A lab-on-a-chip called GradientScreen could help doctors see how a child’s leukaemia will respond to drug combinations at diagnosis, rather than after weeks of standard chemotherapy. The assay was developed by scientists and clinicians at Emory University, Children’s Healthcare of Atlanta and Georgia Tech, and is described in a paper in Nature Biomedical Engineering published on 23 October 2026. In a retrospective study of 17 childhood leukaemia samples, its results correlated with how patients actually fared on treatment.
The childhood leukaemia treatment problem
The study focuses on acute lymphoblastic leukaemia (ALL), a rapid-onset blood cancer that stops the body from producing mature, infection-fighting white blood cells. According to Emory, ALL accounts for 75% of all paediatric leukaemia diagnoses, and of about 6,000 new US cases each year, more than 3,500 are in children.
ALL needs immediate treatment, often a combination of chemotherapy, targeted therapy, immunotherapy and bone marrow transplants. Emory says clinicians currently have no way of knowing how a specific drug combination and dose will affect a given patient. Treatment is personalised to some degree by leukaemia subtype and risk group, but most patients still receive similar regimens. Roughly 15% of children with ALL experience relapse, according to the release.
Genomic sequencing has raised hopes of truly personalised therapy, but Emory notes that two patients can carry the same mutation and still respond very differently. That is why the team argues for functional tests that expose live cancer cells directly to treatments.
How GradientScreen works
GradientScreen is a microfluidic assay that measures how a patient’s leukaemia cells respond to drug combinations from a single blood or bone marrow sample. Inside the device, overlapping gradients of three drugs form across the cells, so each location holds a slightly different mix and ratio.
Rather than testing every drug pair and dose ratio one by one, the chip evaluates the full range of combination ratios within a three-drug regimen in a single experiment. By comparing how cells respond to single drugs versus combinations, it shows which combinations and ratios work better together. The result is a set of “response patterns” that reflect how a patient’s leukaemia might react to treatment. The paper’s supplementary material notes that the device needs over an order of magnitude fewer cells than a conventional well-plate assay for a comparable three-drug test.
What the 17-sample study showed
The researchers tested 17 leukaemia cell and bone marrow samples from ALL patients, provided by the Aflac Cancer and Blood Disorders Center of Children’s Healthcare of Atlanta. Using a statistical method called partial least squares regression, they found that GradientScreen’s combination response measures correlated with:
- minimal residual disease, the amount of leukaemia left after induction, the first month of chemotherapy
- later clinical outcomes, including whether the patient relapsed
The paper also shows the system could be used to test experimental drug regimens for patients who did not respond to standard treatment.
“One of the Holy Grails in oncology is the ability to identify which patient is likely to be unresponsive to standard treatment before initiation of therapy begins,” said Wilbur Lam, a professor of paediatrics and biomedical engineering at Emory and Georgia Tech, and the study’s principal investigator.
Limits and next steps
The study was retrospective and small, with 17 samples, and it reports correlations with outcomes rather than a test that has been used to choose treatment. The authors describe GradientScreen as a research-enabling tool with potential to fit into clinical pathology workflows.
Next, the team plans to expand to other leukaemia types, including acute myeloid leukaemia, and to adult as well as paediatric patients. It also plans to establish the assay within the Emory Lab for Innovative Assay Development, which supports diagnostic testing for patient care. The custom code has been released openly.
As a writer, I feel the idea of testing a child’s own cancer cells against a full spread of drug ratios is compelling, but 17 samples is a starting point. A larger, prospective study will show whether this can truly change treatment decisions.
The paper is dedicated to co-first author Dan Zhang, who passed away while it was being prepared.
FAQ
What is GradientScreen?
It is a microfluidic “lab-on-a-chip” from Emory, Children’s Healthcare of Atlanta and Georgia Tech that tests a patient’s leukaemia cells against a continuous range of three-drug combinations and ratios.
How was it tested in childhood leukaemia?
In a retrospective study of 17 ALL samples, its results correlated with residual disease after the first month of chemotherapy and with relapse.
Is GradientScreen used to treat children today?
Not yet. The authors call it a research tool, and the team is working towards bringing it into Emory’s lab for diagnostic testing.