Journal of Gastrointestinal Surgery

, Volume 17, Issue 2, pp 244–256

Predictors of Unsuccessful Laparoscopic Resection of Gastric Submucosal Neoplasms

Authors

  • Sabha Ganai
    • Department of SurgeryThe University of Chicago Medical Center
  • Vivek N. Prachand
    • Department of SurgeryThe University of Chicago Medical Center
  • Mitchell C. Posner
    • Department of SurgeryThe University of Chicago Medical Center
  • John C. Alverdy
    • Department of SurgeryThe University of Chicago Medical Center
  • Eugene Choi
    • Department of SurgeryThe University of Chicago Medical Center
  • Mustafa Hussain
    • Department of SurgeryThe University of Chicago Medical Center
  • Irving Waxman
    • Center for Endoscopic Research and TherapeuticsThe University of Chicago Medical Center
  • Marco G. Patti
    • Department of SurgeryThe University of Chicago Medical Center
    • Department of SurgeryThe University of Chicago Medical Center
    • The University of Chicago Medicine
2012 SSAT Plenary Presentation

DOI: 10.1007/s11605-012-2095-z

Cite this article as:
Ganai, S., Prachand, V.N., Posner, M.C. et al. J Gastrointest Surg (2013) 17: 244. doi:10.1007/s11605-012-2095-z

Abstract

Background

While laparoscopy has become integral to the performance of foregut surgery, its optimal use in resection of gastric submucosal neoplasms, including gastrointestinal stromal tumors (GISTs), remains uncertain. Concern exists for technical feasibility related to tumor size and location, as well as oncologic outcome.

Methods

From 2002 to 2012, 106 patients underwent resection for gastric submucosal neoplasms, comprising 79 laparoscopic and 27 open resections. Median follow-up was 15 months.

Results

Patients were 62 ± 14 years and 56 % male. Mean tumor size was 5.5 ± 4.3 cm, with 76 % being GISTs. A total of 8 (10 %) conversions occurred in the laparoscopic cohort. On multivariate analysis, conversion was predicted by size greater than 8 cm, while recurrence was predicted by mitotic index (p < 0.05). Laparoscopic resection resulted in better perioperative outcomes, with less morbidity, operative time, blood loss, and length of stay (p < 0.05). No significant difference was seen in survival, with 90 % and 81 % alive 3 years after laparoscopic and open resection, respectively (HR 0.4; 95 % CI 0.1–1.3; p = 0.13).

Conclusions

Laparoscopic resection is feasible and effective in the management of gastric submucosal neoplasms, including GISTs. Caution should be reserved for tumors greater than 8 cm. Oncologic outcome appears to be predicted by tumor biology as opposed to surgical approach.

Keywords

LaparoscopyGastric neoplasmGISTSubmucosal tumors

Introduction

In 1992, the first report of laparoscopic resection of a gastric stromal tumor was published, describing the firing of an endoscopic linear stapling device across the base of an incidentally noted exophytic mass identified during cholecystectomy.1 Since then, the integral role of laparoscopy in foregut surgery has become well-established. With optimization of instrumentation, energy sources, and stapling devices and their use in conjunction with flexible endoscopy and intracorporeal suturing techniques, laparoscopic gastric resection has become a part of the repertoire of general surgeons in managing both benign and malignant conditions.2,3 Moreover, the classification of the gastrointestinal stromal tumor (GIST) as a distinct pathologic entity has led to numerous reports assessing not only feasibility, but confirming the longer-term oncologic efficacy of laparoscopic resection.411 However, the ability to generalize laparoscopic resection techniques to the spectra of gastric submucosal neoplasms remains uncertain, including concerns of the ideal approach based on size6,1214 and location1113,15 of tumors. In order to better evaluate selection for a minimally invasive approach to resection of gastric submucosal neoplasms, we present our experience with both laparoscopic and open techniques. We hypothesized that there are predictors of unsuccessful laparoscopic resection, with failures defined by conversions, complications, and poor oncologic outcomes.

Materials and Methods

A retrospective medical record review was conducted on 106 consecutive patients with gastric submucosal tumors who underwent either open or laparoscopic resection between October 1, 2002 and March 31, 2012, at the University of Chicago Medical Center (UCMC) in Chicago, Illinois. Ethical standards were followed under the guidance of the UCMC Institutional Review Board (protocol 11-060) to ensure protection of patient privacy and confidentiality.

Patient demographics, clinical presentation, and findings from preoperative imaging and endoscopy were extracted from a prospectively maintained electronic medical record (EMR). Pretreatment size of tumor was determined from clinical data, prioritizing measurements obtained via endoscopic ultrasound (EUS) over other imaging modalities. Location of tumor was determined from descriptions within operative reports. To be initially classified in the laparoscopic cohort, laparoscopic techniques were required to be used for at least part of the mobilization of the lesion and/or stomach. Successful laparoscopic resections required completion of the resection and any additional reconstruction via laparoscopic techniques. Conversions were defined as any cases that initially started laparoscopically and proceeded to open resection and/or reconstruction. Cases utilizing laparoscopy for purely diagnostic purposes (i.e., to rule out metastatic disease prior to resection) were classified in the open cohort.

