Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Vol.:(0123456789)
Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6
Discover Oncology
Research
Multicenter cohort analysis ofanoikis andEMT: implications
forprognosis andtherapy inlung adenocarcinoma
LuYin1· ZhanshuoZhang2· ZhuYan3· QiuyueYan4
Received: 5 July 2024 / Accepted: 2 September 2024
© The Author(s) 2024 OPEN
Abstract
Background Anoikis and epithelial-mesenchymal transition (EMT) are pivotal in the distant metastasis of lung adeno-
carcinoma (LUAD). A detailed understanding of their interplay and the identication of key genes is vital for eective
therapeutic strategies against LUAD metastasis.
Methods Key prognostic genes related to anoikis and EMT were identied through univariate Cox regression analysis. We
utilized ten machine learning algorithms to develop the Anoikis and EMT-Related Optimal Model (AEOM). The TCGA-LUAD
dataset served as the training cohort, while six additional international multicenter LUAD datasets were employed as
validation cohorts. The average concordance index (c-index) was used to evaluate model performance and identify the
most eective model. Subsequent multi-omics analyses were conducted to explore dierences in pathway enrichment,
immune inltration, and mutation landscapes between high and low AEOM groups. Experimental validation demon-
strated that RHPN2, a key biomarker within the model, acts as an oncogene facilitating LUAD progression.
Results The AEOM displayed superior prognostic predictive performance for LUAD patients, outperforming numerous previ-
ously published LUAD signatures. Biologically, the AEOM was notably associated with immune features; the high AEOM
group exhibited decreased immune activity and a tendency towards immune-cold tumors, as well as a higher tumor
mutational burden (TMB). Subgroup analysis revealed that the low AEOM + high TMB group had the most favorable
prognosis. The high AEOM group was primarily enriched in cell cycle-related pathways, promoting cancer cell prolifera-
tion. RHPN2, a crucial gene within the AEOM (correlation = 0.85, P < 0.05), was linked to poorer prognosis in LUAD patients
with elevated RHPN2 expression. Further invitro experiments showed that RHPN2 modulates LUAD cell proliferation
and invasion.
Conclusion The AEOM provides a robust prognostic model for LUAD, uncovering critical immune and biological path-
ways, with RHPN2 identied as a key oncogenic driver. These ndings oer valuable insights for targeted therapies and
enhanced patient outcomes.
Keywords Anoikis· Epithelial-mesenchymal transition· Lung adenocarcinoma· Tumor microenvironment· Prognosis
Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1007/ s12672- 024-
01293-6.
* Zhu Yan, 1332080365@qq.com; * Qiuyue Yan, 351124959@qq.com | 1School ofMathmatic andInformation, Nanjing Normal
University ofSpecial Education, Nanjing, China. 2Department ofLung Cancer, Tianjin Lung Cancer Center, National Clinical Research Center
forCancer, Key Laboratory ofCancer Prevention andTherapy, Tianjin’s Clinical Research Center forCancer, Tianjin Medical University Cancer
Institute andHospital, Tianjin, China. 3Department ofRespiratory Diseases, The Aliated Huai’an Hospital ofXuzhou Medical University,
Huai’an Second People’s Hospital, Huai’an, China. 4Emergency Medicine Department, The Aliated Huai’an Hospital ofYangzhou
University, Huai’an Fifth People’s Hospital, Huai’an, China.
Vol:.(1234567890)
Research Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6
1 Introduction
Lung cancer is the leading cause of cancer related deaths in China and worldwide. Compared to other malignant tumors,
it not only has a high incidence but also has subtle early symptoms, often leading to diagnosis at an advanced stage
and reducing the eectiveness of treatment [1, 2]. Non small cell lung cancer (NSCLC) is the most common histologic
subtype,with adenocarcinoma being the predominant type [3]. Currently, the treatment of Lung adenocarcinoma (LUAD)
includes surgery, radiotherapy, chemotherapy, molecular targeted therapy, and immunotherapy, but due to high metas-
tasis rate, resistance to radiotherapy and chemotherapy, and lack of systemic treatment, the 5year survival rate of patients
remains very low [4–6]. The occurrence and development of lung cancer involve interactions among various molecules,
cells, and tissues, therefore exploring potential molecular targets is urgently needed to improve the treatment of lung
cancer.
Epithelial mesenchymal transition(EMT) refers to the process of epithelial cells losing their connections and apical
basal polarity, reorganizing their cytoskeleton, changing dened cell shape, and reprogramming the signal program
of gene expression to transform into mesenchymal cells [7]. Research has shown that EMT play a crucial role in tumor
proliferation, invasion, and metastasis (8). Cao etal [9]. Developed a signature based on EMT related genes that has good
predicting in the prognosis of bladder cancer.
