Skip to main content

Cellanyx Live Tumor Cell Phenomics Dx May Predict Adverse Pathologies in Prostate, Breast Cancer

Premium

NEW YORK (360Dx) – Cellanyx, a startup diagnostics firm based in Boston, has developed an automated personalized medicine tool that uses machine vision to analyze living tumor biopsy cells and gauge aggressiveness of a patient's solid tumor.

The method, described today in Nature Biomedical Engineering, generates objective risk stratification scores in about 72 hours with a sensitivity and specificity of more than 80 percent when compared to gold standard post-surgical adverse pathology findings.

The technology was also shown in the analytical validation to be able to predict some adverse pathologies — things like lymph node positivity, lymphovascular invasion, or extra-prostatic extension in prostate cancer, and positive surgical margin in breast cancer — with a sensitivity and specificity above 95 percent. 

This predictive power is "a big advance beyond current technologies," said Ashok Chander, CEO and cofounder of Cellanyx, in an interview, adding that these numbers could translate into individual lives saved.

The gold standard Gleason scoring system for prostate cancer uses histopathology of formalin-fixed, paraffin-embedded tumor biopsy samples to determine the aggressiveness of a tumor and guide treatment decisions. However, most prostate cancers have a low risk of mortality, and the method tends to lead to overtreatment, particularly for low-grade cancers, as well as undertreatment of more aggressive tumors. In breast cancer the situation is similar, with ducal carcinoma in situ (DCIS) being an example of a low-risk type of breast cancer that tends to be overtreated using current methods.

Furthermore, methods such as liquid biopsy may be good for screening and diagnosis of late-stage metastatic disease, but in determining prognosis and patient risk stratification it may be less reliable. The live-cell phenomics method may be particularly useful for the risk-stratification aspect of cancer care, Chander said. And, by culturing and examining the majority of cells in a biopsy, it can also effectivly cope with the heterogeneity of most tumors.

"The high sensitivity and specificity with the phenotypic test, both exceeding 80 percent, reported in these initial validation studies suggest great promise as a risk stratification tool," said David Albala in a statement. Albala is Chief of Urology at Crouse Hospital in Syracuse, New York, an author on the paper and a member of the Cellanyx Scientific Advisory Board.  

While conventional histopathology relies on FFPE tissue, the field of oncology research has also gathered a wealth of knowledge about living cancer cells. Previously, this knowledge was not very accessible, Chander said, because of the challenge of performing primary live cell culture on biopsy tissues, particularly prostate cells, as well as analyzing them quickly enough to be meaningful for treatment decisions.

Chander developed Cellanyx's core technology during his PhD research at Columbia University, where he studied extracellular matrix components critical for cell survival in culture. Along with three other cofounders, Chander started Cellanyx in 2013.

One component of the Cellanyx technology is the enabling of rapid culture of biopsy tissue, which Chander described as "a major advance." The culture method was published last year in the journal Urology.

The analysis method for determining a patient's risk of aggressive cancer pathologies from the cultured cells, called stratification of adverse pathology, or STRAT-AP, was trained and tested in the study using 59 prostate biopsy samples and 47 breast tissue biopsy samples.

The method uses special transport media and cell dissociation steps on tissue from a standard solid tumor biopsy, then uses a protein formulation of extracellular matrix to coat culture surfaces and help cells survive.

More than 70 percent of the biopsy-derived cells survive transport using the method, according to the study, and more than 80 percent of the living cells then adhere to the ECM-coated surface of a proprietary microfluidic device.

The technology then uses machine vision to measure dynamic cellular phenotypic biomarkers of these living cells, a method touted in a recent review in npj Precision Oncology.

Specifically, once cells are cultured in a single monolayer in the microfluidic, each cell is assigned an identification number and automatically imaged in a label-free way using machine-vision software and differential interference contrast microscopy.

The system tracks dynamic phenotypic markers over 26 time points, including measures such as cell migration velocity, area, and perimeter, as well as cellular-nucleus area, cell-spreading velocity, the curviness of the cell surface, and cell height and adhesion. A somewhat analogous live-cell phenotypic analysis method, called morphokinetic cellular analysis, is used by Accelerate Diagnostics to assess antibiotic susceptibility of bacterial cells, but there does not to seem to be anything similar in the oncology space currently.

In addition to these dynamic markers of cell behavior, the cells are subsequently fixed and stained to perform fixed-cell imaging of a suite of biomarkers with an emphasis on oncoproteins that localize to focal adhesions.

In the Nature Biomedical Engineering study, the Cellanyx machine-learning technology was trained using 70 percent of the single cells from the entire patient population, comparing these to known pathology results from the patients. It was then tested on the remaining 30 percent of cells.

Overall, more than 300 measurements are made on each of about 5,000 single cells from a patient's biopsy using the proprietary machine-vision software.

Combining all the measurements, each cell is then scored positive or negative for general adverse pathology potential (GAPP), local adverse pathology potential (LAPP), and metastatic adverse pathology potential (MAPP). The machine then assigns a clinical score, or prediction value, for GAPP, LAPP, and MAPP, based on machine learning-based thresholds of the number of positive cells in each pathology domain required to indicate a risk of that adverse pathology.

Although what is going on in culture is not necessarily exactly what is happening in situ in the patient, the method provides a reference standard. "Any kind of sampling is going to be inherently biased, even [FFPE-based] current technology," which can vary with time of fixation and may not be an exact replication of what is going on in the body, Chander said. The Cellanyx method, however, also has the advantage of being a "totally objective analysis," he added.

Cellanyx owns the IP for the technology, including proprietary transport and culture media, microfluidics, live and fixed cell imaging of several hundred biomarkers, as well as automated analysis driven by machine learning. The firm is now in the process of obtaining patents on the technology and the initial patent has been accepted by the EU, Chander said.  

The firm is developing solid tumor prognostics for both prostate cancer and breast cancer, and these will most likely be used as lab-developed tests initially, Chander said. The technology could also be used as a research and development tool, such as by pharmaceutical companies for patient selection and stratification in drug studies, he said, noting that the firm has ongoing strategic conversations around both clinical and R&D applications.

Cellanyx has raised approximately $2 million to date with investments from TIE Boston Angels, Treehouse Health, and other angel investors. The firm is now raising a new round of funding to support clinical development of the prostate and breast cancer tests.