Clinical Relevance of Somatic Mutations in Chinese Lung Adenocarcinoma and Their Prognostic Implications for Survival

doi:10.21203/rs.3.rs-3305508/v1

Download PDF

Research Article

Clinical Relevance of Somatic Mutations in Chinese Lung Adenocarcinoma and Their Prognostic Implications for Survival

https://doi.org/10.21203/rs.3.rs-3305508/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 01 May, 2024

Read the published version in Cancer Medicine →

Version 1

posted

You are reading this latest preprint version

Background: To comprehensively elucidate the genomic and mutational features of LUAD, it is imperative to conduct ongoing investigations into the genomic landscape. In this study, we aim to analyze the somatic mutation profile and assessed the significance of these informative genes utilizing a retrospective LUAD cohort.

Methods: A total of 247 Chinese samples were analyzed to exhibit the tumor somatic genomic alterations in patients with LUAD. The Cox regression analysis was employed to identify prognosis-related genes and establish a predictive model for stratifying patients with LUAD.

Results: In the Dianjiang People's Hospital (DPH) cohort, the top five frequent mutated genes were EGFR (68%), TP53 (30%), RBM10 (13%), LRP1B (9%), and KRAS (9%). Of which, EGFR is a mostly altered driver gene and most mutation sites locate in tyrosine kinase regions. Oncogene pathway alteration and mutation signature analysis demonstrated the RTK-RAS pathway alteration and smoking were the main carcinogenic factors of the DPH cohort. Furthermore, we identified 34 driver genes in the DPH cohort, including EGFR (68%), TP53 (30.4%), RBM10 (12.6%), KRAS (8.5%), LRP1B (8.5%), and so on, and 45 Clinical Characteristic-Related Genes (CCRGs) were found to closely related to the clinical high-risk factors.

We developed an MPGM risk model by integrating critical genes and oncogenic pathway alterations in LUAD patients from the DPH cohort. Based on publicly available LUAD datasets, we identified five genes, including BRCA2, ALK, BRAF, EGFR, and PDGFRA according to the multivariable Cox regression analysis. The MPGM-Low group showed significantly better overall survival (OS) compared to the MPGM-High group (P<0.0001, AUC=0.754). The robust performance was validated in 55 LUAD patients from the DPH cohort and another LUAD dataset. Immune characteristics analysis revealed a higher proportion of primarily DCs and mononuclear cells in the MPGM-Low risk group, while the MPGM-High risk group showed lower immune cells and higher tumor cell infiltration.

Conclusions: This study provides a comprehensive genomic landscape of Chinese LUAD patients and developsan MPGM risk model for LUAD prognosis stratification. Further follow-up will be performed for the patients in the DPH cohort consistently to explore the resistance and prognosis genetic features.

Lung adenocarcinoma

Mutation profiles

Clinical characteristic-related genes

MPGM risk model

Targeted sequencing

As is reported in Global cancer data, lung cancer is the leading cause of cancer-related death (Sung, Ferlay, Siegel et al. 2021). Based on the histological classification, non-small cell lung cancer (NSCLC) accounts for approximately 85% of lung cancer cases, and lung adenocarcinoma (LUAD) is the most prevalent subtype of NSCLC (Herbst, Morgensztern, Boshoff 2018).

Numerous studies and expert consensus have provided compelling evidence that the survival of cancer patients is closely related to clinical and pathology characteristics, such as age, clinical stage, smoking status, family history, and so on (Mitsudomi, Tan, Yang et al. 2023). Tu et al. prospectively studied the association of smoking characteristics with all-cause mortality in 23 cancer types. They proved the significant harms of current smoking on survival after a cancer diagnosis. In lung cancer, current smokers (HR = 1.22, 95% CI = 1.12–1.32) and former smokers (HR = 1.18, 95% CI = 1.10–1.26) were at greater risk of death (Chen, Liu, Wen et al. 2022; Tu, Ye, Huang et al. 2022). Guerreiro et al. proposed that with the increasing age (> 50) and clinical stage at diagnosis, relative survival decreased progressively based on a large population study (Chen, Liu, Wen et al. 2022; Guerreiro, Forjaz, Antunes et al. 2021). In addition, Tumor spread through airspaces (STAS) and vascular invasion are aggressive clinical pathology characteristics, independent poor prognosis factors for recurrence-free survival (RFS) in NSCLC according to the multivariate Cox regression analysis (Han, Kim, Mino-Kenudson et al. 2021; Lee, Chung, Cho et al. 2022). Abundant clinical features and prognostic biomarkers are critical in guiding precise therapy decisions and improving survival outcomes. However, most studies primarily focus on the prognostic significance of these clinical features, with limited evidence regarding the association between somatic tumor alterations and clinicopathologic features. Therefore, there is an urgent need for robust genomic evidence to unravel the molecular mechanisms linking gene alterations to clinicopathologic characteristics in LUADs.

With continual advancement and iteration of next-generation sequencing (NGS) technology, abundant driver gene mutations are identified, leading to a more comprehensive genomic landscape of LUAD (Martínez-Jiménez, Muiños, Sentís et al. 2020). Several studies have verified that tumor genome alterations are closely related to cancer occurrence and development. In addition, driver gene and oncogenic pathway alterations could also provide effective targeted therapy opportunities (Tamborero, Rubio-Perez, Deu-Pons et al. 2018). The receptor tyrosine kinase (RTK)/RAS/RAF pathway genes are frequently mutated in LUAD (Network 2014). Recurrently somatic genomic alterations in genes such as EGFR, KRAS, BRAF, and others have emerged as conventional therapeutic targets, leading to revolutionary improvements in clinical outcomes (Ettinger, Wood, Aisner et al. 2023).

