Subtype-specific RNA sequencing using micro-dissection revealed extracellular matrix alterations as key factors in lung adenocarcinoma invasion

Introduction

Lung cancer is the leading cause of cancer-related mortality worldwide, accounting for over one million deaths annually. Among its histological types, lung adenocarcinoma (LUAD) is the most prevalent (1,2). LUAD is further classified into distinct subtypes based on histological characteristics, each exhibiting unique biological behaviors and clinical outcomes (2). Among these, micropapillary and solid predominant (SPA) adenocarcinomas are associated with a poor prognosis, whereas the lepidic predominant adenocarcinoma (LPA) subtype correlates with a significantly more favorable outcome (3,4). The coexistence of multiple subtypes within a single tumor is common. This is particularly evident in invasive LUAD (IAC), which often presents as a mixed ground-glass opacity (mGGO) on computed tomography (CT) scans, and where non-invasive and invasive components frequently coexist (5).

The coexistence of different subtypes, particularly of non-invasive and invasive subtypes, captures the biological characteristics of different tumor stages simultaneously, thereby providing an excellent model for investigating the LUAD tumorigenesis process. Tumor invasion is a critical event in progression, as breaching the basement membrane endows cancer cells with the potential to metastasize. The biological characteristics, treatment strategies, and clinical prognoses of pre-IAC and IAC are fundamentally distinct. Pre-IAC, which typically manifests as a small or pure ground-glass opacity (pGGO) on CT scans, is associated with a nearly 100% 5-year overall survival (OS) and progression-free survival (PFS) rate (6). In contrast, the prognosis for IAC patients is significantly worse and is highly dependent on the disease stage (7). Therefore, suppressing tumor invasion is a crucial therapeutic objective. However, current treatment strategies remain unsatisfactory, and the overall survival rate for lung cancer patients remains low. Consequently, elucidating the regulatory mechanisms of invasion and identifying key drivers is of paramount importance. Tumor invasion is a complex biological process involving multiple steps, including cell proliferation, morphological changes, extracellular matrix (ECM) remodeling, and signal transduction, many aspects of which remain poorly understood in the context of LUAD (8).

In this study, we utilized micro-dissection and next-generation sequencing (NGS) to dissect the transcriptomic alterations in 21 paired non-invasive and invasive subtypes from the same tumors. Our results indicate that ECM remodeling plays a crucial role in LUAD invasion. Subsequent validation using RNA sequencing (RNA-seq) data from independent LUAD tissue samples identified several biomarkers, including AGER and CGNL1, which are strongly associated with LUAD prognosis. We present this article in accordance with the STREGA reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-520/rc).

Methods

Patient enrollment and samples collection

A total of 21 patients who underwent radical surgery at Fudan University Shanghai Cancer Center between 2017 and 2019 were enrolled in this study. Immediately following resection, precise intraoperative diagnosis was conducted using frozen sections. The specimen was serially sectioned along its largest diameter for sampling. Hematoxylin and eosin (H&E) staining was performed on each slide to facilitate an accurate pathological diagnosis according to the International Association for the Study of Lung Cancer (IASLC)/American Thoracic Society (ATS)/European Respiratory Society (ERS) classification standard (2). Only lesions diagnosed as IAC, containing both non-invasive and invasive subtypes, were included. According to the IASLC/ATS/ERS classification, the lepidic subtype was defined as a non-invasive subtype, whereas acinar, papillary, micropapillary, and solid subtypes were classified as invasive subtypes. The abundance of each subtype was evaluated to ensure sufficient RNA quantity and quality for subsequent sequencing.

Frozen tissue samples from an additional 17 patients who underwent radical surgery during the same period were also collected. All H&E-stained sections were re-examined to confirm that the proportion of the predominant subtype exceeded 90%.

Micro-dissection and sample collection

Three sequential tissue sections were prepared and immersed in RNAlater solution to prevent RNA degradation. The same H&E staining procedure was performed on these sections. Subsequently, the boundaries between subtypes were meticulously delineated by experienced pathologists under the microscope. The targeted subtype components were then meticulously micro-dissected, collected, and immediately immersed in RNAlater solution. These samples were stored at 4 ℃ for less than 24 hours prior to RNA extraction.

