A four-marker blood panel that includes two newly identified proteins detected stage I and II pancreatic cancer with 87.5 percent sensitivity at a 5 percent false-positive rate, according to findings released by the National Cancer Institute. The panel, developed under lead investigator Kenneth Zaret, combines the proteins ANPEP and PIGR with two previously studied markers, CA19-9 and THBS2, and achieved 91.9 percent discrimination across disease stages. The results arrive at a moment when roughly 95 percent of pancreatic cancers that begin in exocrine cells are still diagnosed too late for surgery to help.
Why two new markers change the calculus for pancreatic screening
Pancreatic cancer kills most patients because it is almost always found after it has spread. CA19-9, the only blood marker widely used in clinical practice, has well-documented shortcomings as a screening tool. It misses many early tumors and produces false positives in patients with benign bile-duct conditions. Earlier research established THBS2 as a complementary plasma marker and showed that pairing it with CA19-9 improved detection at the time of diagnosis through multi-phase validation cohorts. But a later study using prediagnostic samples from the PLCO cohort found that the THBS2 and CA19-9 combination dropped to AUCs near 0.55 overall, performing better only in blood drawn close to the eventual diagnosis date. That gap between performance at diagnosis and performance months or years earlier has been the central barrier for any blood-based pancreatic cancer screen.
The addition of ANPEP and PIGR to the panel represents an attempt to close that gap. By expanding the marker set, Zaret’s team reported substantially higher sensitivity for early-stage disease than CA19-9 alone has ever achieved in comparable studies. If those numbers hold in prospective testing, the panel could shift the detection window earlier, when tumors are still small enough to remove surgically. That shift is what separates a useful screening tool from a laboratory curiosity.
Whether adding serial blood testing with this new panel to existing high-risk surveillance protocols could increase the proportion of stage I and II detections by 20 percent or more within three years, compared with CA19-9 alone, is a testable question. One active clinical trial registered as NCT06388967 is already evaluating an exosome-based early detection approach for pancreatic cancer, and its endpoints could serve as a framework for measuring exactly that kind of stage-shift improvement. But no published trial data yet confirm such a gain for the four-marker panel specifically.
What the Zaret panel measured and how it performed
The study, backed by the National Cancer Institute, tested blood samples from patients with confirmed pancreatic ductal adenocarcinoma against samples from healthy controls. The four-marker panel achieved 91.9 percent discrimination at a 5 percent false-positive rate across all stages of the disease. For the subset of patients with stage I or II tumors, sensitivity reached 87.5 percent. Those figures represent a significant step beyond what CA19-9 delivers on its own, where sensitivity for early-stage disease has historically been far lower.
The foundational work on THBS2 followed a structured path through Phase 1 discovery and Phase 2a/2b validation cohorts, with performance quantified using ROC curves and c-statistics across multiple patient groups. That earlier research demonstrated the value of combining THBS2 with CA19-9 but also exposed the limits of a two-marker approach, particularly in prediagnostic settings. The PLCO cohort analysis showed that markers performing well when cancer is already present can fall short when the goal is to catch the disease before symptoms appear. The new panel’s reported sensitivity for early-stage tumors at diagnosis is encouraging, but the critical test will be whether it maintains that performance in prediagnostic blood samples collected months or years before clinical detection.
The NCI summary emphasizes that the panel was evaluated using banked plasma from patients whose cancers had already been identified clinically. That design allows precise matching between blood draws and tumor stage but does not replicate the conditions of population screening, where incidence is low and most individuals tested will never develop pancreatic cancer. In such settings, even modest drops in specificity can generate large numbers of false positives, leading to unnecessary imaging, invasive procedures, and anxiety.
Another methodological consideration is spectrum bias. Patients whose samples enter research biobanks often have more advanced or symptomatic disease than those who might be found through true screening. If the training and validation sets over-represent larger or biologically aggressive tumors, a panel may appear more accurate than it would in a community setting where many cancers are indolent or very small. The NCI release does not yet detail tumor sizes, resectability rates, or the proportion of incidentally detected lesions in the study cohort, all of which could influence measured performance.
Gaps between diagnostic accuracy and real-world screening value
Several important questions remain open. The published results describe performance at the time of diagnosis, not in a true screening population where the disease is rare and blood is drawn long before any symptoms emerge. The PLCO experience with THBS2 and CA19-9 showed how dramatically performance can erode in that prediagnostic window. No publicly available data yet confirm whether ANPEP and PIGR maintain their discriminatory power under those tougher conditions.
Exact patient cohort sizes, demographic breakdowns, and prospective follow-up outcomes for the four-marker panel have not been detailed beyond the NCI press release summary. Without those specifics, independent researchers cannot fully evaluate how generalizable the 87.5 percent sensitivity figure is across different populations, age groups, or genetic risk profiles.
For people at elevated risk of pancreatic cancer, including those with strong family histories or inherited genetic mutations, the practical question is straightforward: when might a blood test like this realistically change their care? Current high-risk surveillance programs typically rely on annual or semiannual imaging, often alternating MRI and endoscopic ultrasound, in specialized centers. Adding a four-marker blood panel to that regimen could, in theory, help triage which patients need more urgent imaging or closer follow-up between scheduled scans.
However, the balance of benefit and harm will depend on how the test performs when used repeatedly over years. A panel that looks promising at a single diagnostic time point might generate fluctuating results in longitudinal use, with small biologic variations triggering false alarms. To be clinically useful, thresholds may need to incorporate not just absolute marker levels but also trends over time, age-adjusted baselines, and individual risk factors such as germline mutations or chronic pancreatitis.
Health systems will also have to consider cost and access. Multiplex assays that measure four distinct proteins are inherently more complex than a single-analyte test like CA19-9. If the panel ultimately requires specialized platforms or centralized laboratories, implementation could be limited to large academic centers, at least initially. That would constrain its impact on broader population outcomes, even if performance metrics remain strong in trials.
Regulatory pathways add another layer of uncertainty. To move from research use to routine screening, the panel would need robust evidence from prospective studies demonstrating not only earlier detection but also improved survival or reduced mortality. Surrogate endpoints such as stage shift or resection rates are helpful, but regulators and guideline bodies typically look for harder outcomes before recommending widespread screening, especially for a relatively rare cancer with high morbidity from diagnostic procedures.
Ethical questions will surface as well. If a blood test flags a high probability of early pancreatic cancer but imaging cannot localize a lesion, clinicians may face difficult decisions about how aggressively to pursue further workup. Patients might undergo repeated endoscopic ultrasounds or even exploratory surgery based on biomarker signals alone. Clear protocols will be needed to govern how positive results are confirmed, how often testing is repeated, and when to stop surveillance in the absence of radiologic findings.
Despite these caveats, the Zaret panel underscores a broader shift in pancreatic cancer research toward multi-marker, multi-modal strategies. Instead of seeking a single perfect biomarker, investigators are increasingly combining proteins, exosomal signatures, and imaging features into composite risk scores. The four-marker panel fits squarely within this trend, providing a proof of concept that adding carefully chosen proteins can meaningfully boost sensitivity for early-stage disease at acceptable false-positive rates.
For now, the most defensible interpretation is that the panel marks an important advance in diagnostic accuracy at the time of clinical presentation, while its value as a true screening tool remains unproven. The next phase will require large, carefully designed prospective cohorts that mirror real-world populations, with standardized blood collection prior to symptom onset and long-term follow-up. Only then will researchers know whether the impressive numbers reported so far translate into more patients diagnosed at operable stages-and ultimately, more lives saved.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