Nevertheless, targeted therapy's vulnerability, characterized by the emergence of recurrent drug resistance and adverse reactions, highlights the imperative to gain a deeper understanding of tumor-specific somatic alterations and the underlying molecular mechanisms in LUAD. Despite the significant advancements in clinical trials for targeted therapy and immunotherapy, which have improved clinical outcomes for patients with LUAD, the practical implementation of these effective treatment options remains restricted to a minority of patients. Notably, the five-year survival rate for patients with advanced clinical stages (III-IV) remains below 60% (Detterbeck, Boffa, Kim et al. 2017; Sung, Ferlay, Siegel et al. 2021). Significant heterogeneity was observed among LUAD patients in histological subtypes and in genomic and cellular characteristics (Li, Wang, Yang et al. 2022). Therefore, we aim to develop a prognosis risk model based on mutation characteristics to enable risk stratification for patients, mainly when treatment options are restricted.

In this study, we comprehensively analyzed genomic mutation characteristics in a Chinese LUAD cohort. We explored the significantly different mutated genes between clinical pathology risk factors, which may play an essential role in the occurrence of specific clinical and pathological characteristics. Subsequently, a comprehensive screening was performed, incorporating these genes and hotspot mutations, driver gene alterations, and oncogenic pathway genes to identify a gene set that exhibits the strongest correlation with patient survival prognosis in LUAD. Leveraging these patient-specific genomic alterations, we developed a predictive prediction model to assist clinicians in patient risk stratification. Our study provides new insight into the genomic alterations and prognosis stratification for LUAD.

2.1 Data

The Dianjiang People's Hospital (DPH) cohort consisted of 247 LUAD patients. The baseline information of patients in the DPH cohort was in Supplementary Table 1, and the somatic mutation data were in Supplementary Table 2.

The EAS cohort (Chen, Yang, Teo et al. 2020) included 299 Chinese patients with LUAD and was used as a training cohort to develop the prognosis risk model. The corresponding clinical information and mutation data were summarized in Supplementary Table 3 and Supplementary Table 4, respectively. The RNA-seq data of 169 patients with LUAD from the EAS cohort were also acquired from cBioPortal and published in the previous article.

The validation cohort (MSK-LUAD) (Nguyen, Fong, Luthra et al. 2022) comprised 394 LUAD samples obtained from an Asian population within the MSK-MET dataset (Memorial Sloan Kettering-Metastatic Events and Tropisms, integrated pan-cancer cohort of tumor genomic and clinical outcome data, containing 3853), which was used to validate the predictive performance of risk model. The mutation and corresponding clinical data of EAS and MSK-LUAD cohorts were obtained from the cBioPortal (https://www.cbioportal.org/). The clinical information and mutation data were comprehensively summarized in Supplementary Table 5 and Supplementary Table 6, respectively.

Among 247 LUAD patients from the DPH cohort, 55 were treated with targeted or/chemotherapy /and immunotherapy and participated in regular follow-up examinations. The detailed prognosis and therapy (drug type and adverse reaction) information are in Supplementary Table 7. The last follow-up was performed on 2nd June 2023, and progression-free survival (PFS) was used as the survival endpoint.

2.2 Targeted sequencing

The formalin-fixed paraffin-embedded (FFPE) sample of the DPH cohort was profiled with targeting sequencing (1123 gene panel) (ChosenMed, Beijing, China), and germline variation was synchronously sequenced using the paired adjacent tissues or peripheral blood samples. First, the Maxwell RSC FFPE Plus DNA Kit (Promega, Cat no.AS1720) was used to extract DNA from tumor specimens. Then, Genomic DNA (gDNA) was sheared to about 200 bp fragment sizes with a Covaris E210 system (Covaris, Inc.). The KAPA HyperPrep Kit (Roche, 07962312001) and Agilent SureSelect XT kit (Agilent, G9702C) were used to construct the NGS library. The prepared library was quantified using the Qubit 3.0 Fluorometer (Life Technologies, Inc.), and quality and fragment size were measured with an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.). Finally, targeted sequencing of gDNA was performed, and the samples underwent paired-end sequencing on an Illumina NovaSeq 6000 platform (Illumina Inc) with a 150-bp read length.

2.3 Data processing

Burrows-Wheeler Aligner (BWA version 0.7.11) alignment algorithm was used to align the human reference genome (UCSC hg19). Next, the IndelRealigner tool from Genome Analysis Toolkit (GATK, version 3.6) (McKenna, Hanna, Banks et al. 2010) and VarScan software (Koboldt, Zhang, Larson et al. 2012) were used to call somatic mutations, and ANNOVAR annotated all variants. The following filtering criteria were applied to the mutation candidates to identify SNVs and Indels: (a) variants within intron were deleted; (b) mutations reported in more than 1% of the population in the 1000 Genomes Project (1000gAUG_2015ALL); (c) Mutations were then filtered against common single nucleotide polymorphisms (SNPs) found in dbSNP (http://www.ncbi.nlm.nih.gov/SNP); (d) synonymous variants were excluded; (e) variants with less than 50 supporting reads were removed; (e) mutations in the black flag were excluded; (f) the tumor cell content ≥ 10%, with an average sequencing depth ≥ 2000X, and DNA insert fragment size ≥ 140bp.

