Dieulafoy-Like Lesion Bleeding: In the Loop

Eddi and associates present a patient with a Dieulafoy-like lesion in the afferent loop of a Billroth II (BII) gastric resection.¹ This case is both novel and interesting, as it covers much clinical turf. Acute upper gastrointestinal hemorrhage has an incidence of 50–150/100,000 cases each year and a mortality rate of 7–10%.^2,3 The etiology of the vast majority of upper gastrointestinal hemorrhages is peptic ulcer disease. In approximately 5% of cases, the cause of bleeding cannot be identified with a standard diagnostic method, such as endoscopy, in which case the condition is referred to as “obscure gastrointestinal bleeding.”²

When a Dieulafoy lesion (DL) is suspected, the following diagnostic—though relatively nonspecific—criteria may be used: “active arterial spurting or micropulsatile bleeding from a minute mucosal defect less than 3 mm or normal mucosa; visualization of a protruding vessel with or without bleeding within a small mucosal defect or normal mucosa; and the appearance of a fresh, adherent clot to a minute mucosal defect or normal mucosa.”^4,5,6 For definitive pathologic diagnosis of a DL, it is necessary to obtain a surgical specimen, and herein lies the rub; most recent reports rely almost exclusively on endoscopic criteria, as surgery is not usually needed in these patients.

The case report by Eddi and coworkers examined a patient's disease course, which started as locally obscure bleeding that can be defined as: “bleeding from the gastrointestinal tract that persists or recurs after a negative initial evaluation using endoscopy and radiologic imaging.”^1,7 This definition of obscure bleeding is modified from the definition used by Pasha and colleagues, which applied to the entire gastrointestinal tract.^1,7 In the case study by Eddi and colleagues, endoscopic diagnosis of the DL was established during push enteroscopy just prior to emergency surgery, which would have likely yielded a definitive pathologic diagnosis.¹ Surgery in this case would not have been trivial, as the bleeding site was putative and an intervention likely would have involved either bowel resection in the antemesenteric duodenum or jejunum or the mesenteric side of the duodenum, as well as oversewing of the lesion with or without limited local excision. Performing these interventions in the previously surgerized abdomen of a blood-depleted older patient would have been a daunting prospect. It is therefore not surprising that the mortality associated with DLs exceeds the average mortality for gastrointestinal bleeding more than 2-fold.⁸

Issues with definitions have troubled this condition since its beginning. The lesions were first reported by Gallard in 1884; however, these lesions were named after Dieulafoy—whose own report came more than a decade later—because Gallard's description was deemed lacking.⁹ Dieulafoy used the term “exulceratio simplex” as the key descriptor of the lesion not merely because of its small size and large artery, as suggested by Eddi and associates, but also because he thought that the lesion represented an initial stage of a gastric ulcer.^1,9 In the modern endoscopic era, his description would be called into question, as the term “ulceration” connotes a component of chronic inflammation that is absent from DLs by definition.

Now that the diagnosis and definitions of DLs have been outlined and the gravity of the patient's initial condition has been placed into context, we can consider the novel and evocative elements of the case report by Eddi and colleagues.¹

The occurrence of a DL in the BII afferent limb appears to be novel, but as the authors correctly affirm, no region of the gut is immune. DLs may also occur in the bronchus of the respiratory system, which is not surprising given the fact that the gastrointestinal tract and pulmonary system are evaginations of the same embryologic bud.^10,11 DLs have been found after gastric resections, but the lesions are usually located proximal to the anastomotic site in the stomach.^6,12 In a Swiss series of nonvariceal bleeding in 480 patients, 5.8% of all cases of bleeding were attributed to a DL; in 1 case (4%), 3 separate endoscopic attempts were needed to establish a DL diagnosis, as in the case report by Eddi and coworkers.^1,12 The occurrence of DL bleeding after a BII procedure has been reported in a series from Greece that consisted of a slightly smaller number of patients.⁶ Twelve instances of bleeding were reported in patients who underwent a BII procedure; of these reports, 10 occurred at the anastomotic site. The locations of the remaining 2 cases were not described, making it impossible to confirm the novelty of this aspect of the case report. Nevertheless, the Greek study was germane to the present case report, as patients in the study bled from a different location in at least 10 of the 12 BII reconstruction patients. Some of the reasons advanced for the pathogenesis of the DL are based on this surgically altered anatomic location.