Laparoscopic gastric resection techniques, including laparoscopic-assisted endoscopic submucosal resection, have been described in detail in prior publications.1,2,16 Except for one case, laparoscopic transgastric resection was performed through an anterior gastrotomy17 rather than via intragastric placement of balloon trocars.18 Anterior gastrotomy defects were closed using intracorporeal suture techniques. Hand-assisted laparoscopic surgery was not utilized in this series. The decision to proceed with an open resection was based on factors including surgeon expertise and preference, tumor location, tumor size, and multivisceral involvement as demonstrated on preoperative axial imaging, if performed. The decision to provide neoadjuvant imatinib was based on tumor size, multivisceral involvement, and a diagnosis of GIST. Most patients were discussed in a multidisciplinary tumor board where both biologic and technical issues were considered.

Pathologic data were extracted from pathology reports. A diagnosis of GIST was confirmed via immunohistochemical staining for CD117, supplemented by CD34 and/or DOG1 immunohistochemistry. Determination of pathologic size was based on the greatest gross dimension reported by the pathologist. Determination of resection status was based on a combination of pathologic determination of margins and clinical data as provided within operative reports. All enucleations were thus considered R1 resections (microscopically positive), despite no remark on examination by the pathologist. Any patient with gross residual disease at completion of surgery was considered to have undergone an R2 resection (macroscopically positive). Pathologic staging was performed in accordance to the American Joint Committee on Cancer (AJCC) 7th edition staging system for gastric GIST.19

Postoperative complications were scored according to the Expanded Classification of the Accordion Severity Grading System of Surgical Complications.20 Time to last follow-up was based on the date of last clinical and/or imaging examination as reported in the EMR, with censoring for patients noted as alive (overall survival) or alive with no evidence of disease (disease-free survival). Date of recurrence was based on the date of a positive biopsy or unequivocal imaging study. Date of death was supplemented from data obtained from the Social Security Death Index.

Data are reported as means with standard deviations (SD) or medians with interquartile ranges (IQR), with the exception of Table 7, which reports means with the standard error of the mean. Comparative statistics were performed using Student’s t test or ANOVA for normally distributed data and the Mann–Whitney rank sum test for ordinal variables and non-normally distributed data. Comparisons of categorical variables were performed using chi-square or Fisher exact tests. Unconditional multiple variable logistic regression was performed to determine independent predictors of conversion and recurrence, with inclusion in the model based on a univariate p value less than 0.10. Odds and hazard ratios are reported with 95 % confidence intervals (CI). Time-to-event analysis was performed using the Kaplan–Meier method, with the log rank test used for comparisons between groups. A Cox proportional hazards model was used to identify independent predictors of survival using stepwise inclusion of multiple variables. Significance was determined according to an α of 0.05. Statistical analysis was performed with EpiInfo version 3.5.3 (Centers for Disease Control, Atlanta, GA). Survival curves are depicted using GraphPad Prism version 5.04 (La Jolla, CA).

Results

Table 1 summarizes the demographics and preoperative characteristics of patients with gastric submucosal neoplasms who underwent an attempt at laparoscopic (n = 79) or open resection (n = 27). Patients undergoing open surgery were more likely to present with abdominal pain than those undergoing a laparoscopic resection (Fig. 1a; 52 vs. 26 %, p < 0.05). Despite similar proportions of elective cases, patients in the open cohort were less likely to have preoperative EUS (67 vs. 87 %, p < 0.05). There was significantly greater use of neoadjuvant imatinib in the open group (26 vs. 5 %, p < 0.01), with an approximately 6-month greater time interval from diagnosis to operative intervention in the group undergoing open surgery (p < 0.05). A similar distribution of lesions throughout the stomach was noted between cohorts, with a slightly greater proportion of posterior-wall tumors in the open group.
Table 1

Patient demographics and preoperative characteristics

 

Laparoscopic cohort (n = 79)

Open cohort (n = 27)

p

Age (y)

62.2 ±14.9

62.7 ± 9.6

0.91

Gender (male)

43 (54 %)

16 (59 %)

0.66

BMI (kg/m2)

29.8 ± 7.3

28.1 ± 5.5

0.40

Ethnicity

  

0.36

  -Caucasian

53 (68 %)

16 (60 %)

 

  -African-American

18 (23 %)

7 (26 %)

 

  -Hispanic

4 (5 %)

2 (7 %)

 

  -Asian

3 (4 %)

2 (7 %)

 

Recurrent disease

2 (3 %)

3 (11 %)

0.07

Preoperative EUS

69 (87 %)

18 (67 %)

0.02

Pretreatment size (cm)

3.9 ± 2.7

9.5 ± 6.8

<0.0001

Elective case

72 (91 %)

22 (82 %)

0.18

Prior laparotomy

21 (27 %)

