Hepatic portal venous gas in a patient with diabetes on GLP-1 receptor agonist and alpha-glucosidase inhibitor

Abstract

Hepatic portal venous gas (HPVG) is a rare and potentially fatal condition. We report the case of an elderly patient with type 2 diabetes mellitus who developed HPVG. An 84-year-old woman, treated with oral hypoglycemic agents for 40 years and insulin therapy for 24 years, presented with frequent constipation due to diabetic autonomic neuropathy. She also experienced recurrent hypoglycemia resulting from blood glucose fluctuations caused by poor dietary habits. She was admitted for the management of type 2 diabetes. To prevent hypoglycemia, insulin therapy was discontinued, and treatment with dulaglutide, a glucagon-like peptide-1 receptor agonist (GLP-1 RA), was initiated. Following the addition of acarbose, an alpha-glucosidase inhibitor (αGI), the patient developed abdominal pain and nausea. Abdominal computed tomography revealed HPVG without intestinal necrosis or signs of peritoneal irritation. Conservative therapy, along with the discontinuation of dulaglutide and acarbose, resulted in gradual improvement. We suspect that HPVG in this case was caused by dulaglutide-induced suppression of intestinal peristalsis combined with acarbose-induced elevation of intestinal pressure. To date, no cases of HPVG have been reported in patients receiving combination therapy with dulaglutide and acarbose. Clinicians should be vigilant regarding abdominal symptoms in patients receiving both GLP-1 RAs and αGIs.

Introduction

Hepatic portal venous gas (HPVG), also known as portal emphysema, is a radiographic finding defined as gas accumulation in the portal venous system of the liver [1], typically identified using imaging techniques such as abdominal computed tomography (CT) [2]. HPVG is frequently associated with bowel necrosis, ulcerative colitis, intra-abdominal abscesses, bowel obstruction, and gastric ulcers [1]. It is prevalent in patients with intestinal necrosis and is associated with extremely poor clinical outcomes. However, recent advances in imaging technology have led to increased identification of HPVG in patients with various benign conditions [3]. Although the etiology of HPVG remains elusive, 3 main mechanisms have been proposed: failure of mucosal defense mechanisms due to mucosal damage, increased intraluminal pressure, and the presence of gas-producing bacteria [1].

Management of the underlying cause and clinical symptoms remains the primary strategy for HPVG. Treatment modalities vary based on severity and etiology. For example, severe HPVG caused by perforation or ischemia requires surgical intervention, whereas mild HPVG can be managed conservatively with close monitoring and supportive care [4].

HPVG is a known adverse effect of alpha-glucosidase inhibitor (αGI) use in patients with type 2 diabetes mellitus [5]. However, reports of HPVG caused by glucagon-like peptide-1 receptor agonists (GLP-1 RAs) are limited. In this report, we present a case of HPVG attributed to the combined use of a GLP-1 RA and an αGI.

Case presentation

An 84-year-old woman with type 2 diabetes mellitus complicated by diabetic autonomic neuropathy, hypertension, and dyslipidemia was admitted to our hospital for inpatient diabetes management and education. Her medical history included appendicitis, ischemic heart disease, and retinal photocoagulation for pre-proliferative diabetic retinopathy. Her medications on admission included insulin aspart (4 units before breakfast, lunch, and dinner), insulin glargine (5 units once daily in the evening), and sitagliptin (50 mg/day). She experienced recurrent episodes of hypoglycemia and increased appetite, reporting an inability to stop eating between meals. The patient expressed concerns about continuing insulin therapy due to her age and was admitted for blood glucose management. On admission, her body weight was 52.6 kg, height was 137.9 cm, and body mass index was 28.0 kg/m². Her vital signs were stable, with a body temperature of 36.5 °C, a pulse rate of 75 beats/min, and a blood pressure of 148/80 mmHg. She had no abdominal symptoms, though a surgical scar from a previous appendectomy was noted. No formal diagnostic evaluation for pre-existing gastroparesis was performed on admission. She was unable to perceive a 10-g Semmes–Weinstein monofilament, and her Achilles and patellar reflexes were absent bilaterally. Cardiovascular autonomic reflex testing showed reduced heart rate variability, with a coefficient of variation of R-R intervals of 0.94%, consistent with diabetic autonomic neuropathy. Her glycated hemoglobin (HbA1c) level was 7.7% (60 mmol/mol) (reference range: 4.9-6.0%; 30-42 mmol/mol).

