Anticoagulation-Amplified Recurrent Esophageal Hemorrhage in Triple-Positive Antiphospholipid Syndrome: The Inescapable Anticoagulation Dilemma

Abstract

Introduction

Antiphospholipid syndrome (APS) is an acquired thrombophilia. Triple-positive APS, defined by lupus anticoagulant, anticardiolipin, and anti-β2-glycoprotein I antibody positivity, usually requires long-term vitamin K antagonist therapy; direct oral anticoagulants are discouraged, and low-molecular-weight heparin is reserved for selected situations.

Case Presentation

A 52-year-old female with systemic lupus erythematosus and triple-positive APS on warfarin (International Normalized Ratio [INR] 2.9) presented with massive hematemesis requiring intubation. Serial esophagogastroduodenoscopy showed extensive esophageal clot burden followed by circumferential friability, sloughing, and necrotic/exudative esophagitis without varices or gastric/duodenal bleeding. Hemostatic powder was applied, fresh frozen plasma partially corrected the INR to 1.8, and anticoagulation was held. Therapeutic heparin bridging on hospital day 8 was followed by recurrent melena, hemoglobin decline from 8.3 to 5.5 g/dL, and gross hematuria; INR at rebleeding was 2.2. Hematology evaluation showed factor VIII activity 87% and negative Bethesda assay, arguing against acquired hemophilia A. Repeat endoscopy showed Los Angeles grade C esophagitis with clean-based linear ulcers and a 7-cm sliding hiatal hernia.

Conclusion

Severe diffuse esophageal mucosal injury can cause recurrent major bleeding when anticoagulation is resumed before mucosal recovery. In high-risk APS, anticoagulation resumption should be individualized with hematology and gastroenterology input, and repeat endoscopy may help guide timing.

Introduction

Antiphospholipid syndrome (APS) is an acquired autoimmune thrombophilia associated with arterial, venous, and microvascular thrombosis [1]. APS is defined by clinical thrombosis and/or pregnancy morbidity with persistent antiphospholipid antibodies (lupus anticoagulant, anticardiolipin, and/or anti–β2-glycoprotein I) [1]. Vitamin K antagonists (VKAs) remain standard for thrombotic APS, particularly in high-risk (triple-positive) patients [2]. In high-risk APS, rivaroxaban was associated with increased recurrent thrombosis compared with warfarin, and the European Alliance of Associations for Rheumatology (EULAR) recommends against direct oral anticoagulants (DOACs) in triple-positive APS [2, 3].

Gastrointestinal (GI) involvement in APS most often reflects thrombotic/ischemic complications, but hemorrhagic presentations have been reported, including GI bleeding associated with intestinal ischemia/ulceration and catastrophic APS [4–6]. Reported cases include GI hemorrhage with intestinal ischemia and duodenal ulceration in association with anticardiolipin antibodies [4], giant gastric ulceration with occlusive vasculopathy [5], and catastrophic APS presenting with severe GI bleeding [6].

While hemorrhagic complications of esophageal ulceration in the setting of anticoagulation have been described [7], the intersection of diffuse desquamative esophageal hemorrhage with the need for long-term anticoagulation in triple-positive APS creates a difficult management problem. In triple-positive APS, defined by simultaneous positivity for lupus anticoagulant, anticardiolipin, and anti-β2-glycoprotein I antibodies per the 2023 ACR/EULAR classification criteria [1, 8], long-term VKA therapy remains the primary standard, as DOACs have demonstrated increased recurrent thrombosis in this population and are explicitly contraindicated by EULAR guidance [2, 3]. Although long-term low-molecular-weight heparin has been described as an alternative in selected APS patients who fail or cannot tolerate VKA therapy [3, 9], this approach requires indefinite subcutaneous injections, carries long-term risks including osteoporosis, and lacks prospective evidence specifically in triple-positive APS with prior arterial thrombosis. Unlike anticoagulation in atrial fibrillation or venous thromboembolism, where temporary interruption or straightforward agent substitution is permissible, triple-positive APS limits therapeutic options and makes prolonged anticoagulation interruption clinically concerning. The management problem is therefore timing resumption safely: it is not whether to resume anticoagulation but how and when, in the context of diffuse, incompletely healed esophageal mucosa that hemorrhages with each resumption attempt.

