Dieulafoy's Lesion in an Infant: A Rare Cause of Massive Gastrointestinal Bleeding

Acute life-threatening hemorrhage in an infant, evidenced by bright red blood in the oral cavity, requires thorough evaluation of the naso-oropharynx, respiratory tract, and upper gastrointestinal tract. This is the report of an infant for whom a Dieulafoy's lesion was found as the cause of a massive gastric hemorrhage. Dieulafoy's lesion has not been reported previously in an infant. CASE REPORT An 8-week-old, previously healthy, girl had difficulty breathing and approximately a tablespoon of bright red blood in her mouth 4 hours after she was placed in bed between her parents. She was exclusively bottle fed, using a cow's milk–based formula (Enfamil with iron; Mead Johnson Nutritionals, Evansville, IN, U.S.A.). Upon arrival in the emergency room, she was pale, mottled, and in severe respiratory distress with bloody oral secretions and was responsive only to painful stimuli. While she was supported with 100% oxygen, arterial blood gas showed uncompensated metabolic acidosis (pH 7.18, Paco2 42 mmHg, Po2 252 mmHg, HCO3 15) and hematocrit of 32%. The infant was intubated, given 20 ml/kg intravenous saline and transferred to the authors' facility. Tonic/clonic seizures developed during transfer, which were stopped with lorazepam. The child had been born at term with an uneventful perinatal course and had been receiving no medication. The mother was a 16-year-old, woman with slight mental retardation, who had been physically and emotionally abused by her father. The father was a 19-year-old man who smoked cigarettes. There was no family history of any inherited disorders. At the time of admission, the paternal grandparents voiced concern that the parents might have rolled on the infant while she was sleeping in bed with them. During laryngoscopy, no blood was observed in the oropharynx, and the vocal cords appeared normal. Suctioning of the endotracheal tube yielded a moderate amount of bright red blood and clots. A naso-gastric (NG) tube aspirated approximately 30 ml of “coffee ground” and bright red bloody secretions. A physical examination revealed a pale, afebrile child with normal growth parameters. The child had tachycardia and slight hypotension, with delayed capillary refill. A lung examination showed scattered rhonchi bilaterally. Her abdomen was slightly distended but soft, with positive bowel sounds. The liver was palpable 1 cm below the right costal margin, and the spleen was nonpalpable. Copious melanotic stool was noted. No bruising, petechiae, or signs of trauma were observed. Antibiotics and cimetidine were begun. A truncal radiograph revealed atelectasis of the right middle and lower lobes, with hyperaeration of the left lung. Dilated intestinal loops were noted, and there was no free intraperitoneal air. Initial laboratory evaluation included: hematocrit 28%, hemoglobin 9.6 g/dl, mean corpuscular volume 90 pg, mean corpuscular hemoglobin concentration 33.7 g/dl, erythrocyte distribution width 14.6, ferritin 139 ng/ml, platelets 170/μl, and leukocytes 7900/μl (21% segmented neutrophils, 35% band forms, 38% lymphocytes, 5% eosinophils, and 1% metamyelocytes). No abnormalities were noted on the peripheral smear. Prothrombin time (normal range) was 12.7 sec (11.3–13.9 sec), activated partial thromboplastin time 21.0 sec (21.7–34.9 sec), and fibrinogen level was 157 mg/dl (150–450 mg/dl). Urinalysis results were unremarkable. Serum electrolytes were normal except for decreased serum bicarbonate. Aspartate transaminase, alanine transaminase, alkaline phosphatase, [gamma]-glutamyl transferase, total bilirubin, cholesterol, and ammonia were normal. Levels of lactate dehydrogenase, 686 U/l (180–430 U/l); total protein, 4.7 g/dl (6.2–8.0 g/dl); albumin, 3.5 g/dl (3.8–5.4 g/dl); and uric acid, 5.9 mg/dl (2.0–5.5 mg/dl) were slightly abnormal. Blood and urine cultures were negative. The child received an erythrocyte transfusion. A head computed tomographic (CT) scan was interpreted as normal, and the skeletal survey was negative for fractures. An abdominal ultrasound revealed only a small amount of ascites. The ventilator was quickly weaned to minimal support settings. The secretions from the endotracheal tube cleared. The nasogastric tube drained 20 ml of bright red blood during the next 24 hours. No further seizure activity was observed. The following morning hematocrit was 22%, despite an additional erythrocyte transfusion during the night. Endotracheal suctioning produced scant, old clots, and chest radiograph was clear. Esophagogastroduodenoscopy was performed approximately 24 hours after admission by an adult gastroenterologist (LM) familiar with the lesion, using a GIF-N30 and subsequently a GIF-PQ10 endoscope (Olympus America, Inc., Melville, NY, U.S.A.). Findings included: a normal esophagus and duodenum and an oozing white vessel without surrounding inflammation and high in the body of the stomach, consistent with a Dieulafoy's lesion (Fig. 1). Sclerotherapy of the lesion, using eight injections of 0.5 ml 3.6% NaCl, successfully stopped the bleeding. Extubation was performed quickly, and the child was discharged home with a prescription for cimetidine on the fifth hospital day. She was subsequently lost to follow-up.FIG. 1.: Appearance of lesion in the body of the stomach is shown after sclerotherapy (A-B) and before intervention (C-D).FIG. 1.: ContinuedFIG. 1.: ContinuedFIG. 