Evaluation of endogenous insulin secretion on admission revealed impairment: the serum C-peptide level was slightly below the normal range at 0.23 nmol/L (0.7 ng/mL) (reference range: 0.26-0.83 nmol/L; 0.8-2.5 ng/mL), and the C-peptide index was reduced to 0.74. Urinary C-peptide excretion was decreased to 15.3 μg/day (reference range: 22.8-155.2 μg/day). Tests for glutamic acid decarboxylase antibodies were negative. On day 5 of hospitalization, insulin therapy was gradually tapered and subsequently discontinued to prevent hypoglycemia. Dulaglutide, a long-acting GLP-1 RA (0.75 mg once weekly), was initiated. The patient tolerated dulaglutide without adverse effects. On day 9, oral acarbose (150 mg/day) was added to treat postprandial hyperglycemia. However, on day 12, the patient developed nausea and abdominal pain. The longitudinal changes in blood glucose levels during hospitalization are shown in Fig. 1.

Figure 1.

Time course of blood glucose levels during hospitalization. Blood glucose levels during hospitalization are shown in relation to the initiation of dulaglutide, the addition of acarbose, and the onset of hepatic portal venous gas. Blood glucose levels were measured using a capillary blood glucose meter. Fasting glucose represents early-morning levels, and random glucose represents pre-lunch glucose.

Diagnostic assessment

At symptom onset, her temperature was 36.2 °C, pulse rate was 70 beats/min, and blood pressure was 111/62 mmHg. Abdominal examination revealed moderate distention without tenderness or rigidity. Laboratory testing revealed a white blood cell count of 3.77 × 10⁹/L (reference range: 3.3-8.6 × 10⁹/L) and C-reactive protein of 1.3 mg/L (0.13 mg/dL) (reference range: 0-1.4 mg/L; 0-0.14 mg/dL). A plain abdominal CT performed 9 months previously showed no HPVG (Fig. 2). Due to renal impairment, she underwent an urgent plain abdominal CT on day 13, which revealed extensive gas within the portal venous system, mesenteric venous gas, and thickening of the ascending colon wall (Fig. 3A and 3B). Consequently, the patient was diagnosed with HPVG, possibly associated with the combined use of a GLP-1 RA and an αGI.

Figure 2.

Plain abdominal CT obtained 9 months before the index admission. No hepatic portal venous gas is observed.

Figure 3.

Plain abdominal CT performed during the index admission (day 13 and day 18). Hepatic portal venous gas is present on day 13 (A, B) but resolves by day 18 following the discontinuation of acarbose and dulaglutide (C, D).

Treatment

Although we considered intestinal necrosis as a potential cause of HPVG, the abdominal pain was mild, and there were no signs of peritoneal irritation. Furthermore, imaging studies showed no ascites or free intraperitoneal air suggestive of bowel ischemia. Conservative treatment was initiated following consultation with the gastroenterology department. Dulaglutide and acarbose, the drugs suspected of causing HPVG, were discontinued, and fasting with peripheral parenteral nutrition was initiated. Laboratory results before and at the onset of symptoms are summarized in Table 1. A plain abdominal CT performed on day 18 revealed resolution of HPVG and mesenteric venous gas; however, thickening of the ascending colon wall and a transient worsening of a right-sided pleural effusion were noted. These findings likely reflected residual inflammatory changes and fluid shifts during the recovery phase, although intestinal ischemia could not be completely excluded (Fig. 3C and 3D). The patient's white blood cell count rose to 10.35 × 10⁹/L, and C-reactive protein increased to 137.7 mg/L (13.77 mg/dL), indicating a worsening inflammatory reaction. Empirical antibiotic therapy with cefoperazone sodium/sulbactam sodium (2 g/day) was initiated to cover potential bacterial translocation associated with the colonic wall thickening. The patient's condition subsequently improved, and oral intake was gradually resumed.

Table 1.