RNA extraction and NGS

An equal volume of ice-cold phosphate-buffered saline (PBS) was added to each micro-dissected sample to dilute the RNAlater. The samples were then centrifuged at 2,000 ×g for 5 min at 4 ℃ to pellet the tissue, and the supernatant was carefully removed. The cell pellets were homogenized in lysis buffer, and total nucleic acid was extracted according to the manufacturer’s instructions for the AllPrep DNA/RNA Mini Kit (Qiagen, Hilden, Germany; Cat. No. 80204). For the bulk frozen tissues, an appropriate amount of sample was processed for RNA extraction using the same kit immediately after retrieval from liquid nitrogen. RNA samples were then used for library construction using the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, UK; Cat. No. E7760S). The quantity of RNA and library DNA was measured using a Qubit 3.0 Fluorometer with the appropriate assay kits. The quality and integrity of the RNA and final libraries were assessed using an Agilent Bioanalyzer 2100 system or a QIAxcel Advanced system. NGS was performed on an Illumina HiSeq X10 system.

Chromatin immunoprecipitation sequencing (ChIP-seq) in tissue

ChIP-seq for the H3K27ac histone mark was performed using an anti-H3K27ac antibody (Abcam, Cambridge, UK; ab4729). Briefly, frozen LUAD and paired normal lung tissues were homogenized on ice in RPMI-1640 medium (Corning, Corning, NY, USA; 10-040-CVR) using a Dounce homogenizer. The homogenate was filtered through a 70 µm cell strainer. Cells were fixed in 1% formaldehyde at room temperature for 10 min. The cell pellet was collected by centrifugation, washed twice with ice-cold PBS, and resuspended in lysis buffer for incubation on ice for 20 min. The lysate was sonicated to shear the DNA to an average fragment size of 100–300 bp. Approximately 10% of the chromatin was set aside as the input control. The remainder was precleared with uncoupled Protein A Dynabeads (Novex, Waltham, MA, USA; 10002D) and subsequently incubated with H3K27ac antibody-coupled Dynabeads overnight at 4 ℃ with rotation. The beads were collected using a magnetic rack, and the chromatin was eluted. Both the immunoprecipitated and input samples were reversed cross-linked by incubation at 65 ℃ overnight. They were then treated with RNase A and proteinase K, and the DNA was purified using a commercial kit (Tiangen, Beijing, China; DP214-03). The purified DNA was used for NGS library preparation following the manufacturer’s protocol and was subsequently sequenced.

Bioinformatics and statistics analysis

Raw RNA-seq reads were initially assessed for quality using FastQC and subsequently trimmed to remove adapter sequences with Cutadapt (Table S1). The cleaned reads were aligned to the human reference genome (GRCh38/hg38) using STAR. Gene-level counts were quantified using HTSeq, and differential gene expression analysis was performed with DESeq2. Differentially expressed genes (DEGs) were defined using an adjusted P value cutoff of <0.05. Functional enrichment analysis was conducted using the ClusterProfiler package in R and Metascape. Protein-protein interaction (PPI) network analysis was performed using the STRING database.

ChIP-seq raw reads were aligned to the human reference genome (GRCh38/hg38) using BWA. Peak calling was performed using Model-based Analysis of ChIP-Seq 2 (MACS2) to identify genomic regions with significant enrichment compared to the input control (Appendix 1). A stringent P value threshold of 10⁻⁹ was used for peak calling. The Integrative Genomics Viewer (IGV) was used for visualization of the ChIP-seq results. Differential binding analysis was performed using the DiffBind package in R. De novo motif enrichment analysis was conducted on H3K27ac peaks using the findMotifsGenome.pl script from the HOMER suite.

Clinical information and RNA-seq data for a cohort of 501 LUAD patients were downloaded from The Cancer Genome Atlas (TCGA) database. Survival analysis was performed using the survival and survminer packages in R. A log-rank P value <0.05 was considered statistically significant. Paired and unpaired Student’s t-tests were applied for comparisons between the two groups using GraphPad Prism. Immune cell infiltration profiling was estimated from the micro-dissected RNA-seq data using CIBERSORT (9).

