Keywords
gastrosplenic fistula - subtotal gastrectomy - splenic infarct - pancreatitis
Introduction
A gastrosplenic fistula is a very rare pathological condition. It was first described
in two patients with splenic lymphoma by de Scoville et al in 1962.[1] “Aerosplenomegaly” is the radiological term used to describe an enlarged spleen,
which appeared to be filled with air, which would suggest an enterosplenic fistula.
A gastrosplenic fistula can form as a complication of a variety of conditions, although
splenic lymphoma has most frequently been reported in previous literature.[2] We discuss a nonmalignant pediatric case of a gastrosplenic fistula which was treated
successfully by surgical intervention.
Case Report
Our patient is a 16-year-old male with a history of type 1 diabetes mellitus that
presented to the emergency department with diabetic ketoacidosis and shock. His blood
glucose was 1,037 mg/dL with a pH of 6.95. He was resuscitated, started on an insulin
drip, given vasopressors, and admitted to the pediatric intensive care unit (PICU)
where he was intubated. During his stay in the PICU, he abruptly had 1.5 L of dark
brown to bloody drainage from his nasogastric tube. The dark blown to bloody fluid
did not persist and the patient never had a significant drop in his hemoglobin. Had
either of these factors developed, an earlier endoscopy would have been warranted.
Because of an elevated creatinine (3.4 mg/dL), an abdominal magnetic resonance imaging
(MRI) was obtained which showed a small amount of perisplenic fluid and evidence of
a splenic infarction. Intravenous antibiotic therapy was initiated with clindamycin.
The patient recovered and was discharged from the hospital on day 9.
The patient presented to the emergency department 2 weeks later with severe epigastric
pain, poor appetite, and weight loss of 20 pounds since discharge. He also presented
with labored breathing and a chest X-ray showed a left pleural effusion. Initial laboratories
were as follows: C reactive protein, 5.5 mg/dL; white blood cell count, 16.10[3]/µL; platelet count, 1,206 × 103/µL; hemoglobin, 9.1 g/dL; and serum lipase 890 U/L. On examination, the patient was
ill-appearing and pale with tenderness in the epigastric region. During his hospital
stay, the patient's oral intake decreased, and persistent abdominal pain prompted
a CT of chest, abdomen, and pelvis with intravenous contrast. The results showed bilateral
pleural effusions, left pelvic abscess, pancreatic pseudocyst, and a splenic infarction.
Additionally, the computed tomography (CT) was suspicious for a gastrosplenic fistula,
as the coil of the patient's transpyloric feeding tube appeared to lie within the
upper pole of the spleen ([Fig. 1]). To further evaluate the possibility of a gastrosplenic fistula, an esophagogastroduodenoscopy
was performed, which noted multiple dispersed nonbleeding erosions of the gastric
wall with two visualized openings of the posterior gastric wall ([Fig. 2]). The time between the first and second endoscopy was 3 weeks. There is a growing
body of literature to support the use of endoscopic closure of gastric perforations;
however, this was not initially considered because of the degree of inflammation and
friability of the gastric mucosa.
Fig. 1 CT scan images demonstrating gastrosplenic fistula. (*) shows fistulous communication
between stomach and spleen. Arrowhead represents coiled feeding tube within the capsule
of the spleen. CT, computed tomography.
Fig. 2 Endoscopic images of gastric inflammation and gastrosplenic fistula. (A and B) Images show pictures of two separate ulcers within there gastric wall. (C) Image shows generalized immflamation and mucosal erosions of the gastric wall.
Due to concern for malignancy, tumor markers were checked and were within normal limits.
A nonoperative approach would have been to support this patient with adequate nutrition
and allow for time for the inflammatory process to subside; however, the patient had
persistent epigastric pain and feeding intolerance.
The exact etiology of the gastroplenic fistula was not clear and malignancy had not
been completely excluded. The history and workup for other causes of a gastrosplenic
fistula did not support another etiology. The patient denied a toxic ingestion. Also,
none of his imaging on CT scan or MRI showed any evidence of inflammatory bowel disease.
He did have an elevated lipase but there were no signs of acute or chronic pancreatitis
on his imaging. A positron emission tomography (PET) scan could not be obtained because
of the patient's brittle diabetes and insulin requirements. Because of the patient's
failure to improve with nonoperative management, surgical exploration was recommended.
