lower incidence of cardiac arrhythmiasith the use of RG tubes. Common etiologies of postoper-

tive arrhythmias after esophagectomy include catechol-mine release, hypoxemia, or acidosis [7]. These perturba-ions often accompany inflammation or infection; thus, theowered incidence of pneumonia among patients receivingetrograde decompression may explain the reduced rate ofrrhythmia. No specific protocol was in place for prophy-actic measures against postesophagectomy arrhythmiasuring the study period.The insertion of a retrograde drainage tube adds

bout 15 minutes to the duration of an esophagectomy.he initial learning curve involves the ability to guide

he retrograde tube around the C-loop of the duode-um and beyond the pylorus into the conduit. We

ound that the tube is easier to manipulate into theonduit if the pyloric drainage procedure (pyloroplasty-yloromyotomy) has already been performed. We no-

ticed 4 minor complications during the course of thisretrospective study. Two patients had tubes that brokeat the level of the intestinal insertion site and thus

Table 4. Multivariate Analysis of Predictors of PostoperativeDysrhythmia (n � 87)

Variable No. OR 95% CI p Value

Age 306 1.044 1.018–1.070 �0.001Nasogastric tube 193 2.573 1.358–4.878 0.003History of tobacco use 239 2.061 1.092–4.282 0.043

Significant predictors are shown. Nonsignificant variables include gen-der, neoadjuvant chemotherapy, neoadjuvant radiation, pyloroplasty-pyloromyotomy, prior thoracic surgery, and use of a thoracotomy inci-sion.

CI � confidence interval; OR � odds ratio.

503Ann Thorac Surg PURI ET AL2011;92:499–503 CONDUIT DECOMPRESSION AFTER ESOPHAGECTOMY

GEN

ERA

LT

HO

RA

CIC

required endoscopic snaring for retrieval of the tube.We have since taken care to ensure that the bowel wallpursestring suture and skin-level securing sutures arenot too tight. In our experience, retrograde conduitdecompression provides significant advantages overnasogastric decompression after an esophagectomyand we are now applying this technique in all patientsundergoing open esophageal resection.

References

1. Urschel JD, Antkowiak JG, Takita H. Gastric distention exac-erbates ischemia in a rodent model of partial gastric devas-cularization. Am J Med Sci 1997;314:284–6.

2. Shackcloth MJ, McCarron E, Kendall J, et al. Randomizedclinical trial to determine the effect of nasogastric drainage on

Terrorism

to applications, examinations, and interruption of training.

© 2011 by The Society of Thoracic SurgeonsPublished by Elsevier Inc

3. Cheatham ML, Chapman WC, Key SP, Sawyers JL. A meta-analysis of selective versus routine nasogastric decompres-sion after elective laparotomy. Ann Surg 1995;221:469–78.

4. Urschel JD, Antkowiak JG, Anderson TM, Takita H. Retro-grade jejunogastric tube decompression after esophagec-tomy. J Surg Oncol 1998;68:204–5.

5. Nelson R, Edwards S, Tse B. Prophylactic nasogastric decom-pression after abdominal surgery. Cochrane Database SystRev 2007:CD004929.

6. Schuchert MJ, Pettiford BL, Landreneau JP, et al. Transcervi-cal gastric tube drainage facilitates patient mobility andreduces the risk of pulmonary complications after esophagec-tomy. J Gastrointest Surg 2008;12:1479–84.

7. Leibovitz A, Plotnikov G, Habot B, Rosenberg M, Segal R.Pathogenic colonization of oral flora in frail elderly patientsfed by nasogastric tube or percutaneous enterogastric tube. JGerontol A Biol Sci Med Sci 2003;58:52–5.

8. Joshi N, Localio AR, Hamory BH. A predictive risk index for

tracheal acid aspiration following oesophagectomy. Br J Surg2006;93:547–52.

nosocomial pneumonia in the intensive care unit. Am J Med1992;93:135–42.

Notice From the American Board of Thoracic SurgeryRegarding Trainees and Candidates for Certification WhoAre Called to Military Service Related to the War on

The Board appreciates the concern of those who havereceived emergency calls to military service. They may beassured that the Board will exercise the same sympatheticconsideration as was given to candidates in recognition oftheir special contributions to their country during theVietnam conflict and the Persian Gulf conflict with regard

If you have any questions about how this might affectyou, please call the Board office at (312) 202-5900.

Valerie W. Rusch, MDChair

The American Board of Thoracic Surgery

Ann Thorac Surg 2011;92:503 • 0003-4975/$36.00