Comprehensive characterization of early-onset lung cancer, in Chinese young adults

Abstract

Early-onset lung cancer in young adults represents a less studied clinical entity with increasing incidence which still affects a large number of cancer patients. Here we perform a comprehensive analysis of early-onset lung cancer for the clinicopathological features, genomic alterations, gene expression, and immune landscape by establishing a cohort enrolling 421 non-small cell lung cancer (NSCLC) patients from ten medical centers in China. Comparative analysis reveals a distinct genomic alteration between younger and elder patients with NSCLC, with ERBB2 mutations and ALK-rearrangement strikingly more frequent in younger group. Transcriptome profiling indicates altered cellular metabolism and immune-related genes in tumors from younger patients. Immunological analysis reveals a decreased infiltration of immune cells (notably T cells) in tumors from younger patients. Cellular and mechanistic studies show that the prevalent ERBB2 mutants in cancer from younger patients can indeed drive tumorigenesis by elevating AKT signaling. Importantly, meta-analysis of clinical trials and our clinical practice further validate the promise of HER2-targeted therapy to treat early-onset NSCLC in East Asian patients. Our comprehensive and integrative analysis not only reveal multiple unrecognized characteristics of early-onset lung cancer, but also inform actionable therapeutics to manage this type of cancer.

Fig. 1. Analysis of genetic mutations between younger and elder patients.

a The mutation frequency of the top 10 mutated genes in younger group (YG, n = 215) and elder group (EG, n = 206). b The detailed mutational profile and clinical characteristics for each patient in both cohorts. The upper part displays the mutation frequency and types of 11 driver genes recommended by the NCCN guidelines (Point, point mutation; Del, deletion; Ins, insertion; CNV, copy number variation). The lower part displays the corresponding clinical information for each patient. c Comparative analysis of mutation frequencies in 11 driver genes across two cohorts (EG, n = 206; YG, n = 215). The P values were calculated by two-sided chi-square test or two-sided Fisher’s exact test as appropriate. d The heatmap illustrates the co-mutation and exclusion of 11 driver genes in YG (left, n = 215). The P values were obtained from two-sided Fisher’s exact test, and the color intensity reflects the magnitude of the P value with * indicating the statistical significance of the P value. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Based on the odds ratio (OR) calculated by two-sided Fisher’s exact test, the relationship between two genes can be determined: OR < 1 indicates an exclusionary relationship between two genes, while an OR > 1 indicates a co-mutation relationship (right). The forest plot displays the OR values (dots) and 95% confidence intervals (whiskers). e The mutation subtypes and locations of ERBB2 mutations in YG. f Differences in ERBB2 mutation types between the EG and YG groups, and the ERBB2 exon 20 mutation subtypes in YG (n = mutational events). g Differences in ALK mutation types between the EG and YG groups, and the types of fusions and the subtypes of EML4::ALK fusion in YG (n = mutational events). Source data are provided as a Source Data file.

Fig. 2. Comparative analysis between public database and our data.

a–c Overview of data and case selection process from database (a), and age distribution of patients with age data (b). Proportion of race and ethnicity among with primary lung adenocarcinoma (pLUAD) patients and those with age data in the database, Other: American Indian or Alaska Native, Native Hawaiian or Other Pacific Islander; NA: race and ethnicity details unavailable (c). d Top 10 mutated genes in all patients (n = 3322) or Asian patients (n = 1748) in the public database. e Comparison of EGFR, ERBB2, ALK, and KRAS mutations between EG (n = 1137) and YG (n = 30) in public database. f Comparison of mutations in four genes between EG (n = 1137) and YG (n = 30) in public database and our cohort (EG, n = 206; YG, n = 215). g Comparison of mutations in four genes between Asian patients in public database (n = 117) and elder patients in our data (n = 206). For (e–g) P values were calculated by two-sided chi-square test or two-sided Fisher’s exact test as appropriate. Source data are provided as a Source Data file.

Fig. 3. Transcriptomic and immunological analysis of NSCLC of younger and elder patients.

