Radically different
Nature Medicine, Published online: 09 July 2026; doi:10.1038/s41591-026-04490-8
The 2024 revision of the McDonald diagnostic criteria is an important step toward earlier and more inclusive diagnosis of multiple sclerosis. Realizing its full potential will require continued refinement of biomarkers, disease stratification and clinical trial approaches.
COMPASS, a novel AI model that combines transcriptomic data with tumor-immune information, predicts immunotherapy response more accurately than existing biomarkers across different cancer types, shows data published in Nature Medicine.
The model also offers insight into why a patient may or may not respond to treatment, giving clinicians a better understanding of why responses might not be as expected.
To date, predicting response to immune checkpoint inhibitors (ICIs) has largely relied on tumor mutational burden (TMB) and PD-L1 immunohistochemistry. In addition, researchers have developed transcriptomic signatures that capture T cell dysfunction or score immune checkpoint activity, while network based and machine learning approaches have tried to incorporate gene interaction data, but each of these approaches has limitations because people undergoing ICI treatment often respond in unexpected ways.
“In our benchmarking against 22 of these methods, previous approaches showed inconsistent performance once tested across different cancer types and treatments, which is the core problem COMPASS addresses,” said study senior author Marinka Zitnik, PhD, associate professor of biomedical informatics in the Blavatnik Institute at Harvard Medical School.
She told Inside Precision Medicine that “most AI models for predicting immunotherapy response are validated within a single cancer type or a single drug class, which limits how much they generalize once you move to a new tumor type, a new therapy, or a different hospital’s data. This is significant because it means we can have a model that predicts response in one population of cancer patients, but fails when applied to a different population, a different tumor type, or a different treatment regimen, which is exactly the kind of gap that keeps promising biomarkers from ever reaching routine clinical use.”
Zitnik explained that “COMPASS is different because it is pretrained on transcriptomic data from over 10,000 tumors spanning 33 cancer types before it ever sees a clinical outcome, so it learns broad, biologically grounded patterns of tumor immune biology.”
The model uses a “concept bottleneck architecture” which means that rather than going straight from gene expression data to a prediction, it forces the information through an intermediate layer, or bottleneck, before making the final prediction. In this case, the bottleneck is biologically defined immune concepts, like T cell exhaustion, macrophage activity, or transforming growth factor (TGF)-β signaling. This means clinicians and researchers can see which biological programs are driving a given patient’s predicted response or resistance.
Zitnik and team tested the performance of COMPASS in 1133 patients from 16 clinical cohorts spanning seven cancers and six ICI regimens. The found that, compared with the second-best performing biomarkers of the 22 tested (these varied by cohort), COMPASS improved accuracy by 8.5 percentage points and area under the precision-recall curve by 15.7 percentage points, on average.
“In a setting where response rates to immunotherapy are already low, a meaningful jump in prediction accuracy translates directly into better patient selection,” said Zitnik. “The improvements we saw mean fewer patients would be steered toward a treatment unlikely to help them, and more responders would be correctly identified for therapies that could extend their lives. This is important in cancer where checkpoint inhibitors carry toxicity risks and cost, so gains in identifying who is likely to benefit can meaningfully change treatment decisions and clinical trial design at scale.”
In survival analyses, the team found that individuals with a COMPASS response probability (PR) of 0.5 or higher had a 1-year overall survival rate of 86% compared with 40% for those with a PR below 0.5, yielding a hazard ratio for survival of 4.7.
Performance varied by cancer type and cohort size, but the researchers note that COMPASS was still able to accurately predict response for cancer types it had never seen.
“When we excluded lung adenocarcinoma entirely from training and tested only on that held out cohort, COMPASS still achieved 76.5% accuracy,” said Zitnik. “We saw similarly strong cross cancer generalization for urothelial carcinoma, melanoma, and other tumor types.”
Furthermore, COMPASS achieved 85.3% accuracy for predicting a response to combination therapies (ipilimumab plus pembrolizumab) when trained only on monotherapy cohorts.
In addition to predicting treatment response, the tool also generates a personalized response map for each patient that traces exactly which genes and immune programs are shaping their prediction.
“This explainable AI feature means a clinician is not just given a predicted ‘yes’ or ‘no’, they can see whether a patient’s tumor looks inflamed but is failing due to TGF-β driven suppression or whether it looks immune desert but still shows residual cytotoxic activity that could respond to treatment,” said Zitnik.
“This kind of mechanistic insight could help with patient stratification in clinical trials, indication selection when a drug is being tested in a new cancer type and generating hypotheses about resistance mechanisms that could point toward combination therapies.”
The investigators are now planning to move from using bulk RNA sequencing data for training to single cell and spatial transcriptomic data. They are also exploring how to connect COMPASS with AI agent systems, to help translate the mechanistic insights into more autonomous research workflows.
The post Predicting Cancer Immunotherapy Response Better with COMPASS AI Model appeared first on Inside Precision Medicine.