2.4 Genomic alterations analysis

Tumor mutational burden (TMB) was calculated according to the number of all nonsynonymous mutations and indels per MB of all the genomic alterations, and the cut-off value for TMB-high and TMB-low was defined as ten mutations/MB. Driver genes of the EAS LUAD cohort were identified using MutSigCV (Lawrence, Stojanov, Polak et al. 2013) method.

2.5 Development of the risk model

Multiple parameter gene mutations (MPGM) were screened with mutation frequency > 0% in the training cohort, and the remaining candidate genes were further screened utilizing univariate and multivariable Cox regression analysis. The multivariable Cox regression was performed to develop the MPGM risk model, and the risk scores were calculated using the following formula: \(\text{M}\text{P}\text{G}\text{M} \text{s}\text{c}\text{o}\text{r}\text{e}=\sum (mutationstatus\times coefficient)\). The mutationstatus was defined as follows: "0" denoted the absence of any mutations, while "1" indicated the presence of mutations. The median cutoff value was set to stratify into MPGM-High and MPGM-Low groups.

2.6 Immune infiltrating analysis of risk model

The relative abundance of immune infiltrating cells were analyzed using a different algorithm, including Cibersort (Chen, Khodadoust, Liu et al. 2018), Timer (Li, Fan, Wang et al. 2017), Quantiseq (Finotello, Mayer, Plattner et al. 2017), and MCPcounter (Becht, Giraldo, Lacroix et al. 2016). Among them, the result from Cibersort provided a relative abundance of immune cells. EPIC was used to estimate the ratio of immune cells to cancer cells (Yeo, Wasser, Kumar et al. 2020). The estimate was used to predict tumor purity and stroma abundance and assess the immune cell invasion level (Yoshihara, Shahmoradgoli, Martínez et al. 2013). Additionally, Xcell was employed to evaluate the relative enrichment level of immune cells within the tumor tissue (Aran, Hu, Butte 2017).

2.7 Statistical analysis

Fisher’s exact test (two-sided) was used to assess the correlation between clinical pathology features and somatic mutations. The chi-square test was used to analyze the differences in the smoking population of two mutation signature groups. Two-sided Wilcoxon rank sum testing was used between two continuous variables, such as TMB level. Kaplan–Meier survival curves and Cox regression (hazard ratio [HR] and 95% confidence interval [CI]) for survival variables were assessed for the OS and PFS estimations. P < 0.05 or FDR < 0.05 were considered statistically significant. The AUC of the ROC curve was calculated to evaluate the performance of the risk model. The statistical analyses were performed using R software version 4.1.2.

3.1 Clinical characters of DPH cohort.

We retrospectively enrolled 247 LUAD patients of Chinese sought treatment at the Dianjiang People's Hospital (the “DPH cohort”, N = 247). We collected comprehensive clinicopathologic features of patients, including age, sex, tumor size, smoking status, Body Mass Index (BMI) index, family history of cancer, ki67 expression, stage, histological classification, histological grade, spread through air spaces (STAS), lymph node metastasis, vascular invasion, TMB value, and so on, to assess its correlation with genome and survival prognosis. All the tumor tissues of LUAD patients in the DPH cohort underwent an 1123 gene panel target sequencing (ChosenMed, Beijing, China).

3.2 Identification of a somatic mutation in LUAD of DPH cohort

We performed a strict filtering standard to screen truthful mutation sites. A total of 1419 single-nucleotide variants (SNVs) and 101 insertions/deletions (indels) were called (see Methods and Supplementary Table 2). The mutation landscape of the top 20 frequently mutant genes and corresponding clinical pathology information is shown in Fig. 1A. The lollipop plot demonstrated the mutation position and types of 20 genes (Supplementary Fig. 1). Based on the TMB cutoff value of ten mutation/Mb, 27 LUAD (10.93%) patients were classified in the TMB-H group. The median TMB value of the DPH cohort was 3.43 mutation/Mb, and the median TMB of smokers and non-smokers were 3.95 and 3.29 mutation/Mb, respectively (Supplementary Table 1).

Driver genes of the DPH LUAD cohort were identified using the MutSigCV method, and the assessed likely driver genes were listed in Supplementary Table 8. A total of 34 driver genes were identified with q value (p-adjusted) < 0.05 (Table 1). The top5 frequently mutated driver genes are EGFR (68%), TP53 (30.4%), RBM10 (12.6%), KRAS (8.5%), and LRP1B (8.5%) (Table 1). The pathogenic significance of 34 driver genes was searched in the Cancer Gene Census (CGC, https://cancer.sanger.ac.uk/census) database (Sondka, Bamford, Cole et al. 2018), 24 of which played a role in cancers. Among these 24 driver genes listed in the CGC, nine have ever been reported in lung cancer. Furthermore, these 24 genes have demonstrated significant involvement as oncogenes (10/34), fusion genes (9/34), and tumor suppressor genes (15/34), highlighting their crucial roles in cancer. The other ten driver genes were not reported (denoted as NR) before, including SPAT1(4%), ANKRD11(3.6%), ERCC3(2.8%), RAD50(2.8%), WRN(2.4%), BLM(2.0%), WISP3(2.0%), CDK8(1.6%), PAK3(1.6%), and WEE1(1.6%), which required additional experiment validation and functional exploration for their clinical significance.

