By: Alina Luchian and Dr. Istvan Petak, Founder and CSO – Genomate
Precision oncology has long promised to revolutionize cancer treatment by tailoring therapies to each tumor’s unique genetic profile. Yet more than a decade into this approach, results have been mixed. While some patients have seen dramatic benefits from targeted drugs, overall cancer mortality hasn’t shifted as much as hoped.
A growing number of studies now point to systemic challenges that are holding back progress: inconsistent expert recommendations, overly simplistic treatment models, low rates of actionable findings, and rising costs. To truly deliver on the promise of precision oncology, we need smarter, more scalable tools. Computational reasoning platforms could be the answer.
1. Inconsistent Recommendations and a Crisis of Confidence
Molecular Tumor Boards (MTBs) were designed to help doctors make informed decisions about complex genomic cases. But studies show that their conclusions can vary widely. In one multicenter study in Japan, doctors at different MTBs gave different treatment recommendations for the same patients. Overall agreement was around 62%, and dropped to as low as 18–30% when the genetic findings were less well understood. This kind of variability can shake patient confidence. If two top cancer centers suggest different treatments for the same person, how reliable is our current system?
Computational reasoning platforms aim to fix this. Instead of relying on subjective opinion, they use structured, logic-based systems to evaluate each case. These platforms apply consistent rules and incorporate the latest evidence, helping ensure that every patient gets a fair and objective treatment review.
2. “One Mutation, One Drug” Is No Longer Enough
Early success in precision oncology came from matching a single gene mutation to a specific drug. But most tumors aren’t that simple. In fact, many cancers are driven by multiple genetic alterations; studies suggest an average of four to five in solid tumors. The I-PREDICT trial showed that patients who received treatments addressing more than one driver mutation had better outcomes than those matched to just one.
Computational reasoning platforms examine the entire genetic picture. They analyze how different mutations interact and rank treatments accordingly. This leads to a more personalized and potentially more effective care plan.
3. Limited Benefit from Genomic Testing
Many cancer centers now routinely perform broad genomic testing, but the number of patients who actually benefit remains small. In the MOSCATO study, only about 15% of patients responded to targeted treatments. The NCI-MATCH trial reported meaningful results in just 3 of the first 11 treatment arms. This raises a hard question: are we overpromising? Most patients who undergo testing won’t end up receiving a helpful new treatment.
But there is hope. In May 2025, Nature Precision Oncology published a study by Dirner et al., analyzing 111 patients with advanced lung cancer using a computational reasoning model (Digital Drug Assignment, commercially known as Genomate). The findings were significant:
- Patients lived four times longer without disease progression when treated with Genomate’s top-ranked therapies
- 33% 5-year survival rate, using approved, available drugs
- More than 50% of patients received high-quality treatment matches
This is one of the clearest demonstrations to date that computational reasoning can expand therapeutic opportunities and predict clinical benefit.
4. Too Much Data, Not Enough Answers
Modern cancer sequencing uncovers millions of mutations across hundreds of genes. But only a small fraction have well-established links to effective treatments. Current guidelines usually evaluate one biomarker at a time, an approach that can’t keep up with the volume and complexity of today’s data. Doctors often face a difficult choice: act on limited evidence or do nothing.
Computational reasoning platforms help bridge this gap. They combine large-scale data sources (including real-world outcomes, lab research, and clinical trials) to build structured, transparent treatment rationales. These systems don’t erase uncertainty, but they make it easier to see, understand, and work with.
5. High Costs and Unequal Access
Precision oncology is expensive. Advanced sequencing costs thousands of pounds, and targeted treatments can run over £20,000 per month. Many are approved based on early signs like tumor shrinkage rather than proven survival benefits.
At the same time, many patients, especially those outside major medical centers or in lower-resource settings, struggle to access these tests or treatments.
Computational reasoning can help level the playing field. By providing expert-quality analysis through software, these tools make personalized care more accessible. A community oncologist can input a patient’s genetic data and receive a clear, evidence-ranked list of treatment options. This helps ensure that quality care isn’t limited to a few elite institutions.
The Goal? Turning Data Into Decisions
Precision oncology is not broken, but it is incomplete. The dream of matching every patient to the right therapy is still alive, but it won’t be achieved through sequencing alone. We need to transform how we interpret and act on molecular data. The evidence is mounting that computational reasoning platforms can provide consistent, actionable, and clinically meaningful insights.
If we want precision medicine to be more than a promise, we must equip clinicians with the intelligence to reason, not just react. The next era of oncology will not be defined by the data we collect, but by how we make sense of it.
If you’re interested in exploring how computational reasoning could improve outcomes for your patients, your organization, or your portfolio, we’d love to talk. Learn more about our precision oncology at genomate.health.
References:
Yasuko Aoyagi et al., Clinical utility of comprehensive genomic profiling in Japan: Result of PROFILE-F study, https://pmc.ncbi.nlm.nih.gov/articles/PMC8970371/
Jason K. Sicklick et al., Molecular profiling of cancer patients enables personalized combination therapy: the I-PREDICT study, Nature, https://www.nature.com/articles/s41591-019-0407-5
Christophe Massard et al., High-Throughput Genomics and Clinical Outcome in Hard-to-Treat Advanced Cancers: Results of the MOSCATO 01 Trial, https://pubmed.ncbi.nlm.nih.gov/28365644/
Peter J O’Dwyer et al., The NCI-MATCH trial: Lessons for precision oncology, https://pmc.ncbi.nlm.nih.gov/articles/PMC10612141/
Anna Dirner et al., Real-world performance analysis of a universal computational reasoning model for precision oncology in lung cancer, Nature, https://www.nature.com/articles/s41698-025-00943-4
