, based on a microsimulation model.
To seize this opportunity, however, a greater understanding of natural disease course is needed, along with more sensitive screening tools, reported Brechtje D. M. Koopmann, MD, of Erasmus Medical Center, Rotterdam, the Netherlands, and colleagues.
Previous studies have suggested that the window of opportunity for pancreatic cancer screening may span decades, with estimates ranging from 12 to 50 years, the investigators wrote. Their report was published in Gastroenterology.
“Unfortunately, the poor results of pancreatic cancer screening do not align with this assumption, leaving unanswered whether this large window of opportunity truly exists,” they noted. “Microsimulation modeling, combined with available, if limited data, can provide new information on the natural disease course.”
For the present study, the investigators used the Microsimulation Screening Analysis (MISCAN) model, which has guided development of screening programs around the world for cervical, breast, and colorectal cancer. The model incorporates natural disease course, screening, and demographic data, then uses observable inputs such as precursor lesion prevalence and cancer incidence to estimate unobservable outcomes like stage durations and precursor lesion onset.
Dr. Koopmann and colleagues programmed this model with Dutch pancreatic cancer incidence data and findings from Japanese autopsy cases without pancreatic cancer.
First, the model offered insights into precursor lesion prevalence.
The estimated prevalence of any cystic lesion in the pancreas was 6.1% for individuals 50 years of age and 29.6% for those 80 years of age. Solid precursor lesions (PanINs) were estimated to be mainly multifocal (three or more lesions) in individuals older than 80 years. By this age, almost 12% had at least two PanINs. For those lesions that eventually became cancerous, the mean time since cyst onset was estimated to be 8.8 years, and mean time since PanIN onset was 9.0 years.
However, less than 10% of cystic and PanIN lesions progress to become cancers. PanIN lesions are not visible on imaging, and therefore current screening focuses on finding cystic precursor lesions, although these represent only about 10% of pancreatic cancers.
“Given the low pancreatic cancer progression risk of cysts, evaluation of the efficiency of current surveillance guidelines is necessary,” the investigators noted.
Screening should instead focus on identifying high-grade dysplastic lesions, they suggested. While these lesions may have a very low estimated prevalence, at just 0.3% among individuals 90 years of age, they present the greatest risk of pancreatic cancer.
For precursor cysts exhibiting HGD that progressed to pancreatic cancer, the mean interval between dysplasia and cancer was just 4 years. Among 13.7% of individuals, the interval was less than 1 year, suggesting an even shorter window of opportunity for screening.
Beyond this brief timeframe, low test sensitivity explains why screening efforts to date have fallen short, the investigators wrote.
Better tests are “urgently needed,” they added, while acknowledging the challenges inherent to this endeavor. Previous research has shown that precursor lesions in the pancreas are often less than 5 mm in diameter, making them extremely challenging to detect. An effective tool would need to identify solid precursor lesions (PanINs), and also need to simultaneously determine grade of dysplasia.
“Biomarkers could be the future in this matter,” the investigators suggested.
Dr. Koopmann and colleagues concluded by noting that more research is needed to characterize the pathophysiology of pancreatic cancer. On their part, “the current model will be validated, adjusted, and improved whenever new data from autopsy or prospective surveillance studies become available.”
The study was funded in part by Maag Lever Darm Stichting. The investigators disclosed no conflicts of interest.