Anoikis, a specialized form of programmed cell death resulting from cell detachment from the extracellular matrix [10,
11] is crucial for preventing unwanted cell colonization. In cancer, tumor cells often evade anoikis, allowing survival in
harsh environments like the bloodstream, thereby aiding metastasis. EMT plays a complementary role by transforming
epithelial cells into a mesenchymal state, enhancing their migratory and invasive properties. Both anoikis resistance and
EMT are critical in metastasis, working synergistically to promote cancer cell dissemination. While studies have devel-
oped prognostic models incorporating anoikis and EMT-related genes in other cancers, like colon adenocarcinoma, the
interplay of these processes in LUAD is less understood (12). In LUAD, understanding this crosstalk is vital for discovering
new therapeutic targets and strategies to mitigate metastasis, highlighting the need for continued research in this area.
This study investigates the roles of anoikis and EMT in LUAD. Using machine learning methods, we developed the
Anoikis and EMT-related optimal model (AEOM) and demonstrated its excellent performance in prognosis prediction.
Additionally, we conducted multi-omics analyses to explore dierences in immune inltration, mutation landscape, and
pathway enrichment between dierent AEOM groups. Finally, we identied RHPN2 as a key marker within the AEOM,
functioning as an oncogene that promotes the progression and metastasis of LUAD, potentially serving as a signicant
therapeutic target.
2 Method
2.1 Dataset source
The transcriptomic, copy number variation (CNV), mutation, and clinical data of LUAD were collected from The Cancer
Genome Atlas (TCGA) database (https:// portal. gdc. cancer. gov) and used as the training set for model construction. Six
transcriptome datasets from the GEO database were used for validation: GSE13213 [13] (n = 119), GSE26939 [14] (n = 115),
GSE29016 [15] (n = 39), GSE30219 [16] (n = 86), GSE31210 [17] (n = 227), and GSE42127 [18] (n = 134). To ensure consistency
and comparability, gene expression data were converted to Transcripts Per Million (TPM) format. The “combat” function
in the “sva” package [19, 20] was used to adjust for potential batch eects. Additionally, all datasets from TCGA and GEO
were log-transformed to create a standardized data format from the start of the analysis. Principal component analysis
(PCA) was performed to assess batch eects between datasets.
Development and Validation of the Anoikis and EMT-Related Optimal Model (AEOM) for Prognosis Prediction.
Univariate Cox regression analysis was performed to identify Anoikis and EMT-Related prognostic Genes (AEPGs) for
LUAD patients. Subsequently, utilizing tenfold cross-validation, we examined 101 combinations of ten machine learning
Fig. 1 Selection of EMT and Anoikis genes. A Univariate Cox Regression Analysis: Identication of genes related to prognosis from the EMT
and anoikis pathways. B Data Integration: Analysis across multiple bulk-LUAD datasets to ensure robustness. C Forest Plot: Visualization of
the univariate Cox regression results, highlighting prognostic signicance of selected genes. D, E Gene Ontology Enrichment: Functional
categorization and enrichment analysis of genes associated with prognosis, illustrating key biological processes involved. F Chromosomal
Alterations: Frequency distribution analysis of chromosomal amplications and deletions among the prognostic-related genes, providing
insights into genetic variations contributing to disease progression ▸
Vol.:(0123456789)
Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6 Research
Vol:.(1234567890)
Research Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6
algorithms: stepwise Cox, Lasso, Ridge, partial least squares regression for Cox (plsRcox), CoxBoost, random survival forest
(RSF), generalized boosted regression modeling (GBM), elastic net (Enet), supervised principal components (SuperPC),
and survival support vector machine (survival-SVM). The aim was to identify the most valuable AEOM characterized by
the highest C-index. The accuracy of AEOM was evaluated using ROC curves and PCA analysis. Additionally, we com-
prehensively reviewed prognostic signatures from published literature, including lncRNA and mRNA signatures, and
compared the performance of AEOM using the C-index as the evaluation metric.
2.2 Biological function andpathway analysis
To investigate the biological functions and pathway processes associated with MPRGs, we conducted Gene Ontology
(GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses using the “clusterProler” package. MPRGs were
converted to Entrez IDs and used as inputs for GO and KEGG enrichment analyses, with an adjusted p-value < 0.05 as the
criterion. Additionally, potential mechanisms were examined using Gene Set Variation Analysis (GSVA). The cancer immu-
nity cycle and pathways anticipated to respond to immunotherapy were assessed according to established methods [21,
22] A comprehensive collection of pathway gene sets was sourced from the Molecular Signatures Database (MSigDB).
