NEW YORK (360Dx) – Recently presenting abstracts of two studies on morphometric detection of lung cancer, oncology diagnostic firm VisionGate aims to detect early stage cancer with its optical computed tomography-based technology.
Lung cancer — the leading cause of cancer deaths in the US, with the highest risk among smokers — is usually detected by low-dose computed tomography (LDCT) chest screenings, which integrates X-ray equipment and computer imaging to produce multiple, cross-sectional images or pictures of a person's body. However, the disease typically doesn’t exhibit symptoms until later stages, when damage is irreversible.
Founded in 2003 in Seattle by CEO Alan Nelson, VisionGate has developed a tool called the Cell-CT platform that quickly identifies small driver mutations by integrating artificial intelligence and cellular imagery. The platform can also detect moderate dysplasia in epithelial cells.
Nelson explained that the firm's Lung cancer Early Detection (LuCED) assay — which runs on the Cell-CT platform — detects small and early-stage tumor and pulmonary nodules. First, a patient ingests a saline solution during a COPD screening, producing a deep cough in 15-20 minutes and producing a sputum sample for their clinician. A patient can also cough up a sputum sample at home with a portable sample collection kit that has a shelf life of up to a month.
"While this is less efficient than induction, we've found that if you do the best cough possible three mornings in a row, you'll [produce] an accurate specimen," Nelson said.
The sample kit is then immediately filled with a fixative that preserves the sample, killing any potential pathogen that could act as a biohazard. The clinician or patient then mails the kit to VisionGate's biosignature's laboratory.
When the kit arrives at the lab, VisionGate researchers initially prepare the sample by enzymatically digesting the mucous with dithiothreitol. After filtering the liquid to remove "common fibers," the team enriches the sample with surface antibodies for epithelial cells while subtracting white blood cells. After staining the sample, researchers inject the cells into a microfluidic cartridge and load the chip onto the CT-platform, where the rest of the process is automated.
Nelson explained that the Cell-CT platform optically scans each cell as it spirals down a plastic tube as thin as a human hair, digitally outlining 500 images of the cell. The platform's artificial intelligence system then produces a high-resolution 3D image of each cell by combining the sections outlined in each image.
"We can create an optical projection with images using photons in the visible range, reconstructed tomographically," Nelson explained.
He noted that the automated portion requires three seconds to produce an image with a resolution of 200 nanometers. From sample processing to answer, the overall process takes between four to six hours. However, VisionGate Chief Technology Officer Michael Myers said that the firm is improving the software and hardware to eventually lower the processing time to four hours.
He noted that the Cell-CT platform uses algorithms that integrate over 700 morphometric cellular "features," primarily focused on mutations in the nucleus. For example, the algorithms can detect "subtle" malignancy-associated changes, or MAC, in the cellular and nuclear morphology of cancer-associated cells (CACs) — cells that might have been affected by the presence of neighboring cancer cells. Researchers then use the features to drive pre-defined classifiers to identify a cancer such as lung carcinoma.
In a 2016 International Association for the Study of Lung Cancer (IASLC) conference abstract, VisionGate used the LuCED assay on a set of patients who were either controls or had biopsy-confirmed lung cancer. The team found the tool had a clinical specificity of 97 percent, with a sensitivity of 88 percent for stage I and stage II cancer.
At the 19th World Conference on Lung Cancer last month (WCLC) last month, VisionGate also presented abstracts on the Cell CT-platform's performance in two separate research studies.
In one, VisionGate's researchers demonstrated the Cell CT-platform's ability to detect mismatch repair protein deficiency (MMR-D) by knockdown of MlH1 expression. They reported that the tool's 3D cell morphometric artificial intelligence classifiers successfully discriminated between cells of parental cell lines from those of MMR-D lines, indicating the tool's use as a means for identifying MMR-D in malignant cells.
In the second study, the team investigated Cell-CT's ability to detect MACs in normal cells near the cancer of interest. They foundthat the platform could "discriminate cell features that are too subtle to distinguish by a human," and that "detection of cells with [MAC] may be used to further enhance the LuCED test's performance beyond published levels."
Nelson acknowledged that his team encountered multiple physical and technical limitations while developing the Cell-CT platform for commercial use. When initially building the physical tool, they struggled to develop an optical microscope that did not require a focal point to track epithelial cells.
"The whole machine is pushing the envelope on microfluidics, including the high-speed detection circuits and the video cameras that collect the images," Nelson said. "It took us a while to develop all that technology … since it was a massive physics and engineering issue."
Nelson explained that VisionGate also trained its AI algorithms to visually distinguish mutant epithelial cells from normal cells. After spending years recording data from multiple retrospective studies, the firm's tool now uses a database containing millions of cellular images to help the algorithm detect aberrations in the target cell.
