Background: Generative artificial intelligence (GenAI) chatbots have the potential to provide personalized mental health support to individuals at scale. Objective: This study evaluates the feasibility and usage patterns of the Safespace GenAI chatbot, an artificial intelligence (AI)–driven smartphone app that offers a large language model–powered interactive chatbot to support mental health. Methods: Using a mixed methods approach, we explored baseline attitudes toward GenAI chatbots and chatbot usage patterns, conducted a qualitative content analysis of participants’ experiences, and descriptively assessed patterns related to preintervention depressive symptoms. The study included an initial sample of 42 university students, 20 of whom actively used the chatbot over 2 to 4 weeks, generating 286 user-chatbot interactions. Results: Preintervention surveys indicated that the majority of participants anticipated that the chatbot would be helpful (27/42, 64%) and that they trusted its privacy safeguards (39/42, 93%). Usage patterns suggested that the highest levels of interaction occurred early in the morning and late at night, when peer and professional support may be inaccessible. The qualitative analysis indicated that participants appreciated using the chatbot for reflection as a blended-care tool between their counseling sessions, while also naming technical barriers and specific design needs required to sustain engagement. In addition, our exploratory analyses descriptively showed that participants with elevated depression scores engaged in emotional disclosure during 99% (38 sessions with 8 participants) of their sessions, compared to 84% (26 sessions of 12 participants) of those with low symptoms. Due to the small sample size, future adequately powered studies are needed to inferentially examine these observed patterns. Conclusions: These findings provide initial insights into the usage and engagement dynamics of the Safespace GenAI chatbot and highlight directions for future research to optimize GenAI-driven mental health interventions. Trial Registration: AEA Registry AEARCTR-0013291; https://doi.org/10.1257/rct.13291-1.0
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/8ea8d6d9a0e05f3fdb856f007cfd4549" />
Fine-Tuning Large Language Models for Motivational Interviewing in Health Behavior Change: Development and Evaluation Study
Background: Motivational interviewing (MI) is an effective counseling approach for promoting health behavior change, but its scalability is constrained by the need for highly trained human counselors. Large language models (LLMs) may provide a scalable way to support MI counseling, but evidence remains limited, especially for Chinese MI resources and evaluations based on standardized MI fidelity frameworks. Objective: This study aimed to develop Chinese large language models for motivational interviewing (MI-LLMs) and evaluate whether MI-focused fine-tuning could improve their ability to generate counseling responses consistent with MI principles. Methods: We first curated 5 publicly available Chinese psychological counseling datasets and assessed sampled conversations in terms of comprehensiveness, professionalism, authenticity, and safety. The 2 highest-scoring datasets, CPsyCounD and PsyDTCorpus, were selected for MI-style data construction. Using GPT-4 with a structured MI-informed prompt, we transformed 2040 multiturn counseling conversations into MI-style dialogs. Among these, 2000 dialogs were used for training and 40 for testing. Three Chinese-capable open-source LLMs (Baichuan2-7B-Chat, ChatGLM-4-9B-Chat, and Llama-3-8B-Chinese-Chat-v2) were fine-tuned with low-rank adaptation on the training dataset and were referred to as MI-LLMs. Automatic evaluation was conducted on the testing dataset using Bilingual Evaluation Understudy–4 (BLEU-4) and Recall-Oriented Understudy for Gisting Evaluation (ROUGE) metrics. Manual evaluation was conducted using the Motivational Interviewing Treatment Integrity Coding Manual 4.2.1. Thirty simulated counseling dialogs generated by the MI-LLMs were compared with 30 real MI dialogs sampled from AnnoMI and translated into Chinese. Two trained graduate student raters coded global scores and behavior counts, from which summary scores were subsequently calculated. Results: In automatic evaluation, fine-tuning substantially improved BLEU-4 and ROUGE scores across all 3 models compared with the base models. In manual evaluation, the MI-LLMs achieved technical and relational global scores, as well as total MI-adherent ratios that approached those of real MI dialogs. The MI-LLM based on ChatGLM-4-9B-Chat showed the strongest overall global performance. However, MI-LLMs produced fewer complex reflections and had lower reflection-to-question ratios than real MI dialogs. Conclusions: This study provides preliminary evidence that MI focused fine-tuning can help Chinese LLMs acquire core counseling behaviors consistent with MI principles. It also offers a scalable approach for constructing MI style dialog resources in Chinese. Nevertheless, current MI-LLMs should be regarded as early-stage tools for supporting, rather than replacing human counselors. Future work should expand real MI training data and strengthen the complex reflective skills of MI-LLMs. Further studies are needed to evaluate their effectiveness, acceptability, and safety in health behavior change settings in the real world.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/1c43f039686485cb31ea711826f22d05" />