6 (22 %)

0.61

Interval from diagnosis to surgery (months)

4.4 ± 8.6

10.1 ± 15.7

0.02

Neoadjuvant imatinib

4 (5 %)

7 (26 %)

0.002

Posterior location

35 (44 %)

16 (59 %)

0.18

Gastric mass location

  

0.93

  -GE junction/cardia

13 (17 %)

6 (22 %)

 

  -Fundus

14 (17 %)

4 (15 %)

 

  -Greater curvature

25 (32 %)

11 (41 %)

 

  -Lesser curvature

17 (22 %)

2 (7 %)

 

  -Antrum/pylorus

10 (13 %)

4 (15 %)

 

BMI body mass index, EUS endoscopic ultrasound, GE gastroesophageal

https://static-content.springer.com/image/art%3A10.1007%2Fs11605-012-2095-z/MediaObjects/11605_2012_2095_Fig1_HTML.gif
Fig. 1

Presenting features of gastric submucosal neoplasms undergoing resection (n = 106). a Symptoms based on presentation among patients undergoing laparoscopic (n = 71), conversion to open (n = 8), and primary open resection (n = 27). Significant differences are noted in the presentation of abdominal pain between laparoscopic and open resection groups (*p < 0.01). b Gastric choropleth maps demonstrating location of tumor based on histologic type of submucosal neoplasm. Gastric anatomical regions include cardia/gastroesophageal junction, fundus, greater curvature, lesser curvature, and antrum/pylorus

A total of 8 (10 %) laparoscopic cases were converted to open, which occurred secondary to a failure to progress rather than for urgent indications. Table 2 summarizes the operative and pathologic data for the laparoscopic (n = 71), converted (n = 8), and open (n = 27) groups. Median pretreatment lesion size was 3 cm (IQR 2–4), 8 cm (IQR 3–12), and 8 cm (IQR 5–11), respectively (p < 0.0001). The largest lesion resected in each group was 10.5, 15.5, and 24.0 cm in greatest dimension on final pathology. Despite theoretical concerns for tumor spillage with laparoscopy, there were no iatrogenic ruptures of the tumor pseudocapsule in the laparoscopic group. Overall, the laparoscopic group had significantly shorter operating times and less blood loss than the open and converted groups (p < 0.0001). Creation of a gastroenteric anastomosis was more frequent in the open group than in the laparoscopic group (37 vs. 5 %, p < 0.0001). Patients were also more likely to require multivisceral resection in the open group, which most frequently involved removing the distal pancreas and the spleen.
Table 2

Operative and final pathological data

 

Laparoscopic (n = 71)

Converted (n = 8)

Open (n = 27)

p

Laparoendoscopic approach

7 (10 %)

0

0

n/a

Operative time (min)

132 ± 52

247 ± 115

230 ± 111

<0.0001

Estimated blood loss (mL)

35 ± 66

342 ± 402

364 ± 279

<0.0001

Gastric procedure

   

0.0004

  -Wedge or sleeve resection

52 (73 %)

6 (75 %)

16 (59 %)

 

  -Transgastric wedge resection

8 (11 %)

1 (13 %)

0

 

  -Perigastric mass resection

2 (3 %)

0

1 (4 %)

 

  -Endoscopic submucosal resection

3 (4 %)

0

0

 

  -Enucleation

2 (3 %)

1 (13 %)

0

 

  -Gastrectomy with reconstruction

4 (6 %)

0

10 (37 %)

 

Multivisceral resection

1 (1 %)

2 (25 %)

11 (41 %)

<0.0001

Pathological size (cm)

4.0 ± 2.1

8.0 ± 4.7

8.6 ± 6.1

0.0001

  -Median size (IQR, cm)

3.6 (2.5–5.1)

8.9 (4.1–10.8)

7.7 (4.0–11.5)

 

  -Maximum size (cm)

10.5

15.5

24.0

 

Histology

   

0.04

  -GIST

50 (70 %)

7 (88 %)

24 (89 %)

 

  -Leiomyoma

7 (10 %)

1 (12 %)

2 (7 %)

 

  -Schwannoma

6 (9 %)

0

0

 

  -Carcinoid

3 (4 %)

0

0

 

  -Lipoma

2 (3 %)

0

0

 

  -Adenomyoma

1 (1 %)

0

0

 

  -Heterotopic pancreatic tissue

1 (1 %)

0

0

 

  -Inflammatory fibroid polyp

1 (1 %)

0

1 (4 %)

 

Resection status:

   

0.07

  -R0 (negative)

68 (96 %)

6 (75 %)

25 (93 %)

 

  -R1 (microscopic)

3 (4 %)

1 (13 %)

1 (4 %)

 

  -R2 (gross)

0

1 (13 %)

1 (4 %)

 