2.3 Comprehensive analysis ofgenomic alterations, immune profiling
Genomic alterations, including recurrent amplication and deletion regions, were identied using GISTIC 2.0 analysis.
Tumor Mutational Burden (TMB) was calculated with the R package ‘maftools’ [23]. The Cancer Immunome Atlas (https://
tcia. at/ home) was utilized to assess the immunophenoscore (IPS) of LUAD patients, identifying those suitable for immu-
notherapy. Additionally, the ssGSEA algorithm was used to evaluate immune cell inltration and the activity of immune-
related pathways in tumor samples. The TIMER2.0 database provided comprehensive data on immune cell inltration
abundance in TCGA, incorporating results from multiple algorithms.
2.4 Cell culture andsiRNA transfection
A549 and H1299 LUAD cell lines were obtained from the Institute of Biochemistry and Cell Biology, Chinese Academy of
Sciences, Shanghai. Cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS) and
1% antibiotics (100 U/ml penicillin and 100mg/ml streptomycin). siRNA transfection was conducted using Lipo2000
reagent (Invitrogen, Shanghai) according to the manufacturer’s instructions. A549 and H1299 cells were seeded onto
coverslips in six-well plates, with siRNA transfection performed the next day.
2.5 Colony formation assay
For the colony formation assay, 5000 cells were seeded in each well of a six-well plate with growth medium, which was
replaced weekly. After two weeks, colonies were xed with methanol for 15min and stained with 0.1% crystal violet
(Sigma) for 30min. Colony formation was then quantied by counting the stained colonies.
Fig. 2 Construction and validation of a machine learning model. A Selecting the optimal machine learning model by computing the aver-
age C-index across six validation cohorts. B Pie chart displaying the seven datasets used in the model construction, with TCGA serving as
the training set and the other six datasets as validation sets. C–I Conducting survival analysis on TCGA, GSE13213, GSE26939, GSE29016,
GSE30219, GSE31210 and GSE42127 cohorts. Based on AEOM scores, patients are categorized into high and low expression groups, with the
high AEOM group exhibiting a notably worse prognosis. J–K IPS scores were utilized to predict dierences in immune therapy responses
between the high and low AEOM groups. N–P TIDE analysis was employed to assess the dierences in immune therapy responses between
the high and low AEOM groups, with the low AEOM group demonstrating better responses to immunotherapy
▸
Vol.:(0123456789)
Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6 Research
Vol:.(1234567890)
Research Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6
2.6 Wound‑healing assay
To evaluate cell migration, transfected cells were cultured in six-well plates until 95% conuence. A uniform scratch
was made using a sterile 20μL pipette tip. Detached cells were removed by rinsing twice with PBS. Wound closure was
documented at 0 and 48h post-scratch using ImageJ software to measure scratch width.
2.7 Invasion andmigration assays
Invasion and migration assays were performed using Corning’s Transwell system (24wells, 8μm pore size). For migration
assays, 5 × 10^4 transfected cells were placed in the upper chambers with 350μL of serum-free medium, while 700μL
of medium containing 10% FBS was added to the lower chambers. For invasion assays, Transwell membranes were pre-
coated with Matrigel (Sigma-Aldrich). After 16h, non-migrated cells on the upper surface were removed, and migrated
cells on the lower surface were xed with methanol and stained with 0.1% crystal violet. Images were captured using an
Olympus inverted microscope (Tokyo, Japan).
2.8 Statistical analysis
Data analysis and visualization were performed using R software version 4.2.0. Kaplan–Meier survival analysis and
log-rank tests were utilized to estimate and compare overall survival (OS) among different subtypes. Differences in
continuous variables between groups were assessed using either the Wilcoxon rank-sum test or the Student’s t-test,
depending on data distribution, while categorical variables were evaluated with chi-squared tests or Fisher’s exact
tests. To address multiple comparisons and reduce type I errors, p-values were adjusted using the false discovery rate
(FDR) method. Pearson correlation analysis was conducted to explore relationships among variables. All statistical
tests were two-tailed with a significance threshold set at p < 0.05, ensuring a rigorous evaluation and robust conclu-
sions from the study findings.
3 Result
3.1 Identification andprognostic analysis ofanoikis andEMT‑related genes
Initially, we performed a preliminary screening of anoikis and EMT-related genes. Figure1A illustrates the specific
process. We integrated data from seven global multicenter LUAD datasets (Fig.1B) and identified prognostic-related
genes using univariate Cox analysis (genes with prognostic significance in at least six datasets were included in sub-
sequent studies) (Fig.1C). Ultimately, we identified 23 AEPGs associated with prognosis (p < 0.05, Fig.1D). Further
GO and KEGG enrichment analyses revealed that AEPGs were primarily enriched in pathways such as oocyte meiosis,
the HIF-1 signaling pathway, and the cell cycle (Fig.1D, E). CNV status analysis indicated frequent changes in AEPGs,
with HMMR and KRT8 showing the most extensive CNV amplification (Fig.1F).