Leveraging Self-Reporting in an Existing e-Cohort to Identify Clinically Relevant Mitral Valve Prolapse: Pilot Questionnaire Study
Background: Mitral valve prolapse (MVP) is a common valvulopathy associated, in a minority of cases, with heart failure, severe mitral regurgitation (MR), and sudden arrhythmic death. Digital tools hold promise for faster and more efficient recruitment of study participants into a large-scale MVP Registry. Objective: This study sought to evaluate the feasibility of surveying participants in an existing e-cohort to identify and clinically characterize MVP cases based on self-reporting and to recruit them in an MVP Registry at the University of California, San Francisco. Methods: We surveyed Northern Californian participants of the Health eHeart Study, a large e-cohort using the Eureka digital research infrastructure, about a prior diagnosis of MVP. MVP-positive respondents were asked to provide relevant medical records to confirm their eligibility and were invited to enroll in an MVP Registry if evidence of MVP was confirmed. A follow-up survey was sent after 1 month and after 5 years to collect data about clinical outcomes, including arrhythmias and the need for mitral valve repair. Results: The survey was delivered to 5746 participants, and 520 completed responses were collected. A prior diagnosis of MVP was self-reported by 16.3% (85/520) of respondents. Echocardiograms were obtained from 51.8% (44/85) of participants, and evidence of MVP was confirmed in 32.9% (n=28) of individuals, all of whom joined the registry. Participants with more severe MR had a higher number of correct responses regarding both MVP (odds ratio [OR] 10.58, 95% CI 3.58‐63.04; <.001) and MR diagnosis (OR 4.86, 95% CI 2.11‐16.14; =.002). Longitudinal data were available from most patients through responses to a follow-up survey sent 1 month and 5 years later (18/28, 64.3% and 17/28, 60.7% of MVP confirmed respondents, respectively). Among the patients with electronic health records available, 75% (3/4) had a correct self-reported diagnosis of arrhythmia. Conclusions: e-Cohort methods with self-reported clinical data can be used to prescreen candidates for a research study of MVP. These methods can rapidly identify and retain, among many cases of benign MVP, the minority with clinically relevant presentations such as significant MR and ventricular arrhythmias. These cases may be missed, especially when asymptomatic, by small-scale clinic-based recruitment or family screening methods.
Beyond Area Under the Receiver Operating Characteristic Curve: Evaluating Predictive Performance Metrics Under Class Imbalance in Real-World Clinical Data
Background: Predictive models increasingly support clinical decision-making, although imbalanced outcome distributions are common in health care datasets and can distort performance evaluation. The area under the receiver operating characteristic curve (AUROC) remains the most frequently reported metric, despite its limited ability to reflect clinically meaningful performance under class imbalance. Objective: This study aimed to examine the influences of metric selection on the clinical interpretation of predictive models in imbalanced real-world health care data. Methods: This was a retrospective cohort study, including 17,018 hospitalized patients with COVID-19. Two predictive models using extreme gradient boosting (XGBoost) were developed to predict kidney replacement therapy (KRT) and mortality. Model performance was assessed using AUROC, macro–score, class-specific precision and recall, calibration (curve, slope, and intercept), decision curve analysis, and learning curves. Standard rebalancing strategies were applied exclusively to the training data to evaluate their impact on performance. Results: KRT occurred in 9.5%, and mortality in 18.0%. Although AUROC values were high (0.928 for KRT and 0.945 for mortality), performance in the minority class was substantially lower. For KRT, precision was 0.539 and recall 0.372; for mortality, precision was 0.725 and recall 0.718. Rebalancing strategies were associated with higher recall for the minority class, but this gain was accompanied by a reduction in precision, with minimal impact on AUROC values. As a result, AUROC remained high despite clinically relevant changes in error distribution between false positives and false negatives. The learning curves show a plateau-like shape, with stable validation performance across all training set sizes for both outcomes. Conclusions: AUROC alone is insufficient to evaluate prediction models in imbalanced health care scenarios, even with rebalancing. Routine reporting of class-aware metrics, alongside learning curve analysis, is essential to support robust and clinically meaningful evaluation of predictive models, rather than their direct translation into practice.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/e0a2583c49e29d41efcc3b688510ec53" />
BIO 2026: CEO Calls for U.S. Biotech Urgency and International Competitiveness
SAN DIEGO — Biotechnology is entering one of the most transformative periods in its history. But, according to Biotechnology Innovation Organization (BIO) CEO John Crowley, outdated regulations, rising development costs, and global competition threaten to slow progress unless policymakers act.
