An uncommon complication during management of gastric varices

Abstract

A 69-year-old woman presented to the emergency department with melena for 3 days. She has a medical history of hepatitis B virus-related cirrhosis. The esophagogastroduodenoscopy showed a gastric varix, GOV-2, with an ulcer on its surface. Histoacryl injection was attempted, but massive bleeding ensued just as the needle touched it. Five milliliters of a 1:1 mixture of N-butyl-2-cyanoacrylate and lipiodol were injected into the varix (1 mm per injection, at the same site). The patient was then transferred to the intensive care unit.

Shortness of breath with frequent dry cough developed 1 day later. The patient was slightly tachycardic at 97 beats per minute, tachypneic at 27 times per minute, and normotensive at 132/54 mmHg, with SpO₂ of 98%. Physical examination showed bilateral clear breath sounds. Laboratory investigations showed a hemoglobin level of 7.1 g/dL, platelet count of 15 × 10³/μL, and D-dimer of 1285.4 ng/ml. Arterial blood gas analysis showed pH 7.436, pCO₂ 34.9 mmHg, pO₂ 120.6 mmHg (under O₂ nasal cannula 3 L/min), and bicarbonate 23.8 mEq/L. The chest X-ray showed branching opacities over left hilum (Figure 1).

What is the diagnosis?

Non-contrast chest computerized tomography showed hyperdense material in the left pulmonary branches (Figure 2), and the diagnosis of acute pulmonary embolism after histoacryl injection was made. The patient was under mechanical ventilation support for total 18 days, with only supportive treatment during this period. She was discharged smoothly without any oxygen support.

Histoacryl injection therapy is effective for the gastric varices. The overall complication rate ranged from 0.5% to 5%,¹ and pulmonary embolism is an unusual but potentially life-threatening one,² especially in patients with large varices requiring large volumes of sclerosant.³ Other contributory factors include the rate of injection and ratio of the constituent components of the sclerosant.³ Although a leak through an arteriovenous pulmonary shunt or an open foramen ovale had been proposed,⁴ there was no such abnormality in our case. Hyperdense segments of the pulmonary arteries with respect to the surrounding vessels are the image features on chest X-ray (the so-called “vascular cast sign”), and a confirmatory non-contrast computerized tomography will be more suitable than a contrasted one in detecting the radio-opaque lipiodol.⁵ All endoscopists should keep this rare complication in mind, and timely recognition—if it occurs, unfortunately—relies on the physician's timely awareness.

The authors declare no conflicts of interest.

According to the institutional review board (IRB), there is no need of IRB approval for an image article in our hospital. A written informed consent was obtained before starting, and the authors followed the principles outlined in the WMA Declaration of Helsinki throughout the writing process.