Brain metastases are among the most serious complications of lung cancer. When a tumor spreads to the brain, stereotactic radiosurgery, or SRS, is frequently used to deliver highly focused radiation to one or more lesions while limiting exposure to surrounding healthy tissue. Gamma Knife is one such system, using numerous precisely aimed beams that converge on a target. The prescription dose is commonly selected using factors such as tumor size, location, number of lesions, prior treatments, and institutional protocols. Yet two metastases that appear similar on a scan can behave very differently. Some remain controlled for years, while others recur locally despite treatment, creating a need for models that can estimate tumor-specific outcomes more precisely.

THINKERS-Lung was created to address that problem by treating radiation dose as an active variable within the prediction process. Instead of producing a single fixed forecast based only on patient and tumor characteristics, the model can repeatedly assess the same metastasis under different candidate margin doses. In practical terms, clinicians could provide the system with information available before or at the time of treatment, then compare the predicted probability of local failure at several dose levels. The goal is not for the algorithm to replace a radiation oncologist or neurosurgeon, but to show how the expected benefit of changing the dose may vary from one lesion to another.

The researchers trained the system using retrospective data from 767 patients who underwent Gamma Knife radiosurgery for a total of 3,728 lung cancer brain metastases. The analysis was limited to variables known by the time of treatment, a design choice intended to make the model more realistic for clinical decision-making. These variables included patient-related information and characteristics of the treated lesions, while the margin dose was entered explicitly into the model. The team used a mixture-of-experts, or MoE, neural network combined with discrete-time survival modeling. This architecture allows multiple specialized neural components, known as experts, to learn different patterns in the data, while a gating mechanism determines how much each expert should influence a prediction for a particular case.

The survival component is important because local control is not simply a yes-or-no event. A metastasis may remain controlled at six months but fail at 14 months, and that timing carries clinical meaning. Discrete-time survival modeling divides follow-up into intervals and estimates the probability that failure will occur during each period, while accounting for lesions that have not failed by the end of observation. This approach enables THINKERS-Lung to estimate both the likelihood of local failure and the expected time until that failure. According to the study, the model achieved a mean area under the curve of 0.876 for predicting 12-month local failure during grouped five-fold cross-validation.

The area under the curve, or AUC, measures how effectively a prediction system distinguishes between lesions that experience an event and those that do not. A value of 0.5 corresponds roughly to random discrimination, while a value of 1.0 represents perfect separation. In a patient-grouped holdout test set, THINKERS-Lung produced an AUC of 0.863, with a 95 percent confidence interval ranging from 0.776 to 0.942. The model’s mean absolute error for time-to-failure prediction was 0.99 months in cross-validation and 1.26 months in the holdout test set. These results suggest that the framework was able to rank relative risk and estimate outcome timing with useful accuracy within the study population.

The investigators also assessed calibration, which asks whether predicted probabilities correspond to what actually happens. A model can have strong discrimination while still producing probabilities that are systematically too high or too low. THINKERS-Lung recorded a Brier score of 0.061, a calibration intercept of 0.18, and a calibration slope of 0.87. The Brier score evaluates the overall accuracy of probabilistic predictions, with lower values generally indicating better performance. A calibration slope close to 1 suggests that the spread of predictions is appropriate, while the intercept reflects whether the model tends to underpredict or overpredict overall risk. Together, these figures indicate favorable internal probabilistic performance, although they do not establish that the system will perform equally well in other hospitals or patient populations.

The study’s most distinctive feature is its proposed use as a comparative dose evaluation tool. A conventional prognostic model might estimate whether a patient is likely to experience recurrence after receiving a particular treatment. THINKERS-Lung instead allows the dose to be changed computationally while the other characteristics remain constant. This creates a series of counterfactual estimates: What might the predicted local-failure risk be at one dose, and how might it change at another? Such estimates could help physicians identify cases in which a higher dose appears likely to offer additional tumor control, as well as cases in which escalating radiation may provide little predicted advantage relative to the potential risks to nearby brain structures.

