The biology underlying the approval is rooted in one of the most consequential discoveries in modern cancer genetics. DNA mismatch repair is a cellular quality-control system responsible for correcting errors that arise when DNA is copied during cell division. When the genes encoding this machinery, most commonly MLH1, MSH2, MSH6, or PMS2, are defective or silenced, errors accumulate in stretches of repetitive DNA known as microsatellites, producing a state called microsatellite instability. The resulting tumors carry an exceptionally high burden of mutations, and those mutations are translated into abnormal protein fragments that the immune system can recognize as foreign. This neoantigen-rich environment explains why dMMR tumors respond unusually well to immune checkpoint inhibitors, drugs that release the molecular brakes, such as the PD-L1/PD-1 axis targeted by atezolizumab, that cancers use to evade immune destruction. Paradoxically, however, dMMR tumors are less sensitive to fluoropyrimidine-based chemotherapy, the historical backbone of adjuvant treatment, leaving a critical therapeutic gap that the ATOMIC trial was designed to close.

Colorectal cancer remains the second leading cause of cancer-related death worldwide, and stage III disease, in which cancer has spread to nearby lymph nodes but not to distant organs, represents a pivotal and potentially curable stage. For decades, the standard adjuvant treatment for all patients with stage III colon cancer, regardless of molecular subtype, has been chemotherapy with a fluoropyrimidine combined with oxaliplatin, typically delivered as the FOLFOX regimen. Yet even with this treatment, a substantial proportion of patients experience recurrence, and the diminished sensitivity of dMMR tumors to fluoropyrimidines meant that this subgroup derived comparatively limited benefit from the existing standard. That unmet need provided a strong scientific rationale for a biomarker-driven approach, one that would match the distinctive immunogenicity of dMMR tumors with an agent capable of exploiting it, rather than treating all stage III colon cancers as a single, uniform disease.

The ATOMIC trial enrolled 712 participants between September 2017 and January 2023 across the United States and in Germany, a scale of accrual that investigators attribute to the infrastructure of the National Cancer Institute’s National Clinical Trials Network. Patients with stage III colon cancer and confirmed mismatch repair deficiency were randomized to receive either standard adjuvant chemotherapy alone or chemotherapy combined with atezolizumab. The trial was designed with disease-free survival as its primary endpoint, while overall survival and safety were designated as secondary endpoints, a structure that enabled a comprehensive assessment of both the efficacy and the tolerability of integrating immunotherapy into the adjuvant setting, where patients have already been rendered surgically disease-free and treatment is intended to prevent relapse rather than shrink measurable tumors.

The results, recently published in The New England Journal of Medicine, were striking. Adding atezolizumab to standard adjuvant chemotherapy reduced the risk of cancer recurrence or death by fifty percent, a magnitude of benefit rarely achieved in the adjuvant treatment of gastrointestinal malignancies. After three years of follow-up, 86.3 percent of patients who received the combination remained disease-free, compared with 76.2 percent of those who received chemotherapy alone. In practical terms, the immunotherapy combination translated into roughly one additional patient in ten remaining free of cancer three years after surgery, an absolute improvement with profound implications for a disease in which recurrence after stage III diagnosis frequently presages incurable metastatic spread. The ten-percentage-point separation between the curves, sustained over years, provided the evidentiary foundation on which the regulatory decision was built.

Leadership of the Alliance for Clinical Trials in Oncology framed the approval as both a clinical and an institutional milestone. Evanthia Galanis, MD, DSc, Group Chair of the Alliance, said that the approval of atezolizumab for patients with stage III dMMR colon cancer is a significant milestone for patients and for the organization, noting that the ATOMIC trial demonstrates the potential of introducing immunotherapy earlier in the course of cancer treatment, when the disease is still potentially curable, with the goal of preventing recurrence and increasing the number of patients who can be cured. She added that the trial also demonstrates the power of the National Cancer Institute’s National Clinical Trials Network to conduct large, definitive trials that answer important questions in cancer management and translate those findings into a new standard of care, and that the results show what is possible when investigators across the country come together around a common goal.

Frank A. Sinicrope, MD, the Alliance Study Chair for ATOMIC and a Professor of Oncology and Clinical Investigator of the Mayo Foundation at the Mayo Clinic Comprehensive Cancer Center, emphasized the clinical and scientific significance of the findings. He said the results provide compelling evidence that immunotherapy can play an important role in reducing the risk of recurrence and death after surgery for patients with stage III dMMR colon cancer, and that the FDA approval is an important step in bringing the benefits demonstrated in the trial into routine clinical practice. He also underscored a broader principle that is increasingly shaping oncology: the approval reinforces the importance of understanding the molecular characteristics of a patient’s tumor and using that information to guide treatment. He expressed gratitude to the patients and investigators who made the study possible and helped establish the new treatment regimen.

The trial’s execution reflected a public-private partnership model that has become central to modern cancer drug development. ATOMIC was sponsored by the National Cancer Institute, part of the National Institutes of Health, and led by the Alliance within the NCI-funded National Clinical Trials Network. Genentech provided support to the study through a Cooperative Research and Development Agreement with the NCI, an arrangement that allowed an academic-public consortium to evaluate a commercial immunotherapy agent in a biomarker-defined population while retaining the scientific independence and broad accrual capacity characteristic of network trials. The Alliance itself unites more than 26,000 cancer specialists at 112 main institutions and nearly 1,500 affiliates across the United States and Canada, and also serves as a leading research base for the NCI Community Oncology Research Program, enabling practice-changing trials to reach patients in both academic centers and community settings.

The broader implications extend well beyond a single drug approval. ATOMIC adds to a growing body of evidence that mismatch repair deficiency is one of the most predictive biomarkers in oncology, a molecular signature that reliably identifies patients likely to benefit from checkpoint blockade across disease stages and tumor types. By moving immunotherapy from the metastatic and locally advanced settings into the adjuvant treatment of early-stage colon cancer, the trial demonstrates that immune checkpoint inhibition can prevent recurrence, not merely control established disease, and that earlier intervention may increase the number of patients who are ultimately cured. The approval also validates the strategy of biomarker-driven trial design, in which a molecularly defined subgroup is studied prospectively rather than discovered retrospectively, and it reinforces the value of disease-free survival as an endpoint capable of supporting regulatory decisions when the biological link to overall survival is well established.

For patients, the practical consequence is a new standard of care: those diagnosed with stage III colon cancer will now undergo molecular testing for mismatch repair status as a routine part of treatment planning, and those whose tumors are dMMR will be eligible to receive atezolizumab alongside fluoropyrimidine and oxaliplatin chemotherapy after surgery. The full description of the clinical trial is available through ClinicalTrials.gov under identifier NCT02912559, and the primary results were published by Sinicrope and colleagues in The New England Journal of Medicine in March 2026. As immunotherapy continues its migration earlier into the treatment course of cancers with recognizable immune vulnerabilities, the ATOMIC trial and the approval it produced stand as a template for how large cooperative-group studies, molecular diagnostics, and public-private collaboration can converge to convert fundamental insights about tumor immunology into tangible gains in survival.