1.: ContinuedDISCUSSION Bright red blood in the oral cavity of an infant always warrants a thorough evaluation of the naso-oropharynx, gastrointestinal tract, and lower respiratory tract. The preliminary diagnostic consideration in this child was pulmonary hemorrhage because of the severe initial respiratory symptoms and the chest radiograph. Concurrently, the constellation of teenage parents with poor mental and socioeconomic capabilities and a family history of physical abuse combined with an infant in extremis and bright red blood in the mouth prompted us to entertain the diagnosis of nonaccidental injury. The patient's endotracheal secretions rapidly cleared and the focus turned to the gastrointestinal tract. Acute severe upper gastrointestinal hemorrhage is unusual in infants. A review of 27 children, age 2 days to 12 months, who undergo endoscopy to diagnose the source of upper gastrointestinal hemorrhage, found hemorrhagic gastritis in 33%, duodenal ulcers in 30%, gastric erosions in 15%, duodenal lymphoid hyperplasia in 15%, esophagitis in 11%, and no identifiable source of bleeding in 15%(1). In this young age group, congenital and hereditary disorders, such as hemophilia B, gastric teratoma, bronchopulmonary malformations, and Osler-Weber-Rendu disease, must also be considered. Dieulafoy's lesion, first described in 1884 by Gallard, and further characterized by Dieulafoy as “exulceratio simplex” (also termed caliber-persistent artery of the stomach, cirsoid aneurysm, and submucosal arterial malformation) is defined as an abnormally large-caliber submucosal arterial vessel that protrudes through the gastric mucosa from a noninflamed base (2). This lesion is typically located in the upper portion of the stomach, but has been reported to occur throughout the gastrointestinal tract (3). Rupture of the lesion causes a massive, frequently fatal, hemorrhage. Dieulafoy's lesion most often is found in older males. A review of 177 cases by Reilly and Al-Kawas (2) revealed an age range of 16 to 93 years, with a mean age at diagnosis of 58 years. The youngest patient reported previously with a Dieulafoy's lesion was 5 years old (3) (Table 1). Bleeding of a Dieulafoy's lesion may be due to rupture of a caliber-persistent artery secondary to its repetitive arterial pulsation without surrounding support tissue, or to vascular thrombosis with necrosis and subsequent bleeding. The cumulative risk of these events obviously increases with age, contributing to a higher risk of bleeding in older individuals.TABLE 1: Dieulafoy's lesion in children aRectal lesion.Numerous treatments have been used for Dieulafoy's lesion. Although surgical resection was common therapy, endoscopic treatment has been safe and effective in many cases. In a review of 480 adults receiving emergent endoscopy for upper gastrointestinal hemorrhage, 5.8% had Dieulafoy's lesion and were treated with sclerotherapy, with subsequent surgery in two patients (4). Endoscopic sclerosing therapies, such as heat-probe thermocoagulation (5), electrocoagulation (6), laser photocoagulation (7), and sclerotherapy with norepinephrine and polidocanol (4), alcohol (5), hypertonic glucose (5), histoacryl (8), and hypertonic NaCl, have been of varying success. Endoscopy and sclerotherapy in a 5-kg infant involves overcoming technical difficulties. The first concern is the size of the endoscopy instrument itself. In neonates and young infants, very small diameter instruments are better tolerated and safer. Excellent diagnostic instruments for this age group include those with external diameter of 5.3 to 7.9 mm, such as the GIF-N30 and GIF-XP20 (Olympus America). These instruments have a biopsy channel of only 2 mm, however, which does not accommodate therapeutic instruments such as sclerotherapy needles, thermocoagulators, or multipolar electrocoagulators. In this case we were obliged to use an endoscope (GIF-PQ10, Olympus America) with an external diameter of 9 mm and a biopsy channel of 2.8 mm, which is adequate for passage of therapeutic instruments. The second problem concerned the technique to use for coaptation of the lesion. In adults with Dieulafoy's lesion of the stomach or rectum, sclerotherapy combined with either thermocoagulation or electrocoagulation is used. Proper coaptation of an artery requires use of very firm pressure and high probe energy levels. The combination of high energy with very firm pressure can easily cause a perforation of a thin wall viscus. In this infant, we decided against such adjuvant therapy because of the very thin gastric wall. The usual sclerotherapy solution for ulcer treatment in the authors' institution consists of hypertonic saline with epinephrine (3.6% NaCl, 1:10,000 epinephrine). In this particular patient, epinephrine was not included in the sclerotherapy solution because of the infant's persistent resting tachycardia. Because the most important factor in obtaining hemostasis in nonvariceal sclerotherapy is the compression of the bleeding vessel by the volume of injectate, the authors believed that hypertonic saline alone was a safe and reasonable choice for this child. The local injection of small volumes of hypertonic saline has no systemic side effects, but theoretically prolongs the time of mechanical compression of the vessel and facilitates sclerosis (9). The risk of rebleeding in children is not known. Of 157 adult cases of hemostasis performed in upper gastrointestinal tract Dieulafoy's lesions, endoscopic hemostasis was initially successful in 156, but recurrent bleeding occurred in 27 (17.3%) (10). Theoretically, follow-up of the lesion, with transendoscopic Doppler ultrasound, would reveal whether complete hemostasis has been achieved (11). However, because of the caliber of endoscopic ultrasound instruments, that procedure was not an option for this infant patient. This case represents the first report of Dieulafoy's lesion in an infant. This lesion should be included in the differential diagnosis of massive upper gastrointestinal bleeding in a child. Diagnosis was made by endoscopy. Although numerous mechanisms are available to treat this lesion in an adult, treatment in an infant is limited by the availability of appropriately sized equipment. Sclerotherapy with hypertonic saline was successful in this child, and surgery was avoided.