At the 2026 BIO International Convention in San Diego this week—which drew “roughly 20,000 attendees,” according to the organizers—Crowley outlined a vision for the future of biotechnology centered on accelerating clinical research, embracing artificial intelligence, and maintaining U.S. leadership in a rapidly evolving global bioeconomy.
The grassroots gauntlet
Crowley’s personal journey as a father shaped his path into biotechnology. In the late 1990s, two of his children were diagnosed with a rare form of muscular dystrophy. He left Bristol-Myers Squibb’s marketing department to co-found a biotechnology company with an Oklahoma academic researcher over scientific progress.
The struggle to get funding was immense. Crowley reflected on his first BIO convention in 2000 amidst the excitement of the Human Genome Project: “I came and there were tens of thousands of people partnering as there is today, still a quarter of a century later. Being the 31-year-old CEO of a small startup in Oklahoma City with no money, literally nobody signed up to meet with me and nobody accepted my meeting request.”
Crowley recalled going to the main stage, where a gentleman, rendered quadriplegic through a horse accident, came out on the stage and said, “Biotechnology—it’s a great big word that just means hope. It’s my hope that someday I can hold my wife’s hand on the beach or throw a ball to my kids.”
Crowley, empty-handed, returned to Oklahoma City and was able to scrounge up the funds for his startup, Novazyme Pharmaceuticals, which was ultimately funded by home equity loans and credit card advances to develop rare disease treatments. Just one year later, Novazyme was acquired by Genzyme Corporation for $225 million.
The experience engrained in Crowley two main concepts: first, developing therapeutics doesn’t always start in big pharma but, rather, often has grassroots origins; second, and relatedly, it’s an almost impossible battle for anyone outside of big pharma to fight.
“That’s the way so much of our science happens,” Crowley said. “It comes out of great universities, and it’s a scientist and entrepreneur—and increasingly, families, patients, and patient advocates—leading the way and going through the whole journey, running that gauntlet of making medicines.”
Modernizing clinical trials and accessible AI
To achieve the vision of maximizing the development and reach of biotechnology, Crowley identified a handful of problems, including the need to change the current system of clinical trials. Crowley praised the FDA’s new “Project Trailblazer” initiative to modernize experimental therapy human testing. He argued that clinical trials have become excessively burdensome and costly, limiting innovation and delaying patient access to new treatments.
Over the past year, Crowley and BIO have worked with regulators and industry stakeholders to identify development bottlenecks. “The FDA needs to continue to be the gold standard of the world,” he said, while emphasizing that modernization is necessary to make the agency a stronger “beacon of innovation.” BIO has proposed several reforms, including measures designed to streamline trial approvals and improve the efficiency of regulatory review.
Describing recent discussions among BIO’s board of directors, which includes executives from both major pharmaceutical companies and small biotechnology startups, Crowley said there were two major strategic topics that emerged that dominated the conversation: China and AI.
For AI, the question wasn’t about whether it could revolutionize biotechnology; rather, it had to do with making AI capabilities accessible to companies of all sizes. Crowley noted a major disparity. “Our biggest companies have the resources and the focus to think about AI. They’ve got hundreds or more people focused on this. Our small companies don’t have those resources,” he said.
Crowly continued, “It’s also a challenge because in our industry we would work on such long timelines, and it’s hard for an entrepreneur and biotech of a small or a mid-sized company who’s invested years to get to…starting Phase III, and all of a sudden you’ve got this massive disruptive technology. That’s exactly what AI is going to be.”
The solution, according to Crowley, is for BIO to be at the forefront to enable the rapid implementation of AI into drug development paradigms, clinical trials, and the regulatory review process.