Development and Validation of an Anoikis and EMT-Related Optimal Model (AEOM) for Prognosis and Immuno-
therapy Suitability
Using the expression profiles of 23 prognosis-related AEPGs, we developed an AEOM through a machine learn-
ing combinatorial algorithm. The TCGA dataset served as the training cohort, while six GEO datasets were used for
Fig. 3 Clinical practice value of model. A Comparing the model’s risk scores with other clinical indicators to demonstrate superior prog-
nostic value. B Distribution analysis of LUAD datasets based on model gene expression using PCA. C ROC curves of the model in the TCGA
cohort and GEO cohorts. D Assessing against published signatures, highlighting the model’s highest C-index scores across multiple datasets
▸
Vol.:(0123456789)
Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6 Research
Vol:.(1234567890)
Research Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6
Fig. 4 Assessment of immune inltration. A Calculation of the disparity in immune cell inltration between high and low AEOM groups
utilizing seven distinct immune inltration assessment algorithms. B Examination of dierences in immune-related gene expression at the
mRNA, methylation, and copy number variation levels between high and low AEOM groups
Vol.:(0123456789)
Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6 Research
validation. The average c-index across the six validation cohorts was the criterion for model selection. Ultimately, the
StepCox [both] + SuperPC algorithm emerged as the optimal model (Fig.2A). The AEOM score distinguished patient
prognosis across all seven cohorts (Figs.2B–I), with patients in the high AEOM group exhibiting worse outcomes
compared to those in the low AEOM group. Additionally, IPS scores calculated using the TCIA website suggested that
the low AEOM group is more suitable for treatment with PD-1, CTLA-4, or their combination therapy (Figs.2J–M).
Further analysis using the TIDE website indicated that the high AEOM group had higher TIDE scores (Figs.2N–P),
implying a greater likelihood of immune escape and unsuitability for immunotherapy.
3.2 Superior predictive performance ofAEOM
To evaluate the predictive efficacy of AEOM, we integrated clinical features from seven datasets. The results showed
that the c-index values of AEOM were higher than those of any other clinical features (such as age, gender, stage,
EGFR status, etc.) (Fig.3A). Subsequently, we conducted a principal component analysis (PCA) based on the expression
levels of the AEOM model genes across all datasets. The results indicated that the expression levels of the model genes
could effectively distinguish LUAD patients, with high and low AEOM groups forming distinct clusters (Fig.3B). ROC
curves further confirmed that AEOM scores could reliably predict the prognosis of LUAD patients, which was validated
across all datasets (with one-, three-, and five-year AUC values generally above 0.65, Fig.3C). Next, we compared
AEOM against numerous previously published LUAD signatures and found that AEOM consistently demonstrated
the best predictive performance across all datasets, achieving the highest c-index values (Fig.3D).
3.3 Immune infiltration andregulatory gene expression inhigh andlow aeom groups
Immune regulatory genes are pivotal in modulating the tumor immune response, inuencing both tumor progression
and patient prognosis. In our analysis, we compared the relative expression levels of these genes between the high and
low AEOM groups. Our ndings revealed distinct immune activation patterns: the low AEOM group demonstrated an
upregulation of MHC class II molecules and co-stimulatory molecules, which are crucial for antigen presentation and the
activation of adaptive immune responses. In contrast, the high AEOM group showed increased expression of MHC class I
molecules and certain co-inhibitory molecules, which may contribute to immune evasion by the tumor (Fig.4A). Further
analysis using the TIMER2.0 database allowed us to assess the abundance of various immune inltrating cells within the
tumor microenvironment. The low AEOM group exhibited signicantly higher inltration of lymphocytes, including T
cells and B cells, indicating a more active immune surveillance and potential anti-tumor response (Fig.4B). These dif-
ferences in immune cell inltration and gene expression proles underscore the complex interplay between the tumor
and the immune system, highlighting potential therapeutic targets for enhancing immune-mediated tumor control.