This study defined Clinical Characteristic-Related Genes (CCRGs) as genes frequently mutated in individuals with higher clinical risk factors. These risk factors include smoking, advanced pathology stage (Stage II − IV), lower historical differential grade, solid historical subtype, age ≥ 50 years old, tumor size ≥ 2cm, ki67 expression proportion ≥ 10%, family history of cancers, BMI index ≥ 24, lymph node metastasis, as well as the presence of STAS and vascular invasion. We identified 45 genes significantly associated with higher clinical risk factors (Fig. 1B). Frequent mutations in TP53 and SPTA1 were observed in the smoking population, while KEAP1 and TAF1 mutations were associated with a higher BMI index. Patients with STAS pathology exhibited a higher frequency of TP53, MTOR, CTNNB1, BARD1, BMPR2, DDX3X, MCL1, and SUFU mutations. Similarly, patients carrying TP53, ERBB2, PDGFRB, POLD1, and PTPRT mutations were likelier to display the pathology feature of vascular invasion. TP53, LRP1B, SETD2, RAD50, ALK, CTNNB1, and BRCA2 were frequently observed in patients with larger tumors. The expression of ki67 in tumor tissues strongly correlated with mutations in TP53, LRP1B, SETD2, ALK, KMT2C, EP300, KMT2B, and PDGFRA. Lymph node metastasis was closely related to mutations in TP53, SETD2, ERCC2, FANCF, and FLI1, while patients with advanced pathology stage exhibited a higher frequency of TP53, LRP1B, CDKN2A, GRIN2A, DICER1, BRIP1, NOTCH1, AND PALB2 mutations. Furthermore, MTOR has a higher frequency in solid tumor subtypes. These findings suggest a potential association between gene mutations and various clinical pathology features, although further experimental evidence is required to confirm the underlying biological mechanisms. We also retrieved the pathogenic significance of these CCRG genes from the CGC database. Of these, 18 were not reported, and the other 27 played essential roles as oncogenes (12/27), fusion genes (9/27), and tumor suppressor genes (17/27) (Table 1).

We combined the genes of the top 20 frequently mutant genes, 34 driver genes, and 45 CCRGs as essential genes in LUAD patients from the DPH cohort. Seven of these 72 critical genes (ERBB2, KRAS, LRP1B, RB1, SETD2, SPAT1, and TP53) exhibited all three above labels (Table 1). Among the seven genes, SPAT1 was never reported as a driver gene in the CGC database, and LRP1B and SETD2 were reported as driver genes for the other cancer type.

Table 1

Significant altered genes in the DPH cohort.
Gene	Top 20 gene	driver gene	CCRG	Roles in cancer (CGC)	mut_freq
ANKRD11	Yes	Yes	No	NR	3.6%
ARID1A	Yes	Yes	No	TSG; fusion	4.5%
ARID1B	Yes	Yes	No	TSG	5.7%
ATM	Yes	Yes	No	TSG	4.5%
CREBBP	Yes	Yes	No	oncogene; TSG; fusion	3.6%
EGFR	Yes	Yes	No	oncogene(NSCLC)	68.0%
ERBB2	Yes	Yes	Yes	oncogene; fusion(NSCLC)	7.7%
FOXP1	Yes	Yes	No	oncogene; fusion	4.0%
KMT2D	Yes	No	No	oncogene; TSG	4.5%
KRAS	Yes	Yes	Yes	oncogene(lung cancer)	8.5%
LRP1B	Yes	Yes	Yes	TSG	8.5%
MED12	Yes	Yes	No	TSG	6.9%
MYC	Yes	Yes	No	oncogene; fusion	4.5%
PIK3CA	Yes	Yes	No	oncogene	5.3%
RB1	Yes	Yes	Yes	TSG(small cell lung cancer)	4.0%
RBM10	Yes	Yes	No	TSG(LUAD)	12.6%
SETD2	Yes	Yes	Yes	TSG	4.0%
SMARCA4	Yes	Yes	No	TSG(NSCLC)	4.0%
SPTA1	Yes	Yes	Yes	NR	4.0%
TP53	Yes	Yes	Yes	oncogene; TSG; fusion(lung cancer)	30.4%
AR	No	Yes	No	oncogene	3.2%
AXIN2	No	Yes	No	TSG	2.4%
BLM	No	Yes	No	NR	2.0%
BRAF	No	Yes	No	oncogene; fusion(NSCLC)	3.2%
CDK8	No	Yes	No	NR	1.6%
ERCC3	No	Yes	No	NR	2.8%
FBXW7	No	Yes	No	TSG	2.0%
NONO	No	Yes	No	fusion	1.6%
NRG1	No	Yes	No	TSG; fusion(NSCLC)	2.0%
PAK3	No	Yes	No	NR	1.6%
PTEN	No	Yes	No	TSG	1.6%
RAD50	No	Yes	Yes	NR	2.8%
WEE1	No	Yes	No	NR	1.6%
WISP3	No	Yes	No	NR	2.0%
WRN	No	Yes	No	NR	2.4%
ALK	No	No	Yes	oncogene; fusion (NSCLC)	2.4%
BARD1	No	No	Yes	TSG	0.8%
BMPR2	No	No	Yes	NR	0.8%
BRCA2	No	No	Yes	TSG	2.0%
BRIP1	No	No	Yes	NR	1.6%
CDKN2A	No	No	Yes	TSG	2.4%
CTNNB1	No	No	Yes	oncogene; fusion	2.0%
DDX3X	No	No	Yes	TSG	0.8%
DICER1	No	No	Yes	TSG	2.0%
DYNC2H1	No	No	Yes	NR	2.0%
EP300	No	No	Yes	TSG; fusion	1.6%
ERCC2	No	No	Yes	NR	1.2%
ETV1	No	No	Yes	oncogene; fusion	1.2%
FANCI	No	No	Yes	NR	0.8%
FANCF	No	No	Yes	NR	0.8%
FH	No	No	Yes	NR	0.8%
FLI1	No	No	Yes	oncogene; fusion	0.8%
GRIN2A	No	No	Yes	TSG (LUAD)	1.6%
HGF	No	No	Yes	NR	1.2%
KDR	No	No	Yes	oncogene(NSCLC)	0.8%
KEAP1	No	No	Yes	TSG(NSCLC)	1.6%
KEL	No	No	Yes	NR	1.2%
KMT2B	No	No	Yes	NR	1.6%
KMT2C	No	No	Yes	TSG	2.4%
MCL1	No	No	Yes	NR	0.8%
MTOR	No	No	Yes	oncogene	2.8%
NOTCH1	No	No	Yes	oncogene; TSG; fusion(lung SCC)	1.2%
PALB2	No	No	Yes	NR	1.2%
PDGFRA	No	No	Yes	oncogene; fusion	2.0%
PDGFRB	No	No	Yes	oncogene; fusion	1.2%
POLD1	No	No	Yes	TSG	1.2%
PTPRT	No	No	Yes	TSG(lung cancer)	1.6%
RICTOR	No	No	Yes	NR	1.6%
SMARCA2	No	No	Yes	NR	0.8%
SOX9	No	No	Yes	NR	0.8%
SUFU	No	No	Yes	TSG	0.8%
TAF1	No	No	Yes	NR	1.6%
Note: CGC: Cancer Gene Census, database for driver genes implication (https://cancer.sanger.ac.uk/census); NR: not reported; TSG: tumor suppressor gene.