Challenging China
Crowley’s most stressed point was that the United States must remain competitive against growing international rivals, particularly China. “Drug development has just gotten too costly and burdensome, and it takes too much time,” said Crowley. In this [global] bioeconomy where we need to compete and outcompete countries like China, these are reforms that are needed.”
He characterized biotechnology as a matter of national security and argued that the United States should treat the industry as a strategic asset. While supporting bipartisan efforts in Washington to strengthen domestic biotechnology capabilities, he cautioned against policies that could create unintended consequences or limit access to potentially life-saving technologies.
“The world is a better, safer, healthier, and more prosperous place when the United States and its allies continue to lead in biotechnology,” Crowley said.
China has identified biotechnology as a strategic priority through multiple national development plans and has invested heavily in scientific infrastructure, manufacturing capacity, and research capabilities. Crowley argued that the most effective response is not isolation but improving the competitiveness of the U.S. innovation ecosystem.
Crowley repeatedly returned to what he described as “man-made problems” holding the industry back. While scientific challenges will always exist, Crowley said barriers such as complex regulations, insufficient research funding, delays in patient access, and rising out-of-pocket healthcare costs are obstacles that policymakers can address. “We can’t come to this convention and cure every cancer,” he said. “But if we get together with policymakers and lawmakers, we can pretty quickly solve a lot of these man-made problems if we have the will.”
50 years down, 50 years ahead
As biotechnology celebrates more than 50 years of innovation, Crowley argued that the industry’s future will depend not only on scientific breakthroughs but also on its ability to modernize the systems that govern how those breakthroughs reach patients.
“I hope you see, when you’re here at this convention, that it captures that entrepreneurial spirit,” said Crowley. “It has to be grounded in great science and research, and it’s an exciting time to be in biotech, not just reflecting about all our successes and our many failures and challenges along the way in 50 years and looking out in the months, years, and next 50 years about what biotechnology can do to extend and enhance life and to alleviate an enormous amount of human suffering.”
With advances in gene editing, genomic medicine, artificial intelligence, and cell therapies accelerating simultaneously, Crowley believes the next era of biotechnology could surpass anything seen before—provided the industry can remove the barriers standing in its way.
The post BIO 2026: CEO Calls for U.S. Biotech Urgency and International Competitiveness appeared first on GEN – Genetic Engineering and Biotechnology News.
AI-CURA Automates Genetic Variant Classification
A new AI framework can classify hundreds of genetic variants as accurately as a human expert in a fraction of the time, research suggests.
Combining AI-assisted CURAtor (AI-CURA) with the latest large language models (LLMs) could streamline the diagnosis of rare genetic diseases.
The workflow system, described in Science Translational Medicine, performed as well as clinical experts in classifying 150 variants, while adhering to complex expert guidelines.
It was also able to categorize 150 variants with conflicting classifications.
“This study pushes the boundaries of fully automated variant interpretation,” commented senior journal editor Catherine Charneski, PhD, from the University of Bath.
Whole genome sequencing (WGM) has proven pivotal in ending the prolonged diagnostic odyssey of many patients with rare genetic disorders.
To manage the huge number of variants identified through WGS, attempts have been made by expert associations and working groups to establish guidelines and recommendations.
These now form a widely adopted classification system that categorizes variant-associated evidence into distinct rule-based categories.
But while some rules can be readily automated, most evidence still needs manual interpretation of the literature. This requires variant curators to possess broad knowledge across various aspects of molecular biology and genetics, as well as a deep understanding of expert recommendations to accurately interpret and score variants.
Wei Ma, PhD, and colleagues from the Hong Kong Genome Project (HKGP) therefore developed AI-CURA, a fully automated framework for variant classification that integrates LLMs to handle both literature-independent and literature-dependent evidence.
The tool integrates the assessment of evidence for non–literature-based criteria, which can be automated using standard bioinformatic tools, with a separate LLM-supported assessment of literature-based evidence.
Two state-of-the-art LLMs—DeepSeek-R1 and o3-mini-high—were tested for their ability to summarize literature-derived evidence relevant to variant classification.
The team found that the open-source DeepSeek-R1 outperformed o3-minihigh and had high sensitivity and 100% specificity in interpreting rules from the American College of Medical Genetics and Genomics (ACMG) that require understanding literature-based evidence.
