SIMULATION OF SECONDARY COLORECTAL CANCER UNDER COLONOSCOPY SCREENINGS

Purpose: Patients who just underwent colorectal cancer (CRC) treatment are recommended for colonoscopy screening to prevent secondary CRC (S-CRC). To the best of our knowledge, there is no study that considers the progression of S-CRC and helps clinicians in their S-CRC colonoscopy schedule selection. We develop a simulation model that mimics the progression of S-CRC under an exogenous colonoscopy screening schedule and we estimate unknown parameters about S-CRC using simulation optimization and clinical data from Mayo Clinic. 

Method: Our simulation model mimics the progression of S-CRC for a patient for five years after their initial CRC treatment. Patients are categorized according to age and gender. The inputs of the simulation model are annual polyp appearance probabilities, annual transition probabilities from polyp to S-CRC and S-CRC to distant S-CRC, annual mortality rates from S-CRC and distant S-CRC, and 1-year mortality rate after S-CRC treatment. We use records of 366 cancer patients treated at Mayo Clinic for whom we have follow up data to generate annual polyp appearance probabilities for five years, 5-year S-CRC occurrence and mortality risks. We use these annual polyp appearance probabilities in our model and calibrate the rest of the inputs so that the actual 5-year S-CRC and mortality risks are within the 95% confidence intervals of our simulated results. Our model uses a colonoscopy screening schedule based on guidelines of the American Gastroenterology Association (AGA).

Results: 5-year S-CRC from clinical data for age groups 50-59, 60-69 and 70-79 are estimated to be 0.104, 0.077, 0.113 for females, and 0.091, 0.087, 0.15 for males, respectively. The corresponding 5-year S-CRC mortality risks are 0, 0.039, 0.086 and 0, 0.029, 0.045. Final values of the calibrated inputs are in the table below.

                              

Conclusion: 5-year S-CRC occurrence and mortality risks do not differ substantially except in the 70-79 age-group in which females' mortality risk is twice that of males. Consequently, calibrated inputs in the table above also do not have a gender specific difference except the post S-CRC treatment annual mortality rate which is almost 1.5 times higher in females than males in the 70-79 age-group. Our results also suggest that the progression of S-CRC on post-treatment patients may be different than the progression of primary CRC on regular or high risk patients.