3.3 Oncogenic signaling pathway alterations and mutation signature enrichment analysis regulated by somatic mutations in the DPH cohort

Genetic alterations in signaling pathways have been implicated in heterogeneous tumors' growth and progression processes. In the present study, we analyzed mutations in ten oncogenic signaling pathways, including RTK-RAS, NRF2, TP53, PI3K, WNT, Cell Cycle, HIPPO, MYC, NOTCH, TGF-Beta (Sanchez-Vega, Mina, Armenia et al. 2018), using data from 247 LUAD patients in the DPH cohort. The RTK-RAS pathway (222/247, 89.88%) and TP53 pathway (82/247, 33.20%) exhibited the highest frequency of pathway alterations (Fig. 2A). Among the RTK-RAS pathway genes, EGFR, KRAS, ERBB2, BRAF, ALK, and PDGFRA were the most frequently altered oncogenes (Fig. 2B).

Mutation signature analysis was performed using somatic mutation data of the DPH cohort. We observed four significantly enriched signatures (Fig. 2C). The most prevalent signatures were cosmic4 (smoking signature, cosine-similarity = 0.736) and cosmic15 (defective DNA mismatch repair signature, cosine-similarity = 0.664). In addition, we compared the smoking statuses in the “smoking” signature group and the rest patients, and the percentage of the smoking population was significantly higher than that of the non-smoking population (p < 0.001, Fig. 2D).

Overall, alterations in the RTK-RAS signaling pathway served as the primary oncogenic molecular driver of LUAD patients in the DPH cohort. At the same time, smoking acted as the primary clinical risk characteristic.

3.4 Development and validation of Multiple Parameter Gene Mutation (MPGM) risk model

In the present study, we investigated the significant gene alterations observed in the DPH cohort, elucidating the intricate relationship between clinical pathology and the oncogenic implications of gene mutations. Now, we aim to explore the survival prognosis roles of these gene alterations. Due to a lack of adequate follow-up survival data for the DPH cohort, we used the EAS cohort to develop the risk model. We validated the predictive performance utilizing the MSK-LUAD cohort.

Seventy-two critical genes (top 20 frequently mutant genes, 34 driver genes, and 45 CCRGs, Table 1) and ten oncogenic pathway gene alterations (at least one pathway gene mutated) in the DPH cohort are defined as multiple parameter gene mutations (MPGM) for LUAD patients, which were used to screen prognosis-related signature genes during the initial stage. Genes with mutation frequency = 0% in the EAS cohort were excluded (Supplementary Fig. 2). Finally, 56 essential genes and ten oncogenic pathway gene alterations were selected for univariate and multivariable Cox regression analysis. Finally, five MPGMs with P < 0.05 in univariate and multivariable Cox regression analysis (Table 2). The candidate MPGMs and the coefficients were BRCA2 (1.041), ALK (1.941), BRAF (1.056), EGFR (-0.536), and PDGFRA (1.142). Therefore, the MPGM risk scores were calculated using the following formula: \(\text{M}\text{P}\text{G}\text{M} \text{s}\text{c}\text{o}\text{r}\text{e}=\sum (mutationstatus\times coefficient)\). The mutationstatus was defined as follows: "0" denoted the absence of any mutations, while "1" indicated the presence of mutations. The cutoff was defined as the median value of the MPGM score in the EAS cohort.

