Background:Early-onset lung squamous cell carcinoma (LUSC) is a rare and poorly characterized entity. Its distinct clinicopathological, genomic, and tumor immune microenvironment (TIME) profiles remain to be elucidated. This study sought to explore the molecular landscape and immunophenotype of early-onset LUSC, with a focus on uncovering potential genomic associations with its clinical features. Methods:In this retrospective, single-center cohort study, we conducted a comprehensive multi-omics analysis on patients with surgically resected, treatment-naïve early-onset LUSC (subjects aged ≤40 years, n=21) between 2015 and 2024 using whole-exome sequencing (WES) and digital spatial profiling (DSP). To contextualize our findings, we conducted a comparative analysis using publicly available LUSC data from The Cancer Genome Atlas (TCGA) via the Xena platform. Results:WES revealed a high mutation frequency of CDKN2A (47.6%), which consistently co-occurred with TP53 mutation. Stratification by CDKN2A status identified two subgroups: CDKN2A-mutant (CDKN2Amut, n=10) and CDKN2A-wild-type (CDKN2Awild, n=11). The CDKN2Amut group exhibited more aggressive clinicopathological features, including higher Ki67 index and frequent vascular invasion, alongside a distinct genomic profile. Despite its aggressive phenotype, the CDKN2Amut group displayed a robust "hot" TIME, characterized by enriched cytotoxic and exhausted CD8+ T cells, a higher intra-tumoral CD8+/Foxp3+ ratio, and significantly elevated programmed death-ligand 1 (PD-L1). Comparative analysis with public LUSC databases indicated that this specific immunogenic association was unique to the early-onset context and not observed in general LUSC populations. Patients in CDKN2Amut group showed a trend toward shorter overall survival (OS). Conclusions:Our study identifies CDKN2A mutation as a hallmark of a unique, aggressive yet immunogenic subtype of early-onset LUSC. This association appears to be specific to young patients, highlighting age of onset as a key biological variable. These findings provide a rationale for further investigation of CDKN2A status as a potential biomarker to guide immunotherapy strategies in early-onset LUSC.

e23515 Background: Accurate risk stratification is essential for guiding postoperative adjuvant therapy in resected gastrointestinal stromal tumors (GIST). Traditional Joensuu criteria rely mainly on mitotic count, together with tumor size and primary site. While immunohistochemistry is routinely used for diagnosis, integrating biomarkers into prognostic assessment may improve identification of patients at high risk of recurrence after curative resection. We developed a prognostic model incorporating a proliferation marker (Ki-67) and a vascular marker (CD34) to improve postoperative risk stratification. Methods: We retrospectively collected 634 patients with localized GIST who underwent curative resection at the National Cancer Center of China. The primary endpoint was disease-free survival (DFS). A multivariable Cox regression model including Ki-67 index, CD34 expression, tumor size, and sex was developed; mitotic count and tumor site were evaluated for inclusion. Discrimination was compared with the modified Joensuu criteria using Harrell’s concordance index (C-index) and log-rank tests. Results: In this cohort, the Joensuu criteria showed limited DFS separation between intermediate- and high-risk groups (log-rank P=0.17). In multivariable analysis, Ki-67 index and CD34 positivity were independent prognostic factors. Mitotic count and tumor site were evaluated but were not retained in the final multivariable model. The integrated model achieved a C-index of 0.766 (95% CI, 0.632–0.877), higher than the Joensuu criteria (C-index 0.694; 95% CI, 0.573–0.795). In multivariable Cox regression for DFS (Table), higher Ki-67 index and larger tumor size were associated with increased risk, whereas CD34 positivity and female sex were protective. The model provided improved discrimination among higher-risk patients, with significant separation between intermediate- and high-risk groups (P=0.002), whereas DFS was similar between low- and intermediate-risk groups (P=0.86). Conclusions: A prognostic model integrating Ki-67 and CD34 may refine DFS risk stratification after GIST resection, particularly at the intermediate- versus high-risk boundary where standard criteria show limited discrimination. This approach may support more tailored selection of patients for postoperative adjuvant therapy. Multivariable analysis of factors associated with DFS. Variable HR 95% CI P value Ki-67 index (per 1% increase) 1.04 1.01-1.07 0.007 Tumor size (per 1 cm increase) 1.08 1.01-1.16 0.019 CD34 positive (vs negative) 0.36 0.16-0.81 0.014 Female (vs male) 0.37 0.17-0.81 0.013

