Hang et al. identify an out-of-frame CBX3::ALK fusion in metastatic melanoma that generates functional ALK isoforms through alternative translation start sites. This study demonstrates that rare noncanonical out-of-frame fusions can be oncogenic and therapeutically actionable, highlighting an analytic blind spot in current gene fusion detection pipelines.
Combination immunotherapy for colorectal cancer: Clinical applications, rationale, challenges, and future perspectives
Fei et al. review combination immunotherapy strategies for colorectal cancer, focusing on overcoming resistance in pMMR/MSS/MSI-L tumors. They highlight integrating ICIs with chemotherapy, targeted therapy, and radiotherapy and discuss insights into the tumor microenvironment and biomarkers to improve outcomes.
STAT+: With successful trials, Roche takes its MS drug to regulators, but safety questions loom
The Swiss drugmaker Roche on Tuesday presented the latest data for its experimental multiple sclerosis drug, setting the stage for the company to seek approval for a medicine that it believes can cut relapse rates and slow the progressive disability the disease causes.
Now the test is whether the drug, called fenebrutinib, can win the regulatory green light.
While three late-stage trials of the drug have shown it to be effective, analysts have homed in on some potentially worrying liver safety signals, an issue that previously prompted the Food and Drug Administration to reject an MS therapy developed by Sanofi. In data released Tuesday, researchers also disclosed that there were two drug-related deaths among patients who took fenebrutinib.
STAT+: From Revolution Medicines, more strong data on KRAS drug and a glimpse of a ‘novel class’ beyond it
SAN DIEGO — Revolution Medicines is already cooking up the next iteration of RAS inhibiting drugs.
At the American Association of Cancer Research annual meeting here, the company is the talk of the town for the clinical trial success of daraxonrasib, its next generation targeted therapy, in advanced pancreatic cancer. And while the company presented more data on that drug Tuesday, showing promising first line and combination data on daraxonrasib, scientists also showed in another session intriguing preclinical data on a completely new compound that may represent what comes after the current lineup.
That drug, currently called RM-055, is what RevMed CEO Mark Goldsmith is calling an entirely “novel class of catalytic inhibitors.” These are targeted therapies that not only block the RAS signaling that drives cancer, but molecularly turn the cancer protein off.
STAT+: Kyverna Therapeutics plans to submit cell therapy for stiff person syndrome for FDA approval
A one-time, personalized cell therapy from Kyverna Therapeutics improved mobility and reduced disabilities in patients with stiff person syndrome, a rare, neurological autoimmune disorder, according to study results presented Tuesday.
Kyverna intends to submit the treatment to the Food and Drug Administration by the middle of the year. If approved, it would become the first treatment for stiff person syndrome and the first personalized CAR-T therapy for an autoimmune disease of any kind to reach the market.
Currently, CAR-T treatments are approved only for blood cancers, but using engineered T cells to deplete B cells — essentially performing an immune system reset inside a patient — has pushed a growing number of biotech companies to shift their CAR-T focus to autoimmune diseases.
STAT+: At AACR, more strong results for Revolution Medicine’s KRAS drug, plus assurance from NCI’s director
You’re reading the web version of STAT’s popup newsletter, AACR in 30 seconds, your guide to what’s happening at the American Association of Cancer Researchers’ annual meeting. Sign up here.
We’re nearing the end of a big AACR. We hope to see everyone at our live event on Tuesday night. Clearly, Revolution Medicines and KRAS have been the big topic of the meeting. Last year, AACR was dominated by big concerns over what cancer research funding would look like in the Trump administration. This year, the new head of the NCI tried to allay researchers’ fears. Read on!
Strong results for Revolution Medicines’ KRAS drug
Last week, researchers working with the biotechnology firm Revolution Medicines presented stunning news: the experimental drug daraxonrasib more than doubled survival in second-line pancreatic cancer compared to chemo — although that only meant increasing median survival in this terrible disease by six months.
Roundtables: Unveiling The 10 Things That Matter in AI Right Now
Listen to the session or watch below
Watch a special edition of Roundtables simulcast live from EmTech AI, MIT Technology Review’s signature conference for AI leadership. Subscribers got an exclusive first look at a new list capturing 10 key technologies, emerging trends, bold ideas, and powerful movements in AI that you need to know about in 2026.
Speakers: Grace Huckins, AI reporter, hosted this session as Amy Nordrum and Niall Firth, executive editors, unveiled the list onstage.
Recorded on April 21, 2026
Related Stories:
AACR 2026: MRI-ctDNA Combo Informs HPV-Related Throat Cancer Treatment
At the 2026 annual meeting of the American Association for Cancer Research (AACR), researchers from Memorial Sloan Kettering Cancer Center (MSKCC) presented new evidence that a blood-based biomarker, combined with advanced imaging, could enable real-time adjustment of cancer treatment in patients with HPV-related throat cancer. Findings from this clinical study (NCT03323463) highlight a potentially important shift in care, particularly for HPV-associated oropharyngeal cancer, a disease with generally high cure rates but ongoing efforts to reduce treatment-related toxicity. Rather than waiting until therapy is complete, clinicians may be able to tailor treatment intensity based on early indicators of response.