Ethical statement

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by Fudan University Shanghai Cancer Center Institutional Review Board (No. 090977-1). Informed consents of all patients for donating their samples to the tissue bank of Fudan University Shanghai Cancer Center were obtained from patients themselves or their relatives.

Results

Patients’ and samples’ characteristics

This study included 21 patients (15 female and 6 male) who underwent surgical resection for IAC between 2017 and 2019. The mean age at diagnosis was 58.95 years. All 21 tumors contained lepidic subtype. Twenty tumors (95.2%) harbored a single invasive subtype, and 1 tumor (4.8%) contained two distinct invasive subtypes (acinar and papillary). Preoperatively, all tumors presented as mGGO nodules on CT scans. The median tumor size was 2.0 cm. Only one patient (4.8%, Patient 21) had lymph node metastasis (pN1). Notably, this patient had a relatively small tumor (1.4 cm) that contained solid subtype, a known risk factor for lymph node metastasis and poor prognosis, as established in our previous work (3). EGFR mutations, the most prevalent driver alteration in Asian LUAD patients, were identified in 14 of the 21 patients (66.7%). The spectrum of mutations included exon 19 deletions (n=9, 64.3% of mutants), L858R (n=4, 28.6%), and L861Q (n=1, 7.1%) (Table 1).

ECM remodeling was crucial for LUAD invasion

To investigate transcriptomic alterations during LUAD invasion, we utilized frozen tissue sections containing paired non-invasive and invasive subtypes from the same tumor. The raw RNA-seq data of 21 pairs of non-invasive and invasive subtypes were processed using a standard pipeline as previously described and analyzed for differential gene expression using DESeq2.

We first assessed the global transcriptomic similarity between all samples. Unsupervised clustering and principal component analysis (PCA) revealed substantial transcriptomic heterogeneity across samples. Consistent with previous studies, inter-individual variation, rather than histological subtype, was the primary source of this heterogeneity (Figure 1A,1B). To mitigate the confounding effect of inter-individual variation, a paired analysis was performed. This identified 128 DEGs between non-invasive and invasive subtypes, with 33 upregulated and 95 downregulated in the invasive components (Figure 1C,1D).

Figure 1 Heterogeneity analysis and DEGs analysis of micro-dissected non-invasive and invasive samples. (A) PCA revealed the individual variation was more significant than histological subtype variation. Different histological types were marked with different traits and different patients were in different colors. (B) The Pearson correlation of each micro-dissected sample indicated no obvious clustering in non-invasive or invasive subtypes. (C) The volcano plot showed the DEGs between invasive subtypes and non-invasive subtype. Genes with adjust P value <0.05 were defined as DEGs. Genes with fold change >2 were marked in red. (D) Heatmap showed the expression level of top 20 DEGs with the most significant differences. L: non-invasive subtype; IAC: invasive subtypes. DEG, differentially expressed gene; PC, principal component; PCA, principal component analysis.

Gene Ontology (GO) and functional enrichment analysis of the 128 DEGs revealed a strong association with biological processes related to the ECM. Notably, these DEGs included key regulators of ECM signaling, such as TGFB2, BMPR2, and WNT7A (10) (Figure 2A,2B). Other significantly enriched terms included those related to the plasma membrane and cell adhesion. PPI network analysis suggested that these DEGs may function cooperatively and highlighted their strong connectivity with established ECM core factors, including BMPR2, CAV1, ICAM1, and RAB11A (Figure 2C).

Figure 2 Functional enrichment analysis and gene interaction analysis of DEGs between non-invasive subtypes and invasive subtype. (A) GO analysis of DEGs showed the ECM was the most significantly enriched item. (B) Functional enrichment analysis results showed enrichment of cell adhesion and extracellular structure organization. Network of enriched items in DEGs colored by item ID, nodes shared in the same cluster were close to each other. (C) The PPI network between DEGs. BMPR2, CAV1, ICAM1, and RAB11A may work as core regulators. DEG, differentially expressed gene; ECM, extracellular matrix; GO, Gene Ontology; PPI, protein-protein interaction.