Laboratory results before onset and at onset

Laboratory results	Before onset	At onset	Reference range
Hemoglobin	113 g/L	113 g/L	116-148 g/L
White cells	3.82 × 109/L	3.77 × 109/L	3.3-8.6 × 109/L
Neutrophils	43.0%	50.5%	38.5-80.5%
Lymphocytes	39.8%	34.20%	16.5-49.5%
Platelets	176 × 109/L	168 × 109/L	158-348 × 109/L
Urea nitrogen	8.2 mmol/L (22.9 mg/dL)	10.3 mmol/L (29.0 mg/dL)	2.9-7.1 mmol/L (8.0-20.0 mg/dL)
Creatinine	83.1 μmol/L (0.94 mg/dL)	100.8 μmol/L (1.14 mg/dL)	40.7-69.8 μmol/L (0.47-0.79 mg/dL)
eGFR	43.0 mL/min/1.73 m2	34.8 mL/min/1.73 m2	≥60 mL/min/1.73 m2
AST	19.1 U/L	19.2 U/L	13-30 U/L
ALT	11.9 U/L	8.7 U/L	7-23 U/L
ALP	318 U/L	276 U/L	106-322 U/L
γ-GTP	12.7 U/L	12.8 U/L	9-32 U/L
Total bilirubin	7.7 μmol/L (0.45 mg/dL)	7.9 μmol/L (0.46 mg/dL)	6.8-25.7 μmol/L (0.4-1.5 mg/dL)
Direct bilirubin	0.9 μmol/L (0.05 mg/dL)	1.2 μmol/L (0.07 mg/dL)	≤3.4 μmol/L (≤0.2 mg/dL)
C-reactive protein	0.5 mg/L (0.05 mg/dL)	1.3 mg/L (0.13 mg/dL)	0-1.4 mg/L (0-0.14 mg/dL)

Data are presented in SI units; conventional units are shown in parentheses. Reference ranges are those of our institutional laboratory. “Before onset” denotes laboratory values obtained during the index admission prior to symptom onset (day 4); “At onset” denotes laboratory values at the time of symptom onset (day 12). Day numbers are relative to the index admission (day 0). Bold values indicate results outside the normal reference range.

Abbreviations: ALP, alkaline phosphatase; ALT, alanine transaminase; AST, aspartate transaminase; eGFR, estimated glomerular filtration rate; γ-GTP, gamma-glutamyl transferase.

Outcome and follow-up

To prevent overeating and stabilize glycemic variability, a low dose of liraglutide (0.3 mg/day) was administered on day 26 to minimize gastrointestinal adverse effects. However, the patient developed nausea at noon on the same day, and the drug was discontinued. She was discharged on insulin lispro mix 50/50 (14 units in the morning and 5 units in the evening). Key events and inflammatory markers are shown in Fig. 4.

Figure 4.

Clinical course of the patient. Key events and inflammatory markers are summarized by day.

Discussion

This case report describes an 84-year-old patient with diabetes and associated complications, including constipation resulting from diabetic autonomic neuropathy and pre-proliferative diabetic retinopathy treated with retinal photocoagulation. Although the patient demonstrated decreased insulin secretion, she experienced recurrent hypoglycemia due to advanced age and poor dietary intake. Therefore, we opted for a pharmacological regimen with a low risk of hypoglycemia. Dulaglutide was initiated to stabilize glycemic variability, reduce the risk of recurrent hypoglycemia, and improve the patient's quality of life. Unlike insulin, which requires daily injections, dulaglutide is administered once weekly via a simple device. Furthermore, GLP-1 RAs, such as dulaglutide, promote insulin secretion in a glucose-dependent manner; thus, the risk of hypoglycemia is low when used as monotherapy [6].

Additionally, GLP-1 RAs improve glycemic control by suppressing glucagon secretion and delaying gastric emptying. Excessive suppression can lead to ileus, a complication also reported in patients treated with dipeptidyl peptidase-4 (DPP-4) inhibitors [7].

