Dear Sir,
We thank Dr. Hassan and colleagues for their interest and their careful reading of our article. In their letter to the editor, Dr. Hassan et al. express their concern with our result that optical colonoscopy (OC) every 10 years was more clinically effective than CT colonography (CTC) every 5 years with a 6 mm cut-off for post-CTC polypectomy. Before responding to their suggestions for possible explanations, we would first like to stress that although 10-yearly OC was indeed more effective, the difference with 5-yearly CTC was negligible (0.002 life-years, less than 2% of life-years gained).
Based on the only slightly lower test characteristics of CTC compared with OC, Dr. Hassan et al. expected superior effectiveness of 5-yearly CTC compared with that of 10-yearly OC because of superior cumulative sensitivity of 2 independent CTC examinations. Two subsequent CTC examinations indeed have superior sensitivity over 1 OC examination. However, when these 2 CTC examinations are 5 years apart, some of the advanced lesions missed at the first examination will have already progressed to clinical colorectal cancer (CRC) before the repeat CTC, making CTC no longer superior.
Dr. Hassan and colleagues suggest that the superiority of 10-yearly OC in the MISCAN model is due to the relative importance of polyps 5 mm in the MISCAN modeling of the adenoma-carcinoma sequence. Although the MISCAN model does have a relatively short duration of the development from adenoma to cancer in comparison with other models, we do not believe that this explains the lack of superiority of 5-yearly CTC. In a comparative cost-effectiveness analysis of CTC screening for the Centers for Medicaid and Medicare Services (CMS), 2 other models (SimCRC and CRC-SPIN) with considerably longer durations of the adenoma-carcinoma sequence also found that 10-yearly OC was more clinically effective than 5-yearly CTC. 1 These models are comparable in their duration assumptions to the model of Dr. Hassan, so there has to be a different explanation for the discrepancy in result between the MISCAN model and Dr. Hassan's model.
Dr. Hassan et al. also suggest that the MISCAN modeling results are not consistent with a very low rate of advanced features detected in diminutive lesions (5 mm) in clinical studies (0.09%). However, the MISCAN results are consistent with these observed results. The MISCAN model incorporates the growth of adenomas, whereas these prevalence studies do not. Growth of adenomas increases their malignant potential, so not accounting for the growth will underestimate the true malignant potential of small adenomas over time. Furthermore, the