Collectively, these results underscore the pivotal role of ECM remodeling in facilitating LUAD invasion.

Validation of ECM remodeling in bulk LUAD tumors

To independently validate the role of ECM alterations in LUAD progression, we obtained a separate cohort of bulk tumor tissue samples, comprising 4 LPAs and 13 acinar predominant adenocarcinomas (APAs). All samples were collected immediately post-resection and subjected to rigorous pathological review to confirm that the predominant subtype constituted >90% of the tumor area. RNA was extracted and subjected to transcriptome sequencing.

Consistent with our findings in the micro-dissected samples and prior literature (11), PCA revealed considerable inter-individual heterogeneity within this validation cohort (Figure 3A). Differential expression analysis between LPA and APA samples identified 292 DEGs, with 218 upregulated and 74 downregulated in the more invasive APA group (Figure 3B,3C).

Figure 3 Heterogeneity analysis and functional enrichment analysis of DEGs in bulk LUAD samples. (A) The Pearson correlation of each tissue sample showed no significant clustering in LPA or APA samples. (B) The volcano plot showed the DEGs between LPA and APA tissue samples. Genes with adjust P value <0.05 were defined as DEGs. (C) Heatmap showed top 100 DEGs with the most differences. L was short for LPA and A was short for APA. (D) GO analysis of DEGs between LPA and APA samples showed that ECM was also the most significantly enriched item. APA, acinar predominant adenocarcinoma; DEG, differentially expressed gene; ECM, extracellular matrix; GO, Gene Ontology; LPA, lepidic predominant adenocarcinoma; LUAD, lung adenocarcinoma.

GO analysis revealed significant enrichment of DEGs in ECM-related pathways. Many of the top DEGs encoded fundamental ECM structural components (e.g., VCAN, COL4A5) or proteins involved in ECM modification and signaling (e.g., MMP11, CXCL14). Strikingly, eight of these DEGs (CGNL1, MMP11, PRTG, LINC01614, UPK3B, AGER, KCNS1, and PIP5K1B) overlapped with the invasion-associated DEGs identified from our micro-dissected cohort, highlighting their potential as core drivers of LUAD invasion. For example, MMP11, a member of the matrix metalloproteinase family, was significantly upregulated in both invasive subtypes (from micro-dissection) and APA samples. This gene is a well-established prognostic biomarker in multiple cancer types. Previous functional studies have demonstrated that MMP11 overexpression dramatically promotes tumor growth and invasion (12,13). As anticipated, GO analysis of these tissue-derived DEGs showed the most significant enrichment for terms related to ECM organization and function (Figure 3D). Collectively, these results from bulk tissue samples provide independent validation that ECM remodeling is a pivotal event in the invasive progression of LUAD.

ECM-related genes AGER and CGNL1 were highly associated with patients’ prognosis

We identified an eight-gene signature that was consistently dysregulated in both our micro-dissected cohort and the independent bulk tissue validation cohort. To assess the clinical prognostic value of these eight genes, we leveraged RNA-seq data and clinical information from 501 LUAD patients in the TCGA-LUAD cohort. For each gene, patients were stratified into high- and low-expression groups based on the median expression value.

Survival analysis revealed that among the eight genes, low expression of two ECM-related proteins, AGER and CGNL1, was significantly associated with poorer OS (Figure 4A). Consistent with this prognostic significance, both AGER and CGNL1 were downregulated in invasive subtypes (compared to non-invasive) and in APA samples (compared to LPA) (Figure 4B). Beyond its role in the ECM, AGER is highly expressed in normal lung tissue and represents a unique innate immune pattern recognition receptor frequently dysregulated in lung cancer (14). Given the immunomodulatory role of AGER, we next performed immune cell infiltration profiling on our 21 paired micro-dissected samples. This analysis revealed a significant reduction in neutrophil infiltration in the invasive subtypes compared to their paired non-invasive counterparts (Figure 4C,4D). Neutrophils are key effector cells of the innate immune system and inflammation, and their decrease was concordant with the downregulation of AGER.

