Researchers from Johns Hopkins Kimmel Cancer Center have developed a new computational and biomolecular tool, dubbed CompCyst, that can identify precancerous pancreatic cysts. The technology provides a new method for reliably identifying cancer-causing cysts from those that are not cancer-causing.
The teams work demonstrates that in over half of patients who undergo cyst removal, the procedures are unnecessary because the cysts are unlikely to cause cancer. This results in additional medical expense and risks to patients without improving the rate of elimination of cancerous cysts.
Pancreatic cysts are commonplace, occurring in 800,000 Americans
every year. Only a small fraction of those cysts will progress to cancer, but it
is difficult to identify if a given cyst will be cancerous, due to limitations
of current clinical and imaging tests. This means nearly all people diagnosed
with a cyst are followed long-term and many may undergo surgical cyst removal,
causing additional expense and burden on the healthcare system and increasing
the risk of patient mortality. To address this issue, the researchers developed
CompCyst is a classification scheme, based on Boolean Set Logic, that utilizes information from molecular tests and imaging data to identify whether a pancreatic cyst may lead to cancer. The data produced by the system was compared to histopathology, the gold-standard to identify pancreatic cysts and an invasive method that is not regularly used in clinical practice.
In this study, the researchers evaluated molecular information
from over 800 different pancreatic cysts, along with clinical and imaging data
into an algorithm known as MOCA: Multivariate Organization of Combinatorial
Alterations. They demonstrated that CompCyst performed better than physicians
in classifying whether cysts were cancerous.
“We think CompCyst has the capacity to substantially reduce unnecessary surgeries for pancreatic cysts. Over the next five years, we hope to use CompCyst in many more patients with cysts in an effort to guide surgical treatment — to determine when surgery is needed and when it is not needed — and evaluate how well the test performs,” says Bert Vogelstein, Clayton Professor of Oncology, co-director of the Ludwig Center at the Johns Hopkins Kimmel Cancer Center and a Howard Hughes Medical Institute investigator. The work demonstrates that less invasive tests, along with an improved classification scheme, can have substantial predictive power for use in clinical practice.
The study in Science Translational Medicine: A multimodality test to guide the management of patients with a pancreatic cyst
Image: CT of pancreatic cyst. Credit: Johns Hopkins Kimmel Cancer Center