The majority of lesions resected were GIST (76 %), followed by leiomyoma (9 %), schwannoma (6 %), and carcinoid tumor (3 %). Figure 1b demonstrates choropleth maps that summarize lesion location within the stomach for each histologic subtype. Notably, among the 10 leiomyoma in this series, 90 % were located in the cardia or gastroesophageal (GE) junction, and the one that was located in the fundus was 9.5 cm and underwent open resection. For the nine patients with leiomyoma that had preoperative EUS-guided FNA biopsies, three were consistent with leiomyoma, three were suspicious for GIST, and three were nondiagnostic due to a hypocellular sample. Median size for leiomyoma was 5.3 cm (IQR 3.6–9.5), reaching a maximum of 15 cm. Schwannoma (n = 6) were smaller at 3.2 cm (IQR 2.0–5.0), reaching a maximum of 6.5 cm, and were most frequently seen in the fundus (50 %), followed by the greater and lesser curvatures. All biopsies of schwannomas were nondiagnostic or suggestive of a diagnosis of GIST. In contrast, the three carcinoids in this series were all amenable to accurate preoperative biopsy, and were located in the lesser and greater curvatures. They also tended to be the smallest of lesions, at a median of 2.1 cm (IQR 1.4–2.2).

Table 3 summarizes the pathologic data for gastric GISTs. The majority of those amenable to laparoscopic resection were between 2–5 cm in size, while the majority of conversions and open cases comprised tumors greater than 5 cm. A diagnosis of GIST on preoperative biopsy had a sensitivity of 90 %, specificity of 36 %, and accuracy of 75 %. Sixty percent of laparoscopically resected lesions were Stage IA (less than 5 cm and less than 5 mitoses per 50 high-power fields [HPF]), indicating a low-risk for recurrence. By contrast, only 23 % of conversions and open cases were Stage IA (p < 0.01).
Table 3

Final pathological data for gastrointestinal stromal tumors

 

Laparoscopic (n = 50)

Converted (n = 7)

Open (n = 24)

p

pT

   

0.001

  -1 (<2 cm)

5 (10 %)

1 (14 %)

3 (13 %)

 

  -2 (2–5 cm)

32 (64 %)

1 (14 %)

4 (17 %)

 

  -3 (5–10 cm)

12 (24 %)

2 (29 %)

7 (29 %)

 

  -4 (>10 cm)

1 (2 %)

3 (43 %)

7 (29 %)

 

Mitotic index (mitoses/50 HPF)

4.6 ± 10.0

0.7 ± 0.5

6.2 ± 12.3

0.46

High grade: >5 mitoses/50 HPF

10 (20 %)

0

10 (44 %)

0.24

Necrosis

4 (8 %)

1 (14 %)

7 (29 %)

0.06

Ulceration

3 (6 %)

1 (14 %)

1 (4 %)

0.62

Pathologic Stage

   

<0.0001

  -0 (ypCR)

0

0

4 (17 %)

 

  -IA (<5 cm, low grade)

30 (60 %)

2 (29 %)

5 (21 %)

 

  -IB (5–10 cm, low grade)

9 (18 %)

2 (29 %)

5 (21 %)

 

  -II (<5 cm, high grade or >10 cm, low grade)

7 (14 %)

2 (29 %)

3 (13 %)

 

  -IIIA (5–10 cm, high grade)

4 (8 %)

0

2 (8 %)

 

  -IIIB (>10 cm, high grade)

0

0

2 (8 %)

 

  -IV (N1 or M1)

0

1 (14 %)

3 (13 %)

 

Neoadjuvant imatinib

2 (4 %)

2 (29 %)

7 (29 %)

0.006

Adjuvant imatinib

15 (30 %)

4 (57 %)

11 (46 %)

0.22

ypCR complete pathological response after neoadjuvant imatinib

Table 4 summarizes the postoperative and follow-up data for the series. Patients undergoing successful laparoscopic resection had a significantly lower hospital length of stay, with a median of 3 days (IQR 2–4) in comparison to 7 days (IQR 5–9, p < 0.0001). Laparoscopic resection was also associated with significantly less surgical site infections and postoperative arrhythmias (p < 0.01). Overall complications were also less severe when scored according to the Accordion Severity Grading System (Table 4; Fig. 2a; p < 0.0001).20 There were significantly more patients with no complications in the laparoscopic compared to the open groups (85 vs. 34 %; p < 0.0001). In-hospital mortality included a massive myocardial infarction in one patient in the laparoscopic group, and hospice care initiated after development of cardiogenic shock in one metastatic patient in the open group. The need for a second operative procedure in the laparoscopic group (n = 2) included management of an incarcerated port-site hernia, as well as an iatrogenic gastric outlet obstruction requiring revision to a Billroth II reconstruction. This patient initially underwent transgastric resection of an endophytic 5.4 cm antral GIST, leaving a markedly narrowed gastric outlet at the level of the incisura. The single re-operation in the open group was related to management of an anastomotic leak.
Table 4

Postoperative morbidity and mortality

 

Laparoscopic (n = 71)

Converted (n = 8)

Open (n = 27)

p

Hospital LOS (days)