Table 2

**Univariable and multivariable analyses of OS of MPGMs of patients in the EAS cohort.**
	Univariate Cox		Multivariate Cox
gene Name	HR (95%CI)	p-value	coef	HR	p-value
BRCA2	4.708 (2.05–10.80)	0.00026	1.041	2.83(1.12–7.19)	0.0285
ALK	6.48(1.58–26.57)	0.0094	1.941	6.966(1.67–29.16)	0.0079
BRAF	2.944(1.28–6.75)	0.01	1.056	2.875(1.23–6.74)	0.0151
EGFR	0.452(0.29–0.71)	0.00048	-0.536	0.585(0.37–0.94)	0.0254
PDGFRA	2.717(1.10–6.72)	0.03	1.142	3.132(1.23–7.95)	0.0163
KEAP1	5.415(2.79–10.49)	P < 0.0001	1.561	4.762(0.56–40.67)	0.154
NRF2 pathway	3.113 (1.65–5.86)	P = 0.0004	-0.673	0.510(0.069–3.766)	0.509
PTEN	3.874(1.69–8.91)	P = 0.0014	0.856	2.352(0.87–6.34)	0.091
PI3K pathway	1.775(1.18–2.68)	P = 0.0061	0.382	1.465(0.95–2.26)	0.08

HR: hazard ratio; CI: confidence interval.

According to the MPGM risk model, a total of 299 patients with LUAD from the EAS cohort were stratified into two groups: MPGM-High (N = 26) and MPGM-Low (N = 273) (Fig. 3A). Survival analysis revealed a significantly favorable prognosis in the MPGM-Low group (P < 0.0001, HR = 3.7, 95% CI [2.2, 6.21], median overall survival [mOS]: not reached vs. 29 months). The MPGM risk model had the best area under the curve (AUC) of 0.754 for predicting the survival of 10 years, indicating its robust performance in predicting outcomes (Fig. 3B). Interestingly, we observed a decreased AUC when integrating the MPGM risk model with two significant pathway alterations in univariate Cox regression analysis: NRF2 and PI3K pathways (Fig. 3C). This finding suggested that combining these specific pathway alterations may affect the predictive accuracy of the MPGM risk model.

Moreover, we compare the predicting performance of the MPGM risk model and the model combined with other clinical factors, including age, gender, smoking, stage, histological, and TMB. MPGM risk model combined smoking factor achieved an optimal AUC value of 0.748 (Fig. 3D). Given the satisfactory predictive performance achieved by integrating the MPGM risk model and clinical characteristics, we performed the survival analysis by combing the MPGM model and clinical factors. In the MPGM risk model, when age was combined, the MPGM-High group consistently demonstrated superior survival compared to the MPGM-Low group, regardless of whether the individuals were above or below the age of 64. When the patients of the MPGM-Low group were younger than 64 years old, they achieved the best OS (Supplementary Fig. 3A).

Additionally, this trend persisted irrespective of the TMB value being high or low (Supplementary Fig. 3D) and regardless of smoking status (Supplementary Fig. 3E). These findings highlight the independent predictive capacity of the MPGM risk model. When the MPGM model was combined with the stage (Supplementary Fig. 3B), gender (Supplementary Fig. 3C), and historical differentiation grade (Supplementary Fig. 3F), it could further facilitate a more comprehensive prognostic stratification of patients, thereby aiding clinicians in making personalized clinical decisions.

Next, we used the MSK-LUAD cohort, which provided genomic alterations and survival data, to validate the predictive performance of the MPGM risk model. Based on the median cutoff value, patients in the MSK-LUAD cohort could be stratified into MPGM-High and MPGM-Low groups (Fig. 3E). Patients in the MPGM-Low group obtained a litter more extended OS benefit (N = 233, P = 0.0055, HR = 1.56, 95% CI [1.14, 2.51], mOS = 49) than those in the MPGM-High group (N = 161, mOS = 34 months). The AUC in the validation cohort was 0.619 (Fig. 3F). Moreover, we tried to perform the validation in the DPH cohort. By the end of the latest follow-up visit on 2nd June 2023, the treatment and PFS information of 55 patients from the DPH cohort were collected for validation analysis. As shown in Fig. 3G, patients could also be divided into MPGM-High and MPGM-Low groups despite no significant PFS difference (P = 0.47).

Furthermore, we conducted a comparative analysis to elucidate the HLA variability (Fig. 4A) and the proportion of distinctive clinical characteristics between the MPGM-High and MPGM-Low groups (Fig. 4B-O). According to the results, Patients in the MPGM-High group had a significantly high proportion of low differentiation (P < 0.01, Fig. 4E), expression of ki67 > 10% (P = 0.018, Fig. 4I), and smokers (P = 0.02, Fig. 4N).

The forest plot demonstrated that age > 64 versus age ≤ 64 (HR = 2.03; 95% CI [1.08–3.8]; p = 0.027), stage III-IV versus stage I-II (HR, 3.00; 95% CI [1.61, 5.9]; p < 0.001), and MPGM-High versus MPGM-Low (p < 0.001, HR = 4.24; 95% CI [1.93, 9.3]) were independently correlated with survival prognosis in EAS cohort (Fig. 5A). Due to the independence of the MPGM risk model, we constructed a nomogram based on the multivariable Cox regression analysis, and age, histological grade, stage, smoking status, sex, and TMB were incorporated (Fig. 5B). The nomogram possessed the integrated effects of the MPGM risk model and clinical factors, which could predict survival for 1-year, 3-year, and 5-year more accurately. The Calibration curves were plotted to show the accuracy of the nomogram in predicting the 2-, 3-, and 5-year survival rates (Fig. 5C).