Subject of Research: Adjuvant immunotherapy with atezolizumab for stage III deficient DNA mismatch repair colon cancer based on the phase III ATOMIC trial

Article Title: Alliance announces FDA approval of atezolizumab based on results from the Phase III ATOMIC trial for patients with Stage III dMMR colon cancer

Article References: Alliance announces FDA approval of atezolizumab based on results from the Phase III ATOMIC trial for patients with Stage III dMMR colon cancer. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: atezolizumab, colon cancer, dMMR, microsatellite instability, immunotherapy, ATOMIC trial, adjuvant chemotherapy, FDA approval, clinical trial, Alliance for Clinical Trials in Oncology, National Cancer Institute, disease-free survival

The research team, led by Chao Yu and Siyu Liu with corresponding authors Yakun Wang and Xiaotian Zhang, assembled a cohort of 563 patients with stage III or IV gastric cancer or gastroesophageal junction adenocarcinoma, all of whom had paired tumor and normal tissue analyzed by targeted sequencing. The investigators call this the QZ563 cohort. Within it, a subset of 391 patients carried detailed clinicopathologic, biomarker, treatment, and survival information, forming the Clinical391 cohort used for deeper outcome analyses. By layering somatic mutations, tumor mutational burden, microsatellite instability status, metastatic patterns, and treatment-related biomarkers onto this clinical foundation, the team built a multidimensional map of advanced gastric cancer as it actually presents in Chinese hospitals.

The mutational landscape that emerged was dominated by familiar names. TP53 was altered in 70.8 percent of tumors, cementing its role as the single most frequently disrupted gene in advanced gastric cancer. Alongside it, ARID1A, CDH1, and PIK3CA ranked among the most commonly mutated genes, a pattern consistent with prior genomic surveys of the disease. But the study’s real contribution lies not in cataloging the usual suspects. Instead, the researchers asked how clinically actionable biomarkers carve up the patient population, and whether the resulting categories carry meaningful biological and prognostic differences that standard staging alone cannot capture.

One of the most clinically consequential findings concerns a small but important subgroup: patients whose tumors lack any of the established treatment-related biomarkers. The team labeled this group treatment-biomarker-negative gastric cancer, and their analysis revealed that these tumors are molecularly heterogeneous rather than uniformly indolent or uniformly aggressive. In other words, patients who test negative for actionable markers are not a single biological entity; they represent a mixed collection of genomic states that current biomarker panels do not resolve. The authors are careful to note that this subgroup warrants further validation, but the implication is provocative. A negative biomarker result may conceal as much biological diversity as a positive one, and refining how these patients are characterized could eventually open new therapeutic doors.

Metastatic phenotype added another layer of clinically relevant heterogeneity. By comparing tumors from patients with peritoneal involvement against those with liver metastases, the researchers identified distinct patterns of pathway-level mutation preference, suggesting that the route a gastric cancer takes when it spreads is written, at least in part, into its genome. This distinction matters because peritoneal and liver metastases carry different prognoses and respond differently to systemic therapy. A genomic framework that recognizes metastatic phenotype as a biologically meaningful axis, rather than a mere anatomical detail, could help clinicians anticipate disease trajectories and select treatments more rationally for stage IV patients.

The study’s most striking molecular discovery, however, centers on two chromatin-regulating genes: KMT2C and KMT2D. Loss-of-function alterations in these genes, which disrupt histone methyltransferase complexes that help orchestrate gene expression, were present in 6.4 percent of the QZ563 cohort. That modest frequency conceals a dramatic association. Among tumors carrying KMT2C or KMT2D loss-of-function changes, 48.3 percent were microsatellite instability-high, compared with just 2.7 percent of tumors lacking these alterations, a difference the authors report as highly statistically significant. In plain terms, when the epigenetic machinery governed by KMT2C and KMT2D breaks down, the genome’s mismatch repair system appears far more likely to falter as well, unleashing the mutational storm that defines MSI-high disease.

Crucially, the association did not stop at microsatellite status. Within both microsatellite-stable and MSI-high strata, KMT2C/D loss-of-function tumors showed higher tumor mutational burden than their wild-type counterparts, indicating that these chromatin alterations push genomic instability upward even within established MSI categories. The researchers then cross-checked this pattern against The Cancer Genome Atlas stomach adenocarcinoma dataset, known as TCGA-STAD, and found the same molecular signature, lending independent support to the finding. Yet a cautionary note accompanies the excitement: KMT2C/D loss-of-function status was not associated with a survival benefit in this cohort. High mutational burden is often assumed to predict better responses to immunotherapy, but the data here suggest that KMT2C/D-linked hypermutation alone does not translate into improved outcomes, at least within this retrospective framework.

To move from association toward therapeutic hypothesis, the team turned to DepMap, a large-scale cancer dependency resource that catalogs which genes cancer cell lines cannot survive without. Their analyses nominated WRN, a DNA helicase gene, as a candidate dependency in KMT2C/D loss-of-function MSI models. WRN dependency in MSI-high cancers has emerged in recent years as one of the most tantalizing synthetic lethal opportunities in oncology, and this study extends that concept into the KMT2C/D-altered subset. The authors are explicit that this remains hypothesis-generating and requires functional validation before any clinical translation, but the nomination of a druggable vulnerability within a molecularly defined subgroup exemplifies how large clinical-genomic datasets can seed the next generation of precision trials.

The study also integrated transcriptomic data from TCGA-STAD to probe what KMT2C/D loss-of-function means biologically within the MSI background, examining immune-related and non-immune signature scores between altered and wild-type tumors. Sensitivity analyses confirmed the directional concordance of co-mutation patterns between the Chinese cohort and TCGA, and the researchers stratified their TMB comparisons by MSI status to ensure the association was not simply an artifact of hypermutation. Methodologically, the work illustrates the value of pairing a well-annotated single-institution cohort with public genomic resources: the clinical depth of the Chinese data anchors the findings in real-world patient care, while the external datasets guard against cohort-specific artifacts.

For patients and clinicians, the takeaway is twofold. First, biomarker testing in advanced gastric cancer is not merely a pass-fail gate; the clinical and genomic context surrounding a biomarker result, including metastatic pattern and chromatin gene status, carries information that current decision trees largely ignore. Second, the KMT2C/D-MSI axis offers a potential new lens for subdividing the immunotherapy-sensitive population, one that could eventually explain why some hypermutated tumors respond spectacularly to checkpoint inhibitors while others do not. The study, approved by the Ethics Committee of Peking University Cancer Hospital and funded by Chinese national and municipal science programs, is retrospective by design, and the authors themselves frame the WRN finding as requiring experimental confirmation. Even so, in a disease where one in five patients may harbor molecular features invisible to standard panels, this 563-patient genomic atlas marks a meaningful step toward treatment maps that reflect the full genetic complexity of gastric cancer.