a The volcano plot displays differentially expressed genes (DEGs) between NSCLC tumors of younger and elder patients by RNA-seq analysis. b The top 10 enriched KEGG pathways and GO (BP) terms of DEGs. Two-sided hypergeometric test or two-sided Fisher’s exact test for P value. Benjamini and Hochberg's method was used for false discovery rate (FDR) correction. c The heatmap illustrates the results of immune infiltration estimation using TIMER2.0 on the expression profiles of 14 tumor samples. The estimation adopts the XCELL algorithm, displaying the infiltration status of important immune cell subtypes in the TME. Color-coded scores represent the infiltration level. d The Tukey’s box plot displays the median (thick line), interquartile range (box limits), 1.5× the interquartile range span (whiskers), and outliers (dots) of positive cells by each IHC staining in the tumor or interface in samples from EG (red, n = 110) and YG (blue, n = 108). T: tumor zone; I: interface zone. Unpaired two-sided t-test for P value. P_CD4-T = 0.0054; P_CD4-I = 0.0032; P_CD8-T = 0.0239; P_CD8-I < 0.0001; P_CD3-T < 0.0001; P_CD3-I = 0.0003. e Heatmap showing immune cell infiltration in the tumor or interface zones of each case in EG and YG samples. Red indicates higher density of corresponding marker staining than the mean level, while blue denotes lower density. f Representative cases showing the IHC staining patterns of biomarkers (CD4, CD8, and CD3) with statistically significant differences in positive cell counts in the tumor or interface zones. The cases were derived from patients who underwent IHC (n = 218). Scale bar, 200 μm. g Representative cases showing the multiplex immunofluorescence staining patterns of biomarkers (CD3, CD4, CD8, FOXP3, and PANCK) with statistically significant differences of intensity and density in composite and single-stained images. The cases were derived from patients who underwent multiplex immunofluorescence (n = 20). Scale bar, 200 μm. h The comparison of CD3⁺(EG: n = 10; YG: n = 8), CD4⁺(EG: n = 10; YG: n = 9), CD8⁺(EG: n = 8; YG: n = 10), FOXP3⁺(EG: n = 10; YG: n = 10) and PANCK⁺(EG: n = 10; YG: n = 9) intensity (H-Scores) between EG and YG patients. i The comparison of CD3⁺(EG: n = 9; YG: n = 10), CD4⁺(EG: n = 10; YG: n = 9), CD8⁺(EG: n = 8; YG: n = 10), FOXP3⁺(EG: n = 10; YG: n = 10) and PANCK⁺(EG: n = 10; YG: n = 9) density (Cells/mm²) between EG and YG patients by cell counts. j Cell density of CD4⁺CD3⁺ T cells (EG: n = 10; YG: n = 9), CD8⁺CD3⁺ T cells (EG: n = 8; YG: n = 7), and CD4⁺CD3⁺FOXP3⁻ T cells (EG^: n = 10; YG: n = 9) between EG and YG patients. For (h–j) the dot-line plots indicate Mean ± SD. The outliers were excluded by applying a normality test. Unpaired two-sided t-test for P value. k The comparison of CD4+, CD8+, and CD3+ T cell infiltration levels between samples with ERBB2 (n = 14) and ALK (n = 11) mutations and the overall levels of EG (n = 110) and YG (n = 108). Tukey’s box plot displays the median (thick line), interquartile range (box limits), 1.5× the interquartile range span (whiskers), and outliers (dots). Unpaired two-sided t-test for P value. Source data are provided as a Source Data file.

Fig. 4. Functional and mechanistic investigation of ERBB2 mutation in NSCLC.

a Immunoblotting analysis of indicated targets in control (GFP) or ectopic ERBB2-expressing NCI-H322 and PC-9 cells. YVMA stands for Y772_A775dup mutant, and insVC represents G776delinsVC mutant. p, phosphorylated; t, total. b Cell growth effect of indicated cell lines in 2D monolayer cell culture condition. Mean ± SD with n = 3 biological replicates. Unpaired two-sided t-test for P value. c 3D colony formation assay for indicated cells. Mean ± SD with n = 3 biological replicates. Unpaired two-sided t-test for P value. d, e Wound healing assay for indicated NCI-H322 cells to examine cell migration ability (d) and histogram shows the statistical analysis of the results (e). Scale bar, 250 μm. Mean ± SD with n = 3 biological replicates. Unpaired two-sided t-test for P value. f Tumor images of NCI-H322 cell-based mice xenografts for each group. g Tumor volume measured at indicated time points after xenograft implantation. n = 12 for each group. (n, number of tumors). h Tumor weight of mice xenografts for each group. (n = 6 for GFP, n = 10 for ERBB2-WT, n = 10 for YVMA) Mean ± SEM. Unpaired two-sided t-test for P value. i The number of DEGs caused by ectopic expression of ERBB2-WT or YVMA mutant in NCI-H322 cells via RNA-seq analysis. j Venn diagrams showing the overlaps of DEGs for the indicated groups, the P values were obtained from two-sided Fisher’s exact test. k Heatmap of DEGs in ERBB2-WT- or YVMA-expressing NCI-H322 cells. I Scatter plot highlighting individual DEGs. m, n Top 10 enriched KEGG categories of ERBB2-WT (m) or YVMA (n) groups. One-sided Fisher’s exact test for P value. Source data are provided as a Source Data file.

Fig. 5. Clinical response of ERBB2-mutant patients to targeted therapy.

a The workflow of meta-analysis for evaluating the clinical response of HER2-targeted therapy in NSCLC patients. b, c Treatment efficacy analysis of pyrotinib and poziotinib on the three most common ERBB2 mutation subtypes in NSCLC (b) and a detailed data breakdown from five clinical trials (c). d Diagram of the treatment scheme for Patient A. IM imaging, SD stable disease, PD progressive disease, PR, partial response. e Representative CT images of tumor lesion of Patient A to therapy. f Primary tumor size of Patient A during the entire treatment process. g Serum NSCLC associated-tumor markers (CEA carcinoembryonic antigen, NSE neuron-specific enolase, CYFRA cytokeratin fragment) of Patient A during the multiple-line therapy. h Representative CT images of tumor lesions of Patient B. i Diagram of the treatment scheme for Patient B. j Metastatic tumor size of Patient B during the pyrotinib treatment process. IM imaging, PR partial response. Source data are provided as a Source Data file.