Figure 4 AGER and CGNL1, as well as neutrophils, were downregulated in invasive subtypes. (A) The K-M plot of 501 TCGA-LUAD patients showed patients with high AGER and CGNL1 expression level had better overall survival. (B) AGER and CGNL1 were both down regulated in invasive subtypes when compared with non-invasive subtype. (C) Immune cell profiling of micro-dissection samples. (D) Neutrophils significantly decreased in invasive subtypes. L: non-invasive subtype; IAC: invasive subtypes. K-M, Kaplan-Meier.

In summary, our results establish AGER and CGNL1 as key ECM-derived biomarkers implicated in LUAD invasion and prognosis. Furthermore, our data suggests that AGER may influence LUAD invasion by modulating the immune microenvironment, particularly neutrophil infiltration.

Epigenetic regulation of ECM gene expression during LUAD invasion

To investigate the epigenetic drivers of ECM alterations during LUAD invasion, we performed H3K27ac ChIP-seq on 2 LPA and 4 APA tissue samples. Peak calling was performed using MACS2, identifying a total of 42,818, 76,531, 88,930, 81,931, 79,706, and 60,831 high-confidence peaks in the six individual samples, respectively. Differential binding analysis between the LPA and APA groups identified 97 peaks with significantly altered H3K27ac enrichment. These differential peaks were annotated to the nearest transcriptional start site (TSS) using HOMER (15). The vast majority of these differential peaks (94/97, 96.9%) were located in intronic or intergenic regions, suggesting they may function as distal enhancers (Figure 5A).

Figure 5 Histone modification alternations may induce ECM changes during the invasion process of LUAD. (A) The H3K27ac peaks distribution. Most differential peaks were located at intergenic zones and introns. (B) Functional enrichment analysis revealed differential H3K27ac peaks were highly connected to genes associated with collagen biosynthesis and modification. (C) GO analysis showed differential H3K27ac peaks were highly associated with collagen biosynthesis. (D) The gene expression (upper left panel) and H3K27ac modification (lower panel) were both up regulated in APA tissue samples on VCAM1 gene locus. VCAM1 up-regulation was also observed in micro-dissected invasive components, but not statistically significant (P=0.06, upper right panel). (E) H3K27ac modification was up-regulated on COL15A1 and RUNX2, and down-regulated on DLC1 in APA tissue samples. (F) Motif analysis at differential peaks showed GLi2, PRDM15 and FOXA1 may be the major transcription factors and controlling down-stream ECM gene expression. L: non-invasive subtype; IAC: invasive subtypes. APA, acinar predominant adenocarcinoma; ECM, extracellular matrix; GO, Gene Ontology; LPA, lepidic predominant adenocarcinoma; LUAD, lung adenocarcinoma.

Functional enrichment analysis was performed on the genes associated with these 97 differential peaks. This analysis revealed significant enrichment for terms related to the plasma membrane and collagen biosynthetic process. Notably, several genes encoding key ECM components, including VCAM1 and COL15A1, were among the top enriched genes (Figure 5B,5C). Consistent with its role in invasion, VCAM1 expression was significantly upregulated in APA compared to LPA samples. This upregulation was associated with a corresponding increase in H3K27ac modification at its genomic locus (Figure 5D). A similar upregulation trend for VCAM1 was observed in invasive subtypes from our micro-dissected samples, although it did not reach strict statistical significance (P=0.06) (Figure 5D). Survival analysis was also performed based on VCAM1 level, but the result only showed a trend of poor prognosis for patients with higher VCAM1 level, which was not statistically significant (Figure S1). Beyond ECM-related genes, our analysis also implicated differential H3K27ac peaks near known oncogenes and tumor suppressor genes (TSGs), such as RUNX2 and DLC1 (Figure 5E). The H3K27ac signal at the DLC1 locus was decreased in the APA group, which is consistent with its characterized role as a TSG. This finding aligns with clinical data showing that high DLC1 expression predicts a better prognosis in non-small cell lung cancer (NSCLC) (16-18).