Subject of Research: Clinical-genomic stratification of treatment biomarkers and KMT2C/D-associated microsatellite instability heterogeneity in advanced Chinese gastric cancer

Article Title: Clinical-genomic stratification of treatment biomarkers and KMT2C/D-linked MSI heterogeneity in Chinese gastric cancer: a retrospective cohort study

Article References: Yu, C., Liu, S., Zhou, Z., Qin, N., Yang, L., Wang, J., Ding, M., Chong, X., Wang, Y., & Zhang, X. (2026). Clinical-genomic stratification of treatment biomarkers and KMT2C/D-linked MSI heterogeneity in Chinese gastric cancer: a retrospective cohort study. BMC Cancer. https://doi.org/10.1186/s12885-026-17051-6

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17051-6

Keywords: gastric cancer, KMT2C, KMT2D, microsatellite instability, tumor mutational burden, WRN dependency, treatment biomarkers, tumor heterogeneity, precision oncology, targeted therapy, immunotherapy biomarkers, retrospective cohort study

The urgency behind the effort is easy to grasp. Intrahepatic cholangiocarcinoma accounts for roughly 8 to 15 percent of all primary malignant liver tumors, second only to hepatocellular carcinoma, and its incidence has been climbing. Five-year overall survival sits at approximately nine percent. Surgery remains the only curative option, yet 70 to 80 percent of patients arrive at the clinic with disease that is either locally unresectable or already metastatic. For them, systemic therapy can delay progression but typically extends survival to only about a year. Against that grim backdrop, the discovery that roughly 40 to 50 percent of these tumors carry actionable genetic alterations has transformed the conversation, turning molecular profiling from an academic exercise into a clinical necessity.

The guideline, developed under the auspices of the Chinese Anti-Cancer Association’s liver cancer and pathology societies with international co-authors from Singapore, Australia, the United States and China, was registered on a transparency platform for practice guidelines and graded its evidence using the GRADE system. Its central message is blunt: molecular testing is recommended for all patients with intrahepatic cholangiocarcinoma, and it is essential for those with unresectable or metastatic disease, because the tumor’s genetic landscape differs markedly from that of extrahepatic bile duct cancers and gallbladder cancers, and even different pathological subtypes of the tumor behave differently at the DNA level.

At the top of the target list sits FGFR2, a receptor tyrosine kinase gene that is rearranged or fused in between 6.6 and 20 percent of Chinese patients with the disease, particularly in the small-duct subtype. These fusions typically break the gene between exons 17 and 19, leaving the receptor’s kinase domain intact while deleting regulatory elements that normally switch the receptor off, resulting in constitutive growth signaling. More than 140 partner genes have been identified, with BICC1 the most frequent. Two drugs, pemigatinib and futibatinib, are now approved by regulators in the United States and China for previously treated patients whose tumors harbor FGFR2 fusions, and both are recommended as second-line options in major treatment guidelines.

Choosing the right detection method for FGFR2 turns out to matter enormously, and the consensus devotes unusual technical detail to the question. Fluorescence in situ hybridization with break-apart probes can flag rearrangements but cannot identify fusion partners and may miss closely spaced intrachromosomal events. DNA-based next-generation sequencing can simultaneously detect mutations, amplifications and fusions across many genes, but it cannot confirm that a detected fusion actually produces a functional RNA transcript. RNA-based sequencing, by contrast, provides direct evidence of functional fusions and can uncover novel partners, with concordance between the two sequencing approaches reaching 98 percent. The panel therefore recommends combining DNA- and RNA-based sequencing, reserving FISH as a fallback when sequencing is unavailable, and explicitly discourages FGFR2 immunohistochemistry, which shows poor agreement with molecular methods. In a striking practical touch, the guideline even borrows interpretation thresholds from ALK testing in lung cancer, since no standardized cutoff for FGFR2 break-apart positivity exists.

A second pillar of the guideline concerns IDH1, a metabolic enzyme whose mutations occur in 4.9 to 20 percent of Chinese patients, again concentrated in the small-duct subtype. The inhibitor ivosidenib received United States approval in 2021 for previously treated, IDH1-mutant cholangiocarcinoma, with a companion diagnostic test cleared alongside it. Mutations cluster at position 132, most commonly R132C, a detail with real diagnostic consequences: the commercial immunohistochemistry antibody targets the R132H variant common in gliomas and cannot recognize R132C, limiting staining’s usefulness here. Sequencing, preferably by next-generation platforms that can capture multiple loci, is the preferred route, and the panel notes that secondary resistance mutations such as D279N, or oncogenic IDH2 mutations like R172K, can emerge under treatment pressure, making comprehensive sequencing valuable even after therapy begins.

The guideline then marches through a roster of additional targets. BRAF V600E, present in a subset of the 4.2 percent of Chinese patients with BRAF mutations, is sensitive to the approved dabrafenib-plus-trametinib combination, while non-V600 variants respond to MEK inhibitors but not BRAF inhibitors, so the panel urges attention beyond the flagship site. HER2 overexpression and ERBB2 amplification, found in 1.8 to 8 percent of patients, open doors to trastuzumab-based regimens including trastuzumab deruxtecan, with immunohistochemistry prioritized and equivocal cases confirmed by FISH or sequencing, interpreted for now by adapting breast and gastric cancer criteria. Rarer but druggable alterations receive their due as well: NTRK fusions in under one percent of patients, RET fusions in 1.8 percent, KRAS mutations in 12.4 to 25 percent, and NRG1 fusions in roughly two percent, each with preferred platforms, mostly RNA-based sequencing for fusions and broad DNA panels for point mutations.

Immune checkpoint eligibility also earns a formal recommendation. Deficient mismatch repair or high microsatellite instability, present in 1.6 to 6 percent of Chinese patients, predicts response to immunotherapy, and several checkpoint inhibitors are approved for such tumors in both the United States and China. The panel endorses mismatch repair immunohistochemistry or polymerase chain reaction-based microsatellite testing as primary methods, and adds a cautionary note drawn from a study of 1,942 solid tumors: sequencing-based microsatellite calls are fully concordant with conventional methods only at the extremes, so borderline results must be validated by immunohistochemistry or PCR before treatment decisions rest on them.

Equally pragmatic are the recommendations about samples themselves. Because intrahepatic cholangiocarcinoma is stroma-rich, tumor cell content in biopsies is often low; in one series of 123 advanced biliary tract cancers, more than a quarter of samples were unsuitable for sequencing due to insufficient tumor content. The consensus requires pathologists to verify at least 20 percent tumor cellularity and a minimum of 50 tumor cells before testing, and urges clinicians to obtain enough tissue in a single procedure for both diagnosis and molecular workup. Tissue remains the gold standard, with cytology cell blocks as the fallback and liquid biopsy of circulating tumor DNA as a last resort at accredited laboratories, since concordance between blood and tissue varies dramatically, from 87 to 100 percent for point mutations down to just 18 percent for FGFR2 fusions. Primary lesions are preferred for initial testing, though metastatic sites may be sampled when the primary is inaccessible, and repeat biopsy after progression on targeted therapy is explicitly encouraged to map resistance mechanisms.

The panel closes by sorting biomarkers into essential and optional categories, the latter including emerging targets such as PTEN loss, Claudin 18.2 expression and BRCA1/2 mutations that may guide trial enrollment, and it commits to periodic revisions as new drugs and data accumulate. For a cancer with a nine percent five-year survival rate, the stakes of getting testing right could hardly be higher. What this guideline offers is a shared playbook: a single, evidence-graded document telling laboratories and clinicians worldwide which genes to interrogate, which platforms to trust, how to interpret ambiguous signals, and when to re-biopsy, so that every patient with this aggressive tumor has the best possible chance of finding a therapy matched to the specific molecular engine driving their disease.

