Latest News

Pancreatic Cancer Pain Points 

Pancreatic cancer is a very challenging disease. It’s currently the 3rd leading cause of cancer deaths in the US and the only major cancer with 5-year survival under 20%. While major gains have occurred for other cancers—all-cancer 5-year survival increased from 49% in the mid-1970s to 70% from 2015-2021—survival of pancreatic cancer has seemingly plateaued at 13%. Ibsen explained that these tumors are pretty deep and difficult to detect, and by the time “you get these clinical symptoms, the tumor is already pretty advanced … surgery and therapy won’t be as effective.” Patients who develop jaundice or abdominal pain often discover advanced metastatic disease, and the odds are already against them. Ibsen explained that finding early-stage pancreatic cancer usually occurs only because a radiologist notices an irregularity in imaging collected for other reasons. Finding ways to reliably detect this cancer sooner would certainly improve matters. 

Tumor Signature 

Although pancreatic tumors easily escape early notice, there is a silver lining. “Tumors are shedding lots of different kinds of nanoparticles into blood throughout their development,” Ibsen explained. “They do so to try to communicate with the surrounding tissue, trying to suppress the immune system … trying to weaken the cells around them so it's easier to invade, easier to spread,” he said. These nanoparticles include extracellular vesicles (EVs), such as exosomes, and particles carrying DNA. “That gives us an advantage because these tumors are over-secreting these little particles relative to the healthy tissue,” Ibsen said. The difficulty, then, is finding and characterizing them. “Your doctor can't order an EV nanoparticle test, and the main reason for that is because it's very difficult to recover these particles from blood,” he said. “The traditional way of getting access to these nanoparticles, recovering them from blood or plasma, is very labor-intensive and takes a lot of time, and so it doesn’t lend itself very well for being used in the hospital clinic.”  

Dielectrophoresis Capture 

To improve matters, Ibsen and colleagues developed a blood test for quantifying tumor-secreted nanoparticles. The underlying technology, dielectrophoresis (DEP), dates to the 1950s, when Herbert Pohl used it to separate suspended particles. He was also first to use the technique, in 1966, to separate live and dead yeast cells. Dielectrophoresis works by creating a non-uniform electric field, so it’s stronger in some regions than others. The electric field induces a dipole in polarizable particles, and because the field is non-uniform, the two ends of an induced dipole sit in regions of different field strength, such that attractive and repulsive forces do not cancel as they would in a uniform field. Provided the particle is more polarizable than surrounding media, the net force can pull it toward the high field at electrode edges. 

In work performed during a postdoc at the University of California, San Diego with Michael Heller (co-author on the Small publication), Ibsen significantly advanced the technique to enable EV isolation from biological samples. In this setup, the non-uniform electric field is created by a microarray of ~60 µm circular pads. Using alternating current (AC) (unlike the direct current that Pohl originally used), the frequency of oscillation determines the size of molecules that migrate; the AC also cancels out traditional electrophoretic (charge-based) force on charged molecules. The result is the specific collection and concentration of nanoparticles at the periphery of the circular electrodes.  

A Better Non-Invasive Test 

For their pancreatic cancer test, the researchers looked at blinded, banked samples that went on to be diagnosed as pancreatic cancer, pancreatitis, benign cysts, and precancerous lesions. They combined analysis of two different biomarkers, glypican-1 (GPC1) and cell-free DNA (cfDNA). The DEP method, tuned to a specific AC frequency of 14 kHz, captures nanoparticles centered around 100 nm diameter, and holds them with sufficient affinity to perform washing and immunoblotting on the chip. Fluorescent antibodies reveal GPC1 levels, and a fluorescent DNA intercalator shows cfDNA. Using both biomarkers, the test showed sensitivity of 92%, specificity of 83%, and overall area under the curve (AUC) of 0.93, meaning it will rank a cancer sample higher than a non-cancer sample 93% of the time. Impressively, this performance is superior to the gold-standard method for diagnosis, the fine needle aspirate biopsy, which has AUC of about 0.79. The AUC for the DEP test increased to 0.97 when considering only patients over 50. 

