P-loop and αC-helix compressing (PACC) mutations comprise a structurally distinct subgroup of epidermal growth factor receptor (EGFR) mutations in non-small cell lung cancer (NSCLC) with limited treatment options. Firmonertinib is a once-daily oral, brain-penetrant, mutant-selective EGFR tyrosine kinase inhibitor with a unique trifluoroethoxy side chain. Firmonertinib had high anti-proliferative efficacy in EGFR PACC-mutant cells and xenografts. In the phase 1b FURTHER study (NCT05364073), patients with metastatic EGFR PACC-mutant NSCLC randomized to receive first-line firmonertinib 160 mg (n = 25) or 240 mg (n = 22) had confirmed objective response rates (ORRs) of 43% and 68%, median duration of response not reached and 14.6 months, and median progression-free survival of 11.1 and 16.0 months, respectively; treatment-related adverse events led to dose reductions (19% and 24%) and discontinuation (3% and 0%). Intracranial ORR among patients with brain metastases at baseline was 47% (8/17). Results support further study of first-line firmonertinib in patients with EGFR PACC-mutant NSCLC.

EGFR PACC-mutant non-small cell lung cancer (NSCLC) has limited treatment options. In this issue of Cancer Cell, Le et al. report preclinical profiles and promising systemic and intracranial efficacy of the broad-spectrum CNS-penetrant tyrosine kinase inhibitor firmonertinib in PACC-mutant NSCLC, supporting its further clinical development in this challenging mutation class.

With seven products approved by the US Food and Drug Administration (FDA) for 15 indications and more than 2,600 active clinical trials, chimeric antigen receptor (CAR) T cell therapy is firmly established as an effective therapeutic modality for cancer treatment, and promising early data demonstrate its potential in autoimmune diseases. Key challenges facing the field include extending efficacy to solid tumors and reducing the cost and complexity of manufacturing to improve patient access. In this review, we summarize the clinical lessons learned from the now substantial experience with commercial CAR T cells and clinical trials and discuss three waves shaping next-generation development: multi-engineered and logic-gated CAR T cells, rational combination strategies, and in vivo CAR T cell generation. By integrating clinical experience with emerging strategies, we highlight where CAR T cell therapy has succeeded, where current approaches remain insufficient, and how next-generation cell therapies may be designed to overcome these barriers.

In this issue of Cancer Cell, Shi et al. report encouraging early-phase results combining HRS-4642, an allele-specific KRAS G12D inhibitor, with the PD-L1 inhibitor adebrelimab in previously treated metastatic pancreatic cancer. The intravenous combination was safe and well tolerated, notably sparing patients the gastrointestinal toxicity common to oral KRAS inhibitors, and achieved a promising 43% response rate at the recommended dose.

While tertiary lymphoid structures (TLSs) broadly correlate with immunotherapy efficacy, their functional heterogeneity can translate into divergent clinical outcomes. In this issue of Cancer Cell, Hu et al. uncover a microbiota-orchestrated stem-like lymphoid niche in which CD70[+] ILC2s sustain progenitor exhausted T cells, and their findings redefine the functional organization underlying TLS-associated antitumor immunity.

Neuroblastoma is a pediatric cancer arising from the developing sympathoadrenal lineage. Although cell lines derived from neuroblastoma patients have features of either adrenergic (ADRN) or mesenchymal (MES) cells, identifying MES cells in patient tumors is challenging and controversial. Here, to characterize intra- and inter-patient cellular heterogeneity, we analyze 54 neuroblastoma tumors spanning broad clinical, genetic, and histologic features using single-cell/single-nucleus RNA sequencing (sc/snRNA-seq), bulk RNA sequencing, spatial transcriptomics, and spatial proteomics. We show that tumor cells occupy distinct cellular neighborhoods separated from nonmalignant immune and stromal cells, yet established cell line-derived gene expression signatures fail to identify MES neuroblastoma cells. We therefore develop ADRN/MES signatures from early-passage orthotopic patient-derived xenografts, enabling the identification of MES tumor cells, validated by transcriptomic, epigenomic, proteomic, ultrastructural, and functional assays. In patients, high MES gene expression correlates with worse event-free and overall survival. Together, this study provides a framework for defining clinically relevant neuroblastoma cell states.

Chromatin dynamics shape gastric cancer at every stage. In this issue of Cancer Cell, Wang et al. uncover a premalignant chromatin state primed for oncogenic activation alongside a pseudo-gastric subtype that hijacks acid secretion to evade antitumor immunity, revealing gastric proton pumps as a potential therapeutic target.