Histone modifications such as H3K27ac facilitate an open chromatin state and enhance the binding of transcription factors, thereby potentiating the expression of target genes. Motif analysis of the differential peaks identified significant enrichment for binding sites of transcription factors, including GLi2, PRDM15, and FOXA1 (Figure 5F). FOXA1 is a lineage-specific oncogene that promotes tumorigenesis in part through complex interactions with other transcription factors and miRNAs (19,20). In LUAD, FOXA1 expression is often driven by the master regulator NKX2-1 and is a known promoter of invasion (21).

Collectively, our ChIP-seq analysis suggests a model whereby epigenetic alterations, through modulating the expression of ECM genes directly or via ECM-related transcription factors (e.g., FOXA1), drive ECM remodeling and contribute to LUAD invasion.

Discussion

Malignant proliferation and metastasis are two hallmarks of cancer, with tumor cell invasion representing the critical initial step in the metastatic cascade (22). The prognosis and management of pre-IAC versus IAC are fundamentally different, as pre-invasive cells remain confined by the basement membrane and thus lack metastatic potential (23). However, the mechanisms driving LUAD invasion remain poorly understood, partly due to a lack of research models that faithfully recapitulate the natural ECM alterations and tumor microenvironment (TME) dynamics. In this study, we applied micro-dissection to isolate histologically distinct components from the same LUAD tumor, thereby preserving the native ECM and TME context for analysis. In this study, we identified several ECM genes that may regulate LUAD invasion, especially CGNL1. CGNL1 was reported to influence cancer progression in oral squamous cell carcinoma and head and neck squamous cell carcinoma, but not well studied in LUAD, which was a potential therapeutic target deserving further exploration (24,25).

A major challenge in studying LUAD invasion is confounding inter-tumor heterogeneity. Previous comparisons between pre-IAC and IAC from different patients identified only focal adhesion as a significantly altered pathway (26), likely because pronounced intrinsic differences between individuals can mask subtle but critical pro-invasive transcriptional changes. To overcome this limitation, we employed a paired analysis strategy, comparing non-invasive and invasive subtypes within the same tumor, thereby controlling for inter-individual genetic and microenvironmental variability.

A limitation of our approach is potential RNA degradation. While essential for histological guidance, the formaldehyde fixation and prolonged H&E staining required for traditional micro-dissection are known to compromise RNA integrity (27), which could introduce bias. We mitigated this issue by using RNA protective agents and optimizing the staining protocol on a consecutive section, dedicating other sections for RNA extraction. Nevertheless, the RNA integrity number (RIN) remained lower than that obtained from fresh, bulk tissue samples. Emerging technologies like single-cell RNA-seq (scRNA-seq) and spatial transcriptomics offer powerful solutions to these limitations, enabling the dissection of LUAD invasion at single-cell resolution within intact spatial contexts (28,29). Integrating these multi-omics approaches in future studies will be crucial to precisely delineate TME remodeling and cell-cell communication networks underlying invasion.

There are four major invasive subtypes in LUAD: acinar, papillary, micropapillary, and solid. However, the predominant invasive subtype in our sample is acinar (17/21). Analyzing all patterns could yield more generalizable findings. However, it is biologically uncommon to find a high proportion of both lepidic (typically early-stage) and highly aggressive (e.g., micropapillary or solid, typically late-stage) components within the same tumor, reflecting the natural history of disease progression.

Our integrated multi-omics analysis strongly demonstrates that ECM remodeling plays a critical role in LUAD invasion, at both transcriptomic and epigenetic levels. While numerous studies have associated ECM alteration with lung cancer progression (8), our study provides more direct and compelling evidence by pinpointing these changes specifically to the invasion front within matched samples. Although different invasive subtypes exhibit distinct biology, our study provides a simple and feasible method—micro-dissection of multi-subtypes tumors—to investigate LUAD progression and evolution. We have identified several potential therapeutic targets, however, their clinical translational value requires further functional validation.