We report life-threatening upper GI bleeding from diffuse severe esophageal mucosal injury on serial endoscopy, complicated by recurrent hemorrhage after therapeutic anticoagulation was resumed in a patient with triple-positive APS. To our knowledge, diffuse esophageal sloughing/friability as an apparent bleeding source in APS requiring anticoagulation is uncommon in the literature. The CARE checklist has been completed by the authors for this case report, attached as online supplementary material (for all online suppl. material, see https://doi.org/10.1159/000552878).

Case Presentation

A 52-year-old female with systemic lupus erythematosus (SLE) and triple-positive APS presented with acute hematemesis. Her triple-positive APS profile was based on persistently positive lupus anticoagulant, anticardiolipin IgG, and anti-β2-glycoprotein I IgG. The highest documented solid-phase antibody titers were anticardiolipin IgG 82 GPL units and anti-β2-glycoprotein I IgG 68 units, measured by enzyme-linked immunosorbent assay and confirmed positive on repeat testing at least 12 weeks apart. Anti-phosphatidylserine/prothrombin IgG/IgM antibodies were also positive, but these were considered non-criteria antibodies and were not used to define triple positivity. Her medical history was notable for prior ischemic stroke with branch retinal artery occlusion, Libman-Sacks endocarditis, chronic kidney disease stage III, obesity (body mass index 47.4 kg/m²), and gastroesophageal reflux disease. She denied nonsteroidal anti-inflammatory drug use, and no history of recent doxycycline, bisphosphonates, or potassium chloride. She also denied caustic ingestion and recent nasogastric tube placement. Three weeks before admission, her warfarin dose was increased from 7.5 mg to 10 mg daily after a subtherapeutic International Normalized Ratio (INR) of 1.6. In addition, she was on mycophenolate mofetil 1,000 mg twice daily and hydroxychloroquine 200 mg twice daily for the treatment of SLE. Three months prior to this admission, she was seen by a gastroenterologist for asymptomatic elevation of liver transaminases with normal abdominal imaging and negative infectious and autoimmune testing. She denied drinking alcohol, taking herbal supplements, or having taken antibiotics recently. She denied any history of GI bleeding or esophageal disease.

One hour after dinner on the night of admission, the patient developed acute upper back pain followed by profuse hematemesis associated with diaphoresis. On presentation, she was hemodynamically stable with ongoing hematemesis. Laboratory workup revealed a hemoglobin of 12.4 g/dL, an INR of 2.9, a creatinine of 1.47 mg/dL, and an activated partial thromboplastin time of 38 s (reference range: 25–35 s). She had persistent bleeding and could not be positioned in a recumbent position, so she underwent endotracheal intubation for airway protection and was admitted to the intensive care unit. Fresh frozen plasma was administered for partial warfarin reversal; four-factor prothrombin complex concentrate was not administered due to concern for paradoxical thrombosis in a patient with triple-positive APS and prior arterial thrombotic events. An intravenous proton pump inhibitor was started twice daily, and warfarin, mycophenolate, and hydroxychloroquine were held. Sequential compression devices were applied bilaterally for mechanical thromboprophylaxis. After fresh frozen plasma administration, the INR decreased from 2.9 to 1.8 on hospital day 2; it did not fully normalize.

