MISCAN-ESO (Erasmus/UW)

The Microsimulation Screening Analysis of Esophageal Cancer (MISCAN-ESO) model was developed at Erasmus University and the University of Washington. It is a semi-Markov micro simulation model, as described below. The population is simulated individual by individual, and each person can evolve through discrete disease states (Figure 1). However, instead of modeling annual transitions with associated transition probabilities, the MISCAN-ESO model is semi-Markov in that it generates durations (called sojourn times) in distinct health states. We assume that esophageal adenocarcinoma (EAC) develops through precursor Barrett’s Esophagus (BE). For each individual in the simulated population, a personal risk index is generated. A minority of the population has symptomatic gastroesophageal reflux disease (GERD), giving them a higher risk of developing BE during their lifetime. The sequence from the onset of BE to EAC diagnosis is continued by sojourn times between the different states. BE starts in a phase without dysplasia (ND); after that, dysplasia can develop.

Two stages of dysplasia are defined: Low Grade (LGD) and High Grade Dysplasia (HGD). From HGD, malignant cells may arise and transform to preclinical localized EAC, which can sequentially progress into Regional and Distant preclinical EAC. However, there is also a possibility that regression from HGD to LGD and from LGD to ND will occur. In each of the three preclinical cancer stages, there is a probability of the cancer being diagnosed. The survival after clinical diagnosis depends on the cancer stage and the year of diagnosis (Figure 1).

Additional modules for modeling the characteristics of endoscopic ablation were inserted into the model. For endoscopic ablation, the outcome of the initial 2-year endoscopic treatment for each individual patient is randomly drawn when treatment starts. If treatment fails, the patient remains in endoscopic surveillance at an interval in accordance with their pre-ablative dysplastic grade. If treatment succeeds, the patient will experience complete eradication of dysplasia with persistent metaplasia (CE-D) or complete eradication of dysplasia and intestinal metaplasia (CE-IM) after 2 years. In the first case, we assume that the patient is in the BE non-dysplastic (ND) phase, having the same assumptions as our natural history model. In the latter case, the patient stays in the CE-IM state for sojourn time that is randomly selected from an exponential distribution (which assumes a constant rate for transition to the next state). If the patient experiences a transition to the next state (recurrence/progression), he/she will be reclassified immediately into the state of histological recurrence. A new life-history will be generated for this individual following recurrence with the model simulating surveillance according to the inputs after RFA. Surveillance can detect recurrent stages of BE, dysplasia and EAC. If the patient experiences recurrent stages of BE, a new endoscopic ablation sequence will be inserted, and the surveillance and treatment process will be repeated. This will generate a new life-history for the patient and he/she will be considered to have durations and probability of developing EAC that are based on the new life-history. After determining this new life-history, surveillance is inserted in the model according to the post-RFA surveillance intervals dictated by the common input parameters. A maximum of three touch-ups are allowed.