Ultimately, combining our micro-dissection approach with single-cell and spatial transcriptomics holds great promise for precisely elucidating the molecular alterations and cellular events that drive LUAD pathogenesis.

Conclusions

This study, utilizing an approach that combines micro-dissection with RNA-seq, demonstrates that ECM remodeling is a critical driver of invasion in LUAD. Furthermore, AGER and CGNL1 emerged as significant prognostic biomarkers, underscoring their potential as novel therapeutic targets for LUAD.

Acknowledgments

We sincerely appreciate the assistance provided by the pathologists and technicians in specimen collection and sectioning. We express our sincere gratitude to the members of the Meng lab (Fei-Long Meng, Shanghai Institute of Biochemistry and Cell Biology, Center for Excellence in Molecular Cell Science, Chinese Academy of Science) for their assistance during the research period.

Footnote

Reporting Checklist: The authors have completed the STREGA reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-520/rc

Data Sharing Statement: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-520/dss

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-520/prf

Funding: This work was supported by the Chinese Minister of Science and Technology (Nos. 2017YFA0505501 and 2016YFA0501800), the National Natural Science Foundation of China (Nos. 31701140, 81802273, and 82172744), and the Science and Technology Commission Shanghai Municipality (Outstanding Academic Leader) (No. 19XD1401300).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2025-520/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by Fudan University Shanghai Cancer Center Institutional Review Board (No. 090977-1). Informed consents of all patients for donating their samples to the tissue bank of Fudan University Shanghai Cancer Center were obtained from patients themselves or their relatives.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Siegel RL, Kratzer TB, Giaquinto AN, et al. Cancer statistics, 2025. CA Cancer J Clin 2025;75:10-45. [Crossref] [PubMed]
2. Travis WD, Brambilla E, Noguchi M, et al. International association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol 2011;6:244-85. [Crossref] [PubMed]
3. Zhang Y, Li J, Wang R, et al. The prognostic and predictive value of solid subtype in invasive lung adenocarcinoma. Sci Rep 2014;4:7163. [Crossref] [PubMed]
4. Zhang Y, Wang R, Cai D, et al. A comprehensive investigation of molecular features and prognosis of lung adenocarcinoma with micropapillary component. J Thorac Oncol 2014;9:1772-8. [Crossref] [PubMed]
5. Ye T, Deng L, Wang S, et al. Lung Adenocarcinomas Manifesting as Radiological Part-Solid Nodules Define a Special Clinical Subtype. J Thorac Oncol 2019;14:617-27. [Crossref] [PubMed]
6. Zhang Y, Ma X, Shen X, et al. Surgery for pre- and minimally invasive lung adenocarcinoma. J Thorac Cardiovasc Surg 2022;163:456-64. [Crossref] [PubMed]
7. Rami-Porta R, Nishimura KK, Giroux DJ, et al. The International Association for the Study of Lung Cancer Lung Cancer Staging Project: Proposals for Revision of the TNM Stage Groups in the Forthcoming (Ninth) Edition of the TNM Classification for Lung Cancer. J Thorac Oncol 2024;19:1007-27. [Crossref] [PubMed]
8. Walker C, Mojares E, Del Río Hernández A. Role of Extracellular Matrix in Development and Cancer Progression. Int J Mol Sci 2018;19:3028. [Crossref] [PubMed]
9. Chen B, Khodadoust MS, Liu CL, et al. Profiling Tumor Infiltrating Immune Cells with CIBERSORT. Methods Mol Biol 2018;1711:243-59. [Crossref] [PubMed]
10. Doyle AD, Nazari SS, Yamada KM. Cell-extracellular matrix dynamics. Phys Biol 2022;