The results were quite impressive, then; however, the study group was small—only 36 patient samples, 10 with pancreatic cancer, making the generalizability of results a bit difficult to predict. The test was not perfect; two stage 2 pancreatic ductal adenocarcinoma (PDAC) cases scored benign, whereas two pancreatitis patients scored as cancer. There was also one high-grade pre-cancerous lesion that clustered with cancer cases. The authors did not consider this a false positive, because its likelihood to develop into PDAC means it is treated as cancer and surgically removed. Identifying these high-risk cases is not a major limitation. 

Finally, the patient with the highest overall GPC1 (and intermediate cfDNA) levels did not have pancreatic cancer. “Their biomarker levels, however, were so high that we asked the physicians to do another chart review on this patient later,” Ibsen said. “Several months later, this individual actually got diagnosed with stage two liver cancer.” Again, the authors consider this a feature; finding such high levels of EV-associated GPC1 provided, and would generally provide, a strong case for further patient monitoring and scrutiny.  

Overall accuracy was reported as 89% given these classification caveats. The researchers also performed a leave-one-out analysis, modeling with 35 of the data points and asking if the remaining data point would classify appropriately. Under these conditions, accuracy drops to 75%, demonstrating the need for a larger sample set to clarify assay performance. However, Ibsen said, “even with this small cohort, you can see pretty significant differences between the cancer and non-cancer controls.” 

While the cfDNA level was instrumental in improving diagnostic accuracy, it could give more information. A 2024 paper from Ibsen’s group demonstrated that on-chip PCR can be run with the captured DNA to detect KRAS mutations—the mutations that drive about 90% of pancreatic cancers. One could imagine a future version of this test that combines these results into a more powerful analytic.  

There’s also an interesting indication in the paper about the cfDNA origin. Because sequencing revealed DNA fragments typically 250-800 bp, “the idea is that [in] the necrotic region of the tumor, the cells are falling apart, the chromosomes are coming out, and they break down into little nanoparticles as they go through circulation,” Ibsen said. In contrast, Qiagen kit-collected DNA—shorter with quantized peaks—is consistent with more apoptotic remnants. Further work is necessary to fully characterize the particles carrying DNA. 

Can We Just Screen Everyone? Like Cholesterol? 

It might seem like the best approach for detecting pancreatic cancer sooner is to screen everyone, especially if it’s just a blood test, in this case one requiring a volume (30 µl plasma) on the order of a finger prick (Ibsen says they are looking into this version). However, “that’s really far out in the future,” Ibsen said. “Even if the test is really, really specific and really, really high levels of specificity, sensitivity, there’s going to be enough false positives that it will cause too much trouble.” Given the relative rarity of pancreatic cancer (~20 cases per 100,000 Americans), a 99% specificity test gives a 98% false alarm rate: about 1,000 false positives to catch the 20 true cases, and that’s assuming 100% sensitivity. That’s pretty unacceptable, given that you’d send these individuals for further invasive and potentially dangerous testing. 

The situation improves as you select for a population with higher prevalence. “The idea is to have a blood test that you could apply to people who are at high risk, like people who have other family members who have had pancreatic cancer or all of a sudden just develop diabetes out of nowhere,” Ibsen explained. It would also be helpful when someone shows up at the hospital with unexplained symptoms, as the patients involved in this study did—Ibsen expressed his gratitude to the Brenden-Colson Center for Pancreatic Care at OHSU for their collaboration. As an example, if a cyst is found, this test “can help determine whether this is a benign cyst or if it’s something that you really need to cut out right away,” Ibsen said. Adding one of these blood tests to a physician’s diagnostic toolkit could help identify who should and who should not be subjected to biopsy. 

Ultimately, this study was only designed to differentiate pancreatic cancer from other pancreatic conditions; no healthy controls were included. The test is not designed to provide an early warning the way creeping LDL might portend arterial disease. Still, the test “does a really good job of detecting pancreatic cancer and differentiating it from other kinds of non-cancerous pancreatic diseases like pancreatitis,” Ibsen said. 

This test also does not show clear correlation with disease stage, unlike CA19-9, which is FDA-approved for monitoring progression and has been explored for earlier detection in combination with another glycan. Ibsen stressed that the test “is picking up the early-stage cancers, and we have new biomarkers that should be able to help improve that even more.”