Hepatocellular carcinoma (HCC) was characterized by a highly complex genome, with structural variations (SVs) playing a significant role in its development. In this study, we employed Oxford Nanopore Technology long-read sequencing in paired tumor and adjacent normal liver tissues from 74 Chinese HCC patients to thoroughly characterize the landscape of somatic SVs. Our analysis revealed that somatic SVs were more prevalent in hepatitis B virus (HBV)-related HCC, with chromosome 1 emerging as a major hotspot, and several members of the chromosome 1 open reading frame (C1orf) family genes expression level exhibited significant age-related difference. Notably, HBV-related HCC cases exhibited a higher frequency of deletions, particularly among younger ones (≤ 35 years old). In addition, we observed an increased burden of HBV integration events in younger ones. Remarkably, the divergent-paired related homeobox ( DPRX ) loci was identified as a novel gene for HBV integration in younger patients. Together, these findings delineated the somatic SV landscape in HCC and underscored age-associated HBV-related genomic alterations as key pathological features of hepatocarcinogenesis.

Molecular subtyping plays a critical role in breast cancer management, yet immunohistochemistry (IHC) methods for detecting protein expression levels are hindered by time constraints and interpretational challenges. The association between molecular expression and pathological morphology suggests the potential to predict subtypes from haematoxylin and eosin (H&E) images, offering deeper molecular insights beyond protein expression. However, acquiring molecular information from whole slide images (WSIs) of breast H&E slides remains challenging due to gigapixel resolution and weak labelling. Here, a weak-supervised learning framework based on multi-attribute feature embedding for IHC subtype prediction directly from breast H&E images is presented. Leveraging a multi-centre breast cancer dataset comprising 3,721 H&E images from 1,688 patients, the model achieves the highest mean area under the ROC curve (AUC=0.868) compared to state-of-the-art methods. Validation through IHC testing on 20 cases demonstrates the potential clinical utility of pathological deep features. Patch probability visualization reveals potential morphological associations with molecular subtypes, which generally aligns with protein expression truth in IHC images. Furthermore, multi-omics analysis involving pathological deep features facilitates the exploration of important molecular properties. Global feature visualization elucidates pathological heterogeneity, while differential gene expression identifies potential gene markers and functional heterogeneity among breast cancer subtypes. By combining pathology and transcriptome analysis, the study reveals correlations between pathological features and gene expression. Finally, survival analysis suggests that subtyping probabilities and pathological deep features are potential survival indicators for clinical prognosis.

Epidermal growth factor receptor (EGFR) exon 19 deletions (19del) in non-small cell lung cancer (NSCLC) confer sensitivity to tyrosine kinase inhibitors (TKIs). However, the efficacy of the individual EGFR 19del variant remains unclear. This study aims to dissect the structural disparities of various EGFR 19del mutations and associate these variations with clinical outcomes. Here, a total of 399 patients from two cancer centers covering 35 distinct EGFR 19del variants were included in this study. Through molecular docking analysis, we identified a structure-based subgroup of EGFR 19del mutations, termed CLASS 3G (E746_A750>X, E746_S752>V, L747_A750>P, and L747_T751>P). In silico and in vitro analysis demonstrated that CLASS 3G mutations had reduced binding affinity to third-generation EGFR-TKIs. Clinically, patients with CLASS 3G showed inferior PFS to non-CLASS 3G after receiving third-generation EGFR-TKIs (first-line progression-free survival: 12.8 vs. 25.6 months, P=0.045; second-line progression-free survival: 7.7 vs. 15.2 months, P=0.039). This finding was further validated in an independent cohort, confirming consistent results with the primary analysis. Additionally, we found that CLASS 3G mutations derived significant benefits from third-generation EGFR-TKIs plus chemotherapy compared with third-generation EGFR-TKIs monotherapy (32.5 vs. 12.8 months, P=0.0038). Our study identified a subgroup of EGFR 19del that responded poorly to third-generation EGFR-TKIs based on structural features of the EGFR tyrosine kinase domain. These patients benefit substantially from the combination treatment of third-generation EGFR-TKIs plus chemotherapy.

