A blood test that may be able to detect cancer years before any symptoms appear is under development. The PanSeer assay, which detects methylation markers in blood, was used in healthy individuals and successfully detected five cancer types in 91% of samples from individuals who were diagnosed with cancer 1 to 4 years later.
“We can’t say for sure that the patients didn’t have any symptoms, but we detected the cancer years before they ever walked into the hospital,” said study author Kun Zhang, PhD, a professor of bioengineering at the University of California, San Diego. “We were also able to follow up with patients, so we actually knew that they had cancer.”
Zhang noted that they also followed the individuals whose tests indicated they were cancer free. “They were healthy at the time the samples were obtained, and they remained healthy,” he said. “Follow-up was key to validating these data.”
The PanSeer test is being developed by Singlera Genomics. Zhang is a cofounder and a paid consultant of the company.
The study was published online July 21 in Nature Communications.
Unique among tests
Several blood tests for the detection of cancer have been reported in recent years. A test developed by Grail was able to detect more than 50 types of cancer and also identified the tissues where the cancer originated. The CancerSEEK test identified eight common cancers by measuring circulating tumor DNA from 16 genes, as well as eight protein biomarkers.
Findings regarding the CellMax Life FirstSight blood test were released last month. The test detected all 11 cases of colorectal cancer in a cohort of 354 patients and detected 40 of 53 advanced adenomas.
The latest study with the PanSeer assay is unique, say the investigators, because they had access to blood samples from patients who may have been completely asymptomatic and had not yet been diagnosed with cancer. Other blood tests have typically involved the use of specimens from people with a known diagnosis.
The specimens were collected as part of a 10-year longitudinal study that began in 2007 in China. Zhang and his team were able to test blood samples before individuals had experienced any signs or symptoms of cancer, and they were able to conduct long-term follow-up of the cohort.
Study details
The PanSeer assay uses DNA methylation analysis and screens for a DNA signature called CpG methylation. The results of an early-stage proof-of-concept study were published 3 years ago (Nat Genet. 2017;49:635–642).
For the current study, data were drawn from the Taizhou Longitudinal Study, which included 123,115 individuals aged 25 to 90 years who provided blood samples for long-term storage from 2007 to 2014. Participants were monitored indefinitely for cancer occurrence using local cancer registries and health insurance databases.
The team identified 575 individuals who were initially asymptomatic and healthy but were subsequently diagnosed with one of five common cancer types (stomach, esophagus, colorectum, lung, or liver cancer) within 4 years of their initial blood sampling. The authors selected these five cancer types to study because the incidence rates of these cancers in this population are high and, taken together, account for the highest mortality.
The study design allowed the authors to evaluate specimens both from patients with cancer and from those who were healthy within the same cohort. Using 191 prediagnosis samples, 223 postdiagnosis samples, and 414 healthy samples, they created a training set and validation model.
A machine learning method was created to classify samples as being either from healthy individuals or from those with cancer, using blood samples from the training set. The final classifier achieved 88% sensitivity for postdiagnosis samples and 91% sensitivity for prediagnosis samples at a specificity of 95%.
Zhang feels that initially it would be more appropriate to use the test for high-risk patients and to then evaluate the clinical benefit. “For any test, it is always more prudent to begin with a high-risk population,” he said. “You want to see some benefit with the high-risk population first, and then it can slowly be extended to others at lower risk.”
He emphasized that more rigorous testing is needing before the PanSeer assay is ready for clinical use. The logistics of designing and conducting a clinical trial that would include more than one cancer type would be very complicated. “The option was to break it down to five different studies,” he said. “We decided to begin with colon cancer, and we are currently in the process of talking with the FDA and designing the study.”