Amy Carpenter Aquino
February 2020—New microfluidic and nanotechnologies could take liquid biopsy to the next level as a tool to gauge cancer progression and treatment response.
Sunitha Nagrath, PhD, associate professor of chemical engineering, College of Engineering, University of Michigan, and co-director of the Single Cell Analysis Resource, University of Michigan Rogel Cancer Center, described, at the 2019 AMP annual meeting, technologies that can strengthen liquid biopsy’s role in complementing tissue biopsies.
The focus of her laboratory’s work is the potential of peripheral whole blood circulating tumor cells to be diagnostic and prognostic. “We are increasingly putting efforts into understanding CTCs at a single-cell level,” she said. Dr. Nagrath is also a researcher in the university’s Biointerfaces Institute.
CTCs are shed by the primary tumor, intravasate into the peripheral circulation, and then extravasate into distal sites, causing metastasis. They are characterized by having undergone the epithelial-to-mesenchymal transition (EMT) process and by their ability to “effortlessly” invade the immune system, “withstanding circulation pressures and immune system attacks,” Dr. Nagrath said.
Use of CTCs in the clinical laboratory has been slow, she said, in part because of the rarity of CTCs in blood. It was once believed there were only one to five CTCs among billions of other cells in 7.5 mL of blood (Jack RM, et al. Adv Sci (Weinh). 2016;3[9]:1600063). However, “with advances in technology and our success in detecting these cells, we now realize a patient could have from 10 to 20 or even 100 CTCs per milliliter of blood.”
Another challenge to isolating CTCs is their biological heterogeneity. “CTCs present in many different states,” Dr. Nagrath said. Most CTCs exist in an EMT state where they have semi-epithelial and mesenchymal markers, such as vimentin and E-cadherin, but some CTCs lose their epithelial markers and present as purely mesenchymal phenotype (Bednarz-Knoll N, et al. Cancer Metastasis Rev. 2012;31[3-4]:673–687).
Microfluidic platforms for isolating circulating leukemia cells
These variances affect the success of the immunoaffinity-based approaches in isolating CTCs, the majority of which rely on targeting known cell surface markers, such as epithelial cell adhesion molecule (EpCAM) glycoprotein (Nagrath S, et al. Gastroenterology. 2016;151[3]:412–426).
Dr. Nagrath’s laboratory designs and develops microfluidic technologies for liquid biopsy using immunoaffinity- and physical-property–based approaches, each of which has its advantages, she said. “The immunoaffinity-based approaches are specific and give high purity of the cells that you want, though the flow rate could be slow, up to 10 mL per hour.” Physical-property–based approaches to liquid biopsy, such as simple filtration and inertial fluidic separation, use cell size as a criterion and offer higher throughput and label-free cell capture.
Dr. Nagrath said her laboratory integrates immunoaffinity- and size-based liquid biopsy approaches to isolate CTCs. She and her team developed a microchip with gold posts coated with the nanomaterial graphene oxide (GO chip), which captures CTCs from whole blood with high sensitivity and purity. The combination of the nanomaterial and the chip’s flat surface helps preserve live cells once they are captured. “We can retrieve the live cells, which are vital and able to grow, which also enables downstream analysis,” she said. The CTCs are characterized based on their expression of cytokeratin—“the most tumor-oriented marker”—while CD45 expression is used to distinguish the contaminating white blood cells.
In one investigation, her laboratory team used the GO chip assay to examine how CTC enumeration and characterization can support the prognosis and treatment selection for metastatic breast cancer patients. They performed liquid biopsies on peripheral whole blood from 47 metastatic breast cancer patient samples, using parallel GO chips to isolate and analyze the CTCs (Kim TH, et al. Adv Biosys. 2019;3[2]:1800278).
A HER2 protein expression pattern analysis showed an overall discrepancy of 66.67 percent in the HER2 status of the primary tumor versus corresponding CTC HER2 expression. “Based on the percentage of CK-positive versus CK- and HER2-positive patients, there are CTCs which are only HER2 positive in this cohort,” Dr. Nagrath said.
“We looked at the EMT markers, such as vimentin, and at N-cadherin and E-cadherin to characterize EMT versus MET (mesenchymal-to-epithelial transition) states,” Dr. Nagrath said. They analyzed the proportion of EMT and MET CTCs in ER/PR-positive, HER2-positive, and triple-negative breast cancer patients.