3.3 ± 3.7

6.0 ± 1.6

8.4 ± 5.7

<0.0001

Surgical site infections

1 (1 %)

1 (13 %)

6 (22 %)

0.002

  Superficial

1 (1 %)

1 (13 %)

3 (11 %)

 

  Deep space

0

0

3 (11 %)

 

Anastomotic dehiscence

0

0

2 (7 %)

0.05

Postoperative arrhythmia

0

0

4 (15 %)

0.002

Myocardial infarction

1 (1 %)

0

1 (4 %)

0.70

Pneumonia

1 (1 %)

0

0

0.78

Deep venous thrombosis

2 (3 %)

0

3 (11 %)

0.18

Pulmonary embolism

0

0

1 (4 %)

0.23

Accordion severity grading system

   

<0.0001

  0—No complications

60 (85 %)

3 (38 %)

9 (33 %)

 

  1—Minor (wound, foley)

3 (4 %)

2 (25 %)

5 (19 %)

 

  2—minor (PRBCs, TPN, antibiotics)

4 (6 %)

3 (38 %)

8 (30 %)

 

  3—Endoscopic/radiologic intervention

1 (1 %)

0

2 (7 %)

 

  4—Operative intervention

2 (3 %)

0

1 (4 %)

 

  5—MSOF

0

0

1 (4 %)

 

  6—In-hospital death

1 (1 %)

0

1 (4 %)

 

LOS length of stay; PRBCs packed red blood cells; TPN total parenteral nutrition; MSOF multi-system organ failure

https://static-content.springer.com/image/art%3A10.1007%2Fs11605-012-2095-z/MediaObjects/11605_2012_2095_Fig2_HTML.gif
Fig. 2

Outcomes from successful laparoscopic resection of gastric submucosal neoplasms. a Accordion Severity Grading System comparing complications after laparoscopic resection (n = 71), conversions (n = 8), and open resection (n = 27; p < 0.0001). b Overall survival comparing successful laparoscopic resections (n = 71) with conversions and open resections (n = 35). Median survival is 106.1 months for the open group compared to undefined in the laparoscopic group. Three-year survival was 90 and 81 % for the open and laparoscopic groups, respectively (hazard ratio, 0.42; 95 % CI, 0.14–1.30; p = 0.13). Median follow-up is 15 months

In the subgroup of successful laparoscopic resections, 28 % of posteriorly based lesions were approached by transgastric wedge resection, while 65 % were amenable to standard extraluminal wedge resection, and 6 % required laparoscopic resection with gastroenteric anastomosis. There was a trend toward an increased rate of conversion with lesions at the lesser curvature (18 %) compared to other sites (8 %), although this was not significant (p = 0.35). Non-standard approaches (e.g., transgastric resection, gastrectomy) were used in 36 % of lesser-curvature lesions that were successfully managed laparoscopically. With univariate analysis, predictors of conversion included tumor size greater than 8 cm and multivisceral tumor involvement (p < 0.05). Table 5 summarizes the multivariate model where the only independent predictor of conversion was tumor size greater than 8 cm (p < 0.01).
Table 5

Multivariate Model for Predictors of Conversion

Variable

Conversions (n = 8)

Laparoscopic success (n = 71)

Univariate p

Odds ratio (95 % CI)

Multivariate p

Anastomosis

0

4 (6 %)

0.49

Abdominal pain

4 (50 %)

16 (24 %)

0.11

BMI

31.9 ± 8.1

30.0 ± 7.2

0.38

Prior laparotomy

3 (38 %)

18 (26 %)

0.50

Posterior lesion

6 (75 %)

29 (41 %)

0.07

5.03 (0.66–38.57)

0.12

Multivisceral

2 (25 %)

1 (1 %)

0.001*

7.95 (0.13–467.73)

0.32

Size > 8 cm

5 (63 %)

5 (7 %)

<0.0001*

18.48 (2.29–149.38)

0.006*

*p < 0.05

Recurrences in the laparoscopic group occurred in three individuals, including one patient with a locally recurrent gastric carcinoid, one patient with a low grade, 3.4 cm GIST who developed liver and pelvic recurrences, and one patient with a high-grade, 8.0 cm GIST who developed a mass in the hepatorenal fossa. Recurrences in the GIST patients who had open resections were associated with large size (maximal dimension of 9.3 ± 2.3 cm) and high mitotic rate (mean 13 ± 14 mitoses per 50 HPF). Liver metastases were found in 71 % of open recurrences, while the remaining two recurrences were local. Mean time to recurrence was 36.5 ± 27.1 months for laparoscopic cases and 23.5 ± 13.8 months for open cases. With univariate analysis, predictors of recurrence included tumor size, grade, and an open procedure (p < 0.05). Table 6 summarizes a multivariate model for recurrence, where the only independent predictor of recurrence was a tumor mitotic index greater than 5 per 50 HPF (p < 0.05).
Table 6

Multivariate Model for Predictors of Recurrence

Variable

Recurrence (n = 10)

No recurrence (n = 96)