Moreover, a comprehensive comparison was conducted to evaluate the disparities in ten oncogenic pathway alterations and TMB levels between the MPGM-High and MPGM-Low groups in EAS and DPH cohorts. The alteration frequency of ten oncogenic pathways was higher in the MPGM-Low group than in the MPGM-High group (Fig. 6A). However, a significantly higher TMB level was observed in the MPGM-High group than in the MPGM-Low group, both in the DPH cohort (Fig. 6B) and the EAS cohort (Fig. 6C). This observation was primarily attributed to the prevalence of sporadic mutations in the MPGM-Low group. In contrast, the MPGM-High group predominantly exhibits specific carcinogenic gene alterations, which led to higher hazard risk. The detailed gene mutations in the ten oncogenic signaling pathways of EAS and DPH cohorts are plotted in Supplementary Fig. 4. To assess the therapy decision, we analyzed the targetable gene mutation site referred to the NCCN guideline (Ettinger, Wood, Aisner et al. 2023). The results showed ERBB2_20ins, KRAS_G12/13X, and BRAF_V600E mutations were more frequently mutated in the MPGM-High group, and EGFR mutations, including EGFR_L858R, EGFR_19del, EGFR_20ins, and EGFR_T790M, EGFR_S768I, EGFR_L861Q, and EGFR_G719A have occurred more in the MPGM-Low group (Fig. 6D). We conducted an RNA fusion analysis and identified oncogene fusions, including KIF5B-RET, CCDC6-RET, EZR-ROS1, EML4-ALK, and CD74-NRG1, in eight patients from the DPH cohort commonly. Notably, all these eight patients belonged to the MPGM-High group (Supplementary Table 9). The analysis of copy number variations (CNVs) in the DPH cohort revealed a notable gene amplification (Supplementary Table 10). Furthermore, we performed a Gene Ontology (GO) enrichment analysis on the five model genes (Supplementary Fig. 5), revealing a significant enrichment of the regulation of the MAPK cascade. This finding highlights the potential molecular mechanism underlying the MPGM risk model.

3.5 Prognosis stratification and immune infiltrates analysis for different LUAD subgroups using the MPGM risk model

In addition, we examined the clinical applicability of the MPGM risk model in different LUAD subgroups of the EAS cohort. First, we utilized the MPGM model to stratify early-stage LUAD patients (stage I-II, N = 191), identifying 12 individuals in the MPGM-High group who require closer follow-up (P = 0.011, HR = 2.9; 95% CI [1.22, 6.87], mOS = 91 months vs. not reached) (Fig. 7A). For advanced-stage patients (stage III-IV, N = 106), 92 individuals were classified as MPGM-Low patients using the MPGM risk model (P = 0.00015, HR = 3.34; 95% CI [1.73, 6.47], mOS = 50.9 vs. 18.1 months), which needed comprehensive assessments and avoiding overtreatment (Fig. 7B). These findings underscore the clinical advantage of the MPGM model in guiding risk-based decision-making and optimizing patient management. Subsequently, in the EAS cohort, we identified 50 patients with LUAD who received tyrosine kinase inhibitors (TKIs) treatment. Notably, one patient was classified as an MPGM-High individual using the MPGM risk model, indicating the potential of the MPGM risk model in identifying treatment hazards (Fig. 7C). Following the standard treatment guidelines, patients with EGFR mutations are eligible for corresponding TKI therapy. On the contrary, treatment options are limited for patients without EGFR mutations, and conventional chemotherapy remains the primary choice. We investigated whether the MPGM model could offer additional prognostic stratification for patients without EGFR mutations (N = 154). Based on the model, our findings revealed that 21 patients without EGFR mutations were classified into the MPGM-High group (P = 0.00012, HR = 2.99; 95% CI [1.66, 5.37]), which needed risk assessment when considering conventional chemotherapy (Fig. 7E).

Furthermore, we investigated whether the MPGM model could be used to identify further patients who are suitable or unsuitable for chemotherapy among those who had undergone chemotherapy (N = 68). Five patients with a median OS of 18.1 months were classified into the MPGM-High group, showing a higher risk when receiving chemotherapy treatment (P < 0.0001, Fig. 7D). These above findings highlighted the clinical utility of the MPGM risk model in guiding clinical decision-making regarding drug selection and potentially avoiding adverse outcomes.

Finally, we identified 169 LUAD patients with corresponding RNA-seq data in the EAS cohort. Survival analysis based on the MPGM model revealed significant differences in OS between the two groups (P = 0.0004, Fig. 7F). Therefore, we further investigated these two groups' immune microenvironment features based on the RNA-seq data of 169 LUAD patients in the EAS cohort.

The proportions of immune infiltrating cells in the MPGM-High and MPGM-Low groups were compared using several immune infiltration analysis methods, including EPIC, Quantiseq, Timer, Xcell, Estimate, Cibersort, and Mcpcounter (Fig. 8). The ESTIMATE method assessed the proportion of non-tumor components, including stromal and immune cells. The MPGM-High group exhibited a higher proportion of tumor purity and a lower ESTIMATEScore, indicating a higher degree of tumor invasiveness. Conversely, the MPGM-Low group showed significant enrichment of dendritic cells (DC) and monocytic lineage cells and a higher MicroenvironmentScore and ImmuneScore (Fig. 8), suggesting an enhanced immune anti-tumor activity.