They then tested it using 150 variants curated by ClinGen experts, with 150 expert-curated variants and 150 variants with conflicting classifications from the Clinical Genome Resource.
The open-source LLM DeepSeek-R1 showed high concordance with ClinGen experts in establishing a final diagnosis.
“In this study, DeepSeek-R1 demonstrated high accuracy (89.3 to 100%) in determining the application of seven literature-dependent ACMG rules,” the authors reported.
They added: “Our use of LLMs substantially streamlined the variant analysis and interpretation process. LLMs can finish summarizing the literature evidence in minutes.
“In comparison, curators in the HKGP typically spend around four hours per patient on WGS curation, with most of this time dedicated to reviewing literature.”
The post AI-CURA Automates Genetic Variant Classification appeared first on Inside Precision Medicine.
Social Determinants of Health Improve Disease Risk Prediction Beyond Genetics Alone
A study by researchers at the Icahn School of Medicine at Mount Sinai has found that social determinants of health—including environmental conditions, health behaviors, access to resources, and social well-being—can contribute as much as or more than genetic risk in predicting several common diseases. The research, published in The American Journal of Human Genetics, showed that incorporating social, behavioral, and environmental information into disease-risk models improved prediction when incorporated with genetic information for conditions including asthma, chronic kidney disease, coronary heart disease, high cholesterol, breast cancer, and prostate cancer.
“Genes are an important part of the equation, but they do not determine destiny,” said senior author Samira Asgari, PhD, an assistant professor of genetics and genomic sciences at Mount Sinai. “We found that the circumstances of people’s lives—their environments, behaviors, and social experiences—can contribute as much as genetics to predicting disease risk. To truly understand health, we have to look at the whole person, not just their DNA.”
According to the researchers, complex diseases arise through the interaction of genetic predisposition with environmental, behavioral, and social influences, yet these factors are often studied in isolation. Existing genetic models often rely on polygenic risk scores, while epidemiological approaches focus on lifestyle, environmental exposures, or social factors independently. The researchers sought to bridge that gap by integrating both types of data into a single risk prediction framework.
To conduct the study, the team analyzed data from 413,457 participants in the All of Us Research Program, a nationwide research effort in the U.S. supported by the National Institutes of Health. The team combined genetic information, electronic health records, and survey responses, with more than 100 environmental, behavioral, and social variables were evaluated, to create a broad picture of the different factors that may influence health.
Rather than selecting a limited number of known social risk factors in advance for their survey, the researchers used a statistical technique called multiple correspondence analysis, or MCA. The approach converted more than 100 categorical social, environmental, and behavioral measures into low-dimensional representations that helped identify patterns of non-genetic risk.
The choice to use MCA distinguished the study from many previous approaches. Past methods often have depended on selecting a small set of established risk factors or using statistical procedures that prioritize only the strongest predictors. By contrast, MCA identifies patterns across many correlated variables simultaneously, allowing researchers to examine broader social and environmental variables without assuming beforehand which factors have the most influence on health.
The analysis found known contributors to disease risk such as economic status and smoking, but also identified factors that receive less attention in published studies, including loneliness and spirituality. First author Abhijith Biji, a PhD candidate at Icahn School of Medicine, said the data showing associations involving loneliness was particularly notable.
“Some risk factors, such as smoking, have been studied extensively for decades,” Biji said. “What is especially intriguing is that we also observed associations involving factors like loneliness. Understanding how these experiences may become biologically embedded could open new avenues for research and ultimately improve our understanding of disease.”
When the researchers incorporated the MCA-derived measures into prediction models alongside demographic information and polygenic risk scores, predictive performance improved across all six diseases studied. For four of the six diseases, the gains from the MCA-based measures exceeded those attributable to polygenic risk scores.
The findings also suggested that genetic and non-genetic influences generally act independently rather than modifying one another. The researchers found little evidence for broad gene-environment interactions. Instead, inherited genetic risk and social, behavioral, and environmental context appeared to contribute additively to disease risk.
“This additive relationship suggests that interventions targeting social and behavioral factors can reduce disease risk regardless of genetic background, offering hope for broadly applicable public-health strategies,” the researchers wrote.
