Studies have revealed that non-small cell lung cancer (NSCLC) with epidermal growth factor receptor (EGFR) mutations has a high incidence of brain metastases (BMs). However, the association between EGFR mutations and BMs remains unknown. This review summarizes detailed information about the incidence of BMs, clinical and imaging characteristics of BMs, brain surveillance strategies, influence of treatments on BMs, prognosis after BMs, and differences in EGFR mutations between paired primary tumors and BMs in EGFR-mutated NSCLC. The prognostic results demonstrate that patients with mutated EGFR have a higher incidence of BMs, EGFR tyrosine kinase inhibitors (EGFR-TKIs) (afatinib and osimertinib) delay the development of BMs, and patients with mutated EGFR with synchronous or early BMs have better overall survival after BMs than those with wild-type EGFR. The EGFR mutation status of BM sites is not always in accordance with the primary tumors, which indicates that there is heterogeneity in EGFR gene status between paired primary tumors and BMs. However, the EGFR gene status of the primary site can largely represent that of BM sites. Among patients developing synchronous BMs, patients with mutated EGFR are less likely to have central nervous system (CNS) symptoms than patients with wild-type EGFR. However, the possibility of neuro-symptoms is high in patients with metachronous BMs. Patients with mutated EGFR tend to have multiple BMs as compared to patients with wild-type EGFR. Regarding very early-stage NSCLC patients without neuro-symptoms, regular neuroimaging follow-up is not recommended. Among advanced NSCLC patients with EGFR mutation, liberal brain imaging follow-up in the first several years showed more advantages in terms of cost.

Objective: Overall survival is the gold-standard outcome measure for phase 3 trials, but the need for a long follow-up period can delay the translation of potentially effective treatment to clinical practice. The validity of major pathological response (MPR) as a surrogate of survival for non small cell lung cancer (NSCLC) after neoadjuvant immunotherapy remains unclear. Methods: Eligibility was resectable stage I–III NSCLC and delivery of PD-1/PD-L1/CTLA-4 inhibitors prior to resection; other forms/modalities of neoadjuvant and/or adjuvant therapies were allowed. Statistics utilized the Mantel–Haenszel fixed-effect or random-effect model depending on the heterogeneity (I ²). Results: Fifty-three trials (seven randomized, 29 prospective nonrandomized, 17 retrospective) were identified. The pooled rate of MPR was 53.8%. Compared to neoadjuvant chemotherapy, neoadjuvant chemo-immunotherapy achieved higher MPR (OR 6.19, 4.39–8.74, P<0.00001). MPR was associated with improved disease-free survival/progression-free survival/event-free survival (HR 0.28, 0.10–0.79, P=0.02) and overall survival (HR 0.80, 0.72–0.88, P<0.0001). Patients with stage III (vs I/II) and PD-L1 ≥1% (vs <1%) more likely achieved MPR (OR 1.66,1.02–2.70, P=0.04; OR 2.21,1.28–3.82, P=0.004). Conclusions: The findings of this meta-analysis suggest that neoadjuvant chemo-immunotherapy achieved higher MPR in NSCLC patients, and increased MPR might be associated with survival benefits treated with neoadjuvant immunotherapy. It appears that the MPR may serve as a surrogate endpoint of survival to evaluate neoadjuvant immunotherapy.

Agentic artificial intelligence (AI) systems, designed to autonomously reason, plan, and invoke tools, have shown promise in healthcare, yet systematic benchmarking of their real-world performance remains limited. In this study, we evaluate two such systems: the open-source OpenManus, built on Meta’s Llama-4 and extended with medically customized agents; and Manus, a proprietary agent system employing a multistep planner-executor-verifier architecture. Both systems were assessed across three benchmark families: AgentClinic , a stepwise dialog-based diagnostic simulation; MedAgentsBench , a knowledge-intensive medical QA dataset; and Humanity’s Last Exam (HLE), a suite of challenging text-only and multimodal questions. Despite access to advanced tools (e.g., web browsing, code development and execution, and text file editing) agent systems yielded only modest accuracy gains over baseline LLMs, reaching 60.3% and 28.0% in AgentClinic MedQA and MIMIC, 30.3% on MedAgentsBench, and 8.6% on HLE text. Multimodal accuracy remained low (15.5% on multimodal HLE, 29.2% on AgentClinic NEJM), while resource demands increased substantially, with >10× token usage and >2× latency. Although 89.9% of hallucinations were filtered by in-agent safeguards, hallucinations remained prevalent. These findings reveal that current agentic designs offer modest performance benefits at significant computational and workflow cost, underscoring the need for more accurate, efficient, and clinically viable agent systems.

Background: Significant progress has been made in the investigation of neoadjuvant immune-chemoradiotherapy (NICRT) and neoadjuvant immune-chemotherapy (NICT) on the outcomes of esophageal cancer patients. To summarize the current developments, a systematic review and meta-analysis were conducted to evaluate the efficacy and safety of neoadjuvant immunotherapy combined with chemoradiotherapy or chemotherapy.Methods: A search strategy of prospective studies on esophageal cancer receiving neoadjuvant immunotherapy was predefined to scan PubMed, Embase, Cochrane, and additional major conferences for prospective studies. Efficacy was assessed by pathological complete response (pCR), major pathological response (MPR), and R0 resection rates. Safety was evaluated based on the incidence of grade ≥ 3 treatment-related adverse events (TRAEs), neoadjuvant therapy completion rate, surgical resection rate, and surgical delay rate. Differences between the NICRT and NICT groups were also analyzed.Results: A total of 38 studies qualified for the analysis. The pooled pCR, MPR, and R0 resection rates were 30, 58, and 99%, respectively. The pCR and MPR in the NICRT vs. NICT group were 38% vs. 28% (p=0.078) and 67% vs. 57% (p=0.181), respectively. The pooled incidence of grade ≥ 3 TRAEs was 24% (NICRT,58%, I² = 61% vs. NICT,18%, I² = 79%; p<0.001). In addition, the pooled neoadjuvant therapy completion and surgical resection rates were 92% and 85%, respectively; the difference was not statistically significant between the NICRT and NICT groups.Conclusions: Neoadjuvant immunotherapy combined with chemoradiotherapy or chemotherapy is effective and safe in the short term for locally advanced esophageal cancer. However, further randomized trials are needed to confirm which combined model is more favorable.Systematic review registration: https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42021284266, identifier CRD42021284266.