Subject of Research: Precision molecular biomarker testing guidelines for intrahepatic cholangiocarcinoma

Article Title: Guideline of precisional testing in intrahepatic cholangiocarcinoma: an international expert consensus

Article References: Zhang, X., Han, J., Shi, R., Yu, B., Zhang, X., Li, B., Sheng, X., Li, Z., Zou, Y., Sun, H., Shi, G., Wang, H. L., Zhou, J., Fan, J., Cong, W., & Ji, Y. (2025). Guideline of precisional testing in intrahepatic cholangiocarcinoma: an international expert consensus. Clinical Cancer Bulletin, 4(1), Article 9. https://doi.org/10.1007/s44272-025-00036-0

Image Credits: AI Generated

DOI: 10.1007/s44272-025-00036-0

Keywords: intrahepatic cholangiocarcinoma, molecular testing, FGFR2 fusions, IDH1 mutations, next-generation sequencing, targeted therapy, biomarkers, precision oncology, bile duct cancer, HER2, microsatellite instability, expert consensus

The motivation behind the study lies in a practical bottleneck of modern oncology. Molecular profiling, which guides targeted therapies and immunotherapy decisions, currently depends on genomic or transcriptomic assays such as next generation sequencing. These tests are expensive, require specialized infrastructure, and can add days or weeks to the diagnostic timeline, a delay that matters enormously for patients with aggressive disease. If a machine could reliably estimate molecular features from the slide that is already being examined for diagnosis, the turnaround time for biomarker information could shrink dramatically and the cost per test could fall to nearly nothing beyond the computational expense. That is the promise the authors set out to test on a pan cancer scale rather than in a single tumor type.

The system, whose name stands for Rapid and Intelligent Detector for Genetic Estimation, is built on weakly supervised learning, a strategy that sidesteps one of the most stubborn obstacles in computational pathology. Whole slide images are gigantic, often exceeding one hundred thousand by one hundred thousand pixels, and labeling specific regions of interest by hand is prohibitively laborious. RIDGE instead learns from slide level labels only, using a multiple instance learning formulation in which the slide is treated as a bag of smaller tissue patches and only the overall slide carries a molecular label. The architecture employs fully convolutional networks to aggregate patch level information, allowing the model to identify which morphological patterns are associated with a given biomarker without ever being told where to look.

Several technical choices distinguish the framework. The authors incorporate clustering guided contrastive learning, a self supervised pretraining approach that teaches the network to group visually and biologically similar tissue patterns together before any biomarker prediction begins. Attention mechanisms, including a focused linear attention module, allow the model to weigh the contribution of thousands of patches efficiently, while depthwise convolutions and a mixture of experts design help the network specialize across the heterogeneous landscape of tumor types. Multi task learning enables a single model to predict multiple biomarkers and gene expression signatures simultaneously, sharing learned representations across related prediction problems. The result is a general purpose system intended to work across cancers rather than a bespoke model for each disease.

The training data were substantial. The team developed and validated RIDGE using 4,983 whole slide images from 4,680 patients spanning 12 solid tumor types in The Cancer Genome Atlas, including breast, lung, colon, rectal, stomach, liver, pancreatic, cervical, head and neck, and three kidney cancer cohorts. Performance was measured with the area under the receiver operating characteristic curve, the standard metric for binary classification tasks in medicine. Across all 12 cancer types, RIDGE achieved an overall AUC of 0.763, with a 95 percent confidence interval of 0.724 to 0.802. In a field where individual biomarker prediction models often hover in a similar range, a single framework reaching this level across such a diverse set of tumors and molecular targets is a meaningful benchmark.

Perhaps the most important result concerns generalization beyond the training data. Machine learning models in medicine frequently fail when moved to new datasets, a phenomenon driven by differences in staining protocols, scanners, and patient populations. To test reproducibility, the researchers applied RIDGE to an external validation cohort of 221 whole slide images from 105 colorectal cancer patients in the Clinical Proteomic Tumor Analysis Consortium, a completely independent resource, asking the model to predict microsatellite instability status. MSI is a critical biomarker because microsatellite unstable tumors respond well to immune checkpoint inhibitors. RIDGE achieved an AUC of 0.769, with a confidence interval of 0.686 to 0.841, closely matching its internal performance and suggesting the learned features reflect genuine biology rather than dataset specific artifacts.

Beyond raw accuracy, the authors emphasize interpretability, a persistent concern for clinicians asked to trust black box algorithms. The study reports that the model captures morphological visual characteristics that make gene expression signatures detectable directly from the slides, and supplementary analyses include explainability experiments examining how molecular features manifest in gastric cancer tissue. In effect, the network learns to associate particular cellular and tissue architectures, such as the appearance of tumor infiltrating immune cells, gland formation patterns, or nuclear features, with underlying molecular states. This capacity to quantify genotype phenotype relationships from images could itself become a research tool, allowing investigators to map molecular biology onto tissue morphology at a scale that manual review could never achieve.

The clinical implications, if the approach matures, are considerable. Because H&E stained slides are universally produced, an AI layer on top of standard pathology workflows could provide preliminary biomarker estimates within minutes of slide scanning, flagging patients who should receive priority for confirmatory molecular testing. In hospitals without access to sequencing facilities, such predictions could guide referral decisions and broaden equitable access to precision oncology. The authors argue that RIDGE could significantly expedite cancer screening and personalized therapy, and the pan cancer design means a single deployment could serve pathology departments handling many tumor types rather than requiring separate pipelines for each indication.

Caution is nonetheless warranted before such systems reach the clinic. An AUC of roughly 0.76, while respectable, indicates imperfect discrimination, meaning the model would need to function as a triage and prioritization tool rather than a replacement for definitive molecular assays. The study relied on publicly available, de identified data from TCGA and CPTAC, which, although multi institutional, may not capture the full heterogeneity of staining practices and patient demographics seen in routine global practice. Prospective clinical validation, regulatory review, and demonstration of impact on patient outcomes remain necessary steps. The published work was exempt from additional ethics approval because it used existing consented datasets, but real world deployment would face a new and more demanding evaluation landscape.

Even with those caveats, the study adds to a rapidly growing body of evidence that the humble pathology slide is an information rich object whose molecular content can be unlocked computationally. By demonstrating a single weakly supervised framework that predicts multiple biomarkers across a dozen cancer types and reproduces its performance on external data, the RIDGE team has moved the field closer to a future in which every diagnostic slide yields both a morphological diagnosis and a molecular profile. For patients, that could mean faster answers at lower cost; for researchers, a powerful new lens on the relationship between how a tumor looks and what it is, at the level of its genes.

Subject of Research: Deep learning prediction of pan-cancer molecular biomarkers from H&E-stained whole slide pathology images

Article Title: Deep learning-based large-scale pan-cancer multiple biomarkers prediction using RIDGE with pathological images

Article References: Xi, H., Feng, X., Lu, Y., Li, G., Zhang, Y., Li, J., Wang, Y., Xu, J., Zhang, Y., Sha, C., & He, M. (2026). Deep learning-based large-scale pan-cancer multiple biomarkers prediction using RIDGE with pathological images. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02695-4

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02695-4

Keywords: computational pathology, deep learning, whole slide imaging, biomarker prediction, weakly supervised learning, multiple instance learning, microsatellite instability, TCGA, CPTAC, precision oncology, gene expression, pan-cancer

In an effort to improve on those dismal numbers, veterinary oncologists have tested a series of combination chemotherapy protocols incorporating agents such as vincristine, cyclophosphamide, and dacarbazine alongside doxorubicin. None has proven clearly superior to single-agent doxorubicin, and most carry greater toxicity. One combination, however, generated genuine excitement: doxorubicin paired with dacarbazine, administered as a true concurrent combination for dogs with advanced-stage hemangiosarcoma, produced an objective response rate of 47.4 percent in a prospective study, although its impact on overall survival remains undefined. The catch is logistical and toxic. Dacarbazine must be given as an eight-hour intravenous infusion every three weeks, and beyond the usual myelosuppressive risks of neutropenia and thrombocytopenia, it is notorious for causing acute gastrointestinal upset that requires prophylactic antiemetic therapy.

