According to the American Cancer Society, lung cancer is "by far" the leading cause of cancer death in the United States, accounting for nearly 25 percent of all cancer deaths, and more than colon, breast, and prostate cancers combined. Outcomes for lung cancer patients have improved, however, with better screening practices and earlier detection playing a key part in increasing survival rates.
Noting the risk of overtreating lung cancers identified through screening, a team led by researchers from Moffitt Cancer Center conducted a study relying on radiomics-a tool for extracting clinically relevant information from radiologic imaging-in an effort to reveal tumoral patterns and identify lung cancer patients who are at higher risk of poor outcomes.
"Image-based biomarkers could have translational implications by characterizing tumor behavior of lung cancers diagnosed during lung cancer screening," the authors recently wrote in Cancer Biomarkers (2022; doi: 10.3233/CBM-210194). For the study, the researchers used peritumoral and intratumoral radiomics and volume doubling time (VDT) to identify high-risk subsets of lung patients diagnosed in lung cancer screening that are associated with poor survival outcomes.
The team acquired data and images from the National Lung Screening Trial, calculating VDT between two consequent screening intervals approximately 1 year apart. Peritumoral and intratumoral radiomics were extracted from the baseline screen, while overall survival was the main endpoint. Classification and Regression Tree analyses identified the most predictive covariates to classify patient outcomes.
The decision tree analysis stratified patients into three risk groups (low, intermediate, and high) based on VDT and one radiomic feature. Survival rates among the high-risk patient group were significantly lower, with high-risk patients demonstrating a 5-year overall survival rate of 25 percent, compared to a 5-year survival rate of 83 percent among the low-risk patient group.
Among early-stage lung cancers, high-risk patients had poor survival outcomes (44% 5-year overall survival rate) in comparison to those in the low-risk group (90% 5-year survival rate). For VDT, the decision tree analysis identified a novel cut-point of 279 days, and using this cut-point VDT alone discriminated between aggressive (45% 5-year overall survival) versus indolent/low-risk cancers (83% 5-year overall survival).
"We utilized peritumoral and intratumoral radiomic features and VDT to generate a model that identifies a high-risk group of screen-detected lung cancers associated with poor survival outcomes," the authors wrote. "These vulnerable subset of screen-detected lung cancers may be candidates for more aggressive surveillance/follow-up and treatment, such as adjuvant therapy."
"Though lung cancer screening is a lifesaving modality, there are some potential limitations, including the identification of the indolent cancers, that would not likely become symptomatic in a patient's lifetime and would not contribute to death," stated study co-author Matthew B. Schabath, PhD, Associate Member with joint appointments in the Departments of Cancer Epidemiology and Thoracic Oncology at Moffitt Cancer Center.
With these potential limitations in mind, the researchers sought to determine "whether we can use image-based biomarkers...to discriminate between low-risk or indolent cancers and aggressive high-risk cancers associated with rapid and lethal outcomes," he noted.
Such a biomarker could be used to determine the appropriate treatment for lung cancers identified in lung cancer screening, Schabath said, noting that low-risk patients with very good outcomes would likely have a better chance of cure by surgery and may not need adjuvant therapies such as chemotherapy and/or radiation therapy.
"Conversely, patients with the high-risk tumors would need more aggressive treatment options and would need to be more closely followed," he said.
All the study participants were part of the National Lung Screening Trial and were screened for lung cancer using a low-dose CT scan. "Thus, from those low-dose CT scans, we calculated several hundred image biomarkers," Schabath said. "These radiomic features describe shape and size, and textural heterogeneity by calculating pixel by pixel differences, which are not discernable by the naked eye."
Once these radiomic features were calculated, the researchers used artificial intelligence methods to determine which radiomic features could predict low-risk patients, high-risk patients, and any patients in between. Ultimately, "these radiomic biomarkers could be used as a pre-treatment biomarker to aid in clinical decisions to support [and] personalize treatment options for lung cancers diagnosed in the lung cancer screening setting," Schabath concluded.
Mark McGraw is a contributing writer.