TI-GI CMV disease was defined as an evidence of CMV infection either from (1) typical CMV cytopathic change demonstrated on hematoxylin and eosin (H&E) staining or (2) positive staining for CMV antigen by immunohistochemistry (IHC) in the tissue specimens from gastrointestinal tract (i.e., esophagus, stomach, small intestine, and colon), in which the specimens were obtained from surgical resection or gastrointestinal endoscopy with biopsy. All tissue histology specimens were reviewed by single clinical pathologist (P.W.) who was blinded to clinical characteristics of the patients to confirm the diagnosis of TI-GI CMV before including the patients into the analysis.

All eligible patients were classified into immunocompetent and immunocompromised groups. Patients who were concomitant with the diagnosis of HIV infection, recipients of solid organ or bone marrow transplantation, neutropenia or receiving immunosuppressive agents, i.e., systemic corticosteroid dose ≥20 mg/day of prednisolone or equivalent for more than two weeks, chemotherapy, and immunomodulating agents (e.g., methotrexate, thiopurine, cyclosporine, and tacrolimus) within six months before the time of TI-GI CMV diagnosis were categorized as immunocompromised host. The remaining patients were considered to be in the immunocompetent group, including those with inflammatory bowel disease (IBD) and low dose or a short course corticosteroid administration.

The treatment outcomes were classified in terms of symptomatic, endoscopic, and histological improvement. The improvement in symptoms was defined as the resolution of the patient's presenting symptoms. The endoscopic improvement was divided into two subcategories: complete or partial endoscopic improvement. Disappearance of all abnormal endoscopic lesions seen on prior endoscopy in the follow-up endoscopy was the definition of complete endoscopic improvement, while the amelioration of the endoscopic findings in comparison with the previous study, e.g., smaller ulcer size or number and resolution of mucosal bleeding, but not completely resolute was characterized as partial improvement. Lastly, the histological improvement was defined by an absence of CMV staining on the follow-up tissue specimen.