Circulating tumor DNA (ctDNA) has already shown promise for detecting minimal residual disease (MRD), but its role in guiding treatment decisions during therapy remains largely unexplored. To address this gap, a research team led by Bill H. Diplas, MD, PhD, a radiation oncology fellow at MSKCC, investigated whether serial ctDNA measurements, paired with weekly MRI scans, could provide a more precise and dynamic view of tumor response. In collaboration with Labcorp and Biocartis, the MSKCC researchers developed a personalized ctDNA assay that combined two strategies: detection of patient-specific tumor mutations and quantification of DNA from high-risk HPV strains, particularly HPV-16 and HPV-18, using anchored multiplex PCR and high-throughput sequencing.
The study enrolled 158 patients with HPV-associated oropharyngeal cancer who had undergone primary tumor resection followed by risk-adapted chemoradiotherapy guided by hypoxia assessment. MRIs were performed pretreatment and weekly following treatment to determine tumor volume, and blood samples were collected before treatment and weekly during therapy, yielding nearly 1,000 samples from 119 patients (mean 8.2 samples/patient) up to 126 weeks.
At baseline, ctDNA was identified in 93.9% of patients—outperforming either mutation-based (89.4%) or HPV-based (80.3%) methods alone. ctDNA levels also correlated with tumor size and biological features such as cell death and viral load. Notably, ctDNA emerged as a faster and more sensitive indicator of treatment response than imaging. Changes in ctDNA levels appeared earlier and across a broader dynamic range than tumor size reductions observed on MRI. By the second week of therapy, ctDNA measurements could already distinguish patients likely to require more intensive treatment.
The identification of patients with high-risk disease was significantly improved by combining on-treatment ctDNA assessment with imaging in a multimodal model, outperforming any modality alone. These results underscore the complementary nature of molecular signals in blood and structural changes seen on imaging.
Similar multimodal strategies that integrate ctDNA with imaging have been explored in other cancers, including breast and lung, primarily in research settings. Studies suggest that combining these approaches can improve prediction of treatment response and enable earlier detection of resistance. Broader analyses across colorectal, lung, and breast cancers further support the value of integrating molecular and imaging data to refine models of response and survival. However, most of these approaches remain investigational, and the use of ctDNA to guide real-time treatment decisions is only beginning to be tested in prospective trials.
Although further validation is needed, this study establishes a framework for real-time, personalized treatment in oropharyngeal cancer. If translated into clinical practice, such an approach could accelerate the shift toward adaptive therapy—where decisions are guided not only by how tumors appear on imaging, but by how they respond at the molecular level throughout treatment.
The post AACR 2026: MRI-ctDNA Combo Informs HPV-Related Throat Cancer Treatment appeared first on Inside Precision Medicine.
<![CDATA[Emerging student researchers share their award-winning neuroscience and psychiatry projects.]]>
Autoencoder-Enhanced Convolutional Neural Networks for Plantar Pressure–Based Gait Pattern Recognition: Model Development and Cross-Validated Evaluation Study
<strong>Background:</strong> Plantar pressure imaging is a stable modality that reflects gait-related biomechanical characteristics and has been used increasingly for gait assessment and recognition. However, plantar pressure images are high dimensional and nonlinear, making manual feature engineering and conventional machine learning insufficient to capture discriminative patterns. <strong>Objective:</strong> This study aimed to develop a gait pattern recognition model based on plantar pressure using an autoencoder (AE)-enhanced convolutional neural network (CNN) and to evaluate its performance against baseline deep learning and classical machine learning approaches. <strong>Methods:</strong> A total of 13 healthy volunteers (aged 18-24 years) were recruited. Plantar pressure data were collected during treadmill walking using an in-shoe pressure measurement system and converted into frame-wise plantar pressure images. We compared a lightweight CNN (Light CNN), an AE-CNN cascade model, and an encoder-augmented CNN with an additional bottleneck layer. Model development used participant-wise data partitioning, and performance was evaluated using accuracy, precision, recall, and <i>F</i><sub>1</sub>-score. <strong>Results:</strong> The proposed encoder-augmented CNN achieved the best overall performance (<i>F</i><sub>1</sub>-score=96.20%), outperforming the Light CNN (<i>F</i><sub>1</sub>-score=94.44%) and AE-CNN cascade (<i>F</i><sub>1</sub>-score=92.45%). Confusion matrices and learning curves further indicated stable training behavior and consistent classification performance across gait patterns. <strong>Conclusions:</strong> Integrating representation learning (AE-based compression) with CNN-based classification improved the recognition of gait patterns from plantar pressure images. This pilot study included only healthy participants. Future work should validate generalizability in larger and clinically diverse cohorts and further investigate participant-level evaluation and model interpretability, as well as deployment feasibility.
