Trained immunity of natural killer (NK) cells has shown great potential in the treatment of cancers by eliciting enhanced effector responses to restimulation by cytokines or cancer cells for long time periods after preactivation. However, the human NK cells responsible for the generation and maintenance of trained immunity are largely unknown. We hypothesized that heterogeneous human NK cells would respond differentially to stimulation with a combination of IL-12, IL-15, and IL-18, and that an NK cell subset might exist that is mainly responsible for the induction of trained immunity. On the basis of our hypothesis, we aimed to identify the subset from which cytokine-trained human NK cells originate and to explore possible regulatory targets for drug intervention.

Flow cytometry assays were performed to analyze the functions of cytokine-trained NK cells and examine cell division and protein expression in NK cell subsets. Single-cell RNA sequencing (scRNA-seq) plus TotalSeq^TM technology was used to track the heterogeneity of NK cells during the induction of trained immunity.

Traditional developmental markers for peripheral NK cells were unable to identify the precursors of human NK cells with trained immunity. Therefore, we used scRNA-seq plus TotalSeq^TM technology to track the heterogeneity of NK cells during the induction of trained immunity and identified a unique cluster of CD57⁻NKG2A⁺EZH2⁺IFNG⁺MKI67⁺IL12R⁺IL15R⁺IL18R⁺ NK cells. Enrichment and pseudotime trajectory analyses suggested that this cluster of NK cells contained the precursor of trained NK cells. We then used flow cytometry to further investigate the role of EZH2 in trained NK precursors and found that CD57⁻NKG2A⁺EZH2⁺ NK cells had faster cell cycles and an enhanced trained phenotype, and EZH2 inhibition significantly impaired the induction of trained immunity in NK cells. These results suggested that EZH2 is a unique epigenetic marker of precursors of human NK cells with trained immunity.

Our work revealed human NK heterogeneity in the induction of trained immunity, identified the precursor subset for trained NK cells, and demonstrated the critical role of EZH2 in the induction of trained immunity in human NK cells.

Distinguishing immune-related adverse events (irAEs) caused by immune checkpoint inhibitors (ICIs) from the AEs caused by chemotherapy, targeted therapy, or infection is highly difficult. This study offers new insights into evaluating the diagnosis, differential diagnostic, and prognostic value of ferritin for irAEs induced by ICIs.

From December 1, 2018, to April 1, 2019, we examined 318 patients with malignant tumors who received serum ferritin monitoring. The cohort comprised 231 patients treated with PD-1 inhibitor or combination with chemotherapy, and 87 patients treated with chemotherapy. Of the 231 patients, 90 had irAEs (irAE group), 70 had non-irAEs (non-irAE group), 67 had no AEs (no irAE-non irAE group), and 4 had unclassified AEs. In the 87 patients, 60 had AEs (AE group), and 27 had no AEs (no AE group). Statistical analyses were conducted with nonparametric Mann-Whitney tests.

At the onset of AEs in the irAE group, ferritin (normal range, 35–150 µg/L) rose to a median of 927 µg/L (range, 117–17,825 µg/L) from 86 µg/L at baseline (range, 29–421 µg/L) (P < 0.001). Ferritin levels at the onset of AEs in the irAE group were significantly higher than those in the non-irAE group (median, 81 µg/L; range, 32–478 µg/L) (P < 0.001) and the AE group (median, 103 µg/L; range, 23–712 µg/L) (P < 0.001). After treatment in the irAE group, ferritin continuously decreased to a normal range in recovered patients, showed no significant changes in stable patients, and continued to rise in patients who died.

Ferritin can be used as a diagnostic, differential diagnostic, and prognostic marker for irAEs in patients treated with ICIs.

Lung cancer is the most common cause of cancer-related deaths worldwide. Somatic copy number alterations (SCNAs) have been used to predict responses to therapies in many cancers, including lung cancer. However, little is known about whether they are predictive of radiotherapy outcomes. We aimed to understand the prognostic value and biological functions of SCNAs.

We analyzed the correlation between SCNAs and clinical outcomes in The Cancer Genome Atlas data for 486 patients with non-small cell lung cancer who received radiotherapy. Gene set enrichment analyses were performed to investigate the potential mechanisms underlying the roles of SCNAs in the radiotherapy response. Our results were validated in 20 patients with lung adenocarcinoma (LUAD) receiving radiotherapy.

SCNAs were a better predictor of progression-free survival (PFS) in LUAD (P = 0.024) than in lung squamous carcinoma (P = 0.18) in patients treated with radiotherapy. Univariate and multivariate regression analyses revealed the superiority of SCNAs in predicting PFS in patients with LUAD. Patients with stage Ⅰ cancer and low SCNA levels had longer PFS than those with high SCNA levels (P = 0.022). Our prognostic nomogram also showed that combining SCNAs and tumor/node/metastasis provided a better model for predicting long-term PFS. Additionally, high SCNA may activate the cell cycle pathway and induce tumorigenesis.

SCNAs may be used to predict PFS in patients with early-stage LUAD with radiotherapy, in combination with TNM, with the aim of predicting long-term PFS. Therefore, SCNAs are a novel predictive biomarker for radiotherapy in patients with LUAD.

Patients with cancer pain are highly dependent on morphine analgesia, but studies have shown a negative correlation between morphine demand and patient outcomes. The long-term use of morphine may result in abnormally elevated serum morphine-3-glucuronide (M3G) levels. Hence, the effects of M3G on tumor progression are worth studying.

The effects of M3G on PD-L1 expressions in human non-small cell lung cancer (NSCLC) cell lines were first evaluated. Activation of TLR4 downstream pathways after M3G treatment was then determined by Western blot. The effects of M3G on human cytotoxic T lymphocytes (CTL) cytotoxicity and INF-γ release was also detected. Finally, the LLC murine lung adenocarcinoma cell line were used to establish a murine lung cancer model, and the effects of M3G on tumor growth and metastasis were determined.

M3G promoted the expressions of PD-L1 in the A549 and H1299 cell lines in a TLR4-dependent manner (P < 0.05). M3G activated the PI3K and the NFκB signaling pathways, and this effect was antagonized by a TLR4 pathway inhibitor. A PI3K pathway inhibitor reversed the M3G-mediated PD-L1 upregulation. M3G inhibited the cytotoxicity of CTL on A549 cells and decreased the level of INF-γ. Repeated M3G intraperitoneal injections promoted LLC tumor growth and lung metastasis through the upregulation of tumor expressed PD-L1 and the reduction of CTL in the tumor microenvironment.

M3G specifically activated TLR4 in NSCLC cells and upregulated PD-L1 expression through the PI3K signaling pathway, thereby inhibiting CTL cytotoxicity and finally promoting tumor immune escape.