Pathways Case Record: A Patient With Arteriovenous Malformations

A 68-year-old man with coronary artery disease, heart failure, atrial fibrillation, and recurrent melena was admitted to an outside hospital with multiple episodes of dark, tarry stool and anemia requiring blood transfusions. He was transferred to Massachusetts General Hospital for further care and management. At Mass General, he underwent CT angiography of the abdomen and pelvis that did not reveal an active bleeding source from the gastrointestinal (GI) tract.

Subscribe to the latest updates from Advances in Motion

Thank you for subscribing!

Error: Please enter a valid email address.

Email Address Submit

His history was notable for prior hospitalizations for melena. He was found to have small bowel arteriovenous malformations (AVMs), abnormal blood vessels in which arteries and veins are connected in an irregular manner. These AVMs were the likely source of the bleeding in our patient, occurring in response to anticoagulation medications used to reduce stroke risk in the setting of his atrial fibrillation. As there was concern that his anticoagulation medications were increasing his risk of bleeding, he was taken off these medications several months prior to his current presentation but, unfortunately, continued to have episodic melena. His medications included cardiac-related medications for his heart failure and atrial fibrillation. The patient had no known family history of GI issues or bleeding disorders.

The Pathways Service in the Department of Medicine at Mass General was consulted and focused on the etiology of GI AVMs in our patient, driven by the following questions:

1. What is the etiology of small bowel AVMs in an older patient without clear genetic or acquired risk factors?
2. Do GI AVMs show loss of zonation, similar to cerebral AVMs?
3. Do GI AVMs also show infiltration of distinct immune cell states?
4. Are there genetic factors that correlate with altered gene expression in AVMs?

Background and Diagnosis

While AVMs contribute to half of GI bleeds in patients older than 50, little is known about the pathophysiology of GI AVMs (American College of Gastroenterology). GI AVMs are commonly associated with disorders causing alterations in the coagulation pathway, such as von Willebrand disease, end-stage renal disease (from uremia-induced platelet dysfunction), or mechanical forces that consume clotting factors required to stop bleeding (aortic stenosis, where calcified aortic valve leads to shearing of clotting factors). They are also commonly found in patients with hereditary hemorrhagic telangiectasia (HHT; widened blood vessels on the skin), an inherited disorder predisposing patients to form AVMs throughout their bodies. Lastly, they often develop in the colon due to normal aging.

No unifying mechanism explains why AVMs develop, but several hypotheses exist. One common theory to explain the prevalence of colonic AVMs in the elderly is that as we age, we have increased contractility in the muscularis propria (a layer of smooth muscles near the submucosa of the GT tract), leading to intermittent obstruction and congestion of the submucosal veins. This congestion ultimately leads to the development of new vessels that form AVMs (Gastroenterology). Other researchers have hypothesized that AVMs develop in cases of aortic stenosis and von Willebrand disease via increased vascular endothelial growth factor (VEGF) signaling and angiogenesis (Aliment Pharmacol Ther). It is important to note that prior studies have determined that the pathology underlying AVM development lies within the endothelial cells (Circ Res, Proc Natl Acad Sci U S A). Mutations in key signaling pathways responsible for endothelial cell differentiation, such as TGF-β, VEGF, and NOTCH have been shown to cause AVMs. For example, mutations in ENG, GDF2, ACVRL1, and SMAD genes, common germline mutations seen in HHT, result in abnormal TGF-β signaling, ultimately leading to AVM formation (Hematology Am Soc Hematol Educ Program).

It is not understood why an individual without the diseases listed above would develop numerous AVMs. We hypothesize that it is an acquired problem within the signaling pathways responsible for endothelial cell differentiation. We propose that changes in gene expression within endothelial cells create a unique endothelial cell that expresses both venous and arterial genes. A similar type of endothelial cell was recently reported through a study of cerebral AVMs with the first single-cell atlas of the normal and malformed human brain vasculature (Science). Using single cell RNA sequencing (scRNA-seq), the authors identified more than 40 molecularly defined cell states of vascular, immune, and neighboring glial or neuronal cells from the human adult cerebrovasculature and AVMs (n = 5 samples per group). They found a loss of normal zonation among endothelial cells in AVMs and observed the emergence of a distinct transcriptomic state that corresponded to the presumed site of AVM formation, the so-called 'nidus.' Specifically, the nidus was characterized by heightened angiogenic potential and immune cell crosstalk. They also found that infiltration of distinct immune cell states, such as GPNMB⁺ monocytes, contributed to depletion of stabilizing smooth muscle cells in AVMs that bled. Overall, this paper implicated immune cell infiltration and dysregulated transcriptomic resulting in altered cell fate in the formation of AVMs.

Summary and Future Steps

While cerebral AVMs have been studied, little focus has been devoted to GI AVMs. Consequently, it is unknown if GI AVMs show cell-level loss of zonation or infiltration of distinct immune cell states, like cerebral AVMs. It is also unknown if additional, unknown genetic factors correlate with altered gene expression within AVMs. To address these knowledge gaps, we propose a scRNA-seq experiment focused on GI AVMs. Specifically, we would seek to surgically extract a sample of the patient's GI AVMs including the 'nidus' along with a sample of normal GI vasculature for comparison. We would then conduct scRNA-seq on the healthy and AVM tissue to compare the prevalence of aberrant cells and immune cell infiltration. One limitation of this study is that, by comparing tissue from the same individual, we may not be able to detect baseline changes unique to this patient that are present in both normal and abnormal tissue that increase his risk of developing AVMs. Finally, we would perform whole genome sequencing to identify variants of interest that may modulate gene expression in GI AVMs. These studies could guide the development of novel and targeted therapeutics for the prevention and treatment of GI AVMs.

Current therapeutics for patients like this are, unfortunately, limited. First line treatment is endoscopic intervention via cautery or mechanical hemostasis. Second line treatments include angiography for embolization of bleeding vessel or vasopressin infusion and surgical resection. Some medical interventions exist but have limited data and efficacy. One medical treatment is Octreotide, a somatostatin analog that inhibits the release of vasodilator hormones, such as glucagon, indirectly causing splanchnic vasoconstriction and decreased portal inflow. Alternative agents include angiogenesis inhibitors such as bevacizumab (antibody against VEGF) and thalidomide (downregulates VEGF expression). However, there are currently no proven therapeutics for the prevention of AVMs. By elucidating the pathophysiology of AVM development, we hope to identify new targets for intervention.

Overall, this case demonstrates that there is significantly more to discover and learn about the etiology of AVMs in the GI tract, that cell fate is an important pathophysiological process for many disease states including AVMs. Furthermore, scRNA-seq is a novel technology that can provide insight into cell states and may help identify targeted therapies for AVMs.

Learn about the Pathways Consult Service

Refer a patient to the Pathways Consult Service