Radiomics can predict chemo response rate in lung cancer
Machine learning of radiomic data extracted from baseline CT scans can help determine which lung cancer patients will benefit from chemotherapy.
Platinum-based chemotherapy is the standard of care for first line treatment of advanced state non-small cell lung cancer (NSCLC). However, only about 25 percent of patients respond to this therapeutic regimen.
CT studies are routinely used as a clinical diagnostic tool for cancer tumor staging and monitoring treatment response.
Radiomics is a field of study that extracts quantitative data from medical images using algorithms. This data, called radiomic features, can uncover disease characteristics not apparent to the naked eye.
The researchers, from several U.S. medical and academic institutions, sought to determine whether radiomic features of the texture and size of tumors in CT images could help predict who would respond to chemotherapy. They retrospectively analyzed 125 patients who had been treated with platinum chemotherapy at Cleveland Clinic.
A computer analyzed the CT images of lung cancer to identify unique patterns of heterogeneity inside and outside the tumor. It compared the CT scans of patients who did and did not respond to the chemotherapy; the feature patterns were then used to train a machine-learning classifier to identify the likelihood that a patient would respond to chemotherapy.
The features found differences in shape and texture in lesions between those who had responded to chemotherapy and those who had not. For instance, there was more textural pattern disorder or heterogeneity within and around the lesions in the CT images of the patients who didn’t respond to chemotherapy.
"When we looked at patterns inside the tumor, we got an accuracy of 0.68. But when we looked inside and outside, the accuracy went up to 0.77," said Mohammadhadi Khorrami, a doctoral candidate from the Department of Biomedical Engineering at the Case Western Reserve University School of Engineering in Cleveland, who was a co-leader of the study.
The study was published in Radiology: Artificial Intelligence.
"This is the first study to demonstrate that computer-extracted patterns of heterogeneity, or diversity, from outside the tumor were predictive of response to chemotherapy," said Monica Khunger, MD, from the Department of Internal Medicine at Cleveland Clinic, who co-led the study. "This is very critical because it could allow for predicting in advance of therapy which patients with lung cancer are likely to respond or not. This, in turn, could help identify patients who are likely to not respond to chemotherapy for alternative therapies such as radiation or immunotherapy."