Published Nov 20, 2026 | 7:00 AM ⚊ Updated Nov 20, 2026 | 7:00 AM
Image used for representational purpose. (Ai generated image)
Synopsis: An Indian study involving 1,480 participants across 43 hospitals found that higher concentrations of cell-free DNA were associated with stronger cancer-related signals in blood samples. The findings could help improve AI-assisted multicancer detection tests, but researchers say validation in asymptomatic populations is essential before screening claims can be made.
Artificial intelligence (AI) could help doctors identify signs of multiple cancers through a single test by analysing molecular changes in blood samples.
An Indian multicentre study has identified a factor that can influence how effectively such blood-based tests detect cancer: the amount of DNA fragments circulating in the bloodstream. Researchers found that higher concentrations of cell-free DNA were associated with stronger cancer-related molecular signals, including in patients with early stages of the disease.
The study, published in the Proceedings of the National Academy of Sciences (PNAS) on 3 November 2026, involved 1,480 participants recruited from 43 hospitals across India. These included 950 patients with 14 types of solid cancers, 80 people with symptomatic benign conditions and 450 healthy participants.
The authors said their findings challenge the assumption that higher levels of DNA released by non-cancerous cells would dilute tumour-derived DNA and make cancer detection more difficult.
“We found that, as ccfDNA (circulating cell-free DNA) concentration rises, so does the proportion of it carrying cancer-linked methylation changes,” the authors wrote.
They added that the additional DNA circulating in cancer patients was “not just noise” but was biologically linked to the presence of a tumour, even though most of it did not originate directly from cancer cells.
Cell-free DNA consists of small fragments of DNA released into the bloodstream as cells die or break down. These fragments are present in healthy people and originate predominantly from immune cells.
In people with cancer, a small proportion also comes from tumour cells. These fragments, known as circulating tumour DNA, carry molecular changes that can help identify cancer.
One such change is DNA methylation, which involves chemical marks that influence how genes function. Cancer development can produce abnormal methylation patterns, creating molecular signals that researchers can look for in blood samples.
The study used a targeted methylation sequencing panel covering approximately 5.5 lakh genomic regions and 3.98 million CpG sites, locations in DNA where methylation can occur. Researchers identified 10,225 regions that were more heavily methylated in tumour tissue than in normal tissue.
These molecular patterns can be analysed using computational models trained to distinguish cancer samples from non-cancerous samples. AI-based systems can help identify patterns across large volumes of sequencing data and classify samples according to whether they contain cancer-associated signals.
The commercial CancerSpot test, described in material accompanying the study, uses DNA methylation analysis and an AI-trained classifier to identify cancer signals. The paper itself, however, primarily investigates the relationship between cell-free DNA concentration and cancer-associated methylation patterns.
A positive result from such a test is not a cancer diagnosis. It requires further clinical assessment and, where indicated, imaging or biopsy.
One of the study’s important findings was that cancer detection sensitivity improved with higher cell-free DNA concentrations, even among patients diagnosed at the same disease stage.
For stage I cancer, sensitivity increased from 55% among patients with the lowest cell-free DNA concentrations to 85% among those with the highest concentrations. For stage II cancers, it increased from 59% to 86%.
Sensitivity refers to the proportion of people with cancer who are correctly identified by a test.
For stage IV cancers, the increase was smaller, from 83% to 88%.
The researchers found that the association remained significant after accounting for cancer type, disease stage, age, sex, body mass index, tobacco and alcohol use, and technical factors related to sample collection and analysis.
Cell-free DNA concentration alone explained 19.1% of the variation in cancer-associated methylation signals, compared with 15.2% explained by disease stage alone.
The authors described this as a connection between the amount of DNA circulating in the blood and tumour biology.
“Contrary to the prevailing view that ccfDNA concentration is merely a preanalytical technical variable coarsely related to tumor presence, we demonstrate a deeper connection between ccfDNA concentration and tumor biology,” they wrote.
They said the findings could help explain why blood-based cancer detection tests sometimes report different levels of sensitivity across studies, even when researchers compare patients at similar disease stages.
The authors suggested that cell-free DNA concentration should be considered when evaluating and comparing multicancer early detection tests.
Most of the increase in cell-free DNA observed in cancer patients does not come directly from tumour cells. Previous research has suggested that immune cells, particularly white blood cells, contribute substantially to this increase.
The researchers explored whether inflammation around tumours could explain why higher DNA concentrations were associated with stronger cancer-related signals.
One possible explanation involves neutrophil extracellular trap formation, or NETosis. Neutrophils are immune cells that can release web-like structures made of DNA and proteins as part of the body’s immune response.
Tumours can attract and activate these cells through inflammatory signals. When the structures break down, they release additional DNA into the bloodstream.
The authors proposed that the same inflammatory environment could also influence methylation patterns in tumour cells and surrounding tissues.
However, they were careful not to present this as an established mechanism.
“Our data are observational and do not establish causality or provide a mechanistic explanation for the association between ccfDNA concentration and cCHyF,” they wrote.
They said further research would be needed to test whether inflammation and neutrophil activity explain the relationship.
The study also examined 80 people with symptomatic benign conditions, including ovarian cysts, kidney cysts, breast fibroadenomas, uterine fibroids and inflammatory conditions.
Their cell-free DNA concentrations were higher than those observed in healthy participants, although lower than in patients with malignant tumours.
Cancer-associated methylation signals also increased in this group. The detection rate rose from 23% among people with the lowest DNA concentrations to 54% among those with the highest.
This finding matters because it shows that increased cell-free DNA and cancer-associated molecular signals are not exclusive to malignant disease.
The authors said the association observed in people with benign lesions suggested that the underlying biology “may not be unique to malignancy but may instead reflect a broader response to abnormal tissue growth”.
For AI-based tests, this creates an important challenge: models must distinguish cancer-related patterns from signals associated with non-cancerous growths and inflammation.
Despite the findings, the researchers said the results should not be interpreted as proof that AI-assisted blood tests can reliably screen the general population.
Most cancer patients in the study had already developed symptoms and received a clinical diagnosis. The study was not a population-based screening exercise involving healthy people undergoing routine testing.
The authors warned that early-stage cancers detected in symptomatic patients could be biologically different from small, asymptomatic tumours identified through routine screening.
They said: “Prospective validation in a screening cohort will be required before these findings can inform MCED performance claims in that setting.”
MCED refers to multicancer early detection, or tests designed to identify signals from several cancers through a single blood sample.
The study also did not directly measure circulating tumour DNA fractions in its main cohort. Instead, researchers examined previously published datasets to investigate the relationship between total cell-free DNA and tumour-derived DNA.
Differences in DNA extraction and measurement methods could also affect results, making standardisation necessary before the approach can be widely used.
The paper’s authors concluded that total cell-free DNA concentration could be an important variable when evaluating the performance of multicancer detection tests.
The study was conducted under the EMERGE programme, with participants recruited between June 2022 and February 2025. Most authors are employees of Strand Life Sciences Private Limited, while two hold equity in the company. The paper also reports that an Indian patent was granted in June 2026 for a tool used in the research.
The findings offer a potential way to improve how blood-based cancer detection tests are evaluated. But whether these molecular signals and AI-based analysis can reliably identify cancers before symptoms appear, and ultimately improve patient outcomes, remains to be established.
Also Read:
Doors sold, wiring too: Cancer patient and sons’ desperate fight for survival in Keralam
Love your tea/coffee scalding hot? The habit might be tripling your risk of oesophageal cancer
Doctors sound alarm as ‘preventable’ colorectal cancer claims thousands of lives
(Edited by R Rajesh Kumar.)