That is where temozolomide enters the story. Temozolomide is an oral chemotherapeutic closely related to dacarbazine, but far easier to administer and generally well tolerated in dogs, with favorable safety data already available when it is combined with doxorubicin in canine lymphoma patients. In human medicine, temozolomide is the first-line chemotherapy for glioblastoma multiforme, a deadly brain cancer, and its mechanism of action is well characterized. Once in the bloodstream, the drug spontaneously hydrolyzes into its active metabolite, 5-(3-methyl-1-triazeno) imidazole-4-carboxamide, or MTIC. MTIC methylates nucleotide bases in DNA, with guanine as the most important target. The cell’s mismatch repair pathway then detects these methylated bases and attempts to repair them in a futile cycle that ultimately triggers apoptosis, the programmed death of the cancer cell.

That mechanism also explains why some tumors resist the drug. In cells with an intact mismatch repair system, sensitivity to temozolomide is governed largely by a DNA repair protein called O-6-methylguanine-DNA methyltransferase, or MGMT. MGMT works by directly removing the methyl groups that temozolomide installs, effectively erasing the damage before it can become lethal. When MGMT is overexpressed, the drug’s fingerprints are wiped away almost as fast as they are written, and the tumor survives. In human glioblastoma, this relationship is so reliable that testing MGMT status, particularly epigenetic silencing of the MGMT promoter, is routinely used to decide whether temozolomide is a viable option for a patient. The question for veterinary researchers was whether the same logic could be imported into canine hemangiosarcoma: if MGMT expression varied among tumors, it might serve as a biomarker to identify which dogs would benefit from adding temozolomide to doxorubicin.

A new study led by Brianna Moore, Julie Nguyen-Edquilang, and Matthew R. Berry at the University of Illinois set out to test exactly that hypothesis in the laboratory. The team worked with five canine hemangiosarcoma cell lines, named DHSA-1426, Emma Brain, Emma Spleen, FITZ, and SBHSA, along with a canine aortic endothelial cell line as a comparison. The cell lines had been shared by collaborators at the University of Minnesota, Colorado State University, and the University of Wisconsin-Madison. To anchor their measurements, the researchers used a human cervical cancer cell line, HeLa, and a pair of mouse glioma cell lines, one wild type and one engineered to carry a human MGMT knock-in, as controls. All cells were cultured under standard conditions at 37 degrees Celsius in a humidified incubator with five percent carbon dioxide.

The researchers measured MGMT at two levels. Quantitative PCR, using canine-specific TaqMan primers with GAPDH as the reference gene, quantified MGMT transcript abundance across three biological replicates with five technical replicates each. Western blotting, performed with a rabbit monoclonal antibody and normalized to the loading control beta-actin, assessed protein expression. The two approaches told a generally concordant story. Four of the five hemangiosarcoma cell lines, DHSA-1426, Emma Brain, Emma Spleen, and SBHSA, showed low or absent MGMT transcript and protein. FITZ was the striking exception, displaying the highest MGMT expression of the panel, roughly twenty-fold greater protein expression than the canine endothelial cell line based on normalized densitometry. The protein bands appeared at the expected sizes, approximately 37 kilodaltons in human controls and approximately 40 kilodaltons in canine cells, a slight species difference confirmed against UniProt database entries.

With the expression map in hand, the team turned to cytotoxicity testing using the sulforhodamine B assay. Cells were seeded in 96-well plates, pretreated for two hours with O6-benzylguanine, an inhibitor that depletes the functional pool of MGMT, and then exposed to temozolomide at concentrations ranging from 0.033 micromolar to 1000 micromolar for seven days. Because temozolomide is unstable in solution, it was dissolved fresh immediately before each experiment. Cytotoxicity values were normalized against a DMSO live-cell control and a raptinal control that induces complete cell death, allowing relative responses to be compared across conditions. The control cell lines behaved exactly as biology predicts. GL261 wild-type cells, which lack MGMT, were sensitive to temozolomide even without the inhibitor, while GL261 MGMT knock-in cells were strongly resistant, with IC50 values comparable to the hemangiosarcoma lines, until O6-benzylguanine pretreatment sensitized them, producing the largest drop in IC50 of any line tested.

The hemangiosarcoma results defied expectations. Every canine cell line proved resistant to temozolomide across the entire tested concentration range, regardless of MGMT status. FITZ, the only line expected to be sensitized by MGMT inhibition based on its high expression, remained stubbornly resistant, showing an undulating dose-response curve with no concentration-dependent effect in any of three biological replicates, and no reliable IC50 could be determined. DHSA-1426 likewise showed minimal cytotoxic response across the range, precluding curve fitting. In other words, MGMT expression status could not predict temozolomide sensitivity in canine hemangiosarcoma cells, and blocking MGMT did not rescue drug sensitivity even where the protein was abundant. The data point instead to an MGMT-independent resistance mechanism, and the authors hypothesize that mismatch repair deficiency is the culprit.

That hypothesis has independent support. A comprehensive analysis of microsatellite instability in canine cancers, a hallmark of mismatch repair defects marked by mutations accumulating in repetitive DNA sequences, found a surprisingly high incidence of 63 percent across canine tumors, with hemangiosarcoma among the tumor types showing elevated instability. If canine hemangiosarcoma is indeed frequently mismatch repair deficient, the implications extend well beyond temozolomide. Mismatch repair deficiency also means deficient methylated DNA goes unnoticed by the cell, undermining the drug’s lethal futile-repair cycle. More intriguingly, tumors with high microsatellite instability carry increased mutational burden and therefore heightened immunogenicity, making them particularly susceptible to immune checkpoint blockade. The authors suggest this provides biological rationale for moving beyond chemotherapy combinations entirely and exploring chemo-immunotherapy, pairing doxorubicin with agents such as PD-1 or PD-L1 inhibitors.

The study’s conclusions are a sober but valuable course correction. MGMT remains a highly promising predictive biomarker for temozolomide selection in human glioblastoma, but this in vitro investigation demonstrates that canine hemangiosarcoma cell lines resist the drug through a mechanism that MGMT testing cannot capture. On that basis, the authors conclude the data do not support clinical exploration of temozolomide, alone or combined with doxorubicin, for treating dogs with hemangiosarcoma. Instead, the findings redirect attention toward characterizing mismatch repair deficiency in canine hemangiosarcoma and toward immunotherapy-based strategies that exploit the high mutational burden such deficiency creates. For a disease that claims most of its canine victims within a year of diagnosis, a result that rules out one dead-end combination while pointing toward a biologically grounded alternative is a meaningful step forward for comparative oncology.