3.4 Exploring therelationship betweentumor mutations andpatient prognosis
Given the signicant correlation between tumor mutations, immune response, and patient prognosis, this analysis delves
into these relationships in detail. The heatmap prominently highlights a markedly elevated tumor mutational burden
(TMB) in the high AEOM group, with TP53, XIRP2, SPTA1, ADAMTS12, and RP1L1 being the most frequently mutated genes
(Fig.5A). Subgroup analysis indicated that the high-mutation, low-AEOM group had a better prognosis (Fig.5B). Further-
more, the high AEOM group exhibited a higher TMB, and AEOM scores were positively correlated with TMB (Figs.5C, D).
3.5 Unraveling themechanistic pathways andimmune responses
To explore the potential pathway mechanisms underlying the dierences between various AEOM groups, GSVA enrich-
ment analysis was conducted. The high AEOM group primarily enriched pathways related to DNA replication, check-
point signaling, and lung cancer poor survival (Figs.6A–H). Abnormal activation or suppression of these pathways
Vol:.(1234567890)
Research Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6
leads to uncontrolled cell cycle progression, increased DNA replication stress, accumulation of genetic variations, and
signicant chromosomal structural changes. These collectively promote unlimited tumor cell proliferation, genomic
instability, and enhanced invasive and metastatic potential. Next, we investigated the associations between AEOM
scores, immune therapy-related pathways, and cancer immunity cycle steps. First, AEOM showed signicant positive
correlations with various steps of the cancer immunity cycle (Fig.7A), such as cancer antigen release (Step 1) and the
recruitment of immunosuppressive cells (e.g., MDSCs and neutrophils). This suggests that the high AEOM score group
may exhibit heightened activity in these immune steps, leading to more eective antigen release and recruitment of
immunosuppressive cells. Additionally, AEOM demonstrated strong correlations with several key biological pathways,
such as DNA replication, cell cycle regulation, and mismatch repair. These pathways may be abnormally activated or
suppressed in the high AEOM score group, thereby aecting tumor progression and immune response. For instance,
abnormally active DNA replication and cell cycle regulation can result in rapid tumor cell proliferation and genomic
instability, while alterations in mismatch repair mechanisms may increase mutation rates. Overall, patients with high
AEOM scores exhibit more pronounced activity in these key pathways and immunity cycle steps, which are closely
related to tumor invasiveness and prognosis. This indicates a complex relationship between AEOM and multiple cancer-
related biological pathways and immunity cycle steps, where high AEOM scores may predict poorer prognosis and
higher tumor aggressiveness.
3.6 RHPN2 identified asakey oncogene inLUAD
Further investigation into the correlation between AEOM and clinical indicators revealed that patients with high
AEOM scores had a higher number of deaths and tended to have more advanced T stage, N stage, and overall stage
(Fig.7B). Additionally, all model genes were highly expressed in the high AEOM group. These findings suggest that
AEOM is an effective prognostic predictor and is significantly associated with clinical characteristics. To identify
biomarkers for AEOM, we conducted a correlation analysis between model genes and AEOM scores (Figs.7C–I), find-
ing that RHPN2 had the most significant correlation (correlation = 0.85, p < 0.05). Previous analyses indicated that
RHPN2 is a risk gene (HR > 1, P < 0.05). Therefore, we performed further experimental exploration of RHPN2. CCK8
assays showed that with RHPN2 knockdown, the OD values of LUAD cells decreased, indicating reduced proliferation
capacity (Figs.8A, B). Colony formation assays revealed a significant reduction in the number of colonies in RHPN2
knockdown LUAD cells (Figs.8C, D). Scratch assays demonstrated that RHPN2 knockdown could inhibit the migra-
tion of LUAD cells (Figs.8E, F). Transwell assays confirmed that RHPN2 knockdown inhibited tumor cell invasion and
migration (Figs.8G, H). These findings indicate that RHPN2 functions as an oncogene, regulating the proliferative,
invasive, and migratory capacities of LUAD cells.
4 Discussion
Lung cancer is the second most common and deadliest malignant tumor in the world [24]. Lung cancer also has the
highest incidence and mortality rates in China, with a 5year survival rate between 10%-15% [25]. Approximately
80%-85% of lung cancer cases are NSCLC, with LUAD being the most common pathological type. Despite some
advancements in diagnosis and treatment techniques, the long-term survival rate of LUAD patients remains low.
Therefore, there is an urgent need to find new and effective biomarkers or treatment targets for LUAD patients.
