What if a simple blood test were enough, one day, to detect lung cancer while the tumor is still tiny and silent? A team from the Johns Hopkins Kimmel Cancer Center has just taken a step in this direction, by focusing not on a specific gene, but on general DNA disorder.
Published on January 27, 2026 in the journal Clinical Cancer Researchtheir study describes a blood test
liquid biopsy capable of identifying certain lung cancers very early by measuring “molecular chaos” in the genome, a particular instability of the chemical marks placed on the DNA.
A blood test that tracks the molecular chaos of DNA
The Johns Hopkins team, under the joint leadership of Dr. Hariharan Easwaran, discovered that cancer manifests itself as inconsistency, that is, a series of changes rather than a predictable pattern. Compared to other tests aimed at spotting something that distinguishes cancer cells from normal cells, they decided to approach the question differently: instead of asking “What has changed?”, they asked themselves “How chaotic are these changes?”.
To do this, scientists looked at the
DNA methylationsmall chemical groups that attach to regions called CpG islands. In healthy cells, these patterns remain fairly regular. Conversely, cancer cells distribute methyl groups (small chemical marks) on their DNA in a very haphazard manner. Some regions are strongly marked while others remain blank, creating a kind of molecular chaos. Knowing that tumors constantly release fragments of DNA into the blood (a tiny fraction of circulating DNA), fragments which reflect this same disorder (unlike the regular DNA of healthy cells).
An epigenetic instability index capable of identifying several cancers
From 2,084 methylation profiles, the team isolated 269 DNA regions whose variability best distinguishes cancerous tissues from normal tissues (on 5 types of cancer: breast, colon, brain, lung and pancreas). “We identified specific genomic regions that tend to be most variable in DNA methylation marks during cancer“, specifies Sara Jayne Thursby, postdoctoral researcher. “In free DNA circulating in the blood, this variability should not be high, but if it is, it is indicative of the development of a developing cancer phenotype“.
From these 269 regions, the researchers created an overall score, theEpigenetic Instability Index (EII)applied to circulating free DNA present in the blood. “This is the first study where we really try to integrate the measurement of this variation, or stochasticity, into a diagnostic tool“says Professor Hariharan Easwaran, associate professor of oncology at Johns Hopkins University School of Medicine and lead author of the study. “We immediately found that measuring variation in DNA methylation works better than simply measuring DNA methylation itself.“.
81% of stage IA lung cancers detected with high precision
Tested on blood samples from patients with lung cancer, this index made it possible to identify adenocarcinomas of stage IA lung cancer
with approximately 81% sensitivity to 95% of specificity. In other words, the test detects a little more than 8 out of 10 early cancers, while only giving around 5 false alarms per 100 healthy people. Early breast cancers were detected at almost 68% with this same high specificity.
In a scenario of screening 1,000 people, a test with 90% specificity can falsely alarm around a hundred people; with 95%, this number is almost halved. For tests that already cause anxiety, such as an abnormal mammogram or an elevated PSA, such a test could serve as a secondary “triage” tool before considering a biopsy.
Hope for multi-cancer screening, but still experimental
The same 269 DNA regions show chaotic patterns in other tumors, including colon, brain, pancreas and prostate. The idea of a single screening test
multi cancers therefore begins to emerge, even if it still remains theoretical. And this test could also be of interest for treatment. “We hypothesize that early-stage tumors and precancerous lesions that exhibit a high degree of methylation variation, or epigenetic instability, may be more resistant to intrinsic cancer-protective mechanisms and progress more rapidly.“, explains Professor Thomas Pisanic, research associate professor of oncology at the Johns Hopkins Institute of Nanobiotechnology and co-author of the study
For the moment, this test remains a proof of concept, resulting from case control studies and machine learning models, and it is not routinely available, neither in the United States nor in France. Large trials, carried out in people who do not know if they have cancer, will have to confirm its performance before considering its integration into screening programs or in addition to low-dose chest scans. The next results will be closely scrutinized by oncologists and health authorities.