Subject of Research: MGMT expression and temozolomide resistance mechanisms in canine hemangiosarcoma cell lines

Article Title: Investigating MGMT expression as a resistance mechanism to temozolomide in canine hemangiosarcoma cell lines

Article References: Moore, B., Nguyen-Edquilang, J., & Berry, M. R. (2026). Investigating MGMT expression as a resistance mechanism to temozolomide in canine hemangiosarcoma cell lines. Veterinary Oncology, 3(1), Article 24. https://doi.org/10.1186/s44356-026-00076-1

Image Credits: AI Generated

DOI: 10.1186/s44356-026-00076-1

Keywords: canine hemangiosarcoma, temozolomide, MGMT, doxorubicin, chemotherapy resistance, mismatch repair deficiency, microsatellite instability, veterinary oncology, DNA repair, immunotherapy, cell lines, dog cancer

PD-L1 is a ligand that sits on the surface of tumor cells and other cells in the tumor microenvironment. When it engages its receptor, PD-1, on T lymphocytes, it dampens the immune attack, allowing malignant cells to survive and proliferate. Immune checkpoint inhibitors, or ICIs, are antibodies that block this interaction, releasing the brakes on T cells. These drugs have transformed the treatment of several cancers, but their effectiveness in gastric cancer varies dramatically between patients. Understanding what drives PD-L1 expression in gastric tissue is therefore not just an academic question; it may determine who benefits from these expensive and sometimes toxic therapies.

The review’s central argument is that H. pylori does not push a single button to raise PD-L1. Instead, it manipulates an interlocking network of signaling cascades, with distinct bacterial virulence factors acting as the ignition keys. The two most important of these are cytotoxin-associated gene A, or CagA, and vacuolating cytotoxin A, or VacA. CagA is delivered into gastric epithelial cells through a syringe-like molecular machine known as the type IV secretion system, and once inside it hijacks multiple host signaling routes. VacA, meanwhile, exerts its own effects on cellular signaling and immune function. The review emphasizes that different virulent strains of H. pylori vary in how strongly they drive PD-L1 upregulation, which may partly explain why infection outcomes differ so widely between individuals.

Among the pathways the authors dissect, the nuclear factor kappa-light-chain-enhancer of activated B cells, or NF-κB, route takes center stage. NF-κB is a master transcription factor of inflammation, normally held inactive in the cytoplasm until bacterial products or inflammatory cues activate it. When H. pylori infection triggers NF-κB, the transcription factor translocates into the nucleus and binds to the promoter region of the gene encoding PD-L1, directly boosting its transcription. This provides a direct mechanistic bridge between chronic bacterial inflammation and immune checkpoint expression, and it explains why long-standing H. pylori-associated gastritis can create a microenvironment that is already immunosuppressed before cancer even develops.

Parallel to NF-κB runs the mitogen-activated protein kinase, or MAPK, cascade, another signaling highway activated by CagA and other virulence determinants. The MAPK pathway funnels signals from the cell surface through a chain of kinases to transcription factors that can also enhance PD-L1 expression. In addition, the review highlights the Janus kinase and signal transducer and activator of transcription 1 route, commonly abbreviated as JAK/STAT1. Inflammatory cytokines released during infection activate STAT1, which again can drive PD-L1 transcription, while regulatory elements such as suppressor of cytokine signaling 3, or SOCS3, modulate how long and how intensely the signal persists. These pathways are not independent wires; the authors stress that they crosstalk extensively, amplifying and modulating one another so that PD-L1 expression reflects the integrated state of multiple signaling circuits rather than any single input.

Perhaps the most intriguing thread in the review is the involvement of the Sonic Hedgehog pathway, a developmental signaling system best known for sculpting embryos but increasingly implicated in adult cancers and tissue repair. Through its downstream Glioma-associated oncogene homolog, or GLI, transcription factors, Sonic Hedgehog signaling appears to participate in the H. pylori-driven upregulation of PD-L1, linking a developmental program to immune evasion in the stomach. The review also points to the phosphoinositide 3-kinase and protein kinase B axis, often extended to the mammalian target of rapamycin, or PI3K/AKT/mTOR, as another contributor to PD-L1 regulation. On the protein level, molecules such as CKLF-like MARVEL transmembrane domain-containing protein 6, or CMTM6, which stabilizes PD-L1 on the cell surface, add a further layer of control. Together these findings sketch a regulatory network in which transcriptional activation, pathway crosstalk and post-translational stabilization all converge on a single immune checkpoint molecule.

What makes the review particularly valuable is its insistence on the dual-edged nature of H. pylori-induced PD-L1 overexpression. On one hand, high PD-L1 in the gastric tumor microenvironment is a marker of immune suppression, a sign that the bacteria and the tumor have successfully blinded the local immune system. On the other hand, elevated PD-L1 can serve as a predictor of sensitivity to immune checkpoint inhibitors, because tumors that are actively using the PD-1/PD-L1 brake are precisely the tumors most likely to respond when that brake is removed. This duality complicates clinical decision-making but also creates opportunity: measuring PD-L1 and understanding how it got there could help clinicians identify which gastric cancer patients are most likely to benefit from ICIs.

The review deepens this picture by examining how two additional molecular features modify and amplify the duality. Epstein-Barr virus infection of tumor cells, one of the recognized molecular subtypes of gastric cancer, is associated with particularly high levels of PD-L1 expression, potentially compounding the effects of H. pylori-driven signaling. Likewise, microsatellite instability, a form of hypermutation that produces abundant abnormal proteins, correlates with higher tumor mutational burden and generally better responses to immunotherapy. The review examines how these factors interact with the bacterial program, suggesting that a stratification of gastric cancer patients by H. pylori status, EBV status and microsatellite stability could yield more precise predictions of ICI benefit than any single biomarker alone.

For clinicians, the translational message is that H. pylori history may belong in the immunotherapy discussion. If bacterial virulence factors from specific strains choreograph PD-L1 expression through defined, druggable pathways, then targeting NF-κB, MAPK, STAT1 or Sonic Hedgehog signaling could theoretically complement checkpoint blockade, either by lowering PD-L1 directly or by reshaping the inflammatory microenvironment in ways that make immunotherapy work better. The review offers theoretical foundations and translational guidance for optimizing immunotherapeutic strategies in gastric cancer management, while cautioning that the network’s redundancy means no single pathway inhibitor is likely to suffice on its own. As the authors note, their synthesis aims to bridge the gap between bench-side signaling maps and bedside therapeutic decisions.

The publication also arrives at a moment when the field is wrestling with how to standardize PD-L1 testing and biomarker selection in gastric cancer. By integrating the differential effects of distinct virulent strains, the modifying influence of viral coinfection and microsatellite status, and the layered molecular machinery from CagA delivery through T4SS to CMTM6-mediated protein stabilization, the review provides a framework that researchers can use to design better stratification studies and clinicians can use to interpret ambiguous biomarker results. It is a reminder that some of the most consequential cancer biology begins not with a mutation in a tumor cell but with a bacterium that has coexisted with humans for tens of thousands of years, quietly rewriting the conversation between our immune system and our stomachs. Decoding that conversation, the authors argue, may finally let oncologists turn one of humanity’s oldest infections into a guide for treating one of its deadliest cancers.