Univariate p

Odds ratio (95 % CI)

Multivariate p

R0 status

9 (90 %)

90 (94 %)

0.65

Adjuvant imatinib

4 (40 %)

26 (27 %)

0.46

Prior recurrence

1 (10 %)

4 (4 %)

0.41

Neoadjuvant imatinib

2 (20 %)

9 (9 %)

0.30

Size (cm)

7.9 ± 3.4

5.3 ± 4.3

0.01*

0.99 (0.85–1.14)

0.86

Mitoses > 5 per 50 HPF

5 (50 %)

14 (15 %)

0.006*

4.68 (1.04–21.06)

0.04*

Laparoscopy

3 (30 %)

68 (71 %)

0.009*

0.21 (0.04–1.04)

0.06

*p < 0.05

Median follow-up was 9 months (IQR 1–32) after laparoscopic resection and 30 months (IQR 4–44) for conversions and open cases. There was no significant difference in overall survival between groups, with 90 and 81 % alive at 3 years after laparoscopic and open resection, respectively (HR 0.4; 95 % CI 0.1–1.3; p = 0.13; Fig. 2b). No significant differences in overall and disease-free survival by procedure type were noted when controlling for tumor size, grade, and imatinib use in a Cox proportional hazards model.

Discussion

Submucosal tumors of the gastrointestinal tract are mesenchymal tumors that can be classified based on myogenic origin (i.e., leiomyomas and leiomyosarcomas), neurogenic origin (i.e., schwannomas and granular cell tumors), vascular origin (i.e., hemangiomas and lymphangiomas), and other, including gastrointestinal stromal tumors, carcinoids, lipomas, metastases, and heterotopic pancreatic tissue.21,22 In the stomach, approximately 70 % of these lesions when evaluated and sampled by EUS will be GISTs.23 While GISTs appear phenotypically similar to myogenic tumors, they demonstrate immunohistochemical characteristics that bear resemblance to the interstitial cells of Cajal, characterized by over-expression of CD117 (a marker of c-kit) in 95 % of cases.24,25 Over the past decade, the expanding role of imatinib and tyrosine kinase inhibitors has allowed GISTs to serve as an ideal model for the molecular-based diagnosis and treatment of cancer.26,27

Table 7 summarizes the experience of 37 independent case series examining outcomes after laparoscopic resection for gastric submucosal neoplasms (n = 1174), including our own.415,2850 While these studies vary in their design and inclusion criteria, the majority focus on laparoscopic GIST resections. Favorable operative times and reduced estimated blood loss are described, as well as hospital length of stays that are less than or at least comparable to open resection. Satisfactory perioperative outcomes have been reported, including conversion rates of 5 %, reoperations in 1 %, and postoperative mortality in 0.4 %. While these may be reflective of optimal selection for laparoscopy based on preoperative size and location of lesions, the biological behavior of these tumors may play an equally important role in long-term outcomes.
Table 7

Laparoscopic resection of gastric submucosal neoplasms

Author

Year

N

GISTs n (%)

Size (cm) n (%)

OR time (min)

EBL (mL)

LOS (d)

Conversions n (%)

Mortality n (%)

Reoperations n (%)

Recurrences n (%)

Follow-up (months)

Geis28

1996

8

NR

4.6

143

NR

3.5

0

0

NR

NR

NR

Buyske29

1997

7

NR

3.9

132

NR

5.8

2 (28)

0

0

0

NR

Basso30

2000

9

NR

NR

NR

NR

4

0

0

0

0

22.8a

Choi YB31

2000

32

NR

4.8

NR

NR

5.9

1 (3)

0

1 (3)

0

NR

Hepworth32

2000

9

9 (100)

NR

NR

NR

3a

2 (22)

0

0

NR

NR

Avital33

2003

7

2 (29)

NR

180a

NR

3a

0

0

2 (29)

NR

NR

Walsh18

2003

14

NR

3.8

186

NR

3.8

0

0

0

0

16.2

Hindmarsh34

2005

30

22 (73)

4.5

74

196

5.0

7 (23)

0

0

2 (7)

18a

Bédard35

2006

14

14 (100)

4.1

175

NR

4.6

2 (14)

0

0

1 (7)

46.5

Berindoague36

2006

22

18 (82)

5.6

NR

NR

6a

1 (5)

0

0

1 (5)

32a

Granger37

2006

12

5 (42)

4.1

169

NR

2.3

0

0

0

0

19

Lai4

2006

28

28 (100)

3.4

191

NR

6.7

0

0

0

0

43.3

Mochizuki38

2006

12

10 (83)

2.7

100

0

7

0

0

0

0

26a

Novitsky5

2006

50

50 (100)

4.4

135

85

3.8

0

0

0

4 (8)

36

Otani6

2006

35

35 (100)

4.3

NR

NR

7.2

0

0

0

0

53a

Choi SM7

2007

23

23 (100)

4.2

104

39

5.2

0

0

0

0

61

Catena39

2008

21

21 (100)