Although we do not know the cause of the bleeding that necessitated the BII procedure in this case report, the greatest delay for a DL rebleed in the Swiss study was 6 months, making it unlikely that this patient's original source of bleeding was a DL. In the Greek study, a majority of patients had gastrectomies, but it appears that none of these patients had prior DL bleeding. In this study, endoscopic banding ligation (EBL) was used to control DL bleeding rather than hemostatic clips. DL treatment with epinephrine vasoconstriction injection and hemostatic clipping are fairly contemporary therapeutic options; the report by Eddi and colleagues outlines all contemporary treatment modalities except for argon plasma coagulation.¹³

The second interesting element of the case report by Eddi and coworkers deals with the nature of the bleeding lesion.¹ A DL diagnosis is likely if a tortuous or ectatic artery is seen via angiography or pathology of the resected sample after laparotomy.⁴ Endoscopic ultrasound can also reliably confirm the diagnosis.

There are 3 common scenarios of DL bleeding. The first scenario involves the presentation of a DL with intermittent, recurrent, profuse bleeding. In this scenario, the initial endoscopy and arteriography commonly fail to detect the bleeding source. If repeated endoscopies and arteriographies fail to diagnose the source of obscure bleeding in an unstable patient, intraoperative endoscopy can be performed with a diagnostic yield of 70–100%; however, bleeding in the small intestine requires a laparotomy or laparoscopy.^2,14 In the case report by Eddi and associates, performing an endoscopic ultrasound in the afferent loop would have been challenging and it likely would have required a miniprobe examination.^1,15

In the second bleeding scenario, only an oozing site is visible—no protruding vessels, clots, or minute mucosal defects are seen—and the source of bleeding is thought to most likely be a DL-like bleed.

In the third scenario, the patient has a history of obscure bleeding that requires gastric resection; there is a high suspicion of a DL-like lesion; repeat surgery is an option due to the failure of endoscopic treatment; and there is a role for preoperative angiography for lesion localization.

In the 2 case series previously discussed, the proportion of DL bleeding episodes at gastrointestinal anastomoses ranged from 10% to 58% and occurred predominantly after BII procedures; the setting was usually massive gastrointestinal blood loss, as in the case report by Eddi and coworkers.^1,6,12 Although the reason for the occurrence of a DL at or near the anastomotic site is unknown, the authors postulated that the explanation may involve bile reflux, an incomplete vagotomy, or vascular malformation after gastrectomy.⁶ These etiologies may have applied to the patient in the case report; however, somewhat unusually, the afferent loop appeared to be the source of the bleeding. The diagnosis of a DL in this scenario was likely, though a DL was unusual enough that other entities be considered in the differential diagnosis.

Other possible etiologies include marginal ulcers, which are increasingly being reported after gastric bypass surgery or vertical-banded gastroplasty.¹⁶ In a study of patients taking nonsteroidal anti-inflammatory drugs who presented with obscure bleeding, 12% had ileal and/or jejunal ulcerations ranging from small to large punched-out lesions with oozing blood or scarring that could be mistaken for a DL.¹⁷

Various other causes of upper gastrointestinal hemorrhage have morphologic features that differ from those associated with DLs; nevertheless, these causes should be considered in the differential diagnosis. They include hemobilia, submucosal tumors, ischemic small-bowel necrosis, benign and malignant tumors, telangiectatic lesions, arteriovenous fistulas, aortoenteric fistulas, and Mallory-Weiss lesions. Primary and secondary aortoenteric fistulas most commonly occur at the distal duodenum or jejunum and can cause massive upper gastrointestinal hemorrhage. The latter group of fistulas is usually associated with prior aortic surgery, aortic aneurysms, and severe atherosclerosis.¹⁸ Arterial-type bleeding from diverticula is also interesting from the point of view of overlap, as a DL within a gastric diverticulum has been reported.¹⁹

Vascular malformations, on the other hand, usually cause chronic blood loss, unlike a DL, which typically has intermittent, recurrent, upper gastrointestinal acute hemorrhage. Vascular ectasias are typically seen in patients with renal failure, aortic valve disease, a history of radiation treatment, scleroderma, collagen diseases, or CREST syndrome (calcinosis, Raynaud phenomenon, esophageal dysmotility, sclerodactyly, and telangiectasia). The patient treated by Eddi and associates did not appear to have features of a connective tissue disorder.¹ Bleeding from a Mallory-Weiss tear—a laceration in the mucosa at the gastroesophageal junction extending into the gastric cardia or, less commonly, the distal esophagus—is often self-limited, compared to the recurrent bleeding usually seen with a DL.

Some studies have shown an association between DL-like bleeding and alcohol intake or advanced liver disease, but no association has been seen with nonsteroidal anti-inflammatory drugs.^4,20,21,22 However, concomitant administration of these agents may conceivably prolong and increase the severity of bleeding in patients with a DL.

