Can a Blood Test Catch Pancreatic Cancer Before It Spreads?
At 95% specificity, a four-protein plasma panel identified 91.9% of pancreatic ductal adenocarcinoma cases across stages I-IV and 87.5% of stage I/II cases. The work came from the University of Pennsylvania and Mayo Clinic, and it was a retrospective phase 2 study: stored plasma from people already diagnosed, tested against healthy and nonmalignant disease controls.
Pancreatic ductal adenocarcinoma, which accounts for 92% of all pancreatic cancers, is usually found too late for treatment to help. More than 80% of patients are already ineligible for surgical resection at diagnosis because the tumor has grown into surrounding blood vessels or spread. A blood test that flags the disease earlier is the point of this line of research.
What Makes This Blood Test Different
The team started with two markers they had prior data on: CA19-9 and THBS2. CA19-9 is widely used to track treatment response in patients who already have a pancreatic cancer diagnosis, but it falls short as a standalone screening test. It rises in benign conditions such as pancreatitis and bile duct obstruction, it has low sensitivity and specificity for early-stage disease, and a patient’s genetics affect the levels. In earlier work this group showed that elevated THBS2 could cover pancreatic cancer patients whose CA19-9 was not elevated.
Using mass spectrometry and ELISA on plasma pools from both institutions, the researchers identified two additional proteins raised in stage I/II pancreatic cancer plasma compared with controls: aminopeptidase N (ANPEP) and polymeric immunoglobulin receptor (PIGR). Combining all four changed the performance of the panel.
What the Data Show
The panel was tested in two separate cohorts, 135 plasma samples from Penn and 537 from Mayo, covering pancreatic cancer at different stages against healthy controls and against controls with nonmalignant disease.
Each new marker held up on its own. Comparing healthy controls with stage I/II cancer, ANPEP gave an AUC of 0.78 (0.68 to 0.86) at Penn and 0.80 (0.74 to 0.85) at Mayo. PIGR gave 0.81 (0.70 to 0.88) and 0.86 (0.82 to 0.90).
In multivariable models, the combinations CA19-9/THBS2/ANPEP, CA19-9/THBS2/PIGR, and all four together produced AUCs of 0.94 to 0.96 at Penn and 0.97 at Mayo. Against the harder comparison, cancer versus nonmalignant disease controls in the Mayo cohort, the four-marker panel reached an AUC of 0.87 for stage I/II and 0.91 for stages I-IV.
Using thresholds of CA19-9 at or above 35 U/mL, THBS2 at or above 42 ng/mL, ANPEP at or above 2995 ng/mL, and PIGR at or above 1800 ng/mL, the panel reached a sensitivity of 91.9% for stages I-IV and 87.5% for stage I/II at 95% specificity.
Dr. Kumar’s Take
This is one of the more convincing early-detection panels I have read. Overall five-year relative survival for this cancer is under 13%, and it rises to 44% when the disease is caught while still localized, which happens in only about 14% of cases. Stage I disease carries a survival rate above 80% in the SEER database when it is found and treated. That gap between 13% and 44% is the whole argument for a test like this.
The part that makes me take the numbers seriously is that two independent cohorts at two institutions produced concordant results, with the added markers coming through a discovery step in one set of plasma pools and holding up in separate cohorts. The authors’ own conclusion is the right one: the next step is testing this panel in pre-diagnostic samples, meaning blood drawn before anyone knew cancer was there. Retrospective performance in known cases sets an upper bound on what a real screening test will do, not a floor.
Who Could Benefit Most
Because pancreatic cancer has low incidence in the general population, routine screening is only indicated for patients at high risk for the disease. Those are the people a panel like this would be aimed at first. If a test can find stage I/II disease while resection is still possible, it addresses the specific problem that leaves more than 80% of patients ineligible for surgery at diagnosis.
Safety, Limits, and Caveats
This was a retrospective phase 2 study. The researchers ran the assays on stored plasma from patients whose diagnosis was already known, comparing them with healthy controls and with controls who had nonmalignant disease. That design cannot tell you how the panel behaves in an undiagnosed population, where the disease is far rarer and where a 95% specificity produces many more false alarms per true case. The authors state that assessments in pre-diagnostic cases are warranted, and until those are done this is a promising panel rather than a screening tool.
Practical Takeaways
- If you have known risk factors for pancreatic cancer, ask your doctor whether you fall into the high-risk category for which screening is indicated, since routine screening is not recommended for the general population.
- CA19-9 is used clinically to monitor treatment response in people who already have a pancreatic cancer diagnosis. It is not a reliable standalone screening test, because it also rises in benign conditions such as pancreatitis and bile duct obstruction and performs poorly in early-stage disease. In this research it works as an anchor inside a multi-marker panel, not on its own.
- Genetics affect CA19-9 levels. A FUT2 and FUT3 genotype test can identify people who cannot express or secrete the Lewis blood group antigens, which allows a personalized reference range when CA19-9 is interpreted.
- This four-marker panel is not available for clinical use. It is a step toward earlier detection, not something to request at your next visit.
Related Studies and Research
- Blood tests for Alzheimer’s: a new study shows 83% accuracy explores another promising blood-based biomarker panel for disease detection.
- Can one simple blood test predict your heart and metabolic health? looks at how blood markers can reveal hidden health risks.
- Sleep-wake cycle controls tau protein clearance in brain and spinal fluid examines how biological processes affect protein levels in the body.
- One week of sleep restriction reduces insulin sensitivity in healthy men shows how metabolic markers shift with lifestyle changes.
FAQs
Is this pancreatic cancer blood test available to patients right now?
No. The four-marker panel is a research assay. This was a retrospective phase 2 study run on stored plasma from patients with an established diagnosis, and the authors conclude that assessment in pre-diagnostic cases is the necessary next step. One component, CA19-9, is already in clinical use, but for monitoring treatment response in patients who have already been diagnosed.
How is this different from existing pancreatic cancer screening methods?
Given the low incidence of pancreatic cancer in the general population, routine screening is only indicated for people at high risk. CA19-9 on its own is not adequate for screening: it is raised in benign conditions such as pancreatitis and bile duct obstruction, it has low sensitivity and specificity for early-stage disease, and patient genetics shift the levels. Adding THBS2, ANPEP, and PIGR raised performance to an AUC of 0.94 to 0.96 at Penn and 0.97 at Mayo against healthy controls, and 0.87 for stage I/II against nonmalignant disease controls.
Who should be most concerned about pancreatic cancer risk?
Screening is currently reserved for people at high risk, so the practical step is asking your doctor whether you meet that definition. The reason it matters: overall five-year relative survival is under 13%, but it reaches 44% when the cancer is diagnosed at an early localized stage, and that happens in only about 14% of cases. Stage I disease carries a survival rate above 80% in the SEER database when accurately diagnosed and treated.
Bottom Line
Adding ANPEP and PIGR to a plasma panel of CA19-9 and THBS2 improved detection of early-stage pancreatic ductal adenocarcinoma. At 95% specificity, the four-marker panel reached a sensitivity of 91.9% for stages I-IV and 87.5% for stage I/II, with concordant results across 135 samples at Penn and 537 at Mayo. It remains a retrospective finding, and testing in pre-diagnostic samples is what comes next.
