Multiomic Map of 2,000 Colorectal Tumors Exposes Druggable Vulnerabilities

An international team led by RCSI University of Medicine and Health Sciences and The Institute of Cancer Research, London, has profiled roughly 2,000 colorectal tumors and used a purpose-built AI tool to expose vulnerabilities that drug combinations — including already-approved agents — could target in microsatellite stable (MSS) colorectal cancer, the hard-to-treat form that accounts for 80–85% of the 2,500 new bowel cancer cases diagnosed each year in Ireland.

The study, published in Cancer Cell under the title "Multiomic characterization of microsatellite-stable colorectal cancer reveals distinct biology and targetable vulnerabilities," integrated bulk genomics, epigenomics, proteomics and metabolomics with single-nucleus and single-cell RNA sequencing plus clinical parameters across 2,000 patient tumor samples. More than 300 of the MSS CRC samples had not been profiled before.

What did the AI tool actually find?

The researchers developed a machine-learning platform called iPhenMap to build biological maps of each tumor. Those maps linked genetic and metabolic features — connections that previous studies, examining each data type in isolation, had missed. The authors state plainly that no prior study has delivered a global integration of this breadth in the MSS CRC setting.

The mapping surfaced previously unidentified vulnerabilities in tumor cells: unrecognized mechanisms by which the cancer cells divide, grow, adapt and interact with their environment. The team then tested those vulnerabilities in laboratory models, where certain drug combinations worked well against the disease.

One concrete result: trametinib, a targeted drug already approved for melanoma, lung cancer and thyroid cancer with a specific mutation, significantly reduced tumor growth and extended survival in the in vivo models when paired with other drugs. That measured outcome is preclinical — no patient data yet exist for these combinations.

Why target MSS colorectal cancer?

MSS CRC responds poorly to many newer therapies, including immunotherapies, and trials combining chemotherapy with targeted drugs have delivered only modest gains over standard treatment. Survival in MSS CRC has not significantly improved in recent decades, the authors write. Between 40% and 50% of MSS cases harbor mutations in the RAS gene superfamily, and RAS-mutant patients consistently show inferior outcomes compared with RAS-wild-type counterparts — a gap of eight to ten percentage points in the authors' reported range.

The study focuses squarely on patients whose options have run out. "Patients with MSS RAS-mutant metastatic colorectal cancer urgently need better treatment options, particularly when their disease no longer responds to standard therapy," said research co-lead Professor Annette Byrne, PhD, head of the RCSI Precision Cancer Medicine Group. "While these findings remain preclinical, they highlight promising therapeutic opportunities and provide a strong foundation for further development. The next step will be to identify the biological features that predict response and determine how these approaches can be advanced safely towards clinical evaluation."

How solid is the dataset?

The scale and diversity of the cohort is the study's core asset, and it depends on European collaboration. "By bringing together data from patients across multiple countries and integrating DNA, RNA, epigenetic, protein, metabolic, single-cell and clinical data at this scale, we were able to expose connections between tumor biology and therapeutic dependency that would otherwise be missed," said co-lead and co-senior author Professor Anguraj Sadanandam, PhD, group leader of the Systems and Precision Cancer Medicine Group at the ICR. Rodrigo Dienstmann, PhD, of the Vall d'Hebron Institute of Oncology, Barcelona, and the University of Vic–Central University of Catalonia, was co-senior author.

Sadananadam also framed the wider methodological bet: AI and machine learning will increasingly let cancer researchers merge complex datasets not just to classify tumors, but to understand their behavior, identify vulnerabilities and guide drug development and repurposing.

What happens next?

For R&D managers, the work offers two parallel value streams. First, a repurposing roadmap: the team argues its results justify moving existing, approved drugs — trametinib among them — toward clinical testing in MSS CRC, a pathway that cuts early development cost and risk. Second, a reusable analytical platform: the group believes the iPhenMap approach can identify more effective drugs for other tumor types. The authors call the dataset "a rich resource for the CRC research community" and say the newly identified vulnerabilities may form "the foundation for precision oncology treatment frameworks aimed at improving outcomes in this intractable disease setting."

The immediate near-term task, per Byrne, is biomarker work: defining which biological features predict response before any combination reaches patients.