NEW YORK – A team led by researchers at the Massachusetts Institute of Technology has developed a method that measures dysregulated protease activity for the early detection of lung cancer.
Detailed in a paper published last week in Science Translational Medicine, the test is one of several protease-based diagnostics developed by the MIT team, which in 2015 spun out a company, Glympse Bio, to commercialize the technology.
While the method has shown promise in mouse models, it is still several years away from potential use as a diagnostic tool, but if successful, it could aid early detection of lung cancer by, for instance, helping evaluate indeterminate nodules detected during computed tomography-based lung cancer screening.
The approach uses what the researchers call "activity-based nanosensors," nanoparticles linked to substrates of proteases known to be dysregulated by a particular disease process. These nanoparticles are introduced into a patient and the linked substrates are cleaved by the target proteases. The cleaved substrate can then be collected in urine and measured to assess the level of protease activity, with abnormal levels of activity indicating the patient has a particular condition.
In the STM study, the MIT team targeted lung cancer and developed a panel of 14 peptide substrates to 15 proteases that they determined were upregulated in lung cancer via an analysis of transcriptomic data from The Cancer Genome Atlas. They coupled these substrates to mass-encoded reporters allowing them to be easily quantified by mass spectrometry.
The researchers tested the nanosensors in a mouse model of lung cancer, tracking the progression of lung tumors longitudinally using both micro-computed tomography (microCT) and the protease-based approach. At 100 percent specificity, microCT showed 33.3 percent sensitivity at five weeks after the initiation of tumor development, 75 percent sensitivity at 7.5 weeks and 100 percent at 10.5 weeks.
Using the nanosensor-based measurements to train a machine learning classifier, the researchers were unable to distinguish between mice with lung cancer and controls at five weeks but were able to distinguish between the two with an area under the curve of .95 at 7.5 weeks and .93 at 10.5 weeks. Applying the approach to a different mouse model of lung cancer, they found it distinguished between cases and controls with an AUC of .96 at 5 weeks, .98 at 7.5 weeks, and .93 at 10.5 weeks.
They also trained a classifier to distinguish the two lung cancer models from healthy controls and mice with benign lung inflammation, which can lead to false positives in CT-based lung cancer screening. That classifier separated the two cancer models from the healthy controls and benign inflammation cases with an AUC of .97 at 7.5 weeks.
Use of the synthetic markers gets around limitations — low abundance, in particular — inherent in endogenous protein markers. The researchers can control the level at which they introduce the synthetic markers, which allows them to make certain they are abundant enough for easy detection. This makes them potentially useful in the case of ailments where existing markers are present only at low levels, or for which there are no known protein markers.
The MIT researchers have traditionally used transcriptomic data to identify proteases to target with its nanosensors, but they have recently begun using proteomic analyses to aid this process, said Ava Soleimany, an author on the STM study and a graduate student in the lab of Sangeeta Bhatia, a professor of engineering at MIT and developer of the technology.
"We're really interested in looking not only at gene expression but also proteomic validation of some of these targets, as well as being able to directly measure the activity of these proteases within a tissue specimen like a biopsy sample," she said. "That would give us a way to validate some of our targets on the activity level directly in human tissues, which we think is a key step for not only showing a degree of validation prior to moving to clinical studies but also to continue to improve our sensors by identifying peptide substrates that are specifically and robustly cleaved."
Lung cancer is one of several conditions that includes other cancers and infectious disease that the researchers are investigating using the technology. At Glympse, the company's lead product is a test for early detection of non-alcoholic steatohepatitis (NASH) along with monitoring progression of liver fibrosis in NASH patients and the effectiveness of drug treatments for the condition.
In October, the company announced it had entered into a strategic collaboration with drugmaker Gilead Sciences to use Glympse's technology to stage NASH patients participating in clinical trials and to track response to treatments being developed for the condition.
While the nanosensors have potential advantages over endogenous markers, they are somewhat more complicated from a regulatory perspective as they require introducing synthetic molecules to patients' bodies.
"One of the very earliest things we had to grapple with was… how should it be regulated?" Bhatia said, noting that she and her colleagues used examples like contrast agents injected into patients for imaging studies as precedents to help guide their efforts.
"There have been a number of sort of combination products that have gone through the [US Food and Drug Administration], and that is how we think about [the nanosensors]," she said. "The first [product] is the biggest hurdle because we need to show that the mixture of these probes is safe and we are well on our way to that now, and then the [Center for Devices and Radiological Health] will help us assess diagnostic efficacy."
Bhatia said her lab sees its role as exploring and evaluating, primarily in mouse models, potential clinical uses for the technology with Glympse then taking those uses "forward into patients."
Glypse raised $6.6 million in a seed funding round in 2015. In 2018 it raised a further $22 million in a Series A financing round led by Venture Capital firms LS Polaris Innovation Fund and Arch Ventures.