Despite the promising results, the model remains an investigational decision-support system rather than a validated treatment guideline. The study was retrospective and conducted at a single center, meaning that its data may reflect local patient-selection practices, imaging protocols, follow-up schedules, and Gamma Knife procedures. The authors performed grouped validation by patient to reduce the risk that multiple lesions from the same individual would appear in both training and testing data, but the model still requires external validation using independent, multi-institutional cohorts. Its predictions may also be affected by changes in systemic lung cancer therapy, including targeted treatments and immune checkpoint inhibitors, which can influence intracranial disease behavior. Before clinical adoption, prospective studies will be needed to determine whether AI-guided dose selection actually improves local control, reduces toxicity, or changes patient outcomes.

THINKERS-Lung is part of a broader effort to combine neural networks with clinically interpretable reasoning in radiosurgery. Rather than treating AI as a black box that produces an isolated risk score, the framework is designed to connect tumor-specific prediction with a treatment variable that physicians can examine directly. If future testing confirms its performance, this type of hybrid intelligence could allow radiosurgical planning to move beyond one-size-fits-all dose prescriptions. For patients with lung cancer brain metastases, the longer-term vision is a treatment strategy in which radiation intensity, expected local control, and the timing of possible failure are evaluated together for each lesion. The new findings do not yet prove that the system should guide routine care, but they offer a glimpse of how machine learning could make highly focused cancer treatment more individualized.

Subject of Research: AI-guided personalized stereotactic radiosurgery dose evaluation for lung cancer brain metastases

Article Title: Therapeutic hybrid intelligence with neural and knowledge-based expert reasoning for SRS (THINKERS): an AI model for lung cancer brain metastases

Article References: Reyes JS, Bouras A, Lunsford LD, et al. “Therapeutic hybrid intelligence with neural and knowledge-based expert reasoning for SRS (THINKERS): an AI model for lung cancer brain metastases.” Journal of Neuro-Oncology 179, article 56 (2026). https://doi.org/10.1007/s11060-026-05771-5

Image Credits: AI Generated

DOI: 10.1007/s11060-026-05771-5

Keywords: Lung cancer brain metastases, stereotactic radiosurgery, Gamma Knife, artificial intelligence, mixture-of-experts neural network, survival modeling, local failure, personalized radiation dose, clinical decision support

Brain metastases occur in approximately 30% of patients suffering from these primary cancers, and despite advances in oncology, the prognosis remains grim, with median survival times falling below one year. Therapeutic options have been severely limited by the unique challenges posed by the brain’s protective barriers and microenvironment. Traditional chemotherapeutic agents and immunotherapies often fail to reach metastatic brain tumors in adequate concentrations due to the restrictive nature of the BBB. Surgical intervention is typically feasible only in select cases, further underscoring the urgent need for innovative strategies that can effectively target and eradicate brain metastases.

Addressing these challenges head-on, the researchers harnessed the natural abilities of macrophages, engineering them to express chimeric antigen receptors (CARs) specific to mesothelin, thus creating mesothelin-targeting chimeric antigen receptor macrophages (CAR-Ms). To bolster their immune efficacy and capacity for tumor cell phagocytosis, these macrophages were further fused with the MyD88 immune signaling domain, a vital adaptor molecule that amplifies inflammatory responses and pathogen defense mechanisms. This fusion gave rise to a new cellular entity described as chimeric antigen receptor macrophages fused with MyD88, or CARMA.

CARMA macrophages exhibit remarkable antitumor activity by selectively recognizing mesothelin on the surface of metastatic tumor cells in the brain. Importantly, their mode of action surpasses mere antigen-specific phagocytosis. Beyond directly engulfing and destroying tumor cells expressing mesothelin, CARMA cells secrete tumor necrosis factor (TNF), a potent cytokine that induces apoptosis in adjacent tumor cells even when they lack the targeted antigen. This dual mechanism endows CARMA with a superior ability to restrain the heterogeneous tumor populations characteristic of metastatic brain disease, addressing one of the central challenges in cancer immunotherapy.

In rigorous preclinical evaluation, CARMA demonstrated a robust capacity to penetrate the BBB—a formidable obstacle for many therapeutics—effectively reaching and infiltrating metastatic lesions within the brain parenchyma. Utilizing a humanized mouse model that closely mimics human immune responsiveness, the engineered macrophages were able to significantly curb tumor growth, exhibiting both antigen specificity and a powerful bystander effect through TNF-mediated cytotoxicity. These findings underscore the potential of macrophage-based immunotherapy in overcoming the current therapeutic inefficacies seen in brain metastases.

