Skip to main content
Premium Trial:

Request an Annual Quote

Multi-Analyte Liquid Biopsy Panel May Aid in Pancreatic Cancer Detection, Staging

NEW YORK – Researchers at the University of Pennsylvania have developed a multi-analyte panel to improve detection and staging of pancreatic adenocarcinoma (PDAC) from patient blood samples.

The researchers have now begun enrolling patients in a clinical validation trial with hopes of eventually commercializing a non-invasive assay based on the biomarker panel through a startup called Chip Diagnostics.

Clinicians normally diagnose patients with PDAC by performing abdominal imaging techniques, such as computed tomography and abdominal ultrasound, to identify the tumor's presence. Abnormal growths like as intraductal papillary mucinous neoplasms and mucinous cystic neoplasm are also monitored and collected in patients over time because they can act as high-risk factors for developing PDAC.

In addition, laboratory tests for specific PDAC biomarkers, such as carbohydrate antigen 19-9 (CA19-9), can also be performed to help detect the presence of PDAC in patients.

However, standard methods of collecting tumor samples from a patient with metastatic PDAC, such as ultrasound endoscopy, can be invasive and difficult to perform, said Erica Carpenter, director of the liquid biopsy laboratory at UPenn and one of the panel's developers. In addition, she argued that single-biomarker assays can be unreliable and miss rare mutations linked to PDAC.

"Compared to other solid tissues, you're flying blind more of the time because you can't perform molecular and other types of analysis that you'd typically perform with the tissue," Carpenter said. "Our aim was [therefore] to come up with a noninvasive way of detecting the disease before sending a patient to curative surgery."

In a proof-of-concept study published on Thursday in Clincal Cancer Research, Carpenter and her colleagues collected 3 ml of whole blood from 204 therapy-naïve subjects (71 healthy, 44 non-PDAC pancreatic disease, and 89 PDAC patients) to analyze plasma for potential biomarkers, including tumor-associated extracellular vesicle (EV) miRNA and mRNA, circulating cell-free DNA, circulating cell-free DNA KRAS mutations, and CA19-9.

"There are a number of medical conditions, such as pancreatitis and diabetes, that are associated with a high risk of PDAC," Carpenter, a senior co-author on the study, explained. "For that reason, we included some of those types in our control group."

David Issadore, co-senior author and bioengineering associate professor at UPenn, explained the team used a method of integrating exosome track-etched magnetic nanopores (TENPO) to isolate EVs in plasma samples from 29 PDAC patients. Previously developed by Issadore's lab, the TENPO device isolates specific subtypes of exosomes by immunomagnetically labeling protein surface markers and capturing these targeted exosomes directly from unprocessed plasma. After lysing the labeled EVs on chip, the team then extracted the resulting RNA, with the EV miRNA and mRNA processed for downstream analysis.

The researchers then used next-generation sequencing (NGS) to identify miRNA that are potentially expressed differently among the patient cohorts. They used Qiagen's QIAseq miRNA library kit to prepare targeted sequencing libraries from isolated EV miRNA and sequenced them using Illumina's HiSeq 2500 instrument.

To identify the most informative EV miRNA biomarkers for PDAC detection, Carpenter and her colleagues then ran a LASSO algorithm on the EV miRNA sequencing results. Validating the candidates with qPCR, the team found five miRNA biomarkers that corresponded well with the sequencing data. The group also included six EV mRNAs that had previously been used to distinguish stage IV PDAC patients from healthy controls, creating a panel of 11 potential EV miRNA biomarkers.

Isolating ccfDNA from the plasma samples using Qiagen's QIAamp Circulating Nucleic Acid kit, the researchers then quantified the concentration of extracted ccfDNA then used a droplet digital PCR platform from RainDance Technology (now part of Bio-Rad Technologies) to detect KRAS mutations in the samples.

Afterward, the team used a clinical research pancreatic cancer sample to validate an electrochemiluminescence immunoassay (ECLIA) and measure CA19-9 as a PDAC biomarker for the larger panel.  

Carpenter and her colleagues therefore developed an initial classification model using the 14-biomarker candidates, including the miRNA/MRNA biomarkers, KRAS mutation, ccfDNA concentration, and CA19-9. The group first trained the machine learning model with a training set of 15 healthy controls, 12 disease controls, and 20 patients with PDAC of various stages.

Carpenter's group then aimed to further optimize this biomarker subset to identify patients with PDAC versus non-cancerous patients. Applying a LASSO algorithm to the training set, the team selected a panel of five biomarkers that had an AUC of .93, which included EV-CK18 mRNA, EV-CD63 mRNA, EV-miR.409, ccfDNA concentration, and CA19-9.