The operation began with a repeat upper endoscopy to assess the condition of the gastric
lumen. However, the scope could not be advanced much beyond the gastroesophageal (GE)
junction as the stomach had become very inflamed at this location. Biopsies were not
taken at the time of the first endoscopy because of the degree of inflammation and
friability of the gastric wall. The changes between the two endoscopies were presumed
to be due to progression of the inflammatory process. A left subcostal incision was
made and extended across the midline partially to the right subcostal region and extended
superiorly to the level of the xiphoid process. Adhesiolysis was performed to take
down very dense adhesions between the omentum and left upper quadrant. The anterior
gastric wall was intimately attached to the anterior abdominal wall, liver, and spleen.
We encountered a very dense inflammatory process involving the upper pole of the spleen.
The inflammatory process involving the upper pole of the spleen was able to be divided
from the remaining portion of the spleen living the majority of the splenic parenchyma
intact. The body of the stomach was completely fibrotic and necrotic. After separating
these structures, 4 to 5 cm of proximal stomach beyond the GE junction was found to
be viable as was 2 to 3 cm of antrum proximal to the pylorus. Thus, a subtotal gastrectomy
was completed by transecting the proximal stomach at the level of viability and transecting
the distal stomach at the level of the antrum. A “Roux-en-Y” reconstruction was performed
as well as placement of a feeding jejunostomy. Total operative time was 5 hours. The
patient initially did well but developed an intra-abdominal abscess in the left upper
quadrant seen on a CT that was obtained secondary to fever. The abscess was drained
by interventional radiology and found to be from a leak from his gastrojejunal anastomosis.
This was managed with external drainage and resolved in 4 weeks. Currently the patient
is doing well, has no epigastric pain, and is maintaining his weight on a combination
of oral and jejunostomy feeds. Pathological evaluation of the resected stomach revealed
marked edema with necrotizing inflammation. There was no evidence of malignancy. A
widely patent fistulous tract was seen extending into the congested and hemorrhagic
omental tissue.
Discussion
The etiology of the case is unclear. Possible etiologies include an infarct of the
upper pole of the spleen and stomach from a low flow state, complications of pancreatitis,
or a toxic ingestion causing necrosis of the stomach. Another cause of this fistula
is nasogastric tube trauma; however, this is less likely due to multiple perforations
seen on endoscopy. Gastrosplenic fistula can arise as a complication of a primary
splenic or gastric malignancy, most commonly diffuse large B cell lymphoma (DLBCL).[2] This may be attributed to this tumor's nature to widely infiltrate the serosa and
failure to initiate a desmoplastic reaction. This fistula may occur spontaneously
or after recent chemotherapy treatment as a possible consequence of tumor lysis syndrome.[3] Although less frequent benign conditions, such as a benign gastric ulcer, splenic
abscess, and Crohn's disease, have also been implicated as a cause.[4]
[5]
[6] More recently, a study reported three cases of gastrosplenic fistula as a complication
of sleeve gastrectomy.[7] A literature review identified a total of 28 cases reports; the etiology in 75%
of cases was determined to be a lymphoma.[8] The most common presenting complaints included abdominal pain (32%), weakness (21%),
and upper gastrointestinal bleeding (14%).[8] Abdominal CT is the most useful radiological test for reaching a definitive diagnosis.[9] In a noncontrast CT, an air-fluid level or free air in the spleen may indicate a
gastrosplenic fistula, although splenic abscess must be ruled out. Contrasted CT may
allow one to visualize the anomalous tract and orally administered contrast may be
seen flowing from the stomach to the spleen. Other tests that may be performed include
upper GI series as well as upper endoscopy to visualize the opening of the fistula
and for tissue sampling.
Surgical resection of the fistula is usually indicated.[9] This is done to prevent leaking of gastric contents that may erode the splenic vessels
resulting in life-threatening hematemesis. The operation may include of partial gastrectomy,
splenectomy, and/or distal pancreatectomy depending on the underlying etiology. A
feeding jejunostomy should also be considered. It is generally done as an open procedure,
although a laparoscopic approach has previously been described.[10]
Conclusion
Gastrosplenic fistulas have most commonly been reported in association with splenic
and gastric malignancies. Abdominal CT is the most helpful radiological test for diagnosing
this because of the ability to show thin cuts and demonstrate very detailed anatomy
in this region. Surgical intervention is usually indicated. We have presented a case
of gastrosplenic fistula in the pediatric population of unknown etiology that required
surgical treatment.