The novelty and success of this approach rest not only on CARMA’s ability to breach the BBB but also on the strategic enhancement of its phagocytic and immune signaling capabilities via MyD88. The MyD88 signaling module intensifies the macrophage’s immune activation state, ensuring prolonged survival, enhanced cytokine production, and a sustained cytotoxic assault on metastatic cells. This molecular synergy within CARMA empowers a level of immune orchestration and tumor targeting previously unattainable using conventional CAR-T cell therapies or unmodified macrophage approaches.

Furthermore, safety considerations, a critical aspect in immunotherapy design, have been judiciously addressed through the antigen specificity of CARMA. By targeting mesothelin—a tumor-associated antigen with limited expression in normal tissues—the therapy aims to minimize off-target effects and systemic toxicity. Also, leveraging macrophages’ natural tropism for tumors may help localize potent immunological actions within the tumor microenvironment, reducing the likelihood of systemic inflammatory responses that have complicated other immune-based therapies.

The clinical implications of CARMA therapy extend well beyond brain metastases from lung, melanoma, or breast cancers. Given macrophages’ ubiquitous presence and ease of manipulation, this platform could be adapted to target a range of other tumor-associated antigens across different malignancies with central nervous system involvement. Additionally, the modular nature of CAR engineering allows customization of immune signaling domains to optimize therapeutic profiles for various tumor types and microenvironments.

While still in preclinical stages, the success of CARMA’s design and function opens an exciting vista for future clinical trials aimed at evaluating its safety, dosing, and therapeutic efficacy in human patients. If translated successfully, CARMA could redefine standards of care for metastatic brain disease, a condition that has long been an unmet medical need due to limited and often ineffective treatment options. The potential to extend life expectancy and improve quality of life for thousands of affected patients worldwide is vast.

This innovation also revives broader discussions about the utility of innate immune cells in adoptive cell transfer therapies. Although CAR-T cell therapies have transformed certain hematological malignancies, their efficacy in solid tumors, especially within the central nervous system, remains limited. The CARMA model propels macrophages into the spotlight as versatile and potent effectors capable of overcoming anatomical and cellular hurdles that impede other immune cells.

Moreover, the inducible signaling from MyD88 within CARMA macrophages exemplifies an intelligent design approach to amplify antitumor immunity without exacerbating systemic inflammation. Leveraging innate immune pathways to coordinate targeted killing and inflammatory signaling marks a paradigm shift, integrating biological insights into the engineering of next-generation immunotherapies that are both effective and potentially safer.

The development of CARMA macrophages underscores a thoughtful and strategic convergence of cellular biology, immunology, and bioengineering aimed at resolving a critical clinical problem. It further epitomizes the potential of marrying innate immune functions with synthetic biology to craft therapeutic solutions addressing diseases located in sanctuary sites protected by formidable physiological barriers.

As the research community lauds CARMA’s preclinical accomplishments, attention now turns toward translational strategies, including scalable manufacturing processes, long-term safety profiling, and understanding interactions within the complex tumor-immune microenvironment of human patients. The implications for personalized medicine are profound, as CARMA therapies could be tailored to specific antigen profiles and disease contexts, offering bespoke immunotherapeutic regimens for individuals suffering from brain metastases and potentially other metastatic cancers.

Ultimately, the promise of CARMA may herald a new era in neuro-oncology and immunotherapy—a future where the immune system’s innate sentinels are endowed with precision-targeted weaponry, navigating the tightly regulated realms of the brain to eradicate metastatic disease and offer renewed hope to patients facing dismal prognoses.

Subject of Research:
Genetically engineered macrophages with Chimeric Antigen Receptors targeting mesothelin and fused with MyD88 signaling domain to treat metastatic brain tumors.

Article Title:
MyD88-mediated chimaeric antigen receptor macrophages suppress brain metastasis using target-specific phagocytosis.

Article References:
Wu, SY., Tyagi, A., Wu, K. et al. MyD88-mediated chimaeric antigen receptor macrophages suppress brain metastasis using target-specific phagocytosis. Nat. Biomed. Eng (2026). https://doi.org/10.1038/s41551-026-01613-x

Image Credits: AI Generated

DOI:
https://doi.org/10.1038/s41551-026-01613-x