A cancer that behaves differently in every child
Some tumors are predictable. Neuroblastoma is not.
It is the most common solid cancer outside the brain in children, accounting for roughly 15% of all childhood cancer deaths. In some children, it disappears on its own without treatment. In others, it spreads rapidly and resists nearly every therapy doctors try.
What makes it so hard to treat is that it doesn't behave the same way twice. Two children with neuroblastoma can have tumors that look almost identical under a microscope but respond to treatment completely differently.
The old way of figuring out which child is which
For years, oncologists relied on standard imaging — CT scans and MRIs — combined with traditional biopsy and pathology to classify neuroblastoma and estimate risk. This approach worked reasonably well for many patients. But it often missed what was happening at the molecular level — the genetic switches inside the tumor driving its behavior.
A key example: a gene called MYCN. When MYCN is amplified (meaning extra copies are present inside the tumor), the cancer tends to be far more aggressive. Knowing whether MYCN is amplified matters enormously for treatment planning. But traditional imaging couldn't tell you that without an invasive biopsy.
A new layer of intelligence added to the toolbox
A review published in Frontiers in Medicine examined how a combination of new technologies — grouped under the umbrella of "multi-omics" — is changing the way neuroblastoma is understood and managed.
Multi-omics means combining multiple types of molecular data at once. Transcriptomics maps which genes are active inside a tumor. Single-cell sequencing looks at individual cancer cells to understand the diversity within one tumor. Radiomics extracts quantitative information from scans — texture, shape, metabolic signal — that the human eye can't detect. Digital pathology uses AI to analyze microscope images of tumor tissue at a scale no human pathologist could match alone.
How AI reads what doctors can't see
Think of a traditional scan as a photograph. Radiomics with AI is like having a program analyze every pixel of that photograph at 1000x zoom, looking for patterns invisible to the naked eye.
In neuroblastoma, AI tools analyzing MRI and PET-CT images have shown the ability to predict MYCN amplification status — without a needle — and to flag which patients are at higher risk of bone marrow metastasis before symptoms appear.
What the research covered
This review synthesized findings from recent studies applying transcriptomics, single-cell sequencing, radiomics, digital pathology AI, and multi-omics fusion to neuroblastoma diagnosis and treatment planning. The focus was on how combining these tools improves accuracy beyond what any single method can achieve alone.
No single technology outperformed the others across every task. But when these tools were combined — integrating genetic, imaging, and pathology data together — the resulting models classified tumors more precisely, predicted outcomes more accurately, and identified potential drug targets that would have been missed by any individual approach.
AI-driven analysis of digital pathology slides proved especially promising for identifying cellular diversity within tumors — which matters because different cell populations within the same tumor may respond differently to the same drug.
These tools are emerging — they are not yet the standard of care in most oncology centers, and many remain in the research and validation stage.
Where this fits in the bigger picture
Pediatric oncology has made remarkable strides over the past 50 years. Many childhood cancers that were once uniformly fatal now have high survival rates. Neuroblastoma remains one of the hardest targets — particularly in its high-risk form — and the field has been searching for smarter tools to close the gap.
Multi-omics approaches represent a shift from treating neuroblastoma as a single disease to treating it as a collection of related diseases, each requiring a tailored response.
What this means for families
If your child has been diagnosed with neuroblastoma, ask your oncologist whether genomic or molecular testing — including MYCN amplification status — is part of their diagnostic workup. At major pediatric cancer centers, many of these tools are already available as part of clinical care or research protocols.
Families can also look for clinical trials through resources like ClinicalTrials.gov, where studies using AI-assisted diagnostics and targeted therapies for neuroblastoma are increasingly active.
The limits of this research
This was a review article, not a clinical trial. The studies reviewed used different populations, different technology platforms, and different outcome measures — making direct comparisons difficult. Many of the AI and omics tools described are still being validated in prospective studies before they can reliably guide treatment decisions in routine clinical settings.
What comes next
The next major step is integration. Research groups are now building clinical decision platforms that bring together imaging, genetic, and pathology data in real time — allowing oncologists to see a unified picture of a tumor's behavior rather than consulting each data type separately. If validated in large prospective trials, these platforms could meaningfully change how neuroblastoma is classified and treated within the next five to ten years.