Abstract
INTRODUCTION: Cytomegalovirus (CMV) viral load detected by real-time polymerase chain reaction (PCR) in plasma or stool may facilitate detection of CMV colitis. METHODS: This prospective study enrolled 117 patients with clinically suspected CMV colitis. Patients presenting with gastrointestinal symptoms and having increased risk of CMV infection were eligible. All participants underwent colonoscopy with tissue biopsy. Five patients underwent colonoscopy twice because of clinical recurrence, resulting in a total of 122 colonoscopies. Stool CMV-PCR and plasma CMV-PCR were performed within 7 days before/after colonoscopy. Twenty asymptomatic volunteers also underwent the same protocol. RESULTS: Twenty-seven (23.1%) of 122 colonoscopies yielded positive for CMV colitis. The sensitivity and specificity was 70.4% and 91.6% for stool CMV-PCR and 66.7% and 94.7% for plasma CMV-PCR, respectively. The sensitivity of either positive plasma or positive stool CMV-PCR was 81.5%, which is significantly higher than that of plasma CMV-PCR alone ( P = 0.045). However, positive results from both tests yielded a specificity of 95.8%, which is significantly higher than that of stool CMV-PCR alone ( P = 0.045). There was a good and significant correlation between stool CMV-PCR and plasma CMV-PCR ( r = 0.71, P < 0.01), and both tests significantly correlated with the cytomegalic cell count ( r = 0.62, P < 0.01 for stool and r = 0.64, P < 0.01 for plasma). There were no positive stool or plasma CMV-PCR assays among volunteers. DISCUSSION: The results of this study strongly suggest that the combination of stool CMV-PCR and plasma CMV-PCR can be used to confidently rule in (both positive) or rule out (both negative) a diagnosis of CMV colitis.