In the Journals

Simulation model provides more accurate estimate of cancer progression in Barrett's esophagus

Simulation modeling revealed that the lower estimates of the rate of progression from Barrett’s esophagus to esophageal adenocarcinoma from population-based studies are likely more accurate than higher estimates from prospective studies due to detection bias from endoscopic surveillance.

Because the true risk of progression from BE to cancer is uncertain, researchers used the ERASMUS/UW esophageal adenocarcinoma model — which was previously developed by Erasmus Medical Center University in the Netherlands and the University of Washington — to more accurately estimate cancer incidence in BE by reconciling published data.

First, they adjusted the model to reproduce the design of population-based studies with realistic inaccuracies in surveillance and diagnoses, then calibrated the model to match the annual progression rate for men reported by population-based studies (0.18%). Then they used the model to simulate the population based on three scenarios: one without surveillance, one with a population-based study design in which 38% of patients with BE and no dysplasia and 52% of patients with low-grade dysplasia received initial surveillance, and one with a prospective design in which 100% of patients received initial surveillance. Finally, they predicted progression rates for population-based and prospective study designs and for different follow-up periods and compared them with published data.

The model reproduced the mean annual progression rate from BE and no dysplasia or low-grade dysplasia to esophageal adenocarcinoma from population-based studies with 5 years of follow-up (0.19%). It also predicted the annual progression rate for prospectively designed studies (0.36%), which was 91% higher than the population-based design.

“These estimates were close to the published estimates of 0.41% annual progression rate, with a 128% difference between designs,” the researchers wrote.

The progression rates for each study design each increased from 0.63% to 0.65% annually after 20 years, corresponding to a 9.1% to 9.5% cumulative incidence of esophageal adenocarcinoma (5% difference between designs). The relative differences between the progression rates of both study designs decreased from 91% after 5 years to 5% after 20 years.

“In conclusion, the published differences between the lower progression rates reported from population-based studies and the higher progression rates observed in prospective studies can be explained by detection bias from endoscopic surveillance in the prospective studies,” the researchers wrote. “For a short time frame, the clinical progression rate from BE to [esophageal adenocarcinoma], reflecting the individual risk, is likely to be closer to the estimations in the population-based studies than prospective studies because less surveillance was performed in the former studies. Clinicians informing their BE patients about their cancer risk can best use this clinical progression rate, which is not influenced by surveillance-detected cancers.” – by Adam Leitenberger

Disclosure: The researchers report no relevant financial disclosures.

Simulation modeling revealed that the lower estimates of the rate of progression from Barrett’s esophagus to esophageal adenocarcinoma from population-based studies are likely more accurate than higher estimates from prospective studies due to detection bias from endoscopic surveillance.

Because the true risk of progression from BE to cancer is uncertain, researchers used the ERASMUS/UW esophageal adenocarcinoma model — which was previously developed by Erasmus Medical Center University in the Netherlands and the University of Washington — to more accurately estimate cancer incidence in BE by reconciling published data.

First, they adjusted the model to reproduce the design of population-based studies with realistic inaccuracies in surveillance and diagnoses, then calibrated the model to match the annual progression rate for men reported by population-based studies (0.18%). Then they used the model to simulate the population based on three scenarios: one without surveillance, one with a population-based study design in which 38% of patients with BE and no dysplasia and 52% of patients with low-grade dysplasia received initial surveillance, and one with a prospective design in which 100% of patients received initial surveillance. Finally, they predicted progression rates for population-based and prospective study designs and for different follow-up periods and compared them with published data.

The model reproduced the mean annual progression rate from BE and no dysplasia or low-grade dysplasia to esophageal adenocarcinoma from population-based studies with 5 years of follow-up (0.19%). It also predicted the annual progression rate for prospectively designed studies (0.36%), which was 91% higher than the population-based design.

“These estimates were close to the published estimates of 0.41% annual progression rate, with a 128% difference between designs,” the researchers wrote.

The progression rates for each study design each increased from 0.63% to 0.65% annually after 20 years, corresponding to a 9.1% to 9.5% cumulative incidence of esophageal adenocarcinoma (5% difference between designs). The relative differences between the progression rates of both study designs decreased from 91% after 5 years to 5% after 20 years.

“In conclusion, the published differences between the lower progression rates reported from population-based studies and the higher progression rates observed in prospective studies can be explained by detection bias from endoscopic surveillance in the prospective studies,” the researchers wrote. “For a short time frame, the clinical progression rate from BE to [esophageal adenocarcinoma], reflecting the individual risk, is likely to be closer to the estimations in the population-based studies than prospective studies because less surveillance was performed in the former studies. Clinicians informing their BE patients about their cancer risk can best use this clinical progression rate, which is not influenced by surveillance-detected cancers.” – by Adam Leitenberger

Disclosure: The researchers report no relevant financial disclosures.