Another newly described entity that can be diagnosed using the previously mentioned endoscopic criteria is DL-like bleeding caused by isolated gut spider nevi in patients with advanced liver disease. Spider nevi are central arterioles with small vessels radiating from the center.^23,24 Both DLs and bleeding from isolated spider nevi may share a common endoscopic appearance in patients with advanced liver disease. Pressure within a central arteriole can reach as high as 70 mmHg, and erosion of thin overlying mucosa can give rise to arteriolar spurting bleeding, which is indistinguishable from a classic DL.²⁵ There are many other forms of arterial or arteriolar bleeding that may qualify as DL bleeding based on endoscopic criteria; examples of these forms include aortoesophageal fistulas and splenic artery aneurysms.^26,27

Because endoscopic features are shared with other entities, many authors now characterize bleeding or lesions as “DL-like.” In fact, a PubMed search of the term “DL-like” found 27 results, more than 10% of the results obtained by a search of the term “Dieulafoy” (240). The patient described in the case report by Eddi and colleagues did not have a previous aortic surgery, past penetrating trauma, or advanced liver disease, making DL the likely diagnosis by exclusion.¹ We suggest that “DL-like” be used to describe bleeding when angiographic, endoscopic ultrasound, or pathologic data are unavailable or when less than 50% of cases in a series follow definitive lines of evidence for a DL. These guidelines would promote a more homogeneous population and allow for the creation of needed meta-analyses, as most series are quite small and unlikely to inform recommendations for definitive treatment in different clinical scenarios.

The distribution of these lesions is important and may help to determine a likely common pathway of causation, as previously discussed for a postgastrectomy DL. Currently, endoscopy is the initial step for diagnosing a DL, as this procedure can be both diagnostic and therapeutic.^28,29 When endoscopic investigation fails, angiography, video capsule endoscopy (VCE), techne-tium-99m–labeled red blood cell scan, and endoscopic ultrasonography can be useful for identifying the correct diagnosis.^2,14,15 In the case presented by Eddi and associates, VCE would not have contributed to the diagnosis.¹ Endoscopic ultrasound can reveal a large-caliber vessel in the submucosa, suggesting a diagnosis of DL in the absence of bleeding.¹⁵ Angiography can particularly aid in the diagnosis of a DL in the colon or rectum, where endoscopic visualization can be obscured due to poor bowel preparation or active bleeding.³⁰

The success rates of various endoscopic interventions for achieving hemostasis of DL bleeding range from 79% to 100% (Table l).^11,29,31,32 Several studies have reported that EBL, one of these interventions, is inexpensive, easy to use, and as effective as bipolar coagulation or injection.^32,33 Likewise, endoscopic hemostatic clip application is as effective as EBL for achieving hemostasis of a DL.³⁴ In difficult-to-treat cases, angiography can not only aid in the diagnosis of a DL, but it can also be useful for achieving hemostasis by embolization.

Table 1.

Initial Success Rates of Various Endoscopic Methods for Treatment of Bleeding Caused by Dieulafoy Lesions

Modality	Hemostatic success rate	Rebleeding rate	Reference*
Ethanol injection	98%	11%	Asaki S, et al36
Norepinephrine and polidocanol	96.4%	11%	Baettig B, et al12
Thermal probe	79%	9%	Norton ID, et al29
Hemostatic clip placement	94.1% 93.8%	9.3% 0%	Yamaguchi Y, et al37 Park CH, et al40
Endoscopic band ligation	100%	None	Alis H, et al31
Argon plasma coagulation	82%	6%	Lacopini F, et al39
Nd:YAG laser	80%	Not available	Skok P38

^*These references are a representative—not an exhaustive—list of applicable studies.

Nd:YAG=neodymium: yttrium aluminum garnet.

When endoscopic and angiographic interventions fail, surgical treatment in the form of wide-wedge resection of the affected portion or oversewing of the lesion can be performed.⁴ Newer, minimally invasive laparoscopic surgery has also reportedly been successful for treatment of a DL located in the jejunum and stomach; preoperative endoscopic tattooing of the DL may be a valuable adjunct to the laparoscopist.^30,35

Conclusion

DLs and DL-like bleeding should be included in the differential diagnosis of patients presenting with recurrent, intermittent, massive upper gastrointestinal bleeding in all age groups as well as in patients with multiple comorbidities. Endoscopy remains the diagnostic and therapeutic tool of choice, with high success rates. The case reported by Eddi and coworkers reflects this statement and serves to reinforce many clinical considerations.¹