The patient underwent an emergent esophagogastroduodenoscopy (EGD) on hospital day #2, which demonstrated an extensive intraluminal blood clot accumulation present within the esophagus, making it difficult to adequately visualize even after aggressive suctioning (Fig. 1a, b). There was mucosal disruption in the distal esophagus noted after partial clot removal, and no evidence of varices or active bleeding was noted. Hemostatic powder (TC-325, Hemospray) was applied circumferentially to the distal esophagus. Computed tomography (CT) angiogram showed no active GI bleeding and demonstrated distal esophageal wall thickening with retained ingested material, interpreted as esophagitis. EGD performed on hospital day 3 again demonstrated an extensive adherent blood clot accumulation to the mid and distal esophagus (Fig. 2a). EGD was repeated on hospital day 5, which showed circumferential exudative and necrotic lesions with markedly friable mucosa and prominent white exudates were noted, consistent with severe esophagitis (Fig. 2b). No gastric or duodenal abnormalities were noted at the time. Due to mucosal friability, biopsies were not obtained, and empiric fluconazole therapy was initiated for possible candidal esophagitis.

Fig. 1.

Esophagogastroduodenoscopy (EGD) findings on hospital days 2 and 3. a Upper-to-mid esophagus: diffuse hemorrhagic ulcerative esophagitis with extensive adherent clot, limiting visualization despite suctioning. No esophageal varices identified. b Distal esophagus: deep ulceration with mucosal disruption felt to represent the primary bleeding source. No pseudolumen or evidence of perforation.

Fig. 2.

Esophagogastroduodenoscopy (EGD) findings on hospital day 5. a Distal esophagus: circumferential inflammatory edema and ulcerative mucosal injury with luminal narrowing and persistent adherent clot, without active bleeding at the time of examination. b Distal esophagus and gastroesophageal junction: extensive mucosal denudation and sloughing involving the distal esophagus and gastroesophageal junction, without black necrosis or pseudomembrane formation.

Her intensive care unit course was complicated by a hemoglobin drop requiring 2 units of packed red blood cells and concerns for aspiration pneumonia treated with ceftriaxone. The patient was successfully extubated on the evening of hospital day 5 and transferred to the general medical unit. Due to her high thrombotic risk, therapeutic anticoagulation was initiated with a heparin infusion as a bridge to warfarin on hospital day 8. Additionally, mycophenolate and hydroxychloroquine were restarted.

On hospital day 10, the patient experienced an acute clinical decline manifested by melena and odynophagia, with a hemoglobin drop from 8.3 g/dL to 5.5 g/dL. The INR at the time of rebleeding was 2.2, within the therapeutic range. Anticoagulation was immediately discontinued, and 2 units of packed red blood cells were transfused. Hematology was consulted to evaluate for acquired coagulopathy; factor VIII activity was 87% (reference range: >50%), and the Bethesda assay detected no factor VIII inhibitor, excluding acquired hemophilia A. Concurrently, she developed gross hematuria with urinary retention. Computed tomography urogram demonstrated bladder wall thickening consistent with cystitis. Renal function transiently worsened, attributed to prerenal azotemia, and improved with intravenous fluid resuscitation.

A fourth EGD on hospital day 12 demonstrated patchy Los Angeles grade C reflux esophagitis with multiple mucosal breaks and several large, superficial, linear ulcers with clean bases (Forrest III) involving the mid and distal esophagus (Fig. 3a, b). A 7-cm sliding hiatal hernia was noted (Fig. 4); however, no active bleeding or stigmata of recent hemorrhage were present.

Fig. 3.

Esophagogastroduodenoscopy (EGD) findings on hospital day 12. a Healing ulcerative esophagitis in the mid-esophagus with patchy erythema and superficial mucosal breaks without active bleeding or adherent clot. b Distal esophagus and gastroesophageal junction on hospital day 12 showing large, superficial linear ulcers with clean bases (Forrest III), consistent with resolving ulcerative esophagitis and no evidence of active or recent bleeding.

Fig. 4.

Esophagogastroduodenoscopy (EGD) findings on hospital day 12 (continued). Retroflexed view demonstrating ∼7-cm sliding hiatal hernia (type I) without Cameron lesions.