The researchers noted that the study does not establish causation. Because many survey responses were collected at a single point in time and some exposures may have occurred after disease onset, the findings should be considered as contributions to disease-risk prediction rather than proof that specific factors cause disease.
Building on this work, the team next will seek to integrate social determinants of health with additional biological measures and look mechanisms that may directly connect social experiences to disease. The investigators will also bring in longitudinal data, harmonize survey instruments across cohorts, and integrating other data types to better understand how environmental, behavioral, and social factors influence disease development and interact with biological processes.
“Our goal is to build a more complete understanding of health and disease,” Asgari said. “By combining genetics with social and environmental context, we can move toward risk models that better reflect the realities of people’s lives and help advance more personalized approaches to health.”
The post Social Determinants of Health Improve Disease Risk Prediction Beyond Genetics Alone appeared first on Inside Precision Medicine.
Segmenting Older Adults by Their Acceptance of Digital Health Care Devices: Cross-Sectional Study Using the Augmented Technology Acceptance Model and K-Means Clustering
Background: Population aging has become a critical global challenge, with South Korea entering a super-aged society and facing rapidly increasing health care demands. In response, digital health care devices have emerged as promising tools for supporting personalized health management and improving health care accessibility among older adults. However, despite their potential, adoption rates among older adults remain relatively low. Prior research based on the Technology Acceptance Model (TAM) has largely relied on variable-centered approaches, overlooking substantial heterogeneity in acceptance patterns among older adults. A person-centered segmentation approach is therefore needed to identify diverse acceptance profiles. Few studies have integrated the augmented TAM with K-means clustering to identify acceptance-based segments in this population. Objective: This study aims to segment older adults based on their acceptance patterns toward digital health care devices by integrating the TAM framework with data-driven clustering techniques. Methods: A cross-sectional survey was conducted with 349 adults aged 65 years and older who were recruited from older adult welfare centers and community facilities in the Seoul metropolitan area of South Korea. We measured 10 constructs within an augmented TAM framework: 2 core constructs (perceived usefulness, perceived ease of use), 6 extended constructs (compatibility, privacy, self-efficacy, price consciousness, health empowerment, attitude toward digital health care), 1 health-related construct (health threat susceptibility), and intention to use as the outcome. Principal component analysis (PCA) and K-means clustering were used to identify latent segments. The number of components was determined using parallel analysis and the Kaiser criterion, and the optimal number of clusters was validated using the silhouette coefficient. Robustness was further assessed through 100-seed stability analysis and PCA sensitivity tests. Results: We identified 2 principal components, and a 4-cluster solution was selected (K=4, silhouette coeffficient=0.383). The analysis revealed 4 distinct segments: core adopters (57/349, 16.3%), who scored highest across all constructs; potential adopters (64/349, 18.3%), who recognized the value of digital health care devices but exhibited low self-efficacy and perceived ease of use; neutral majority (159/349, 45.6%), who showed near-average scores; and rejecters (69/349, 19.8%), who scored negatively across all dimensions. Robustness checks confirmed high clustering reliability (94%‐99% agreement). Notably, potential adopters represented a critical target group, as their acceptance barriers stemmed from capability constraints rather than lack of motivation. This group combined high perceived usefulness (+0.50) with the lowest self-efficacy (−1.07) and perceived ease of use (−0.83). Conclusions: This study demonstrated that technology acceptance among older adults is heterogeneous rather than uniform and highlights the importance of segment-specific strategies. By integrating theory-driven acceptance constructs with unsupervised machine learning, the study provides a practical framework for identifying actionable user segments and designing tailored diffusion strategies. These findings offer important implications for policymakers, technology developers, and health care professionals seeking to facilitate inclusive adoption of digital health care technologies in aging societies.
<img src="https://jmir-production.s3.us-east-2.amazonaws.com/thumbs/263dcb74bd6e914c1992dcea67468d92" />
Peptide-Based mRNA Vaccine Offers New Hope for Neuroblastoma Treatment
In a world’s first, researchers from RCSI University of Medicine and Health Sciences in Dublin, Ireland, have developed an mRNA vaccine for neuroblastoma that has shown promising results in early laboratory testing.
Led by Olga Piskareva, senior lecturer in the RCSI Department of Anatomy and Regenerative Medicine, the study demonstrates the therapeutic potential of the vaccine for treating neuroblastoma and paves the way for further studies.