Vol.:(0123456789)
Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6 Research
Fig. 5 Genomic and transcriptomic proling between high and low-risk groups in the TCGA dataset. A Genomic landscape dierences
between high and low AEOM groups. B Survival curves displaying survival dierences among four subgroups. C Correlation between risk
scores and TMB. D Dierences in standardized TMB between high and low AEOM groups
Vol:.(1234567890)
Research Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6
Fig. 6 Comparing the biological characteristics of high and low-risk groups in the TCGA cohort. (A) GSVA enrichment analysis describes the biological charac-
teristics of high and low AEOM groups. B–H tSNE plots depict dierences in GO and KEGG pathway activities between high and low AEOM groups
Vol.:(0123456789)
Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6 Research
Fig. 7 Correlation analysis of AEOM score and the expression patterns of model genes. A Correlation analysis between AEOM score and
tumor immune cycle pathway activity using ssGSEA for evaluation. B Heatmap displaying dierences in clinical indicators between high
and low AEOM groups, along with dierences in the expression of model genes. C Correlation between expression of model genes and
AEOM score
Vol:.(1234567890)
Research Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6
Anoikis resistance is a critical factor influencing the progression and metastasis of various cancers, including
lung cancer, gastric cancer, colorectal cancer, breast cancer, and prostate cancer, among others [26–28]. This
resistance enables tumor cells to survive detachment from the extracellular matrix, facilitating their dissemina-
tion to distant sites. Tumor cells can acquire anoikis resistance through multiple signaling pathways, enhancing
their invasive and metastatic potential [29]. Research has highlighted the pivotal role of EMT in conferring resist-
ance to anoikis. EMT involves the transformation of epithelial cells into mesenchymal cells, which is character-
ized by reduced cell–cell adhesion and increased migratory capacity. A key marker of this transformation is the
downregulation of E-cadherin, a protein essential for maintaining epithelial integrity. Studies have shown that
decreased E-cadherin expression is strongly associated with a heightened resistance to anoikis in tumor cells. In
LUAD, understanding the interplay between EMT and anoikis resistance is vital for identifying key genes that drive
invasion and metastasis. A comprehensive analysis that integrates both EMT and anoikis pathways can provide
insights into potential therapeutic targets, paving the way for more effective treatments aimed at limiting cancer
spread and improving patient outcomes.
This study systematically analyzed genes related to anoikis and EMT and developed a prognostic model based
on these genes, referred to as the AEOM. By integrating multiple global multicenter LUAD datasets, we identified
23 AEPGs and utilized these genes to develop the AEOM model. Validation results demonstrated that the AEOM
exhibited excellent performance in predicting the prognosis of LUAD patients. Patients with high AEOM scores had
worse prognoses, higher likelihoods of immune escape, and lower suitability for immunotherapy. Furthermore,
experimental validation revealed that RHPN2 functions as a key oncogene in LUAD, regulating the proliferative,
invasive, and migratory capacities of tumor cells. In summary, the AEOM model and its associated biomarkers, such
as RHPN2, provide new perspectives and potential targets for prognostic assessment and personalized treatment
of LUAD patients.
However, this study also has some limitations. First, it is a retrospective study based on gene expression profiles
from the TCGA and GEO databases and a few clinical factors, and some specific clinical information related to lung
cancer may not have been obtained. Second, this study only used cell experiments for the validation of key genes,
lacking further exploration in clinical tissue samples and invivo experiments. Collecting in-house cohorts is neces-
sary to further validate the effectiveness of the model.
In summary, this study identies and validates a novel prognostic model based on anoikis and EMT-related genes
for predicting LUAD patient outcomes, highlighting RHPN2 as a key oncogene and suggesting potential biomarkers for
personalized treatment strategies.
Fig. 8 Exploring the role of the key gene RHPN. A, B CCK-8 analysis: Viability of A549 and H1299 cells signicantly decreases after RHPN2
knockout, used for further invitro experiments. C, D Colony formation assay: cells with reduced RHPN2 expression show signicantly fewer
colony numbers compared to the NC group. E, F Scratch assay: Cells with decreased RHPN2 expression exhibit signicantly slower wound
healing rates. G–I Transwell assay: downregulation of RHPN2 expression inhibits tumor cell migration and invasion capabilities. Notations:
**P < 0.01; ***P < 0.001
▸
Vol.:(0123456789)
Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6 Research
Vol:.(1234567890)
Research Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6
Acknowledgements No funding
Author contributions The study was conceived and designed by LY. Data collection was conducted by ZZ. LY performed the statistical analysis.
The rst draft of the manuscript was written by LY. The experiment was performed by LY. The nal approval of the submitted version was given
by ZY and QY. All authors contributed to the manuscript and approved the submitted version.
Funding No funding
Data availability All datasets pertinent to this study are accessible through the TCGA database (http:// cance rgeno me. nih. gov/), GEO database
(https:// www. ncbi. nlm. nih. gov/ geo/), the Molecular Signature Database (MSigDB) (https:// www. gsea- msigdb. org/ gsea/ msigdb/), TIP (http://
biocc. hrbmu. edu. cn/ TIP/ index. jsp), or the data availability sections of the relevant publications. All data relevant to this investigation, whether
generated or analyzed, are comprehensively detailed in this manuscript and its supplementary materials. For further inquiries or data requests,
interested parties are advised to reach out to the corresponding authors.
