Skip to main content
Premium Trial:

Request an Annual Quote

Metabolic Assay Shows Promise in Predicting Treatment Response in Lung Cancer Study


NEW YORK – Cancer patients with genetic mutations that are thought to be vulnerable to targeted therapies, based on available clinical or preclinical data, often don't respond to those treatments or rapidly develop resistance to them.

This is a challenge that has plagued precision oncology, and researchers from the Institute for Systems Biology and their collaborators in China and the US have developed a diagnostic hoping to address it, starting with non-small cell lung cancer where there are a number of molecularly targeted treatment options.  Specifically, they have developed a single-cell metabolic assay that measures the glycolytic phenotype of cancer cells taken from a blood draw in order to determine whether or not a patient will respond to treatment.

"This work is driven by unmet clinical need for non-small cell lung cancer patients. They need a simple and cost-effective method that can identify nonresponders and short-term beneficiaries to the standard clinical management, prior to the onset of therapy," said Wei Wei, an assistant professor at the ISB and a corresponding author on a paper published this week in Nature Communications. "The current clinical decision-making in non-small cell lung cancer is primarily driven by the tumor genetics. But the problem is that patients with the same targetable mutation have variable responses to the same treatment. Patients have to go through the therapy in order to know if it works or not."

For example, NSCLC patients with EGFR mutations may be prescribed EGFR-tyrosine kinase inhibitors. But about 20 percent to 30 percent of NSCLC patients with EGFR-sensitive mutations do not respond or develop resistance rapidly to EGFR-TKI treatment, even if they have the same EGFR mutations as those who do respond to treatment and don't harbor concurrent resistance mutations, the researchers said.

However, they noted that change in metabolic activity can be used to quickly and reliably determine the response of tumor cells to stressful conditions such as drug treatment. Successful treatment of a tumor cell is usually accompanied by a reduction in the cell's aberrant glycolytic activity.

In the early 1920s, Otto Warburg observed that cancer cells were highly fermentative, and hypothesized that the effect was due to a metabolic injury. In the decades since then, researchers discovered that cancer cells produce large quantities of lactic acid, and that extracellular and intratumoral acidification is a major factor in local tumor growth and metastasis.

Indeed, cancers are marked by a reprogramming of cellular metabolism. Cancer cells use glycolysis more readily than normal cells in order to produce metabolites that could help them proliferate — this reliance on aerobic glycolysis is called the Warburg effect.

The inhibition of glycolysis, assessed by positron emission tomography (PET), has been utilized as an in vivo predictive biomarker of drug response for brain cancer, the ISB researchers and their colleagues noted in their new paper. And new studies have shown that tumor cells can also use amino acids and fatty acids to meet their metabolic needs, which also has implications for using metabolic activity in tumors as a predictive biomarker of therapy efficacy and resistance.

In order to test their hypothesis, the researchers used samples of pleural effusion — excess fluid from the pleural cavity and usually the first sign of lung cancer — containing rare disseminated metastatic tumor cells to analyze the metabolic state of patient tumor cells.

They developed an on-chip metabolic cytometry (OMC) platform and fluorescent metabolic probes to perform metabolic phenotyping on the rare disseminated tumor cells (DTCs) in pleural effusion samples from a cohort of 32 lung adenocarcinoma patients. The probes covered prevalent driver oncogenes and molecular subtypes, and quantified the glucose uptake and mitochondrial oxidation activity of cells at the single-cell resolution. They then performed single-cell sequencing to determine which molecular signatures were associated with distinct metabolic phenotypes in their samples.

A change in the metabolic activity of tumor cells is a very fast and reliable readout of response to drug treatment, Wei said. A successful drug engagement is normally accompanied by the reduction of glucose consumption and sometimes accompanied by a metabolic program switch from glycolysis to respiration.