“We are at the beginning of the mRNA vaccine development journey, but we have successfully completed the first milestone, and we are very proud of it,” Piskareva told Inside Precision Medicine.
Neuroblastoma is an aggressive pediatric solid tumor that accounts for 15% of cancer-related deaths in children. Despite recent advances in treatment options, around 80% of patients with clinically aggressive disease do not show sustained responses, highlighting the need for novel treatments.
Piskareva has worked in neuroblastoma research since 2011 and felt the time was right to develop a vaccine. Her proposal was strongly endorsed in funding calls and supported by the Conor Foley Neuroblastoma Cancer Research Foundation. This support was particularly important to Piskareva as the charity was founded by a family who lost their son after a 14-year battle with neuroblastoma.
Unlike many mRNA vaccines that use lipid nanoparticles to deliver their payload, Piskareva and team instead used self-assembling peptide nanoparticles.
The self-assembling peptide, known as RALA, is composed of a repeating amino acid sequence of arginine (R), alanine (A), leucine (L), and alanine (A) that come together to form stable nanoparticles that protect mRNA encoding glypican 2 (RALA/mGPC2), a potent tumor-associated antigen in neuroblastoma. After entering a cell, the RALA nanoparticles react with the intracellular environment and change their structure, which allows them to deliver the GPC2 mRNA.
Piskareva and co-authors explain in Molecular Therapy Oncology that the RALA technology offers several advantages over more commonly used lipid nanoparticle delivery including high mRNA encapsulation efficiency, straightforward purification, no immune response to RALA itself, no restriction on the size or number of mRNA cargos to be delivered, stability at room temperature, and lower costs.
After initial experiments showing the viability of the vaccine in vitro, the researchers tested its efficacy in mouse models.
They showed that RALA/mGPC2 vaccination generated an antigen-specific cellular immune response against GPC2, with significant increases in interferon-γ and interleukin-2 expression by splenocytes and tumor necrosis factor-α expression by CD4+ and CD8+ T cells.
Investigating tumor control, the team demonstrated that immunization delayed tumor development by 10–11 days and reduced tumor volume by 70% compared with unvaccinated controls in a subcutaneous murine model of neuroblastoma, with the potential further to reduce tumor progression via prolonged administration.
Piskareva noted that as biological ageing in mice does not follow the same pattern as it does in humans, it is fair to assume that a 10–11 day delay in mice would equate to two years in preadolescent humans and one year in adulthood.
“However, the most important clinical message from this number is that there is significant potential to further delay tumor growth by trying a different vaccination schedule or dose, or by co-treating with immune-stimulating drugs,” she remarked.
The vaccine also has the potential to be highly personalized. “We can profile a given patient with neuroblastoma, select its shared and unique targets, design and synthesize mRNA, coat it with peptides, and have a personalized vaccine ready for use,” said Piskareva. “We can also create a pool of the most common targets and have the mRNA vaccine on demand.”
“By developing mRNA for multiple targets, we can increase the vaccine’s ability to help the host’s immune system kill cancer cells. The mRNA vaccine technology is like LEGO bricks. By combining different bricks, we can tailor the vaccine to the individual’s needs with high precision,” she added.
Piskareva and team are now planning further studies to investigate optimal vaccination doses and frequency, and characterize the immune response on a wider scale and in greater detail.
“The move to clinical trials will depend largely on the quality and quantity of data collected in pre-clinical studies. We will closely monitor developments in clinical trials for adult mRNA vaccines, learn from their experience and adopt the best approaches to avoid unnecessary delays,” Piskareva concluded.
The post Peptide-Based mRNA Vaccine Offers New Hope for Neuroblastoma Treatment appeared first on Inside Precision Medicine.
Catheter-Based OCT Imaging Shows Promise for Noninvasive Endometrial Cancer Diagnosis
A research team at Washington University in St. Louis has developed a catheter-based optical imaging method that could be used as an “optical biopsy” for detecting endometrial cancer and its precancerous lesions. The approach, described in the journal npj Imaging, uses three-dimensional optical coherence tomography (OCT) imaging combined with a machine learning algorithm which examines and analyzes the entire endometrial cavity to identify tissue changes associated with endometrial intraepithelial neoplasia (EIN) and endometrial cancer.