Declarations
Ethic approval and consent to participate All experiments conducted in this study were approved by the Ethics Committee of the Aliated
Huai’an Hospital of Xuzhou Medical University.
Competing interests It is hereby declared by the authors that the research was carried out without the presence of any potential conict of
interest arising from commercial or nancial relationships.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which
permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to
the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modied the licensed material. You
do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party
material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If
material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds
the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco
mmons. org/ licen ses/ by- nc- nd/4. 0/.
References
1. Kang Y, Jin Y, Li Q, Yuan X. Advances in lung cancer driver genes associated with brain metastasis. Front Oncol. 2020. https:// doi. org/ 10.
3389/ fonc. 2020. 606300.
2. Zhang L, Cui Y, Mei J, Zhang Z, Zhang P. Exploring cellular diversity in lung adenocarcinoma epithelium: advancing prognostic methods
and immunotherapeutic strategies. Cell Prolif. 2024. https:// doi. org/ 10. 1111/ cpr. 13703.
3. Zappa C, Mousa SA. Non-small cell lung cancer current treatment and future advances. Transl Lung Cancer Res. 2016. https:// doi. org/ 10.
21037/ tlcr. 2016. 06. 07.
4. Succony L, Rassl DM, Barker AP, McCaughan FM, Rintoul RC. Adenocarcinoma spectrum lesions of the lung: detection, pathology and
treatment strategies. Cancer Treat Rev. 2021. https:// doi. org/ 10. 1016/j. ctrv. 2021. 102237.
5. Li Y, Yan B, He S. Advances and challenges in the treatment of lung cancer. Biomed Pharmacother. 2023;169:115891. https:// doi. org/ 10.
1016/j. biopha. 2023. 115891.
6. Zhang P, Pei S, Zhou G, Zhang M, Zhang L, Zhang Z. Purine metabolism in lung adenocarcinoma: a single-cell analysis revealing prognostic
and immunotherapeutic insights. J Cell Mol Med. 2024;28(8):e18284. https:// doi. org/ 10. 1111/ jcmm. 18284.
7. Manoletti G, Fedele M. Epithelial-mesenchymal transition (Emt) 2021. Int J Mol Sci. 2022. https:// doi. org/ 10. 3390/ ijms2 31058 48.
8. Huang Y, Hong W, Wei X. The molecular mechanisms and therapeutic strategies of Emt in tumor progression and metastasis. J Hematol
Oncol. 2022;15(1):129. https:// doi. org/ 10. 1186/ s13045- 022- 01347-8.
9. Cao R, Yuan L, Ma B, Wang G, Qiu W, Tian Y. An Emt-related gene signature for the prognosis of human bladder cancer. J Cell Mol Med.
2020;24(1):605–17. https:// doi. org/ 10. 1111/ jcmm. 14767.
10. Kim YN, Koo KH, Sung JY, Yun UJ, Kim H. Anoikis resistance: an essential prerequisite for tumor metastasis. Int J Cell Biol. 2012. https:// doi.
org/ 10. 1155/ 2012/ 306879.
11. Zhang P, Zhang H, Tang J, Ren Q, Zhang J, Chi H, etal. The integrated single-cell analysis developed an immunogenic cell death signature
to predict lung adenocarcinoma prognosis and immunotherapy. Aging. 2023. https:// doi. org/ 10. 18632/ aging. 205077.
12. Zhou J, Yang S, Zhu D, Li H, Miao X, Gu M, etal. The crosstalk between anoikis and epithelial-mesenchymal transition and their synergistic
roles in predicting prognosis in colon adenocarcinoma. Front Oncol. 2023. https:// doi. org/ 10. 3389/ fonc. 2023. 11842 15.
13. Tomida S, Takeuchi T, Shimada Y, Arima C, Matsuo K, Mitsudomi T, etal. Relapse-related molecular signature in lung adenocarcinomas
identies patients with dismal prognosis. J Clin Oncol. 2009;27(17):2793–9. https:// doi. org/ 10. 1200/ jco. 2008. 19. 7053.
14. Wilkerson MD, Yin X, Walter V, Zhao N, Cabanski CR, Hayward MC, etal. Dierential pathogenesis of lung adenocarcinoma subtypes
involving sequence mutations, copy number, chromosomal instability, and methylation. PloS one. 2012;7(5):e36530. https:// doi. org/ 10.
