MISCAN-Colon Erasmus University Medical Center / Memorial Sloan Kettering

Model Overview

The Microsimulation SCreening Analysis (MISCAN) Colorectal Cancer Model (Erasmus University Medical Center/Memorial Sloan Kettering) describes discrete event state transitions in continuous time. State transitions can depend on the age, sex, and race (black or white) of the individual, lesion location within the large intestine, and calendar time. For each simulated individual, MISCAN first generates a time of birth and a time of death from causes other than colorectal cancer (CRC). Next, MISCAN generates adenomas within the individual using a non-homogeneous Poisson process. The individual-level hazard rate ratio, which is drawn from a gamma distribution, may depend on sex and race. Moreover, the Poisson process takes a hazard function as an input through which the risk of adenoma onset increases with age. Once initiated, each adenoma is assigned a location in the large intestine according to the observed location distribution of CRC.

MISCAN simulates three adenoma size categories (small [1-5 mm], medium [6-9 mm], and large [10+ mm]) in seven locations. All adenomas start small and can transition through larger size categories. MISCAN simulates two types of adenomas: progressive and nonprogressive. Nonprogressive adenomas grow in size but never develop into cancer. Progressive adenomas eventually develop into cancer, although other causes of death may occur before the transition to preclinical (or clinical) cancer. The probability that an adenoma is progressive is a function of the age of adenoma onset. Progressive medium and large adenomas can transition to stage I preclinical (asymptomatic) cancer, although the rate of progression is faster for larger adenomas.

Preclinical cancers progress through stages I through IV. Durations within adenoma categories and within preclinical cancer stages are fully correlated through common random numbers, but there is no correlation between the duration of adenoma stages and preclinical cancer stages. At each preclinical cancer stage, MISCAN simulates whether there is a transition to symptomatic and clinically detected cancer as a function of age, preclinical cancer location, and optionally other factors such as sex and race. After clinical detection, MISCAN simulates the survival time to death from CRC using relative survival estimates from the Surveillance, Epidemiology, and End Results (SEER) program, following an approach that is common across the three models.¹

CRC survival time depends on age and stage at detection, cancer location (colon or rectum), and (optionally) sex and race. For individuals with synchronous CRCs at the time of diagnosis, MISCAN uses the stage-specific survival of the cancer with the highest stage present. For all individuals with CRC, their death date is set to the earliest simulated death date (either due to CRC or other causes).

Risk Factor Modeling

MISCAN can model the impact of risk factors on CRC incidence by changing the age-specific incidence of adenomas. Because risk factors are assumed not to influence the progression rate from adenoma to cancer, higher CRC incidence is an immediate result of the higher adenoma incidence. Therefore, these risk factors only affect the individual’s hazard rate ratio of adenoma onset. Previously modeled risk factors associated with an increased risk of adenomas include obesity, smoking, and red meat consumption. Modeled risk factors associated with a decreased risk of adenomas include fruit and vegetable consumption, physical activity, multivitamin use, aspirin use, and hormone replacement therapy. All risk factors are currently modeled as dichotomous characteristics.

References

1. Rutter CM, Johnson EA, Feuer EJ, Knudsen AB, Kuntz KM, Schrag D. Secular trends in colon and rectal cancer relative survival. J Natl Cancer Inst. 2013;105(23):1806-13.