“Current endometrial biopsy practice has an estimated false-negative rate of about 10% (approximately 90% sensitivity), largely due to sampling limitations and interpretive variability,” said senior investigator Quing Zhu, PhD, a professor of engineering at Washington University. “With our three-dimensional OCT imaging system combined with machine learning, we can image the entire endometrial cavity in two to three seconds and may have a potential to achieve higher sensitivity than random biopsy sampling.”
Endometrial cancer is the most common gynecologic malignancy in the United States, with estimated 69,000 cases projected to be diagnosed in 2025. As with most cancers, early detection has a significant impact on treatment outcomes with five-year survival rates between 80% and 90% when it is diagnosed at stage I.
Existing diagnostic tools have limitations that can impact early and accurate diagnosis. For instance, transvaginal ultrasound is ineffective for early EC, while endometrial biopsy has a 10% false-negative rate due to sampling and interpretive variability.” Although hysteroscopy allows direct visualization of the uterine cavity, it does not provide information about subsurface tissue architecture.
In an interview with Inside Precision Medicine, Zhu said the most widely used diagnostic approaches can miss cancers or depend heavily on operator skill. She noted that the low resolution of transvaginal ultrasound limits detection of early disease, while operative hysteroscopy requires cervical dilation and carries procedural risks. Endometrial biopsy, she added, can miss cancers that occupy less than half of the endometrial cavity surface.
The new approach developed by Zhu and team uses OCT, a light-based imaging technology that creates high-resolution cross-sectional images of tissue. This imaging method uses low-coherence interferometry to measure the echo time delay and intensity of backscattered light, producing real-time images of tissue microstructure with micrometer-scale resolution with tissues penetration depths of approximately one to two millimeters.
To create a method to comprehensively image the endometrium the WashU team developed a custom 3.1-millimeter catheter. Zhu said that the catheter rotates within the endometrial cavity at roughly 600 revolutions per minute while being pulled back automatically at a constant speed. Depending on uterine size, a 3- to 5-centimeter segment of the cavity can be imaged in approximately two to three minutes. The resulting volumetric scans provide three-dimensional views of tissue structure and optical properties throughout the cavity. The team then applied computational analysis to identify functional, structural, and radiomic features based on OCT intensity and scattering images.
To test this OCT/machine learning approach, the researchers evaluated the technology on 57 freshly excised hysterectomy specimens representing a range of conditions, including normal endometrium, benign abnormalities, EIN, and endometrial cancer. OCT identified 34 specimens that contained either high-risk precancerous lesions or early-stage cancers.
The OCT images revealed differences among normal endometrium, benign endometrium, high-risk precancerous lesions, and cancers at different stages. This new method attained an exploratory sensitivity of 94% and specificity of 87%. A cross-validated logistic regression classifier produced sensitivity of 91% and specificity of 83%.
“These findings support catheter-based 3D OCT as a promising noninvasive optical biopsy approach to improve detection of endometrial cancer,” the researchers wrote in the abstract.
The work builds on earlier investigations of OCT in endometrial disease. Previous research had shown that OCT could distinguish endometrial pathologies, but in those studies the imaging was slow or limited to two-dimensional analysis. “This study is the first to combine catheter-based 3D OCT imaging with functional, structural and radiomic feature analysis to assess the endometrial cavity,” the researchers wrote.
Researchers believe the technology could improve patient care by reducing dependence on repeated tissue biopsies. In the introduction, they wrote that “a real-time, noninvasive, high-resolution modality for subsurface imaging could improve diagnostic accuracy, reduce unnecessary biopsies, and support fertility-sparing management.” Such a tool could be particularly useful for women undergoing serial monitoring while receiving hormone-based treatment.
The investigators describe the method as an optical biopsy because it provides diagnostic information without requiring removal of tissue. “Unlike traditional tissue biopsy, it does not require painful physical tissue samples,” Zhu told Inside Precision Medicine.
The technology is still in an early stage of development. Zhu said future development will require a forward-viewing catheter to improve imaging of the uterine fundus and developing methods for faster data acquisition.
Zhu is now looking to secure funding and begin studies in patients to establish in vivo feasibility and to eventually move the technology into clinical trials.
The post Catheter-Based OCT Imaging Shows Promise for Noninvasive Endometrial Cancer Diagnosis appeared first on Inside Precision Medicine.