No clinical signs or symptoms of deep venous thrombosis were observed during the hospitalization. No screening venous Doppler ultrasound was performed in the absence of limb swelling, limb pain, chest pain, hypoxemia, or other clinical features suggestive of venous thromboembolism. The patient was discharged on hospital day 13 to an inpatient rehab unit with a hemoglobin of 9.5 g/dL and an INR of 2.0. Anticoagulation had been discontinued on hospital day 10 following the rebleeding episode; the INR of 2.0 at discharge reflected residual warfarin effect from therapy administered between hospital days 8 and 10 and was anticipated to decline. Warfarin was therefore carefully restarted at discharge, with subcutaneous enoxaparin as bridging therapy to maintain therapeutic anticoagulation during the expected nadir period. This decision followed the hospital day 12 EGD showing clean-based Forrest III ulcers without active bleeding or stigmata of recent hemorrhage, and was made because of her high thrombotic risk from triple-positive APS with prior arterial events. She was started on pantoprazole 40 mg two times per day and a short course of sucralfate. Outpatient post-discharge plans were to include a repeat upper endoscopy in 3 months and a urology visit.

Discussion

This case illustrates a life-threatening presentation of hematemesis due to extensive esophageal mucosal injury with a massive intraluminal clot and widespread circumferential friability, sloughing, and ulceration on serial endoscopy. Despite repeated evaluations, no focal source of bleeding was identified. The extensive intraluminal blood clot accumulation was itself a diagnostic and therapeutic obstacle, as it prevented complete mucosal assessment on the initial two endoscopies and precluded safe biopsy acquisition. Computed tomography angiogram showed no active extravasation; however, CT angiography sensitivity depends on active bleeding rate and timing, and a negative study does not exclude intermittent bleeding [10]. Given the absence of a focal endoscopic lesion and repeated findings of extensive circumferential injury with clot accumulation, the bleeding source was suspected to be from diffuse esophageal mucosal hemorrhage. Although formal endoscopic characterization of ulcerative esophagitis was progressively established over serial procedures (hospital days 2, 3, 5, and 12), the esophageal mucosal injury itself antedated and caused the initial hemorrhage; the delay in diagnosis was attributable to the extensive intraluminal blood clot accumulation that obscured direct mucosal visualization on the earliest endoscopies. This case reveals the difficulty of balancing bleeding risk against thrombosis prevention in a patient with triple-positive APS and a history of arterial thrombosis. In triple-positive APS and in APS with arterial thrombosis, long-term warfarin is generally recommended, while DOACs are discouraged based on randomized data and society guidance [2, 3]. This case shows that widespread esophageal mucosal injury, rather than a focal lesion, can produce major bleeding and can rebleed when anticoagulation is resumed before mucosal recovery. Prior descriptions of esophagitis-related hemorrhage in anticoagulated patients have not addressed the specific context of triple-positive APS, in which the treating physician faces limited anticoagulation options. Warfarin remains the primary standard, DOACs, particularly rivaroxaban, are discouraged in triple-positive APS, and long-term low-molecular-weight heparin, while a potential alternative in selected cases [3, 9], carries practical limitations including the need for indefinite subcutaneous injections and lacks prospective data in triple-positive APS with arterial thrombosis. In this setting, prolonged interruption of anticoagulation carries substantial thrombotic concern, but early resumption may provoke recurrent bleeding before mucosal recovery. This case uses serial endoscopy to show how persistent esophageal mucosal injury complicated anticoagulation resumption in triple-positive APS with prior arterial thrombosis.