"In those cells derived from the pleural effusion of the lung cancer patients, we found that cells are differentially engaged in either glycolysis or respiration. So, they perfectly segregate into two metabolic phenotypes," Wei said. "And we found [that] these two metabolic phenotypes predicted patients' therapy response, clinical performance, and long-term survival."

The reported average survival time for Stage IV lung adenocarcinoma patients with malignant pleural effusion (MPE) is around six months, so the researchers chose to follow up with patients five to seven months after the MPE draw in order to assess whether their assay and the use of glycolysis as a biomarker had correctly predicted the patients' response to therapy.

Of the 32 patients in the cohort, 14 were newly diagnosed and their MPE was analyzed before the onset of first-line treatment. Among them, patients with predominantly glycolytic cells in their effusions all had progressive disease and died before the follow-up. In contrast, patients with predominantly mitochondrial oxidation cells in their effusions were all partial responders with reduced tumor sizes upon follow-up. Further, the researchers found, four of the five patients with a balanced metabolic phenotype had a stable disease upon follow-up.

In addition to predicting short-term therapy response, the metabolic phenotyping approach was also able to predict long-term survival. The patients with tumor cells that had a predominantly mitochondrial oxidation phenotype or balanced phenotype had significantly longer survival times than patients with predominantly glycolytic cells. This held true for all the patients in the study, regardless of when they were diagnosed or whether or not they had EGFR mutations.

When they delved into the molecular underpinnings of this predictive biomarker, the researchers didn't find a clear association between the glycolysis ratio and tumor mutations or CNVs. Instead, Wei said, they found that the biomarker may be related to the cancer cells' epithelial-to-mesenchymal transition (EMT) program. Indeed, he added, glycolytic cells bear more mesenchymal features and they're more invasive, while the cells engaged in respiration had elevated expression of epithelial-related genes and reduced expression of mesenchymal-related genes.

Although the present research is in non-small cell lung cancer, Wei believes these findings will hold true for a number of cancers. "We believe this metabolic response and metabolic activity is a pretty general phenomena across different cancer types, and we are currently testing this [assay] on colorectal cancer and liver cancer," he said. "Preliminary results are promising in terms of if this assay will work on other cancer types, and as far as I can tell right now, our preliminary results support this notion."

He also noted that the assay could determine treatment outcomes for patients who aren't found to have driver mutations or for cancers that aren't driven by oncogenes. Of the 32 patients in the study, some "had the driver mutation and some of them simply did not," he said. "It looks like this type of segregation works both for patients with driver mutations and without driver mutations, [so] this might be translatable to other cancer types without a clear driver mutation. But we need to test it before we can make any solid conclusions."

And although the assay hasn't been tested in pediatric cancers, Wei also noted that such an assay might prove useful in informing treatment decisions for those patients as well, given that most pediatric cancers are driven by changes at the RNA level rather than the DNA level and therefore tend to confound conventional genetic testing meant for adult cancers. However, he added, studies need to be conducted before the assays utility in kids can be determined.

Beyond its accuracy, Wei further pointed to the assay's simplicity as one of its assets. The test was specifically developed to run from liquid biopsy samples, he said, making it less invasive than tests that require solid tumor biopsy samples. And because the assay is less invasive and simpler to run, it can be performed on any standard clinical lab equipment with a fluorescent microscope or other fluorescent imaging system.

"We believe it has the potential to be democratized into regular cytology labs or cytopathology labs," Wei said, adding that the entire assay can be completed within 15 minutes.

Although there's still a lot of work to be done before the assay shows up in hospital labs, Wei hopes that he and his team will be able to develop it into a commercially available test or CLIA-approved assay at some point.

He also highlighted the research's implications for future cancer diagnostics. "Except for PET imaging, functional assays are rarely used as a diagnostic for clinical decision making. Our work tries to highlight the promise of using cellular metabolic function to address some of the challenging questions in cancer diagnostics and provides complementary information to the molecular information you get from tumor genetics or some histology type of testing," Wei said.