Chromatin remodeling is prominent in cancers, yet it remains obscure how chromatin dynamics underlies stepwise tumorigenesis and informs anti-cancer therapeutics. By decoding a large-scale chromatin accessibility atlas of 650 multistage samples from 439 patients, we delineate epigenetic switches spanning multistep gastric tumorigenesis. A cancer-resembling epigenetic priming state in intestinal metaplasia licenses subsequent tumor initiation. In advanced malignancies, inter-tumoral diversity stratifies three epigenomic subtypes. The proliferative and mesenchymal-inflammatory subtypes are sensitive to neoadjuvant immunotherapy, while the pseudo-gastric subtype is intrinsically drug resistant. Mechanistically, the refractory subtype is featured by a unique gastric lineage-mimicking phenotype, governed in part by FOSL2 to up-regulate H[+]/K[+] ATPase transporters and sculpt an acidic milieu suppressing CD8[+] T cells. Inhibition of this phenotype by either FOSL2 depletion or vonoprazan, a clinically available acid blocker, sensitizes immunotherapy. Collectively, these findings highlight that chromatin accessibility dynamics facilitates sequential tumorigenesis from gastric pre-cancer and defines subtypes exploitable to guide immunotherapy and overcome drug resistance.

Therapeutic cancer vaccines are increasingly tested in clinical settings alongside standard-of-care treatments that often include chemotherapy, yet whether chemotherapy synergizes with cancer vaccines remains unclear. Here, we tested heterologous prime-boost viral vector vaccines in combination with various chemotherapy regimens. Both carboplatin plus paclitaxel (CarboTaxol) and cyclophosphamide improve vaccine efficacy and enhance antigen-specific CD8[+] T cell responses. These chemotherapies act as immunological adjuvants independently of tumor presence. Mechanistically, CarboTaxol induces an early, antigen-independent expansion of stem-like T cell factor 1 (TCF1)[+]CD8[+] T cells, an effect also observed in patients with different cancer types. Genetic or pharmacological disruption of TCF1 impairs the immunological adjuvant effect of CarboTaxol. Adding programmed cell death 1 (PD-1) blockade to viral vector vaccines and CarboTaxol further improves tumor control and survival. Together, these findings identify a TCF1-dependent mechanism underlying the immune adjuvant effect of chemotherapy and provide a rationale for clinical evaluation of this triple combination therapy.

In Nature, Desharnais et al. show that in diverse mouse models, anti-PD-1 sensitivity is more closely linked to the diet-microbiome-metabolite ecosystem than to obesity-associated metabolic dysfunction. Their findings reinforce body mass index (BMI) as an incomplete proxy for host biology and suggest strategies to improve anti-PD-1 efficacy without inducing obesity.

In this issue of Cancer Cell, García-Gómez et al. show that brain-seeding cancer cells exploit an MXD4-dependent proliferative pause that enables survival under stress before metastatic outgrowth. By restraining MYC and activating protective stress responses, this pause becomes an Achilles' heel, potentially targetable to eradicate micrometastases before clinical relapse.

Successful metastatic colonization requires cancer cells to survive multiple stresses during early organ adaptation, yet how metastasis-initiating cells overcome this bottleneck remains unknown. Here, we identify a transient MXD4-dependent proliferative pause that enables aggressive cancer cells to survive early brain colonization. Before vascular co-option-associated outgrowth, metastatic cells temporarily restrain proliferation through the MYC antagonist MXD4, buffering stresses encountered after extravasation. Genetic disruption of MXD4 compromises survival of micrometastases, preventing their progression to macrometastases. Analyses of human brain micrometastases validate this transient adaptive state in patients. Exploiting vulnerabilities associated with the proliferative pause reveals preventive therapeutic opportunities across experimental settings, including spontaneous brain metastasis models and minimal residual disease following neurosurgical resection. In addition, patient-derived organotypic cultures confirm the feasibility of targeting this state with clinically available drugs. These findings uncover a previously unrecognized adaptive stage during metastatic colonization and identify actionable vulnerabilities to prevent brain metastasis progression and relapse.

Chimeric antigen receptor (CAR) T cell therapy has shown promising activity in solid tumors, yet translational progress remains limited by preclinical models that incompletely capture human tumor biology. Patient-derived organoids (PDOs) preserve key features of human tumors, including cellular heterogeneity and tissue architecture, thereby enabling functional evaluation of CAR T cell targets, next-generation approaches, resistance mechanisms, and selected safety considerations. We discuss how PDOs complement existing preclinical platforms to support CAR T cell development for solid tumors.

Small cell lung cancer (SCLC) is a highly aggressive neoplasm with limited sensitivity to anti-PD-(L)1 blockade, which is likely caused by the epigenetic silencing of MHC-I. Elucidating MHC-I-independent immune recognition mechanisms is therefore crucial for enhancing treatment responses and improving clinical outcomes in a greater number of patients. Leveraging single-cell approaches, we discovered γδ T cell infiltration in biospecimens from patients with SCLC. Despite PD-1 expression, γδ T cells maintained a cytotoxic transcriptional profile, suggesting an anti-tumor role. Indeed, high γδ T cell infiltration in two practice-changing clinical trials predicted improved response to anti-PD-L1 immunotherapy in patients with SCLC. Moreover, using preclinical models, we demonstrated that γδ T cells are effective at tarlatamab (delta-like ligand 3 [DLL3]-CD3 bispecific T cell engager [BiTE])-redirected SCLC killing and that zoledronate, an FDA-approved compound, can sensitize SCLC cells to γδ T cell-mediated killing. Thus, our findings suggest that engaged γδ T cells are potentially valuable targets for SCLC therapy.