Aside from anticoagulation considerations, the endoscopic findings required careful differential evaluation. There is a broad differential diagnosis for extensive esophageal injury with sloughing and mucosal friability. Severe reflux or ulcerative esophagitis, particularly in a patient with chronic gastroesophageal reflux disease, morbid obesity, and a large hiatal hernia, can cause significant mucosal damage and friability. The endoscopic appearance raised concern for esophagitis dissecans superficialis (EDS). EDS is a clinicopathologic entity characterized endoscopically by sloughing sheets of squamous mucosa and histologically by superficial epithelial separation with histology supporting the diagnosis when biopsy is safe [11]. In our patient, biopsy was deferred due to extreme mucosal friability and the risk of provoking hemorrhage in the setting of a fragile mucosa.

In addition to noninfectious causes, infectious etiologies such as Candida, cytomegalovirus, and herpes simplex virus can also cause esophagitis in immunocompromised patients. Although early visualization was limited by extensive blood clot accumulation, classic endoscopic features such as adherent white plaques suggestive of candidiasis or discrete deep ulcers typical of viral esophagitis were not clearly identified. Infectious esophagitis remained a consideration given immunosuppression and adherent white material; however, tissue sampling was deferred, limiting diagnostic certainty. On subsequent endoscopies, the distribution and sloughing phenotype raised concern for a noninfectious desquamative process. Acute esophageal necrosis (“black esophagus”) is a rare cause of upper GI bleeding characterized by circumferential black esophageal mucosa, typically in critically ill patients with hypoperfusion [12], which was not the case with our patient. Intramural esophageal hematoma/dissection can present with acute pain and hematemesis and may show an intramural hematoma or a double-lumen (“double-barrel”) appearance on imaging/endoscopy, which was not described in this case [13].

In patients with SLE presenting with severe or unexpected hemorrhage, acquired hemophilia A due to anti-factor VIII autoantibodies warrants consideration, particularly when hemorrhage occurs at unusual mucosal sites or is disproportionate to the degree of anticoagulation [14]. Acquired hemophilia A is an autoimmune bleeding disorder that characteristically produces soft tissue and mucocutaneous hemorrhage, with markedly prolonged activated partial thromboplastin time and severely reduced factor VIII activity [14]. In our patient, hematology consultation was obtained, and factor VIII activity was 87% with no inhibitor detected on Bethesda assay, effectively excluding acquired hemophilia A. This evaluation is particularly relevant given the concurrent immunosuppression with mycophenolate mofetil and the presence of SLE, both of which are recognized associations with acquired factor VIII inhibitors [14]. The normal factor VIII activity and negative inhibitor screen argued against acquired hemophilia A as the explanation for bleeding. The clinical course therefore supported anticoagulation-amplified mucosal injury as the main driver, although the absence of biopsy limits certainty about the exact esophageal pathology.

The severity and recurrence of bleeding in this case were likely multifactorial. The rebleeding on hospital day 10 occurred at a therapeutic INR of 2.2, confirming that even standard-intensity anticoagulation (not supratherapeutic levels) was sufficient to provoke hemorrhage from incompletely healed mucosa. This patient had baseline mucosal vulnerability related to chronic gastroesophageal reflux disease, morbid obesity, and a large hiatal hernia. These factors may have predisposed the esophagus to significant reflux-mediated injury, supported by persistent Los Angeles grade C esophagitis and a 7-cm sliding hiatal hernia discovered endoscopically on hospital day 12. Superimposed anticoagulation, including recent warfarin dose escalation followed by heparin bridging, may have amplified hemorrhage from otherwise superficial mucosal breaks, transforming widespread injury into clinically significant bleeding. The observed time course supports this model, with early endoscopic evaluations limited by extensive intraluminal clot accumulation followed by later findings of healing but persistent Los Angeles grade C esophagitis and Forrest III ulcers, consistent with a generalized injury undergoing incomplete resolution at the time anticoagulation was resumed.