Antidepression treatment based on a person’s individual biomarkers could help determine which of the world’s most popular medications to use, a clinical trial suggests.
The SMART Trial to Predict Anhedonia Response to Antidepressant Treatment results suggest that behavioral, brain, and clinical data could together determine the optimal antidepressant to choose before a person starts treatment.
There were no significant primary endpoint differences in depression outcomes between participants who received bupropion or sertraline based on biomarkers identified in the prior Establishing Moderators and Biosignatures of Antidepressant Response in Clinical Care (EMBARC) study.
But people negative for biomarkers with both drugs had significantly worse depression symptom trajectories than those who had at least one positive biomarker, regardless of the drug they received.
Response rates among participants with both biomarkers were almost double that of those without any biomarkers, with those who had at least one biomarker having an intermediate response.
“Our results suggest that we could boost response rate by using two sets of biomarkers previously identified in the EMBARC study, making an important contribution to advancing the goals of precision psychiatry,” reported Peter Zhukovsky, PhD, from Harvard Medical School, and co-workers in Nature Mental Health.
“Ultimately, we strongly hope these advances will enable personalized treatment guidance to accelerate and boost antidepressant benefits.”
Treatment for depression is often still trial and error, with symptoms improving in only half the people taking an antidepressant. This could be due to treatments not being chosen based on people’s individual characteristics.
Finding markers that predict response to different antidepressants could therefore provide patients and clinicians with valuable information to guide treatment choice.
The trial was among the first to investigate how treatment could be guided using clinical information such as responses to questionnaires, behavioral information such as performance in computerized tasks, and brain data such as magnetic resonance imaging (MRI) scans.
It was carried out as part of the Wellcome Leap Multi-Channel Psych Program effort to double the number of people who respond to the first treatment they try for depression via the integration of multimodal biomarkers.
Firstly, the researchers investigated biomarker models that predicted response to the selective serotonin reuptake inhibitor (SSRI) sertraline or the norepinephrine-dopamine reuptake inhibitor bupropion using the EMBARC study.
The treatment-assignment algorithm that was developed generated two marker-based indications for each patient—one for bupropion and sertraline—with the predictive model achieving a cross-validated area under the curve of 0.86 and 0.66, respectively.
The team then examined whether antidepressant response could be boosted using their created biomarker combination of a functional MRI imaging marker, reward learning and sensitivity, cognitive control, the clinical variables of depression severity and neuroticism, and the demographic variable of employment status.
After analyzing these biomarkers among participants, who had major depressive disorder, the group was randomly assigned to receive a full 8-week course of an SSRI or non-SSRI.
The primary outcome was the change in depression severity from pretreatment baseline to eight weeks after the start of treatment, with no significant differences in treatment outcomes for those assigned a drug consistent versus inconsistent with their biomarkers.
This, the researchers say was possibly due to the limited power to detect moderate effects.
However, significant differences emerged in symptom reduction trajectories for those with positive markers for both medications, with a response rate of 71.4% compared with 65.4% for those with a positive biomarker for either drug and 42.9% for those with two negative markers.
The authors concluded: “We found that, relative to patients with two negative markers, those with one or two markers were characterized by significantly larger reduction in depressive symptoms, showing that biomarker-guided treatment selection can boost efficacy for two of the most widely prescribed antidepressants around the world.”
The post Biomarkers Could Help in Antidepressant Choice appeared first on Inside Precision Medicine.
Nature Medicine, Published online: 03 July 2026; doi:10.1038/s41591-026-04493-5
The authors present a framework for the selection of immune aging biomarkers suitable for clinical trials in geroscience, alongside a translational roadmap for biomarker development to support the next generation of clinical studies on immune aging.
A set of blood-based circular RNAs (circRNAs) could change how we diagnose and monitor Alzheimer’s disease, providing a simple, noninvasive test that can detect the disease with remarkable accuracy and predict its progression years before symptoms appear.
Washington University School of Medicine researchers analyzed blood samples from 1,221 individuals, including people with Alzheimer’s disease and cognitively healthy participants, making it one of the largest investigations of blood circRNAs in Alzheimer’s to date. The findings, published in a Nature Medicine study, identified 34 circRNAs whose combined expression patterns accurately distinguished Alzheimer’s disease from healthy aging.
Unlike conventional RNA molecules, circRNAs form single-stranded closed loops that resist degradation and are abundant in the brain. Their stability and ability to cross the blood-brain barrier make them attractive candidates for blood-based biomarkers that reflect changes occurring in the brain.
The research indicated that a predictive model built from the 34 circRNAs achieved an area under the curve (AUC) of 0.945 for identifying biomarker-confirmed Alzheimer’s disease, outperforming the widely used plasma biomarker pTau217 (AUC 0.877). When circRNA measurements were combined with pTau217, diagnostic performance increased further to an AUC of 0.977.