Mixed invasive mucinous adenocarcinoma and non-mucinous adenocarcinoma (mixed IMA/NMA) is a rare subtype of lung adenocarcinoma (LUAD) with limited available data. This study aimed to comprehensively analyze the characteristics of this rare entity. A total of 738 surgical cases were enrolled, including 349 pure invasive mucinous adenocarcinoma (IMA), 61 mixed IMA/NMA and 328 pure non-mucinous adenocarcinoma (NMA) cases. Using amplification refractory mutation system, immunohistochemistry, and DNA-/RNA-based next-generation sequencing (DNA and RNA NGS), distinct molecular features were identified in mixed IMA/NMA cases compared with IMA and NMA cases, particularly in EGFR, KRAS, and ALK alterations and PD-L1 expression status. Paired analysis of the IMA and NMA components within mixed IMA/NMA cases using DNA and RNA NGS revealed one to three shared genomic alterations between the two components in the same tumor. However, significant differences were observed in the levels of cancer-associated fibroblasts, protumor cytokines, MHC-II, coactivation molecules, T cells, and effector cells between the components. Similarly, multiplex immunofluorescence assay demonstrated that immune cell infiltration, including CD4+ and CD8+ T cells, was significantly higher in the NMA components compared to the IMA components. Postoperative follow-up revealed no significant difference in disease-free survival (DFS) or overall survival (OS) between NMA and mixed IMA/NMA cases; however, both groups showed significantly shorter DFS (P = 0.011) and OS (P = 0.027) compared to IMA cases. Together, this study provides a comprehensive characterization of the molecular profiles, clonal relatedness, tumor heterogeneity, and surgical outcomes of mixed IMA/NMA, which may inform diagnostic and therapeutic strategies for this rare LUAD subtype.

BACKGROUND:RET fusions represent actionable oncogenic drivers in lung adenocarcinoma (LUAD). However, the clinicopathological features, co-mutation landscape, and therapeutic outcomes across different fusion partners remain incompletely characterized. METHODS:We retrospectively analyzed 268 patients with RET fusion-positive LUAD diagnosed between July 2017 and December 2024. Clinicopathological features, metastatic patterns, and concomitant genomic alterations were compared between KIF5B and non-KIF5B subgroups. Treatment outcomes of selective RET inhibitors, multikinase inhibitors (MKIs), and immunotherapy combined with chemotherapy were evaluated, with subgroup analyses according to fusion partners. RESULTS:The median age at diagnosis was 58 years; most patients were female (57.0%) and never/light smokers (60.8%). Bone (12.3%), pleural (11.9%), and brain metastases (6.7%) were the most common metastatic sites. KIF5B-RET fusions were more frequently detected in earlier disease stages compared with non-KIF5B (p = 0.0531). Among 274 RET fusion events, KIF5B-RET was predominant (65%), followed by CCDC6-RET (16%) and NCOA4-RET (1%). Concomitant mutations were identified in 28.7% of patients, most commonly TP53 (39.0%) and CDKN2A (13.0%), with CDKN2A alterations predominantly consisting of functionally inactivating SNVs concurrent with shallow deletions more enriched in non-KIF5B (p = 0.0393). Nineteen patients received targeted therapy, including pralsetinib, selpercatinib, and cabozantinib, achieving an overall ORR of 57.9% and mPFS of 12.0 months. Notably, non-KIF5B patients demonstrated longer median PFS than KIF5B patients under pralsetinib (17.0 vs. 5.5 months, p = 0.0473). Fifteen patients received first-line immunochemotherapy, achieving a median PFS of 17.0 months, ORR of 40.0%, and DCR of 80.0%, comparable to targeted therapy (p = 0.3871). CONCLUSIONS:RET fusion-positive LUAD comprises biologically heterogeneous subsets defined by fusion partners. KIF5B-RET tend to occur at earlier stages, whereas non-KIF5B are more frequently associated with CDKN2A co-mutations and may derive greater benefit from selective RET inhibition. Immunochemotherapy demonstrates comparable efficacy regardless of fusion partner, highlighting the need for partner-specific therapeutic strategies in RET fusion-positive LUAD.