1371/ journ al. pone. 00365 30.
15. Staaf J, Jönsson G, Jönsson M, Karlsson A, Isaksson S, Salomonsson A, etal. Relation between smoking history and gene expression proles
in lung adenocarcinomas. BMC Med Genom. 2012;5:22. https:// doi. org/ 10. 1186/ 1755- 8794-5- 22.
Vol.:(0123456789)
Discover Oncology (2024) 15:462 | https://doi.org/10.1007/s12672-024-01293-6 Research
16. Rousseaux S, Debernardi A, Jacquiau B, Vitte AL, Vesin A, Nagy-Mignotte H, etal. Ectopic activation of germline and placental genes
identies aggressive metastasis-prone lung cancers. Sci Transl Med. 2013. https:// doi. org/ 10. 1126/ scitr anslm ed. 30057 23.
17. Okayama H, Kohno T, Ishii Y, Shimada Y, Shiraishi K, Iwakawa R, etal. Identication of genes upregulated in Alk-positive and Egfr/Kras/
Alk-negative lung adenocarcinomas. Cancer Res. 2012;72(1):100–11. https:// doi. org/ 10. 1158/ 0008- 5472. Can- 11- 1403.
18. Tang H, Xiao G, Behrens C, Schiller J, Allen J, Chow CW, etal. A 12-gene set predicts survival benets from adjuvant chemotherapy in
non-small cell lung cancer patients. Clin Cancer Res. 2013;19(6):1577–86. https:// doi. org/ 10. 1158/ 1078- 0432. Ccr- 12- 2321.
19. Zhang Y, Parmigiani G, Johnson WE. Combat-Seq: batch eect adjustment for Rna-Seq count data. NAR Genom Bioinform. 2020. https://
doi. org/ 10. 1093/ nargab/ lqaa0 78.
20. Zhang P, Wu X, Wang D, Zhang M, Zhang B, Zhang Z. Unraveling the role of low-density lipoprotein-related genes in lung adenocarcinoma:
insights into tumor microenvironment and clinical prognosis. Environ Toxicol. 2024. https:// doi. org/ 10. 1002/ tox. 24230.
21. Xu L, Deng C, Pang B, Zhang X, Liu W, Liao G, etal. Tip: a web server for resolving tumor immunophenotype proling. Cancer Res.
2018;78(23):6575–80. https:// doi. org/ 10. 1158/ 0008- 5472. Can- 18- 0689.
22. Hu J, Yu A, Othmane B, Qiu D, Li H, Li C, etal. Siglec15 shapes a non-inamed tumor microenvironment and predicts the molecular subtype
in bladder cancer. Theranostics. 2021;11(7):3089–108. https:// doi. org/ 10. 7150/ thno. 53649.
23. Zhang P, Dong S, Sun W, Zhong W, Xiong J, Gong X, etal. Deciphering treg cell roles in esophageal squamous cell carcinoma: a compre-
hensive prognostic and immunotherapeutic analysis. Front Mol Biosci. 2023. https:// doi. org/ 10. 3389/ fmolb. 2023. 12775 30.
24. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, etal. Global cancer statistics 2020: globocan estimates of incidence
and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–49. https:// doi. org/ 10. 3322/ caac. 21660.
25. Wu F, Wang L, Zhou C. Lung cancer in china: current and prospect. Curr Opin Oncol. 2021;33(1):40–6. https:// doi. org/ 10. 1097/ cco. 00000
00000 000703.
26. Wang J, Luo Z, Lin L, Sui X, Yu L, Xu C, etal. Anoikis-associated lung cancer metastasis: mechanisms and therapies. Cancers. 2022. https://
doi. org/ 10. 3390/ cance rs141 94791.
27. Dai Y, Zhang X, Ou Y, Zou L, Zhang D, Yang Q, etal. Anoikis resistance-protagonists of breast cancer cells survive and metastasize after
Ecm detachment. Cell Commun Signal. 2023;21(1):190. https:// doi. org/ 10. 1186/ s12964- 023- 01183-4.
28. Zhao X, Wang Z, Tang Z, Hu J, Zhou Y, Ge J, etal. An anoikis-related gene signature for prediction of the prognosis in prostate cancer. Front
Oncol. 2023;13:1169425. https:// doi. org/ 10. 3389/ fonc. 2023. 11694 25.
29. Frisch SM, Schaller M, Cieply B. Mechanisms that link the oncogenic epithelial-mesenchymal transition to suppression of anoikis. J Cell
Sci. 2013. https:// doi. org/ 10. 1242/ jcs. 120907.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional aliations.