Beyond diagnosis, the circRNA signature demonstrated exceptional ability to predict disease progression. Individuals with elevated circRNA scores were nearly three times more likely to progress to symptomatic Alzheimer’s disease than those with lower scores. The model also outperformed pTau217 in forecasting progression and remained highly specific for Alzheimer’s, showing limited predictive ability for other neurodegenerative disorders such as Parkinson’s disease, frontotemporal dementia, and dementia with Lewy bodies.
Importantly, the findings were independently replicated in two additional cohorts, including 551 participants from the Knight Alzheimer’s Disease Research Center and 1,767 participants enrolled in the Anti-Amyloid Treatment in Asymptomatic Alzheimer’s Disease (A4) study. This independent validation demonstrates that the circRNA signature is robust across multiple populations and study designs.
The researchers also found evidence that circRNA changes begin approximately two to four years before the onset of clinical symptoms, suggesting that these molecules may capture biological processes closely linked to the transition from silent pathology to cognitive decline. Such biomarkers could become increasingly valuable as disease-modifying therapies enter clinical practice, where monitoring ongoing neurodegeneration is just as important as detecting amyloid pathology.
The work builds on intellectual property protected by patent PCT/US2026/017857, “Blood Circular RNA as a Noninvasive Biomarker of Alzheimer’s Disease,” which Circular Genomics has licensed. The company, based in San Diego, California, is developing next-generation molecular blood biomarker diagnostics for precision neurology. The patent covers the use of blood circRNA signatures for the diagnosis and monitoring of Alzheimer’s disease and supports continued development of clinically deployable blood tests.
While the authors emphasize that larger prospective studies are still needed before widespread clinical implementation, the findings position blood circRNAs as a promising new class of biomarkers for Alzheimer’s disease. By combining high diagnostic accuracy with strong prediction of disease progression using a simple blood sample, circRNA-based testing could help identify patients earlier, improve clinical trial enrollment, and provide physicians with new tools to monitor disease over time.
The post Blood circRNAs Can Predict Alzheimer’s Years Prior to Symptoms Onset appeared first on Inside Precision Medicine.
Researchers at the University of Texas MD Anderson Cancer Center have identified a strategy to reverse resistance to standard treatment in BRAF-mutant advanced melanoma. Their findings, published today in Nature Communications, support using a biomarker-guided approach to improve outcomes for patients with treatment-resistant melanoma.
“Patients whose melanoma has stopped responding to standard therapies currently have very few effective treatment options,” said Vashisht Gopal Yennu Nanda, PhD, associate professor of melanoma medical oncology and translational molecular pathology at UT MD Anderson. “Our findings could help address this critical need for these patients by guiding clinicians toward combinations tailored to each individual’s tumors.”
About 50% of melanoma tumors carry BRAF mutations that drive uncontrolled tumor growth. Although the standard of care, consisting of a combination of BRAF and MEK inhibitors, is initially effective in most patients, about 80% will develop resistance within two years. In many cancers, but especially in melanoma, acquired resistance is often driven by the tumor increasing production of proteins from the BCL2 family, which block apoptosis and support the survival of cancer cells.
Yennu Nanda and colleagues tested the effects of adding a BCL2 inhibitor drug to the standard two-drug regimen in patient-derived xenograft models, which were established using melanoma tumors that had acquired resistance to standard therapy. Results showed that tumors that expressed high levels of BCL2 responded well to the triple combination, reversing resistance and inducing a complete tumor regression.
However, tumors that expressed high levels of MCL1—another protein from the BCL2 family—did not respond to this combination. In these tumors, the researchers tested an alternative treatment course adding an experimental MCL1 inhibitor to standard treatment, which successfully led to complete tumor regression.
“Targeted therapy works by shutting down the main signal driving melanoma growth, but tumors often have backup systems that keep them alive,” said Yennu Nanda. “By identifying which protein a tumor relies on for survival, we may be able to match patients to drug combinations tailored to their specific tumor biology.”
MCL1 inhibitors have previously shown promising antitumor activity, but early clinical trials flagged concerning heart-related side effects that have prevented them from moving through clinical development and receiving approval. In this study, however, the combination of an MCL1 inhibitor with standard BRAF-MEK inhibitors seemed to protect cardiac cells from the harmful effects associated with these experimental drugs.
“We did not anticipate that pairing these drugs would reduce MCL1 inhibitor toxicity,” said Michael A. Davies, MD, PhD, chair of melanoma medical oncology at UT MD Anderson. “If this finding is confirmed in clinical trials, it could give a second life to a class of drugs that has struggled to advance through development. It also reinforces that the most effective combinations are those that eliminate cancer while sparing healthy tissue.”
Building on these findings, the researchers are now working on analyzing tumor samples from a recent Phase II clinical trial in melanoma patients who received standard treatment and a BCL2 inhibitor, with the goal of studying whether MCL1 expression can predict clinical response. Down the line, their goal is to design clinical trials where melanoma patients are matched with drug combinations based on the expression of BCL2 or MCL1 biomarkers.
The post Personalized Therapy Could Overcome Resistance in Metastatic Melanoma appeared first on Inside Precision Medicine.