4.5

151

101

4.8

0

0

0

NR

35

Hiki16

2008

7

6 (86)

4.6

169

7

7.4

0

0

NR

NR

NR

Nakamori40

2008

25

25 (100)

5.0

165

50

6.6

0

0

0

2 (8)

37

Privette15

2008

12

9 (75)

5.1

234

108

4.6

0

0

0

0

NR

Sexton8

2008

61

61 (100)

3.8

152

97

3.9

1 (2)

1 (2)

1 (2)

3 (5)

15

Wilhelm41

2008

93

62 (67)

2.6

93

NR

7.5

6 (7)

0

4 (4)

0

39.5a

Hwang12

2009

63

41 (65)

NR

86

33

5.3

0

0

0

0

15

Silberhumer42

2009

22

22 (100)

3.5

NR

NR

7.8

4 (18)

0

0

0

30

Tabrizianb9

2009

55

55 (100)

3.5

143

138

6

7 (13)

1 (2)

3 (5)

4 (7)

41

Goh10

2010

14

11 (79)

3.1a

150a

0a

4.5a

1 (7)

0

0

0

8a

Sasaki43

2010

45

37 (82)

3.2a

100a

5a

7a

1 (2)

0

0

0

74a

Warsi44

2010

22

17 (77)

3.0

80

NR

3a

1 (5)

0

0

0

12a

Ke45

2010

84

43 (51)

NR

63

86

5.6

0

0

0

0

51

Karakousisc11

2011

40

40 (100)

3.6a

96a

25a

4a

13 (25)c

0

0

1 (3)

28a

Lee13

2011

50

35 (70)

2.9

153

NR

5.7

1 (2)

0

0

0

21

Ma46

2011

56

56 (100)

3.0a

90a

55a

7.0

0

0

0

0

21.5a

Nguyen47

2011

44

31 (70)

NR

97

33

2.6

0

0

1 (2)

1 (2)

51.6

Ryu48

2011

20

16 (80)

2.4

84

NR

4.7

0

0

0

0

NR

De Vogelaere14

2012

31

31 (100)

4.4

99

122

8.5

0

1 (3)

0

0

56.3

Kakeji49

2012

18

10 (56)

3.5

128

25

9.8

0

0

1 (6)

0

46.1

Ganai

2012

79

57 (72)

4.4

142

66

3.6

8 (10)

1 (1)

2 (3)

4 (5)

21

Total

37

1174

902 (77 %)

3.7 ± 0.1

118 ± 7

70 ± 12

5.5 ± 0.3

4.9 %

0.3 %

1.3 %

2.1 %

35 ± 3

GISTs gastrointestinal stromal tumors; OR operative; EBL estimated blood loss; LOS hospital length of stay; NR not reported

aMedian value is reported; all other values represent means

bData also comprised an additional 21 patients undergoing laparoscopic management of small bowel GIST

cConversions (n = 13) are not included elsewhere in data as reported by authors

While gastrointestinal (GI) bleed is commonly described as a presenting symptom for GIST,25 our data suggest that abdominal pain is also a common presenting symptom that was seen in half of patients requiring open resection. This may relate to larger size, posterior location, or multivisceral involvement, which were features of note in the open cohort. Among patients presenting with GI bleed, there was no difference in likelihood of an open vs. laparoscopic approach. In addition, the greater use of neoadjuvant imatinib in the open group correlates with size and the expanded time interval of time seen between diagnosis and surgery. There was less use of endoscopic ultrasound in the open group, which may be explained by (1) a decision to operate without further diagnostic studies due to a large, resectable tumor on axial imaging, or (2) a patient who already had diagnosis of GIST by initial endoscopic biopsy and was being considered for neoadjuvant imatinib. Laparoscopically resected tumors tended to be smaller which may be why EUS utilization was greater; these tumors were either initially detected by EUS or best accessed via EUS-guided FNA.

The overall accuracy of EUS-FNA cytology for malignant disease, including GIST, has been reported as 80–90 %. Accuracy for submucosal tumors, however, is closer to 50 %, with almost half of biopsy specimens being inadequate.21 Indeed, we found indeterminate results in the majority of schwannomas and leiomyomas assessed by EUS-guided FNA in our series, with a presumptive diagnosis of GIST given until pathological analysis after resection. While we report a sensitivity of 90 %, overall diagnostic accuracy for GIST in our series was 75 %, accounting for a high false-positive rate for other stromal tumors deemed suspicious for GIST. This becomes particularly important for leiomyomas, which are often located in the cardia and gastroesophageal junction, and may be amenable to enucleation rather than wedge resection. As not all submucosal lesions are GISTs, it is important to account for the challenge in accurate preoperative diagnosis as this has implications in the planned surgical approach.