"We needed to develop a database of well-characterized cells with clear pathology, and collecting that database and improving the algorithm [has] required a lot of retrospective and prospective specimens all over the world," Nelson said.
According to Nelson, VisionGate has filed for over 200 patents — with 176 already issued — linked to its novel optical microscope without a focal point and the 3D-imaging software.
Since its founding, VisionGate has raised $55 million in private funding. According to Nelson, VisionGate has not pursued any forms of venture capital financing and instead has — until recently — relied on angel funding fueled by friends and academic collaborators. However, the firm recently completed a series A preferred funding round with Blue Cross Blue Shield earlier this year as "an exclusive industry investor."
Nelson said that VisionGate will offer the LuCED assay as a fully commercialized assay by seeking US 510(k) approval from the US Food and Drug Administration, expects to launch the product in about two years. The FDA considers LuCED a Class II medical device, he said.
VisionGate is also actively working with several early users to further improve the LuCED assay, including the University of Colorado Cancer Center, the University of Pittsburgh, and the University of Chicago.
"[The universities] are participating in our clinical studies and will be running our clinical trials designed to obtain FDA approval," Myers said.
According to Nelson, use of the LuCED assay may minimize false-positive and -negative issues associated with LDCT scans. In a study published in June 2011 in the New England Journal of Medicine, researchers assessing results from the National Lung Cancer Screening Trial saw a 20 percent reduction in lung cancer patient mortality when they used the LDCT scanner compared to chest X-ray. However, there was a 44 percent false positive rate of detecting tumors with LDCTs.
Nelson said that the firm does not plan to replace standard pathology-based methods with its Cell-CT platform, but instead would offer the tool as an initial diagnostic before invasive procedures. If clinicians discover that a patient is cancer positive with the assay, they can pursue additional screening and biopsy extraction.
"The LuCED test can add value, as the test is more accurate and, therefore, has a much lower rate of false positive than LDCT, allowing cancer detection to be managed with lower cost and lower risk to patients," Myers explained.
VisionGate is entering a crowded space for the early detection of lung cancer, especially regarding groups aiming to use liquid biopsy as a sample collection method.
Molecular diagnostics startup Grail recently released data regarding early-stage lung cancer analysis using sequencing methods as part of its goal to offer a diagnostic tool for detecting early-stage cancer in patients. Also, after acquiring Armune Bioscience last December, Exact Sciences is now developing a liquid-biopsy based test for early detection of cancers, including lung cancer.
Nelson argued that VisionGate's assay is different from competitors' approaches because it directly measures epithelial cells in the patient's body using sputum. In addition, Nelson highlighted that the 3D imaging tool's high specificity stands out from other tools that use microRNAs, circulating tumor cells, or cell-free DNA in blood samples.
"Every other sample is indirect, and usually is blood-based, although there are a few companies," such as Ancon Medical, "looking at breath for volatile organic compounds," Nelson said. "The specificity is surprisingly low, and you typically need to do something invasive to show that [the patient] doesn't have lung cancer."
According to Myers, the firm's LuCED lung test has performed well across all stages — including stage 1 and 2— of lung cancer, which "has not gone unnoticed by other diagnostic and pharmaceutical companies."
As firms like VisionGate and Grail develop a variety of screening tools to detect different cancers, Epic Sciences Chief of Medical Innovation Ryan Dittamore noted that researchers are looking at issues such as tumor mutational burden in patients. Epic's platform centers around arraying circulating tumor cells for imaging-based analysis and downstream genomics.
Because some groups focus on either patients' cells or the mutated cells, Dittamore believes that selecting specific cells types may be predictive but not yet clinically useful due to being context-specific.
"Ultimately, we've learned that you need to look both at host and tumor cells … as people have different experiences as the host," he said.
However, Dittamore pointed out that there is a big opportunity for noninvasive early detection tools overall, "especially in lung cancer, when you don't have to biopsy the whole region, and not only perform screening tests, which are difficult [for patients] to go through."
While Nelson is still evaluating the commercial pricing of the lung cancery assay, he noted that "certain personalized medicine diagnostics are charging… between $2,000 and $4,000 a test."
VisionGate currently has two offices: one in Seattle that serves as an engineering facility that develops the diagnostic technology, and one that serves as a business facility in Phoenix. This week, it appointed John Glasspool as Chief Operating and Strategy Officer and to its board of directors.
As VisionGate's technology moves toward full commercialization, Myers said that the firm is considering new applications, including urine assessment for bladder collaboration through a partnership with Yale University. In addition, the firm is considering pleural fluid assessment for mesothelioma, oral scrapes for oral cancer, and has begun studies on using esophageal swabs for Barrett's esophagus and esophageal cancer.