Subject of Research: Molecular regulation of PD-L1 expression in gastric cancer under Helicobacter pylori infection and its implications for immunotherapy

Article Title: Molecular regulatory network of PD-L1 expression under Helicobacter pylori infection and its clinical translational value: a review of multi-pathway crosstalk mechanisms

Article References: Wang, Q., Duan, T., Feng, J., Sun, L., Wei, J., Zhang, Y., Song, A., Gao, T., & Feng, Y. (2026). Molecular regulatory network of PD-L1 expression under Helicobacter pylori infection and its clinical translational value: a review of multi-pathway crosstalk mechanisms. Cancer Immunology, Immunotherapy. https://doi.org/10.1007/s00262-026-04581-y

Image Credits: AI Generated

DOI: 10.1007/s00262-026-04581-y

Keywords: Helicobacter pylori, PD-L1, gastric cancer, CagA, VacA, NF-κB, MAPK, STAT1, Sonic Hedgehog, immune checkpoint inhibitors, Epstein-Barr virus, microsatellite instability

The central insight driving the review is that the tumor microenvironment is not a passive backdrop but an active determinant of both treatment response and clinical outcomes. In colorectal cancer, the composition of immune cells infiltrating a tumor, the ratio of tumor cells to stroma, the presence of tertiary lymphoid structures and the burden of tumor-infiltrating lymphocytes all carry prognostic and predictive weight. Yet these features are traditionally measured on tissue samples obtained through biopsy or surgery, which capture only a fragment of the tumor and cannot be repeated freely over time. Radiomics offers an alternative. By extracting hundreds of quantitative features from routine medical images, such as computed tomography and magnetic resonance imaging, radiomics aims to characterize the whole tumor, including its internal heterogeneity, without touching the patient.

Technically, radiomics pipelines convert standard acquisitions such as T2-weighted imaging, contrast-enhanced T1-weighted imaging and diffusion-weighted imaging into high-dimensional feature sets. First-pass texture statistics capture local gray-level variation, while higher-order features extracted through filtered images and deep learning networks probe patterns that the human eye cannot resolve. Apparent diffusion coefficient maps derived from diffusion-weighted imaging, for example, reflect tissue cellularity and can serve as indirect surrogates of tumor density and stromal content. The review organizes the colorectal cancer radiomics literature into three thematic clusters: features that correlate directly with microenvironment components, features associated with vascular-invasion-related phenotypes, and features tied to tumor-intrinsic properties that are themselves shaped by the microenvironment. Each cluster, the authors argue, contributes a different piece of the puzzle linking what radiologists see on a screen to what pathologists see under a microscope and what molecular biologists sequence in the lab.

The vascular-invasion theme is particularly consequential clinically. Extramural venous invasion and microvascular invasion are established markers of poor prognosis in colorectal cancer, signaling that tumor cells have entered the circulatory system and raised the risk of metastasis. Radiomic models trained on CT and MRI can flag these phenotypes before surgery, potentially informing decisions about neoadjuvant therapy and surgical planning. Meanwhile, radiomic signatures predicting microsatellite instability and deficient mismatch repair status offer a non-invasive proxy for the single most important biomarker in modern colorectal cancer immunotherapy, since patients with dMMR or MSI-high tumors are the ones most likely to benefit from immune checkpoint inhibitors. In locally advanced rectal cancer, radiomic models have also been used to predict pathological complete response after chemoradiotherapy, a finding that could eventually help identify patients for organ-preserving strategies.

Pathomics extends the same quantitative logic to the microscopic scale. Whole-slide imaging digitizes histopathology slides, and computational methods then profile tumor architecture and the spatial distribution of immune cells at high throughput. Where a pathologist might estimate tumor-infiltrating lymphocyte density visually, pathomic pipelines can quantify it precisely, map the spatial relationships between tumor nests and stromal compartments, and compute the tumor-stroma ratio automatically. Deep learning models trained on whole-slide images can even predict molecular alterations, such as microsatellite instability, directly from hematoxylin and eosin-stained tissue. Because pathomics operates on resected or biopsied tissue, it provides the microscopic ground truth that radiomics lacks, and the two approaches are natural partners: radiomics sees the whole tumor in vivo, while pathomics resolves the cellular detail of the sampled regions.

The most ambitious portion of the review describes cross-scale integration, in which radiomic and pathomic features are fused with genomics and transcriptomics to trace a continuous chain from macroscopic phenotype to molecular mechanism. Radiomic features that predict microsatellite instability, for instance, can be connected to the immune-inflamed transcriptional programs that accompany deficient mismatch repair, including upregulated checkpoint molecules and enriched cytotoxic T-cell signatures. Consensus molecular subtypes of colorectal cancer, which stratify tumors by their gene-expression patterns, also leave imaging fingerprints, and studies reviewed by the authors show that radiomic models can distinguish between these molecular classes with useful accuracy. At the single-cell and spatial-omics frontier, technologies such as single-cell RNA sequencing and spatially resolved transcriptomics reveal the precise cellular neighborhoods within the microenvironment, including interactions between tumor-associated macrophages, myeloid-derived suppressor cells, cancer-associated fibroblasts and lymphocytes, offering mechanistic explanations for the imaging features that models detect.

Multimodal fusion is the methodological glue holding this framework together. Rather than treating imaging, pathology and molecular data as parallel silos, fusion approaches combine them within a single predictive model, allowing each modality to compensate for the blind spots of the others. Deep learning architectures can ingest radiomic features from CT or MRI, pathomic features from whole-slide images, and genomic or transcriptomic profiles from the same patient, learning joint representations that outperform any single data type. The review highlights early studies demonstrating that such combined models improve prediction of prognosis, immunotherapy response and treatment-related outcomes in colorectal cancer compared with unimodal baselines, suggesting that the cross-scale framework is more than the sum of its parts.

The authors are explicit, however, about the caveats. Much of the evidence reviewed is retrospective, derived from single-center cohorts with limited sample sizes, and few radiomic models have been validated prospectively or across diverse populations. Standardization of image acquisition, feature definitions and model reporting remains inconsistent across the field, raising concerns about reproducibility. The relationship between imaging features and microenvironment biology is often correlational rather than mechanistically established, and the review repeatedly emphasizes that clinical utility, and any role in therapeutic decision-making, remain to be established through prospective validation. These are not trivial hurdles; they are the same obstacles that have slowed the translation of radiomics in other cancer types.

Even so, the trajectory described in the review is striking. What began as an effort to squeeze extra information out of images that radiologists already acquire routinely has matured into a multi-scale program that connects the radiology suite to the pathology lab and the sequencing core. If prospective studies bear out the promise of imaging-driven multi-omics integration, clinicians could one day profile a patient’s tumor microenvironment repeatedly, cheaply and non-invasively, tracking its evolution under therapy and selecting patients for immunotherapy with far greater precision than today’s single-timepoint biomarkers allow. For a disease that kills hundreds of thousands of people each year and frustrates clinicians with its heterogeneity, that would represent a genuinely transformative shift, one that this review maps out with unusual technical clarity and admirable restraint about what has, and has not, yet been proven.