Precision oncology strategies guided by tumor molecular profiling often target key genomic aberrations in patients. Herein, we assembled National Cancer Center-Clinical Diagnostics Knowledgebase, compiling clinical targeted sequencing data from 6935 tumor tissues and matched normal samples, along with available pathological and clinical information. Comprehensive genomic profiling was conducted to characterize tumor type-specific somatic alterations, and comparative analyses were performed across distinct cohorts. Key genomic characteristics included high-frequency alterations in TP53 (57.8%), APC (22.6%), KRAS (21.3%), and EGFR (17.5%), among which EGFR mutations were significantly enriched in lung adenocarcinoma patients. In this cohort, 70.2% of the samples harbored at least one clinically actionable genomic aberration. 14.9% of patients showed high tumor mutational burden (TMB > 10 mutations/Mb), and the TMB level was significantly higher in patients with microsatellite instability-high than in those with microsatellite stability. We also correlated next-generation sequencing (NGS) results with conventional molecular pathology assays. We found high consistency between ERBB2 focal amplification cases determined by NGS and clinically targetable ERBB2 amplification/HER2 overexpression cases. In conclusion, this study constructed a large-scale real-world genomic dataset representative of Chinese cancer patients, spanning multiple tumor types. Moreover, our findings underscore the clinical value of NGS in identifying patients with ERBB2 amplification who may potentially benefit from targeted treatments, particularly in non-small-cell lung cancer cases where NGS panel testing is prioritized.

Certain hereditary syndromes increase the incidence of neuroendocrine tumors (NETs), suggesting germline variation may play a role in NET pathogenesis. However, the germline mutation spectrum in Chinese NET patients remains unclear. This study aimed to investigate the germline genetic variations and associated clinicopathological features in Chinese NET patients. A single-center retrospective study was conducted at the Cancer Hospital of Chinese Academy of Medical Sciences; 453 histologically confirmed NET patients who received germline genetic testing between June 2018 and March 2025 were enrolled. Germline DNA extracted from saliva or blood samples was analyzed using multigene panels and whole-exome sequencing. Results showed 11.5% (52/453) cases carried germline pathogenic or likely pathogenic variants (P/LPVs) across 28 cancer predisposing genes. MEN1 was the most frequently mutated gene, accounting for 3.5% (16/453) of all enrolled patients, followed by PALB2, SDHB, and BRIP1. The most common variant of uncertain significance (VUS) was found in MUTYH. Compared with non-carriers, P/LPV carriers were characterized by a significantly younger age at diagnosis (p < 0.001), male predominance (p = 0.008), higher tumor grade (p = 0.006), a stronger family history of cancer (p = 0.047), different primary tumor locations (p < 0.001), lower somatostatin receptor 2 expression (p = 0.010), and more aggressive tumor behavior, including higher metastatic rate (p = 0.002) and advanced stage (p = 0.011). Furthermore, patients with MEN1 P/LPVs had more mediastinal tumors and a younger age than those with non-MEN1 P/LPVs. This study, which is the largest to date regarding germline variations in Chinese patients with NETs, reveals a distinct mutational profile and identifies the unique clinicopathological features among carriers of germline P/LPVs.

