Teasing pathologic from benign pneumatosis intestinalis

Author(s):
Julia Song; Biqi Zhang; David Mahvi; Mahsa Shariat; Manuel Castillo-Angeles ; Tanujit Dey; Reza Askari

Background:

Pneumatosis intestinalis (PI) is a rare radiographic finding that can range from benign to needing emergent surgery. Sufficiently powered studies are lacking and recommendations for surgical versus non-operative management remain unclear. We sought to identify key predictors of pathologic PI.

Hypothesis:

Physical exam, laboratory, and radiographic findings can be used to distinguish pathologic from benign PI.

Methods:

Patients ≥ 18 years old with radiographic evidence of PI were identified and retrospectively analyzed at two large academic medical centers (2010-2021). Patients were excluded if they pursued comfort-measures only or if there was concurrent radiographic evidence of a vaso-occlusive process. PI was defined as pathologic if ischemic and/or perforated bowel was found on exploratory laparotomy or if the patient died prior to planned surgery. Full, stepwise, and lasso regression models were compared to determine the best fit model for prediction.

Results:

334 patients with PI were identified. 91 (27%) underwent exploratory laparotomy, of which 59 (65%) had ischemic and/or perforated bowel. These latter patients as well as 10 other patients who died before exploratory laparotomy defined the pathologic PI cohort. The stepwise model was found to be the best fit for predicting pathologic PI (Table 1). Significant predictors were presence of portal venous gas (OR 2.23, 95% CI 1.03-4.83), multi-segment PI (OR 2.01, CI 0.97-4.19), vasopressor use (OR 4.56, CI 1.78-11.70), peritonitis (OR 11.46, CI 4.72-27.82), leukocyte count (OR 1.05, CI 1.01-1.10), and end organ injury (OR 1.75, CI 1.16-2.63), which were used to construct a nomogram for clinical use (Figure 1).

Conclusions:

A nomogram score based on presence of portal venous gas, multi-segment PI, vasopressor use, peritonitis, leukocytosis, and end organ injury can help predict the probability of pathologic PI and therefore inform surgical decision-making.