Our series is limited by bias related to selection and its retrospective design. While our attempts at laparoscopic resection occurred in 75 % of cases, this has increased from 50 % of cases seen between 2002 and 2004 to 89 % of cases seen from 2010 to 2012 (p < 0.01). Consideration of imaging characteristics suggesting a complex resection related to size and multivisceral involvement in GISTs may have influenced both the use of neoadjuvant imatinib and the decision to proceed with an open resection. As this study occurred at a tertiary-care comprehensive cancer center, the data are limited in follow-up, especially for the majority of low-risk lesions. The presence of selection bias is important to appreciate, and hopefully accounted for by variables included in the multivariate models.

The role of laparoscopy in the resection of gastric submucosal neoplasms is important. The value of systematic lymph node dissection and anatomical resection has been largely refuted for what was historically known as gastric leiomyosarcoma after numerous reports failed to show lymphatic metastasis or any survival benefit from lymphadenectomy.50,51 In addition, the role of margin status has been shown to not be predictive of survival after primary resection of GIST, allowing for performance of wedge resections to grossly-normal tissue.52 These features have allowed laparoscopic techniques to be both feasible and oncologically safe if performed properly in well-selected patients.

In 2004, the European Society for Medical Oncology presented a consensus conference on the management of GIST that recommended against laparoscopic surgery owing to a higher risk of tumor rupture, but stated that it “might be accepted in cases of small (<2 cm) intramural tumors”.53 In 2008, the Japan Society of Clinical Oncology identified that laparoscopic resection may be safely performed for GISTs or submucosal tumors under 5 cm.54 While concerns about the appropriateness of laparoscopy for colon cancer have prompted randomized-controlled trials demonstrating non-inferiority,55 it is unlikely that a randomized-controlled trial addressing this question will be feasible in the setting of GIST, partly because of its relative infrequency as an entity. While several investigators have made comparisons between open and laparoscopic cohorts,10,40,42,56 there are currently only two size-matched case–control studies comparing laparoscopic and open surgery, demonstrating similar oncologic outcomes with superior postoperative recovery.11,13

In our analysis of unsuccessful outcomes after laparoscopy for gastric submucosal neoplasms, the data suggest that laparoscopic techniques may be less suitable for tumors greater than 8 cm as this was an independent predictor of conversion. Facility with intracorporeal suturing and gastrointestinal reconstruction may have allowed for favorable outcomes in complex laparoscopic cases involving posteriorly based lesions requiring transgastric approaches. These approaches may have also increased the feasibility of laparoscopic excision at complex locations including the lesser curvature and antrum. While conversion to an open approach may reflect sound surgical judgment rather than a “failure” per se, it is important to note that several open cases and conversions required en bloc resection of multiple viscera including the distal pancreas and spleen. Complexity related to multivisceral resection may influence the likelihood of higher-grade morbidity seen with conversions and open surgery, although this may be simply related to the effect of laparotomy. Potential advantages of laparoscopy include reduced incision size, pain, infection risk, and hernia formation, which may be worth an attempt at laparoscopy followed by conversion if unsuccessful. Of interest, patients converted to open surgery had morbidity and length of stay that was intermediate between laparoscopy and open, perhaps reflecting the fact that none of the conversions were emergent in nature and suggesting that outcomes following conversion were not necessarily inferior to open surgery.

Important to this study is the finding of excellent perioperative outcomes in the form of hospital length of stay, operative time, and estimated blood loss after laparoscopic resection. We are first to report complications after resection of gastric submucosal neoplasm using a validated grading system,20 which clearly demonstrates the value of laparoscopic resection. However, a technical failure occurred after laparoscopic resection in a patient with an antral tumor leading to gastric outlet obstruction. Consideration should be given for distal gastrectomy either by laparoscopic or open techniques for large tumors near the antrum and pylorus to avoid narrowing the residual stomach.

Moreover, with regard to oncologic outcomes, we have shown that this may be related more to tumor biology than technical considerations. Our finding that mitotic index was predictive of recurrence is supported by prior studies,25,57 as with the pattern of recurrence favoring hepatic metastasis.53 Survival analysis showed no significant differences in survival between laparoscopic and open cohorts when controlled for biologic variables, but our data are limited by short follow-up, particularly in the laparoscopic cohort. Among those patients who recurred, there was a mean time to recurrence of three years for the laparoscopic group and two years for the open group. However, Miettinen et al. report that recurrences may occur over 20 years after resection, suggesting caution at extrapolating oncologic outcomes in the short term.25

Conclusions

Laparoscopic resection for gastric submucosal neoplasms is clearly feasible, with excellent perioperative outcomes and a reduction in morbidity, length of stay, blood loss, and operative time. Concerns arise in the use of laparoscopy for posterior tumors, which may be mitigated by additional expertise in laparoscopic intracorporeal suturing, allowing for transgastric resection. Moreover, conversions were associated with tumors greater than 8 cm in size, suggesting an open technique may be more appropriate with larger tumors. The biological behavior of GISTs may ultimately be more predictive of long-term outcome than the technique used to resect them, but the principles of oncologic surgery, favoring avoidance of capsular disruption and resection with a grossly negative margin should be maintained.

Copyright information

© The Society for Surgery of the Alimentary Tract 2012