BACKGROUND:The tumor and node metastasis (TNM) staging and pathological grading systems are currently insufficient for accurately predicting recurrence-free survival (RFS) in patients with pathological stage (p-stage) I lung adenocarcinoma (LUAD). Therefore, there is an urgent need for a more economical and applicable clinical prediction model to assess the risk of recurrence and guide clinical postoperative care. METHODS:This retrospective study included 544 patients with p-stage I LUAD who were randomly allocated to development (272 patients) and validation (272 patients) cohorts. Cox regression and backward model selection were used to develop the prediction model. The predictive performance of the model was then compared with that of the current TNM staging system and two major pathological grading systems. The primary endpoint was RFS. RESULTS:A total of 79 out of 544 patients with p-stage I LUAD experienced recurrence after surgery. Four risk factors were incorporated into a weighted risk index-high-grade patterns ratio, epidermal growth factor receptor mutation status, spread through air spaces status and consolidation tumor ratio-to establish the "CEHS" RFS prediction model. This model demonstrated superior predictive accuracy compared with existing staging and grading systems. High-risk patients had significantly shorter RFS than low-risk patients did. An online algorithm based on the CEHS model was also developed. CONCLUSION:We established and validated a novel model that integrates radiological, molecular and pathological features to predict RFS in patients with p-stage I LUAD. This new model exhibited excellent discriminatory power for classifying early-stage LUAD patients at different risks of recurrence.

Squamous cell carcinoma (SCC) is a highly heterogeneous and aggressive cancer type with significant global mortality. While environmental and genetic risk factors contribute to its development, the underlying mutational processes remain poorly characterized. Mutational signatures, which reflect specific patterns of somatic mutations, provide critical insights into the molecular mechanisms driving tumorigenesis. However, a comprehensive analysis of mutational signatures across SCC subtypes and their associations with clinical outcomes is lacking. We conducted a comprehensive analysis from 16 publicly available cohorts representing four SCC subtypes: lung SCC (LSCC), head and neck SCC (HNSC), esophageal SCC (ESCC), and cervical SCC (CESC). Based on COSMIC v3.3, we identified representative mutational signatures of each cancer type. Associations between signatures, clinical parameters, and survival outcomes were evaluated using Kaplan–Meier analysis, Cox regression, and Fisher’s exact tests. Eight representative single-base substitution (SBS) signatures were identified across the four SCC subtypes. Common signatures included APOBEC-associated SBS2/SBS13 and aging-related SBS5, while subtype-specific signatures such as tobacco-related SBS4 and replication-associated SBS16 were also observed. Notably, SBS16 was significantly associated with shorter overall survival in ESCC, HNSC, and LSCC. In contrast, SBS10b was associated with improved survival in CESC. Furthermore, we found significant associations between specific gene mutations and mutational signatures. For example, PIK3CA mutations were positively correlated with APOBEC signatures in LSCC and ESCC, while KRAS exhibited a negative association with APOBEC signatures in CESC. Distinct mutation patterns in genes like PIK3CA and TP53 were observed in tumors with and without specific signatures, highlighting the interplay between mutational processes and driver mutations. Subtype-specific clustering based on signature fractions revealed potential associations with clinical factors, such as smoking history and anatomic site. Our study provides a detailed characterization of shared and subtype-specific mutational signatures across SCCs, offering insights into their molecular heterogeneity and underlying carcinogenic processes. The associations between mutational signatures, clinical outcomes, and gene mutations underscore the potential of mutational signatures as biomarkers for prognosis and personalized therapy. These findings enhance our understanding of SCC biology and pave the way for precision oncology approaches tailored to individual genomic profiles.

