Methodology and Application of Effectiveness and Over-Detection Evaluation for Colorectal Cancer Screening

Background: The evaluation for screening program should need long-term follow-up to get the final results regardless randomized controlled trial or population-based screening program. Therefore, projection of reducing advanced stage of colorectal cancer and mortality for a population–based screening for CRC with Fecal Occult Blood Test (FOBT) requires the elucidation of temporal disease natural history and the sensitivity of FOBT. The multi-step natural course on colorectal cancer (CRC) from CRC free, to the pre-clinical screen-detectable phase (PCDP), and to the clinical phase (CP) plays an important role in designing primary prevention and population-based screening program for reducing the proportion of advanced CRC and mortality from CRC. Therefore, the methodological approach will be prerequisite for screening evaluation. According to the tumor regression theory, slow growing cancers staying in the pre-clinical state (non-progressive type) or progressing too slowly to be diagnosed with symptom before the patient died from other causes would not have been detected had screening not been applied. These cases are so-called over-detected cases. From the clinical aspects, the burden of over-treatment is attributed to unnecessary treatments and therapies administered to these over-detected cases. Population-based screening for common cancers has long been debated on over-detection. Mass screening for colorectal cancer with FOBT on this thorny issue has been scarcely highlighted. Population-based randomized controlled trial on fecal occult blood test in Finland provides an opportunity to estimate transition parameters of the three-state with adjustment for sensitivity. Study Aims: To evaluate and project the effectiveness of CRC screening for reducing proportion of advanced CRC and mortality from CRC. We would apply five-state natural history model and markov cycle tree to achieve this task. To quantify the proportion of over-detection of CRC when a population-based screening for CRC with FOBT test was implemented. Both of methods will support and apply to Taiwan Nationwide CRC screening program with FOBT evaluation and over-detection in young population aged 40-49. We also conduct these two methods applied empirical screening data from Keelung Integrated Screening program to evaluate the CRC screening with different screening interval on general, high risk with diabetes, high risk with family history among aged 40-49. Materials and Methods: We used two rounds of screening together with the control group from Finnish population-based randomized control trial for colorectal cancer screening to estimate the three-state and five-state natural history with pre-clinical incidence rate and also sensitivity of localized or non-localized CRC. About the Bayesian approach, the similar analyses were also applied to the two previous randomized controlled trials (English and Demark) and taken as prior for posterior update. A computer simulation was conducted to predict the relative rate of reducing advanced CRC and mortality from CRC by simulating. The screening strategies with different screening interval would be evaluated using these two outcomes. The three-state Markov model with the consideration of the sensitivity of FOB test was constructed and applied to the data to estimate the attributable proportion of over-detection of colorectal cancer resulting from FOB test. Expected outcome: According to our study aims, the three-state and five-state natural history of CRC will be estimated using Finnish population-based randomized controlled trial. The projection of reducing in mortality from CRC and proportion of advanced CRC would be answered by our simulation methods considering the parameters of screening scenario. Using the three-state natural history, we try to estimate the over-detection proportion which comes from CRC screening implementation. We would apply these methods on Taiwan nationwide biennial screening program and Keelung young aged 40-49 population to evaluate the effectiveness and over-detection among aged 40-49.

Project ID:PC10401-0204
External Project ID:NSC102-2314-B182-028-MY3