The concurrent development of gross hematuria on hospital day 10, coinciding with the second episode of melena after heparin reinitiation, is relevant. CT urogram demonstrated bladder wall thickening without structural lesion, a pattern consistent with hemorrhagic cystitis in the context of therapeutic anticoagulation amplifying underlying mucosal vulnerability. No structural urologic pathology (mass, calculus, or arteriovenous malformation) was identified, and hematuria resolved spontaneously upon anticoagulation cessation, paralleling the esophageal bleeding course. This simultaneous esophageal bleeding and hematuria after heparin bridging suggests that the mucosal bleeding risk extended beyond the esophagus after anticoagulation was resumed. This finding supports repeat endoscopic assessment before future anticoagulation resumption when clinically feasible, as premature reinitiation appears to unmask hemorrhagic potential across multiple mucosal surfaces simultaneously.

The patient’s bleeding risk was further amplified by anticoagulant therapy. Anticoagulants, such as warfarin, decrease clotting factors and increase risk for GI bleeding, especially in the presence of underlying mucosal vulnerability. In largely non-APS cohorts with major GI bleeding, resumption of anticoagulation has been associated with lower thromboembolism and mortality, with variable effects on rebleeding risk [15–17]. In patients with very high thrombotic risk, similar to our patient with triple-positive APS, this individualized decision should be made in a multidisciplinary discussion weighing rebleeding risk against risk of thrombosis.

Endoscopic management in this case displayed distinct challenges due to the friability of the mucosa. The extensive nature of mucosal injury prevented focal hemostatic techniques from being performed. Hemostatic powder was applied because targeted endoscopic therapy was not feasible given diffuse circumferential mucosal injury and blood clot accumulation. Biopsy was deferred throughout the hospital course due to extreme mucosal friability and the risk of provoking hemorrhage in the setting of ongoing anticoagulation and fragile mucosa. Although histopathologic analysis could have clarified the precise underlying pathology, including confirmation of EDS versus severe ulcerative reflux esophagitis versus infectious processes, safety considerations outweighed diagnostic yield in this critically ill patient.

This case has important limitations. In the absence of histological confirmation, it is difficult to prove the underlying etiology of the esophageal mucosal injury. CT imaging findings were nonspecific by describing “esophagitis” without further characterization. Empiric antifungal therapy was initiated without tissue biopsy, limiting causal inference with regard to infection. Despite these limitations, the clinical narrative provides insight into a rare but serious presentation of widespread esophageal injury complicated by anticoagulation management.

Clinicians must recognize that sloughing and ulcerative esophagitis can be a source of life-threatening upper GI bleeding, particularly in patients receiving anticoagulation. In high-thrombotic-risk conditions such as triple-positive APS, the timing and strategy of anticoagulation resumption must be individualized, as premature reinitiation may provoke recurrent hemorrhage. Finally, when endoscopic appearance suggests sloughing esophagitis or EDS, clinicians should cautiously interpret findings and avoid prematurely attributing pathology to infection without histologic evidence when biopsy is unsafe.

Statement of Ethics

This study was conducted in accordance with the principles of the Declaration of Helsinki. Ethical approval was not required for this case report because it does not meet the US federal regulatory definition of research involving human subjects (45 CFR 46) and is therefore not considered human subjects research. This determination is consistent with widely accepted institutional policies for single case reports. Written informed consent was obtained from the patient for publication of the details of this medical case and any accompanying images. All identifying information has been anonymized to protect patient privacy.

Conflict of Interest Statement

The authors have no conflicts of interest to declare.

Funding Sources

This study was not supported by any sponsor or funder.

Author Contributions

A.A.: concept and design, data acquisition, manuscript drafting, critical revision, and final approval. C.Z.: data acquisition, manuscript drafting, critical revision, and final approval. A.G. and H.B.: manuscript drafting, critical revision, and final approval. All authors have read and approved the final manuscript.

Funding Statement

This study was not supported by any sponsor or funder.

Data Availability Statement

All data generated or analyzed during this study are included in this published article. Further inquiries can be directed to the corresponding author (A.A.) at akram.alnounou@wmed.edu.