Background Introducing molecular features into diagnostic criteria in the WHO Classification of Renal Tumours, Fifth Edition (2022), significantly improves the discrimination between morphologically similar neoplasms. Renal cell tumours with eosinophilic cytoplasm often exhibit overlapping histological features, making the differential diagnoses challenging. Methods In this retrospective study, we investigated 16 cases of atypical eosinophilic renal cell tumours (ATERCT), with four renal oncocytoma (RO) and three classic chromophobe renal cell carcinomas (chRCC) as controls. Morphology, immunohistochemistry (IHC), special staining and whole-exome sequencing (WES) were employed to facilitate the subtyping of these cases. Results Based on histomorphological features (tumour growth pattern and nuclear morphology), IHC markers (CK7, CK20, CD117, and mTOR) and special colloidal iron staining, 7 cases were classified including 2 eosinophilic chRCC, 2 RO, 1 eosinophilic vacuolated tumour (EVT), 1 eosinophilic solid and cystic renal cell carcinoma (ESC RCC) and 1 low-grade eosinophilic tumour (LOT).The gene mutational profiles by WES confirmed 5 of the 7 cases with marker gene mutations, and suggested the pathological types of 6 cases in the remaining 9 cases, which left 3 cases unresolved. The landscape of mutational profiling demonstrated an individual-specific pattern in the 23 patients that no single mutation definitively distinguished among RO, eosinophilic chRCC, LOT, EVT, or ESC RCC. A panel covering 69 mutations from 34 genes could segregate the 23 patients into 6 distinct groups by consensus clustering analysis. By integrating morphological, IHC, special staining and mutational features, we found a combination of characters of FGFR1 p.D44del / microscopic tumour borders / solid nest flakes / nucleomorphs / CD117 might represent the strongest power of subtyping the 16 patients by a L1-regularized supportive virtual machine algorithm. Conclusions These findings indicated that genetic mutations were diverse among eosinophilic renal cell tumours and particular mutations might facilitate the differential diagnosis. The integration of morphological, IHC, special staining and mutational features could enhance subtyping diagnostically challenging eosinophilic ATERCT.

Lymph node metastasis (LNM) is a critical prognostic and therapeutic determinant in small cell lung cancer (SCLC), yet its spatial cellular ecosystem remains poorly understood. Here, we perform single-cell spatial transcriptomics using the CosMx Spatial Molecular Imager on 105 primary and metastatic lymph node specimens from 75 SCLC patients, generating a comprehensive atlas of over 600,000 cells. We identify three LNM-enriched malignant subclusters with distinct metabolic and angiogenic programs that spatially correlate with immune exclusion features. Spatial analysis reveals vascular-immune crosstalk, wherein endothelial cells orchestrate immune activation through avoidance of malignant cells while forming functional perivascular niches with cytotoxic T cells during LNM. Cellular neighborhood analysis delineates distinct multicellular niches and identifies a pan-immune hotspot (PIHs-1) whose abundance is an independent predictor of survival. This study provides a high-resolution spatial map of the SCLC tumor microenvironment during LNM and establishes spatially defined architectures as both mechanistic insights and translatable biomarkers.

Small cell lung cancer (SCLC) is a highly aggressive neuroendocrine carcinoma with a poor prognosis and limited therapeutic advances. In this study, we performed deep proteomic profiling of 134 patients with resectable, limited-stage SCLC. We quantified over 10,000 proteins and delineated two distinct proteomic subtypes, PS1 and PS2, which exhibit divergent biological pathways, metabolic-immune features and clinical outcomes. The PS1 subtype exhibited enhanced neuroendocrine differentiation and proliferative signaling, correlating with poorer survival outcomes. In contrast, the PS2 subtype was characterized by concurrent activation of immune pathways and fatty acid metabolism, which was linkied to a more favorable prognosis. Further stratification of PS2 revealed two immune microenvironment clusters, IIS and IAS, with differential prognostic and therapeutic implications. The IAS cluster, characterized by robust adaptive immunity, was associated with favorable prognosis, while the IIS cluster, marked by heightened innate but deficient adaptive immunity, showed superior responsiveness to immune checkpoint blockade. Our study advances the understanding of SCLC heterogeneity and provides a clinically actionable proteomic taxonomy for subtype classification and therapy selection.