11. Comprehensive molecular profiling of lung adenocarcinoma. Nature 2014;511:543-50. [Crossref] [PubMed]
12. Yang H, Jiang P, Liu D, et al. Matrix Metalloproteinase 11 Is a Potential Therapeutic Target in Lung Adenocarcinoma. Mol Ther Oncolytics 2019;14:82-93. [Crossref] [PubMed]
13. Zhuang Y, Li X, Zhan P, et al. MMP11 promotes the proliferation and progression of breast cancer through stabilizing Smad2 protein. Oncol Rep 2021;45:16. [Crossref] [PubMed]
14. Pan Z, Liu L, Nie W, et al. Long non-coding RNA AGER-1 functionally upregulates the innate immunity gene AGER and approximates its anti-tumor effect in lung cancer. Mol Carcinog 2018;57:305-18. [Crossref] [PubMed]
15. Heinz S, Benner C, Spann N, et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol Cell 2010;38:576-89. [Crossref] [PubMed]
16. Sun L, Sun J, Song JD. High expression of DLC family proteins predicts better prognosis and inhibits tumor progression in NSCLC. Mol Med Rep 2019;19:4881-9. [Crossref] [PubMed]
17. Kim TY, Jackson S, Xiong Y, et al. CRL4A-FBXW5-mediated degradation of DLC1 Rho GTPase-activating protein tumor suppressor promotes non-small cell lung cancer cell growth. Proc Natl Acad Sci U S A 2013;110:16868-73. [Crossref] [PubMed]
18. Tripathi BK, Anderman MF, Bhargava D, et al. Inhibition of cytoplasmic EZH2 induces antitumor activity through stabilization of the DLC1 tumor suppressor protein. Nat Commun 2021;12:6941. [Crossref] [PubMed]
19. Peng Q, Zhang L, Li J, et al. FOXA1 Suppresses the Growth, Migration, and Invasion of Nasopharyngeal Carcinoma Cells through Repressing miR-100-5p and miR-125b-5p. J Cancer 2020;11:2485-95. [Crossref] [PubMed]
20. Takaku M, Grimm SA, De Kumar B, et al. Cancer-specific mutation of GATA3 disrupts the transcriptional regulatory network governed by Estrogen Receptor alpha, FOXA1 and GATA3. Nucleic Acids Res 2020;48:4756-68. [Crossref] [PubMed]
21. La'ah AS, Tsai PH, Yarmishyn AA, et al. Neutrophils Recruited by NKX2-1 Suppression via Activation of CXCLs/CXCR2 Axis Promote Lung Adenocarcinoma Progression. Adv Sci (Weinh) 2024;11:e2400370. [Crossref] [PubMed]
22. Quintero-Fabián S, Arreola R, Becerril-Villanueva E, et al. Role of Matrix Metalloproteinases in Angiogenesis and Cancer. Front Oncol 2019;9:1370. [Crossref] [PubMed]
23. Popper HH. Progression and metastasis of lung cancer. Cancer Metastasis Rev 2016;35:75-91. [Crossref] [PubMed]
24. Oshima S, Asai S, Seki N, et al. Identification of Tumor Suppressive Genes Regulated by miR-31-5p and miR-31-3p in Head and Neck Squamous Cell Carcinoma. Int J Mol Sci 2021;22:6199. [Crossref] [PubMed]
25. Huang G, Chen S, He J, et al. Histone Lysine Lactylation (Kla)-induced BCAM Promotes OSCC Progression and Cis-Platinum Resistance. Oral Dis 2025;31:1116-32. [Crossref] [PubMed]
26. Shang J, Jiang H, Zhao Y, et al. Molecular subtyping of stage I lung adenocarcinoma via molecular alterations in pre-invasive lesion progression. J Transl Med 2025;23:263. [Crossref] [PubMed]
27. Morton ML, Bai X, Merry CR, et al. Identification of mRNAs and lincRNAs associated with lung cancer progression using next-generation RNA sequencing from laser micro-dissected archival FFPE tissue specimens. Lung Cancer 2014;85:31-9. [Crossref] [PubMed]
28. Zilionis R, Engblom C, Pfirschke C, et al. Single-Cell Transcriptomics of Human and Mouse Lung Cancers Reveals Conserved Myeloid Populations across Individuals and Species. Immunity 2019;50:1317-1334.e10. [Crossref] [PubMed]
29. Chen J, Suo S, Tam PP, et al. Spatial transcriptomic analysis of cryosectioned tissue samples with Geo-seq. Nat Protoc 2017;12:566-80. [Crossref] [PubMed]