HPVG is relatively rare and is often associated with lethal intra-abdominal conditions such as intestinal necrosis [1]. Highlighting this severity, a recent nationwide study from Japan (2010-2015, n = 1590) reported an overall in-hospital mortality rate of 27.3%. Mortality rates by cause were 26.8% for intestinal ischemia (16.5% for surgically treated cases), 31.1% for gastrointestinal obstruction or dilation, 33.3% for gastrointestinal perforation, 13.6% for gastrointestinal infection, and 56.4% for sepsis [8]. HPVG caused by αGIs has been previously reported [5]. Our literature review revealed no fatal cases associated with αGIs, and several studies demonstrated improvement following discontinuation of these drugs. The primary adverse effects of αGI therapy are gastrointestinal symptoms, particularly bloating and abdominal discomfort. A potential mechanism for these symptoms is carbohydrate fermentation by intestinal microbiota, resulting in intraluminal gas production. This mechanism, in conjunction with diabetes-associated decreased peristalsis, may increase intraluminal pressure, allowing gas-producing bacteria to enter the intestinal wall [5]. In this case, the GLP-1 RA used is an incretin mimetic that inhibits gastric emptying and reduces peristalsis in the small and large intestines [9]. Cases of HPVG following the use of DPP-4 inhibitors have also been reported [10]. While reports on HPVG caused by GLP-1 RAs are scarce, a recent case described HPVG with gastric pneumatosis in a patient treated with semaglutide for type 2 diabetes, suggesting that impaired gastric motility and increased intragastric pressure may contribute to the development of HPVG in patients receiving GLP-1 RA therapy [11].

Our patient had pre-existing gastrointestinal motility dysfunction due to diabetic autonomic neuropathy. We hypothesized that incretin-induced motility suppression and αGI-induced intestinal gas production increased intraluminal pressure, leading to gas entry into the portal circulation (HPVG) (Fig. 5). Although a definitive causal relationship cannot be established, the temporal association and biologically plausible mechanisms suggest that GLP-1 RA–related gastrointestinal hypomotility, potentially augmented by concomitant αGI therapy, contributed to the development of HPVG. According to the Naranjo Adverse Drug Reaction Probability Scale, the association between GLP-1 RA therapy and HPVG in this case was classified as “possible,” supporting a contributory but non-definitive link. Both αGIs and GLP-1 RAs are commonly used in patients with type 2 diabetes; however, their combined use warrants caution, particularly in patients with diabetes complicated by gastrointestinal autonomic neuropathy. Given the rarity of HPVG associated with GLP-1 receptor agonists, further studies are needed to clarify the underlying mechanisms and identify patients at risk.

Figure 5.

Potential pathophysiology of HPVG in this case. In the context of diabetic autonomic neuropathy–related intestinal hypomotility, GLP-1 receptor agonist (GLP-1 RA)–induced motility suppression and alpha-glucosidase inhibitor (αGI)–driven fermentation increase intraluminal gas and pressure, precipitating HPVG.

Learning points

• HPVG is a high-risk condition that requires caution regarding potential ischemia.

• HPVG is a rare complication of αGI therapy; however, the condition typically resolves upon discontinuation.

• The use of GLP-1 RAs and αGIs may increase the risk of HPVG in patients with diabetes and autonomic gastrointestinal dysfunction.

Contributors

All authors contributed to the diagnosis and management of the patient and to the writing and revision of the manuscript. All authors reviewed and approved the final draft of the manuscript. All authors made individual contributions to authorship. T.N., M.O. (Mitsuhiro Okamoto), M.O. (Masahide Okamoto), K.G., T.M., and H.S. were involved in the diagnosis and management of the patient and manuscript submission.

Abbreviations

αGI

alpha-glucosidase inhibitor

ALP

alkaline phosphatase

ALT

alanine transaminase

AST

aspartate transaminase

CT

computed tomography

DDP-4

dipeptidyl peptidase-4

eGFR

estimated glomerular filtration rate

GLP-1 RA

glucagon-like peptide-1 receptor agonist

HPVG

hepatic portal venous gas

γ-GTP

gamma-glutamyl transferase

Funding

No public or commercial funding was received.

Disclosures

H.S. is an Associate Editor of JCEM Case Reports and had no role in the review or decision process for this manuscript. The other authors have no conflicts of interest to disclose.

Informed patient consent for publication

Signed informed consent was obtained directly from the patient.

Data availability

Original data generated and analyzed for this case report are included in this published article.