8615 Background: RET fusions represent actionable oncogenic drivers in lung adenocarcinoma (LUAD). However, the clinicopathological features, co-mutation landscape, and therapeutic outcomes across different RET fusion partners remain incompletely characterized. Methods: We retrospectively analyzed 268 patients with RET fusion–positive LUAD diagnosed between July 2017 and December 2024. Clinicopathological characteristics, metastatic patterns, and concomitant mutations were compared between KIF5B and non- KIF5B subgroups. Treatment outcomes of selective RET inhibitors, multikinase inhibitors (MKIs), and immunotherapy combined with chemotherapy were assessed, with subgroup analyses according to RET fusion partners. Results: The median age at diagnosis was 58 years, and most patients were female (57.0%) and never/light smokers (60.8%). Bone (12.3%), pleural (11.9%), and brain metastases (6.7%) were the most common metastatic sites. KIF5B-RET fusions were more frequently detected in earlier disease stages compared with non- KIF5B fusions (p = 0.0531). Among 274 RET fusion events, KIF5B-RET was predominant (65%), followed by CCDC6-RET (16%) and NCOA4-RET (1%). Concomitant mutations were identified in 28.7% of patients, most commonly TP53 (39.0%) and CDKN2A (13.0%), with CDKN2A more enriched in the non- KIF5B group (p = 0.0393). Nineteen patients received targeted therapy, including pralsetinib (n=12), selpercatinib (n=2), and cabozantinib (n=5), achieving an overall ORR of 57.9% and median PFS of 12.0 months. Notably, non- KIF5B patients demonstrated longer median PFS than KIF5B patients under pralsetinib (17.0 vs. 5.5 months, p = 0.0473). Fifteen patients received first-line immunochemotherapy, achieving a median PFS of 17.0 months, ORR of 40.0%, and DCR of 80.0%, comparable to targeted therapy (p = 0.3871). PD-L1 expression showed no correlation with outcomes. Conclusions: RET fusion–positive LUAD comprises biologically heterogeneous subsets defined by fusion partners. KIF5B-RET fusions tend to occur at earlier stages, whereas non- KIF5B fusions are more frequently associated with CDKN2A co-mutations and appear to derive greater benefit from selective RET inhibition. Immunochemotherapy demonstrates comparable efficacy regardless of fusion partner, highlighting the need for partner-specific therapeutic strategies in RET fusion–positive LUAD.

Human epidermal growth factor receptor 2 (HER2) is a key biomarker and therapeutic target in several malignancies, including breast, gastric, and other solid tumors. Recent advancements in cancer molecular profiling and the Food and Drug Administration's approval of trastuzumab deruxtecan for HER2u2010positive panu2010tumor indications have highlighted the broader relevance of HER2 alterations across diverse cancers. However, the lack of standardized guidelines for HER2 testing in a panu2010tumor context creates variability in clinical practice, hindering the optimal implementation of HER2u2010targeted therapies beyond traditional indications. To address this gap, a multidisciplinary panel of Chinese experts has developed a consensus providing comprehensive recommendations on diagnostic strategies, testing methodologies, and clinical applications of HER2 overexpression detection. By establishing a unified framework for HER2 overexpression assessment, this consensus aims to enhance the precision of HER2 testing, optimize patient selection for targeted therapies, and improve clinical outcomes across a wide spectrum of HER2 overexpression malignancies.

Background:Current evidence on the efficacy and safety of lorlatinib as first-line or subsequent-line therapy for patients with anaplastic lymphoma kinase (ALK)-positive (ALK +) non-small cell lung cancer (NSCLC) in real-world clinical settings remains insufficient. We aim to further evaluate the efficacy and safety of lorlatinib through a real-world cohort study and investigate potential mechanisms of resistance. Methods:This study is a single-center cohort study in China. We retrospectively and prospectively collected data on patients with advanced or metastatic ALK + NSCLC who initiated lorlatinib treatment at National Cancer Center from December 1, 2020. Patients were categorized into two groups based on lorlatinib treatment sequence: first-line cohort and subsequent-line cohort. Demographic characteristics, efficacy, and safety outcomes were comprehensively documented. Survival curves were generated using the Kaplan-Meier method, and group comparisons were performed with the log-rank test. Continuous variables were analyzed using Student's t-tests. The endpoints of this study included measures of both treatment efficacy and safety. Results:As of July 18, 2025, a total of 36 patients were enrolled in the lorlatinib first-line treatment cohort, and 43 patients were enrolled in the lorlatinib subsequent-line treatment cohort for analysis. For the first-line treatment cohort, the median follow-up time was 12.7 months and the median progression-free survival (PFS) had not yet been reached; the objective response rate (ORR) was 82.9%, and the disease control rate (DCR) was 100%. For the subsequent-line treatment cohort, the median follow-up time was 19.9 months and the median PFS was 16.8 months; the ORR was 40.5%, and the DCR was 92.9%. Among all patient groups included in this study, the adverse events associated with lorlatinib treatment predominantly comprised hypercholesterolemia, hypertriglyceridemia, edema, cognitive impairment/mood disorders, elevated transaminases, weight gain, and peripheral neuropathy. No cases of interstitial lung disease were observed. The overall safety profile of lorlatinib is manageable. Analysis of next-generation sequencing (NGS) test results from patients with lorlatinib resistance demonstrated that ALK compound mutations, novel ALK fusions, and MET gene amplification may be potential mechanisms contributing to lorlatinib resistance. Conclusions:Lorlatinib has shown remarkable efficacy in both first-line and subsequent-line treatment settings for patients with locally advanced or metastatic ALK + NSCLC. Effective management of lorlatinib-related adverse events, through close monitoring and timely intervention, is essential to enhance patient tolerance. Lorlatinib has progressively transformed the therapeutic landscape for patients with ALK + NSCLC.