“We found that the percentages of CTCs that are more EMT-like are high in triple-negative breast cancer,” she said. An association between a high proportion of EMT-like CTCs and aggressive disease can be assistive in selecting therapies.
“The GO chip is able to capture the CTCs at high sensitivity,” Dr. Nagrath said. “There are 10 to 20 CTCs per milliliter in these breast cancer patients, and they yield biologically relevant markers to categorize the aggressiveness of the disease.”
[dropcap]D[/dropcap]r. Nagrath presented an example of how monitoring programmed death-ligand 1 (PD-L1) expression in CTCs of non-small cell lung cancer patients undergoing radiation therapy could aid treatment. Radiation has been demonstrated to upregulate PD-L1 expression in tumor cells, but the difficulty of obtaining tissue biopsies has deterred validation in NSCLC patients (Wang Y, et al. Sci Rep. 2019;9[1]:566).
Dr. Nagrath and her team used the GO chip to isolate CTCs in whole blood samples taken from 12 nonmetastatic NSCLC patients undergoing chemoradiation (eight) or radiation only (four). They captured the cells before, during, and after treatment.
Analysis showed CTCs that were CK positive, CD45 negative, PD-L1 positive, and PD-L1 negative.
“When we looked at the number of CTCs in non-small cell lung cancer patients versus a healthy control [n = 6], the number of CTCs is clearly high in these patients. We found greater than two CTCs per milliliter as a threshold, based on the healthy controls.”
They looked at the patient samples taken at three different visits—before and during radiation and at follow-up. “We can see that the PD-L1-positive CTCs and the PD-L1-negative CTCs proportion varies through the treatment,” Dr. Nagrath said. The first patient, for example, had a higher percentage of PD-L1-negative CTCs at the first visit, but as he progressed through treatment, he had a higher number of PD-L1-positive CTCs.
“We found that the percentage of CTCs positive for PD-L1 increased during the treatment,” she said, suggesting that CTC analysis could be used to determine the best time to provide PD-L1-blocking therapies. “Maybe some patients are more prone to being responsive versus nonresponsive based on the timing.”
The number of CTCs was not found to be significant to progression-free survival. “However, patients with fewer PD-L1-positive CTCs had a better progression-free survival compared with the patients who had a high number of PD-L1-positive CTCs.” The PFS difference was more than 12 months.
Messenger RNA (mRNA) profiling showed that PD-L1 expression was higher in visit two than in visits one or three. A comparison of patients based on prognosis groups revealed significantly higher levels of PD-L1 expression in patients with a poor prognosis versus patients with a good prognosis, Dr. Nagrath said. Patients in the poor prognosis group also had markedly higher mRNA levels of PD-L1.
Other relevant markers that showed significant expression variances based on patient prognosis were mTOR, Ki-67, and CD33. “All three were high in the poor prognosis group versus the good prognosis group,” Dr. Nagrath said.
Dr. Nagrath and colleagues also studied castration-resistant prostate cancer patients to see whether the GO chip could be used to analyze CTCs at an mRNA level and to identify genes that could provide information about progression-free survival.
They ran two parallel GO chips—one for enumeration and one for quantitative reverse transcription PCR—and then analyzed the CTCs (Kozminsky M, et al. Adv Sci (Weinh).2018;6[2]:1801254). They conducted the study on 41 metastatic castrate-resistant prostate cancer patients between August 2013 and November 2016. “The median CTCs were 20 CTCs per milliliter in this cohort, which is great,” Dr. Nagrath said. “We had CTCs present in single-cell form and in cluster form in these prostate cancer patients.” Clusters ranged between two and eight CTCs, and some clusters were tagged with WBCs. The high number of CTC clusters detected suggested increased sensitivity of the GO chip, Dr. Nagrath said.
Some CTCs, which came from 10 patients with bone metastases, were positive for the epidermal growth factor receptor marker. “We realized that the EGFR-positive marker was indicative of aggressive disease in prostate cancer,” Dr. Nagrath said, noting that those results were published in an earlier study (Day KC, et al. Cancer Res. 2017;77[1]:74–85).
The RNA expression was conducted on 35 patient samples. “We looked for different genes that give survival predictability,” Dr. Nagrath said. “There are certain genes that are indicative of epithelial nature, such as EpCAM and E-cadherin, where expression at lower levels indicates better survival. But if certain genes, such as CD44 and Ki-67, are expressed at high levels, the patients had a poor survival compared to the other group.”
