Abstract
BACKGROUND: Although cytomegalovirus (CMV) reactivation is implicated in ulcerative colitis (UC) exacerbation, the efficacy of antiviral treatment in hospitalized patients with UC with suspected CMV infection has not been thoroughly investigated. We aimed to investigate the diagnostic accuracies of typical endoscopic findings for CMV reactivation in hospitalized UC patients. METHODS: A total of 143 hospitalized cases due to the exacerbation of UC were retrospectively collected. Sensitivity, specificity and diagnostic accuracy of endoscopic findings (punched-out ulcer, round ulcer, girdle ulcer, longitudinal ulcer, and ulcer with wide mucosal defect)for prediction of CMV colitis (histological CMV positivity) and CMV viremia (antigenemia, serum DNA) was assessed. Endoscopic characteristics were compared between patients who initially received anti-viral treatment and who did not receive. RESULTS: The diagnostic performance of endoscopic findings for histologically confirmed CMV infection varied by lesion type. Punched-out ulcers demonstrated a sensitivity of 66.7%, specificity of 58.3%, and overall diagnostic accuracy of 59.1%. Longitudinal ulcers showed a sensitivity of 58.3%, specificity of 40.0%, and diagnostic accuracy of 41.7%. In contrast, wide mucosal defects exhibited lower sensitivity (16.6%) but higher specificity (75.0%), with an overall diagnostic accuracy of 69.7%. The same tendencies were found as diagnostic accuracies of endoscopic findings of punched-out ulcers, longitudinal ulcers, and wide mucosal defects for CMV antigenemia positivity was 55.6%, 40.0%, 60.4%, respectively. The specificity of girdle ulcers for positivity of serum DNA test was relatively high (84.8%) while sensitivity was 9.1%. In total, the diagnostic accuracies of endoscopic findings with punched-out, round, girdle, and longitudinal ulcers, and those with wide mucosal defects were 46.9%, 42.9%, 33.7%, 51.0%, and 35.7% for positivity of serum DNA test, respectively. Punched-out ulceration (57% vs.13%; p < 0.001), longitudinal ulceration (70%vs.34%; p < 0.001), and ulceration with wide mucosal defects (31%vs.9%; p < 0.001) were higher with ganciclovir treatment than without. CONCLUSION: Endoscopic findings cannot predict CMV antigenemia or CMV colitis. Therefore, antiviral treatment should not be administered without evidence of CMV reactivation using only endoscopic findings.