Background/Objectives: Ovarian cancer has the highest mortality among gynecological malignancies, with platinum resistance significantly contributing to poor prognosis. We aimed to develop a multimodal model (MMHC-OCPR) to predict platinum response and recurrence risk, enabling earlier personalized treatment and improved outcomes. Methods: This multicenter retrospective study included a combined cohort of 431 patients, comprising 1182 whole slide images (WSIs) curated from two independent datasets. The primary cohort consisted of 376 patients from the National Cancer Center (China), which was further partitioned into training, validation and internal test sets to ensure model development and evaluation. An additional external test cohort was incorporated using publicly available data from TCGA, enhancing the generalizability of our findings. We implemented a weakly supervised multiple instance learning framework to integrate histopathological imaging with clinicopathological variables, further strengthened by the incorporation of the transformer-based pretrained encoder UNI2-h, which enhanced the model's predictive performance. Results: All patients in the primary cohort had pathology slides collected from primary ovarian tumors and metastatic tumor, along with clinical factors related to prognosis and treatment response. The baseline platinum response classifier using primary WSIs achieved an AUC of 0.896 in the internal test group and 0.876 in the external test group. Integration of metastatic WSIs and clinical data inputs yielded a superior AUC of 0.914 in the internal test set. The recurrence risk model demonstrated a C-index of 0.801, rising to 0.838 after multimodal enhancement. The model stratified patients into low-, intermediate- and high-risk groups with 2-year progression-free survival rates of 77.3%, 48.0% and 2.0%, respectively. Conclusions: Our model enables the early detection of platinum resistance, guiding timely treatment intensification. The recurrence risk stratification supports personalized management by identifying patients with favorable outcomes following surgery and chemotherapy, potentially sparing them from maintenance therapy to reduce associated toxicity, cost, and enhance quality of life.

Distant metastasis, characterized by organotropism, is a major cause of mortality in lung adenocarcinoma (LUAD). In this study, digital spatial profiling (DSP), multiplex immunofluorescence (mIF), and clinical data from 52 LUAD patients were integrated to develop organ-specific metastasis risk models, and the molecular mechanisms underlying metastatic organotropism were investigated. Random forest models based on primary tumor spatial transcriptomics accurately predicted metastasis to the brain (AUC = 0.974), liver (AUC = 0.975), adrenal gland (AUC = 0.929), and bone (AUC = 0.907). Key compartment-specific gene expression signatures associated with organotropic metastasis were identified, including those expressed in tumors (e.g., FKBP1A for the brain and MOCOS for the liver), immune cells (e.g., ADAMTSL2 for the liver), and stromal cells (e.g., CKAP2 for the brain). Pathway analyses revealed distinct biological processes associated with organotropism, such as enriched cell death pathways in brain metastasis and extracellular matrix (ECM) remodeling in liver metastasis. Postmetastasis survival models highlight stromal gene expression (e.g., PKM for OS and VCAM1 for PFS) and immunosuppressive microenvironments (e.g., M2 macrophage infiltration) as critical prognostic factors. The high-precision prediction models and key molecular signatures identified in this study enhance our understanding of "seed-soil" interaction dynamics and offer promising biomarkers and therapeutic targets for future clinical use.