[dropcap]D[/dropcap]r. Nagrath and colleagues in her laboratory recently developed the microfluidic Labyrinth, which is a size-based CTC isolation technology. It is a take on spiral-based microfluidic technologies but “with a twist.”
“We have seen that the CTCs are larger in size compared to the WBCs,” and using that criterion, she said, “one can isolate these cells.” While smaller cells could be missed, the overwhelming advantage of the microfluidic Labyrinth technology is that it performs at a flow rate of 2.5 mL per minute.
The Labyrinth’s fluidic path has inward and outward turns and several sharp curvatures (Lin E, et al. Cell Syst. 2017;5[3]:295–304.e4). Dr. Nagrath described the design—inspired by Minotaur’s prison in Greek mythology—as “not a curve like a spiral; it has a lot of change of directions.” Instead of leading its target to a dead-end trap in the center, the Labyrinth separates CTCs and WBCs into different streamlines based on their size and drives them into one of four outlets. “The first outlet collects WBCs, and the second outlet collects the CTCs,” she said. Red blood cells and other blood components exit through the other two outlets.
Dr. Nagrath explained that the Labyrinth’s unique design of multiple turns focuses the CTCs into a single streamline. She described a demonstration using WBCs labeled with 4′,6-diamidino-2-phenylindole (DAPI) mixed with green fluorescent protein-labeled CTCs. The larger CTCs were separated first, about 50 mm from the device inlet, she said, while the smaller WBCs traveled about 500 to 600 mm before focusing into a single file.
“I’m talking about 600 mm, which is 60 cm,” she pointed out. “Such a long path on a small fluidic chip the size of a business card, which can fit on a 2- by 3-inch glass slide. That is the beauty of microfluidics.”
Once the CTCs are focused away from the WBCs, they can be collected. “They are continuously flowing through the device, with no antibodies needed.”
When Dr. Nagrath and her team looked at the rate of CTC recovery from the Labyrinth using four different cancer cell lines—breast (MCF-7), pancreatic (PANC-1), prostate (PC-3), and lung (H1650)—“we were able to isolate them at greater than 90 percent efficiency.” (Fig. 1).
Looking specifically at CTC isolation from 100 metastatic breast cancer patients, most patients had about five to nine CTCs per mL, she said. Dr. Nagrath and her team found that using a double Labyrinth procedure, in which the blood is taken from the first Labyrinth and run through a second device, enhanced the purity with regard to WBC contamination. “If you do Labyrinth once, you have contamination of a few thousand WBCs, but if you run the Labyrinth twice, you get a high purity, about 300 WBCs retained along with the CTCs,” she said.
Analysis of the CTCs isolated from 20 pancreatic cancer patients showed another benefit of the label-free method. “We were not only able to capture the cells which were epithelial phenotype based on the cytokeratin expression, but we also captured the cells which were cytokeratin negative but ATDC [ataxia telangiectasia group D] positive, which is more of a mesenchymal marker for pancreatic cancer patients.”
Isolating CTCs along the whole spectrum of epithelial to mesenchymal phenotype “is the real advantage of the label-free method,” she said. “We are able to capture these EMT-like, or cancer stem cell-like, circulating tumor cells” and demonstrate intrapatient CTC heterogeneity.
Pathologists would question how one could know that the CTCs matched the primary tumor, she said. Dr. Nagrath and her team looked at the pancreatic CTCs for known genetic mutations: APC, BRAF, CDKN2A, CTNNB1, KRAS, NRAS, PIK3CA, SMAD4, and TP53. “In 100 percent of the patients we tested, the KRAS gene was mutated in the CTCs, compared with the healthy controls,” she said. They also found a high number of mutations in the TP53 gene, which is a tumor suppressor gene associated with lung cancer and other solid tumors.
A study of 10 NSCLC patients, conducted with Nithya Ramnath, MBBS, medical oncologist at UM’s Rogel Cancer Center, illustrated the potential of CTCs in guiding precision medicine. “These were all targetable patients with either EGFR mutations or ALK rearrangements or ROS1 deletions,” Dr. Nagrath said. The patients’ CTCs were collected for isolation, expanded ex vivo, and cultured. The CTCs were then treated with the same therapy assigned to the patient, and the results were studied to identify resistant mutations. The study was funded by a National Cancer Institute Innovative Molecular Analysis Technologies program grant.
