Reply to the letter by A. S. Ravi Kumar et al. regarding “Detectability of colorectal neoplasia with FDG-PET/CT”

Abstract

We analyzed positive predictive values (PPVs) of fluorodeoxyglucose (FDG)-positron emission tomography/computed tomography (PET/CT) for detecting colorectal neoplasia using our published data [1]. PPVs for any neoplasia ([5 mm) were 46 % in the proximal colon and 70 % in the distal colon. In addition, PPVs for larger neoplasia ([11 mm) were 38 % in the proximal colon and 59 % in the distal colon. Thus, our PPVs were different from those reported by other authors including Lee et al. [2]. Both sensitivity and PPVs were lower in the proximal colon than in the distal colon in our study. PET/CT is occasionally faulted for revealing intense metabolic activity at sites not considered to be related to the malignant process under study. This physiologic uptake in the gastrointestinal tract has been attributed to uptake by smooth muscles or excretion and intraluminal concentration of 18F-FDG [3]. However, the discrepancy in PPVs between Lee et al.’s study and ours could be largely attributed to study design. We interpreted PET/CT images of the patients who underwent both PET/CT and colonoscopy without knowledge of colonoscopy findings, whereas in Lee et al.’s study, colonoscopy was performed to confirm positive findings of PET/CT. Firstly, the anatomical complexity of the distal colon may be an important factor. In Lee’s method, when positive 18F-FDG incorporation was observed in the proximal colon, colonoscopy was performed with the aim of discovering any lesion in the proximal colon. Then because of the anatomical complexity of the distal colon, neoplastic lesions, if present, in the distal colon may easily be missed. On the other hand, colonoscopists in our study were blinded to the results of PET/CT. Therefore, meticulous observation was performed throughout the colorectum because target lesions had not been known prior to colonoscopy. Thus, the aim of colonoscopy may have yielded a difference between the studies. Secondly, the difference in prevalence of neoplastic lesions may also be involved. As is well known, the prevalence of neoplastic lesions is higher in the distal colon than in the proximal colon. It is natural that unbiased interpretation of PET/CT yields a higher false-positive rate at the location where neoplastic lesions are sparser. Lastly, the state of mind of the interpreters of PET/CT images should be noted. In our study, interpreters were likely to overestimate the positivity of PET/CT. A false positive would be better than a false negative, because the latter call would suggest the interpreters were incompetent. In fact, in our study, the SUV max was likely to be lower in the false-positive cases than in the true-positive cases. In contrast, in Lee’s study, interpreters would have been likely to underestimate the positivity of PET/CT. A false negative would be better than a false positive in this This author’s reply refers to the letter to the editor at doi:10.1007s00535-012-0593-0.