MISCAN-gastric Erasmus Medical Center (EMC)

To simulate non-cardia gastric cancer (GC) incidence and mortality based on SEER data. By modeling the continuous progression of individuals through precancerous and cancerous states, MISCAN-gastric evaluates the (cost-)effectiveness of GC screening and prevention strategies such as H. pylori eradication and endoscopic surveillance. The model accounts for age, sex, and race/ethnicity-specific risk factors, including H. pylori infection status, to provide insights into early detection, cancer prevention, and public health interventions. Its adaptability allows for policy-relevant analyses both in the U.S. and globally to help mitigate the burden of gastric cancer.

Contacts:
Iris Lansdorp-Vogelaar i.vogelaar@erasmusmc.nl
Duco T. Mülder d.t.mulder@erasmusmc.nl

Summary

The MISCAN-gastric model is a stochastic, semi-Markov, microsimulation model programmed in Python 3.10. It simulates continuous-time transitions between discrete health states, with transition probabilities and durations that vary stochastically across individuals. These onset and durations of precursor states are further influenced by age, sex, race/ethnicity, birth cohort, and key risk factors such as H. pylori infection. Model parameters are calibrated using a genetic algorithm to optimize fit to observed data.

The MISCAN-gastric model comprises three main modules:

Demography module – Each individual's date of birth and death is determined using life tables. Competing mortality is observed rather than generated stochastically.

Natural history module – Individuals enter the model at birth as part of the general population. The model follows Correa’s cascade, simulating progression from chronic gastritis to atrophic gastritis, intestinal metaplasia, low-grade dysplasia (LGD), high-grade dysplasia (HGD), and ultimately, cancer. Cancer progresses through SEER-defined stages (Localized, Regional, and Distant), and individuals may die from cancer or competing causes at any stage (Figure 1).

Screening and surveillance evaluation module – This module overlays screening and surveillance interventions onto the natural history framework. The model evaluates the cost-effectiveness of different GC screening and prevention strategies, including H. pylori eradication and endoscopic surveillance. Outcomes include cancer cases prevented, deaths averted, quality-adjusted life years (QALYs) gained, and potential harms such as overdiagnosis and adverse effects.

MISCAN-gastric models the duration of each precancerous and cancerous state, using age-specific transition times for intestinal metaplasia (IM) and dysplasia. H. pylori infection increases the risk of developing atrophic gastritis (AG) and accelerates progression through all precursor states, while smoking is not explicitly modeled as a separate risk factor. The model does not allow for regression, meaning once a health state is reached, progression follows a one-directional pathway.

Calibration is performed using SEER data focusing on non-cardia intestinal-type adenocarcinomas and prevalence data from systematic review evidence developed for the CISNET Gastric models. It accounts for sex-specific cancer incidence, stage distribution, and survival rates among Non-Hispanic Black and White individuals. A single calibrated parameter set is determined through the genetic algorithm, optimizing transition probabilities to reflect real-world epidemiological trends.

MISCAN-gastric's ability to model individual life histories makes it a powerful tool for public health decision-making. By providing robust estimates of screening impact and cost-effectiveness, it is able to inform policies aimed at reducing the burden of gastric cancer, particularly in high-risk populations.

Natural history of the MISCAN-gastric model

Figure 1: The arrows represent transitions between health states. The duration between transitions varies by race, sex and/or H. pylori infection status, depending on the color of the arrow. H. pylori, Helicobacter Pylori.