“We were able to isolate the CTCs at a high number using the microfluidic Labyrinth,” Dr. Nagrath said, with an average capture result of 100 CTCs or more per mL. “We were able to isolate the CTCs that were vimentin positive and EpCAM positive or negative,” based on pan-CK-positive CTCs.
“We also found that several CTCs in these metastatic patients were presented in cluster form rather than as a single cell,” she said. The clusters had varied expression of either vimentin or EpCAM, though the majority of the CTCs in the clusters expressed vimentin (Zeinali M, et al. Cancers. 2020;12[1]:127; doi:10.3390/cancers12010127). (Fig. 2).
The most important distinction between the patients with a single CTC and those with CTC clusters was the survival rate. “The patients with the single CTCs had a better progression-free survival than the patients with clusters, and the difference was more than 20 months,” Dr. Nagrath said. “These are all patients with EGFR mutations and/or ROS1/ALK rearrangements, who have a better progression-free survival rate than the other non-targetable metastatic lung cancer patients.”
In a previous study, Dr. Nagrath and her colleagues captured CTCs from early-stage lung cancer patients on the GO chip and used cancer-associated fibroblasts to culture and expand the CTCs (Zhang Z, et al. Oncotarget. 2014;5[23]:12383–12397). They found through next-generation sequencing analysis that the expanded lung CTCs carried the same genetic mutations as the primary tumor.
In another example of CTC use in precision medicine, Dr. Nagrath and her team tested different drugs on expanded CTCs from a lung cancer patient with a ROS1 deletion. The patient had been treated with crizotinib but was switched to entrectinib after developing a brain mass and then to lorlatinib. “By that time, we found that TPX-0005, which is a small-molecule compound, was much more effective,” she said. TPX-0005 (repotrectinib) was not available at the University of Michigan Hospital, but the patient was placed on a clinical trial at another institution and responded well on the drug for 18 months.
This success story “shows the power of a liquid biopsy,” she said, and how the CTCs can be used to determine the biological molecular characteristics of the primary tumor.
Last, Dr. Nagrath shared unpublished data from a study of ex vivo expansion of CTCs from three pancreatic cancer patient samples. The CTCs were captured with the Labyrinth, expanded, and then evaluated for characterization, metastatic capability, spheroid formation, and treatment response.
All three pancreatic patients had high percentages of EMT-like CTCs with vimentin, EpCAM, CD45, and other markers.
When the expanded pancreatic CTCs were injected into patient-derived xenograft mouse models, there was a 100 percent take rate. “It really speaks to the aggressive nature of pancreatic cancer,” Dr. Nagrath said. A similar experiment with the injection of lung cancer CTCs resulted in a take rate of less than 10 percent. The pancreatic CTC take rate and the high percentage of metastases seen in the liver, spleen, pancreas, and more “all clearly showed the invasion of the tumor cells that were injected into these mice.”
[dropcap]N[/dropcap]ext on the horizon are wearable technologies that deliver continuous depletion of CTCs from the blood. “It is an idea we all thought about for some time,” Dr. Nagrath said, adding that it took them five years to come up with a wearable device. “Using a dual lumen catheter, we can continuously draw the blood onto a CTC chip or isolation device.”
The system was developed in collaboration with Daniel F. Hayes, MD, Stuart B. Padnos professor of breast cancer research and co-director of the breast oncology program, Rogel Cancer Center. “You can draw the blood, take out the CTCs, and put the blood back,” Dr. Nagrath said.
The device prototype is not yet small enough to be worn as a wristwatch, “but we are close enough.” The system components are a GO chip in a heparin injector with a small peristaltic micropump that circulates the blood through a chip equipped with flow sensors.
A canine model study done in collaboration with Douglas Thamm, DVM, director of clinical research at the Flint Animal Cancer Center, James L. Voss Veterinary Teaching Hospital, Colorado State University, involved injecting MCF-7 breast cancer cells into a canine subject and monitoring blood draws every 15 to 20 minutes. The system design is modular and allows for continuous manual chip replacement. “We wanted to prove that we can either use one type of antibody or you can use multiple different types of antibodies, and we can replace the microfluidic chips as we sample the blood.”
Amy Carpenter Aquino is CAP TODAY senior editor.