To demonstrate the five-biomarker panel's ability to detect PDAC, Carpenter's team applied it to an independent test set of 136 subjects, achieving a sensitivity of 88 percent and specificity of 95 percent.

Believing that the diagnostic model could be trained to develop a panel that — combined with imaging — would better stage PDAC patients by distinguishing metastatic from non-metastatic disease, the researchers then applied the model on a cohort of 20 PDAC patients to retrain it for disease staging. The cohort included nine resectable patients with no detectable metastasis and 11 patients with metastasis.

In this training group, four biomarkers, including EV-miR.1299, EV-GAPDH, circulating mutant KRAS allele fraction, and CA19-9, proved to have the highest accuracy (91 percent).

To further examine the panel's ability to detect occult metastatic disease, the researchers applied the approach to an independent set of 37 subjects with PDAC as part of a clinical workflow, starting with diagnostic imaging and followed by liquid biopsy. Twelve of the 37 patients that were identified by imaging alone as having metastases had no further evaluation. The remaining 25 patients that were determined by imaging to be resectable had no metastases. After four months, 16 of the patients did not have metastases, while nine were found to have occult metastases.

The panel correctly identified seven of the nine patients with occult metastatic diseases, and 14 of 16 patients as being metastasis-free. The assay had a sensitivity of 78 percent and a specificity of 88 percent, which the study authors believe compares favorably to the accuracy of imaging alone.

Carpenter and her colleagues therefore believe that using the panel may improve PDAC diagnostic accuracy and identify non-metastatic patients who are best suited for surgery.

"We're [examining] clinical classifications that don't have high-performing conventional biomarkers but that are better than being totally random," Issadore said. "By measuring these different biomarkers that you can algorithmically combine together, we can develop a synthetic biomarker [that can] get the job done."

However, Issadore acknowledged that the study had multiple limitations related to the multiple analytes and sample size. Because the panel uses different analytes, Issadore noted that a clinical assay might eventually require multiple blood draws or collection types.

Issadore also noted his lab initially struggled to develop and manufacture enough microfluidic chips to run all the blood samples in the study. While their magnetic nano-fluidic is not yet commercially available, he believes that it could eventually be used as part of a panel for diagnostic purposes.

Commercial path

The investigators are now enrolling patients at UPenn for a larger trial to validate the panel for PDAC detection and disease staging. While patient enrollment is still in its earliest stages, Carpenter's team is currently determining the total amount of PDAC, non-cancer disease, and control samples needed in total to power the validation study. However, Carpenter envisions collecting "certainly in the hundreds, and more than the 204 samples in the last study."

Carpenter highlighted that her team is also collaborating with thoracic surgery colleagues at UPenn to apply the proof-of-concept study's methodologies to metastatic lung cancer patients receiving immunotherapy, with the ultimate goal of providing clinical decision-making support for a cancer type that she noted is difficult to extract tissue from for diagnostic testing.   

Depending on the results of the validation trial, Issadore said that the researchers may commercialize a a cancer diagnostic assay based on the methodology through Chip Diagnostics, a startup he founded at UPenn in 2018.

"We can't say for sure how it'd play out, but there's a huge opportunity to commercialize some of these assays in the academic setting, as the platform can be applied to pancreatic cancer," Issadore said. "We've already applied the [microfluidic chip] to areas such as liver cancer and traumatic brain damage, and we can see the work we've done in pancreatic cancer as one of the many opportunities we can help."

Carpenter said that for PDAC detection and staging, the assay can produce diagnostic results within two to three days. However, she noted that to use the test for clinical decision-making, it would have to be validated in a CAP-accredited, CLIA-approved lab setting.

If the panel is validated, Carpenter believes that it could be ordered as additional bloodwork in the doctor's office, especially if a patient is at high risk of developing pancreatic cancer and has concerning symptoms.  

"Some of these patients who have high-risk diseases, such as diabetes or pancreatitis, are sometimes unfortunately hospitalized for these conditions," Carpenter explained. "If the doctor is concerned that it would lead to cancer, they could test [the blood sample] from there."

Carpenter also envisions a staging assay based off the multi-biomarker panel being used in the clinical space to help identify whether a patient may qualify for PDAC curative surgery. After diagnosing the patient with PDAC using an imaging-based method, Carpenter believes that the clinician could use the test to determine if the tumor is localized or has already metastasized, nullifying the option of surgery.

"We've [now] shown that, for a group of patients with an almost intractable cancer, we could establish an approach for non-invasive testing that, if validated independently could direct clinical decision-making that is not currently possible," Carpenter added.