Subject of Research: Decoding the tumor microenvironment in colorectal cancer through radiomics and multi-omics integration

Article Title: From imaging to multi-omics: decoding the tumor microenvironment in colorectal cancer

Article References: Gong, L., Wu, X., Zhang, W., Lai, B., Yuan, J., Gu, Y., Shen, H., Liu, X., Xiong, Y., Zheng, J., Wang, L., Han, X., Zhang, B., & Zhang, S. (2026). From imaging to multi-omics: decoding the tumor microenvironment in colorectal cancer. Journal of Translational Medicine. https://doi.org/10.1186/s12967-026-08958-6

Image Credits: AI Generated

DOI: 10.1186/s12967-026-08958-6

Keywords: colorectal cancer, tumor microenvironment, radiomics, pathomics, multi-omics, immunotherapy, microsatellite instability, deep learning, spatial omics, tumor heterogeneity, precision oncology, biomarkers

The research, led by Cédric Poyet of Stadtspital Triemli in Zurich and Marie Lork of the University Hospital of Zurich, together with colleagues from Kantonsspital Baden, Luzerner Kantonsspital and University Hospital Zurich, set out to identify molecular correlates of overall survival in muscle-invasive bladder cancer, often abbreviated MIBC. The team analyzed tumor DNA extracted from cystectomy specimens, the surgical samples obtained when the bladder is removed, from 32 patients treated at Swiss centers. The study received ethical approval from the Cantonal Ethics Committee Zurich and was conducted in accordance with the Declaration of Helsinki.

To characterize the genomic landscape of each tumor, the investigators used the FoundationOne CDx comprehensive genomic profiling platform, a targeted next-generation sequencing assay capable of detecting substitutions, insertions and deletions, copy number alterations and selected genomic instability markers across hundreds of cancer-related genes. Patients were then divided into two comparison groups based on a hard clinical endpoint: a favorable outcome group of 14 patients who survived at least 60 months after surgery, and a poor outcome group of 18 patients who survived fewer than 60 months. This dichotomy allowed the researchers to ask a simple but clinically vital question: which genomic features separate long-term survivors from those who die earlier of their disease?

Across the entire cohort, the sequencing effort uncovered 279 pathogenic or likely pathogenic mutations distributed across 88 genes. The most frequently altered genes were familiar names in bladder cancer biology: TP53, the guardian-of-the-genome tumor suppressor whose inactivation is a near-universal event in this disease; PIK3CA, a signaling kinase driving PI3K pathway activation; KDM6A, a histone demethylase involved in chromatin regulation; and FGFR3, a receptor tyrosine kinase that is a well-established oncogenic driver and drug target in urothelial carcinoma. Perhaps surprisingly, the distributions of these frequent alterations were similar between the favorable and poor outcome groups, indicating that the presence or absence of these canonical mutations alone does not explain the dramatic survival differences observed in the clinic.

The team next turned to the standard quantitative indicators of genomic instability that have been proposed as prognostic and predictive biomarkers in many tumor types. Tumor mutational burden, or TMB, reflects the total number of somatic mutations carried by a tumor and is widely used as a proxy for responsiveness to immune checkpoint inhibitors. Microsatellite instability, or MSI, marks defects in DNA mismatch repair and carries prognostic and predictive significance in colorectal and several other cancers. In this MIBC cohort, however, both metrics were comparable between the long-term survivors and the poor outcome group, and neither proved prognostically informative. The finding is a caution against assuming that biomarkers validated in other cancers will translate directly to bladder cancer.

The decisive signal emerged from a different layer of genomic complexity: copy number alterations. Tumors from the poor outcome group exhibited a significantly higher frequency and burden of gene amplifications, events in which segments of DNA containing particular genes are copied multiple times, often massively, driving overexpression of the encoded proteins. Crucially, these amplifications frequently involved known oncogenes and co-amplification hotspots, regions of the genome where neighboring growth-promoting genes are gained together in a single event. In other words, patients whose tumors carried a heavy load of oncogene amplifications were disproportionately represented among those who died within five years of cystectomy.

The biological logic behind this observation is compelling. While point mutations typically disable tumor suppressors or alter the function of a single protein, amplifications act as gene dosage escalators, flooding tumor cells with growth factor receptors, signaling kinases and cell cycle accelerators. High-level amplification of oncogenes can simultaneously promote proliferation, survival under therapeutic stress and metastatic competence. Moreover, co-amplification events can deliver several oncogenic payloads at once, creating tumors that are intrinsically more aggressive and harder to eradicate with a single targeted agent. The Swiss findings suggest that this dosage-driven mode of tumor evolution may be a hallmark of the most lethal forms of MIBC.

The results also carry therapeutic implications. Amplified oncogenes are, in principle, druggable targets. FGFR inhibitors are already approved for metastatic urothelial carcinoma in tumors with FGFR alterations, and agents directed against amplified receptor kinases and downstream signaling nodes are in clinical development across many cancer types. If the association between amplification burden and poor survival is confirmed, comprehensive copy number profiling at the time of cystectomy could help identify patients who warrant intensified treatment, such as perioperative systemic therapy escalation, enrollment in targeted therapy trials or closer surveillance for recurrence. Conversely, the lack of prognostic value for TMB and MSI in this cohort suggests that these markers should not be relied upon in isolation for outcome prediction in MIBC.

The authors are careful to frame the study as exploratory, and the caveats are substantial. The cohort comprised only 32 patients, divided into groups of 14 and 18, a sample size that limits statistical power and leaves open the possibility of confounding by clinical factors such as stage, nodal status and treatment sequence, which the abstract does not address in detail. The use of a targeted panel, while broad, does not capture the full spectrum of structural variants and noncoding alterations that whole-genome sequencing would reveal. The authors explicitly call for validation in larger cohorts to determine whether oncogene amplifications can serve as robust prognostic markers and to explore their potential as therapeutic targets. It is also worth noting that Roche funded the genomic testing through the FoundationOne CDx platform but had no role in study design, data analysis, interpretation or manuscript writing, apart from being granted the opportunity to review the manuscript prior to submission.

Even with these limitations, the study adds an important dimension to the ongoing effort to bring precision oncology to bladder cancer. The field has long focused on the mutational catalog of urothelial carcinoma, one of the most heavily mutated of all common tumors, yet this work suggests that the architecture of copy number gains may carry at least as much prognostic weight as the mutation list itself. For patients facing cystectomy, a procedure with significant morbidity and a five-year survival that remains unsatisfactory for many, any molecular signal that reliably separates indolent from lethal disease is valuable. If larger studies confirm that oncogene amplification burden predicts poor overall survival, clinicians may one day sequence not just for mutations but for the sheer number of oncogene copies a tumor carries, using that information to triage patients toward more aggressive, and hopefully more effective, treatment strategies from the moment of diagnosis.

Subject of Research: Genomic profiling of oncogene amplifications as prognostic markers of overall survival in muscle-invasive bladder cancer

Article Title: Oncogene-driven genomic profiles are linked to poor overall survival in muscle-invasive bladder cancer (MIBC)

Article References: Poyet, C., Franzen, A. S., Bieri, U., Kaufmann, E., Eberli, D., Schmid, M., Zoche, M., Moch, H., & Lork, M. (2026). Oncogene-driven genomic profiles are linked to poor overall survival in muscle-invasive bladder cancer (MIBC). Journal of Cancer Research and Clinical Oncology. https://doi.org/10.1007/s00432-026-06626-2

Image Credits: AI Generated

DOI: 10.1007/s00432-026-06626-2

Keywords: muscle-invasive bladder cancer, oncogene amplifications, genomic profiling, tumor mutational burden, microsatellite instability, TP53, FGFR3, PIK3CA, copy number alterations, prognostic biomarkers, cystectomy, FoundationOne CDx