The latest advance of phase 3 trials ADAURA study (NCT02511106) confirmed Adjuvant osimertinib could get a significant 85% 5-year OS benefit among patients with completely resected, EGFR-mutated, stage IB to IIIA NSCLC, showing the powerful therapy efficiency of targeted therapy (Tsuboi, Herbst, John et al. 2023). Recent clinical trial research on perioperative immunotherapy in early-stage non-small cell lung cancer (NSCLC), including phase 3 trials such as CheckMate 816 and KEYNOTE-671, has yielded remarkable advancements. These findings signify a paradigm shift in research focus from exploring survival benefits solely in advanced metastatic NSCLC to encompassing the early and mid-stage NSCLC patients and establishing a new perioperative treatment pattern (Forde, Spicer, Lu et al. 2022; Wakelee, Liberman, Kato et al. 2023). Despite significant advancements in clinical trials for targeted therapy and immunotherapy, leading to improved outcomes in LUAD patients, a subset of patients cannot undergo surgery or lack targetable biomarkers. Therefore, it is imperative to explore effective biomarkers to stratify patients and identify additional targetable biomarkers (Sun, Bleiberg, Hwang et al. 2023). Consequently, our objective is to identify tumor-specific informative genes based on mutation characteristics, facilitating risk stratification for patients, particularly with limited treatment options.

In this study, we retrospectively collected a moderate-size LUAD cohort and profiled the somatic altered features based on the genomic data of multiple gene sequencing panel. The alteration of the RTK-RAS pathway and smoking are identified as the main carcinogenic factors. We first explore the association between somatic tumor mutations and clinical risk factors. A total of 45 CCRGs are significantly closely related to the high-risk clinicopathological characteristics, including smoking, advanced pathology stage (Stage II − IV), lower historical differential grade, solid historical subtype, age ≥ 50 years old, tumor size ≥ 2cm, ki67 expression proportion ≥ 10%, family history of cancers, BMI index ≥ 24, lymph node metastasis, as well as the presence of STAS and vascular invasion. The TP53 mutation is significantly associated with tobacco consumption (Le Calvez, Mukeria, Hunt et al. 2005). In addition, TP53 mutations are substantially higher in patients with other high-risk clinical factors, including STAS+, high ki67 expression, larger tumor size, and lymph node metastasis. The TP53 mutation in tumors could result in the loss of tumor-suppressing function, promoting tumor proliferation and metastasis (Alvarado-Ortiz, de la Cruz-López, Becerril-Rico et al. 2020). Regarding STAS, Ye et al. conducted a study to investigate the molecular characteristics distinguishing NSCLC patients with STAS + and STAS-. Their findings revealed significant TP53 mutation and ALK fusion occurrences in STAS + NSCLC patients (Ye, Yu, Zhao et al. 2023). In our study, we also observed a correlation between STAS + and several oncogene mutations, namely MTOR and CTNNB1, as well as tumor suppressor genes, including BARD1, DDX3X, and SUFU. However, recurrent mutations of EGFR and KRAS were not found to be associated with STAS+, which is consistent with previous studies (Zeng, Wang, Li et al. 2020; Zhang, Liu, Feng et al. 2020). KEAP1 is a tumor suppressor gene in NSCLC, which is associated with a higher BMI index of LUAD patients in our study. There is no reported relationship between KEAP1 and BMI index. Still, we found KEAP1 mutation and BMI index are proven to be related to survival benefits from immunotherapy in NSCLC, which provides new insight for further study (Di Federico, De Giglio, Parisi et al. 2021; Kichenadasse, Miners, Mangoni et al. 2020). Significant associations were observed between vascular invasion and mutations in oncogenes ERBB2 and PDGFRA, as well as tumor suppressor genes POLD1 and PTPRT. PDGFRA is an angiogenesis-related gene, and the mRNA expression of PDGFRA and vascular invasion are positively related to HIF-1alpha in hepatocellular carcinoma (HCC), which could promote tumor inflammation (Dai, Gao, Qiu et al. 2009). Furthermore, mutations in tumor suppressor genes LRP1B, SETD2, and TP53 are strongly associated with larger tumor size, elevated ki67 expression, lymph node metastasis, and advanced pathological stage in patients with LUAD, implying a significant association with proliferation and metastasis of LUAD. Kadara et al. found that the SETD2 mutations were related to the progression and poor survival of patients with LUAD (Kadara, Choi, Zhang et al. 2017). LRP1B mutations significantly affect the Cell Cycle and Antigen Processing and Presentation pathways, which are crucial in tumorigenesis. Notably, NSCLC patients with LRP1B mutations exhibit high TMB values and derive incredible survival benefits from immunotherapy (Chen, Chong, Wu et al. 2019). These CCRG genes provide new insight into the molecular mechanisms underlying the occurrence of high-risk clinical factors, thus providing valuable avenues for further investigation in this area of research.

In addition to identifying gene mutations significantly associated with these clinical features, we also identified novel driver genes, including SPAT1, ANKRD11, ERCC3, RAD50, WRN, BLM, WISP3, CDK8, PAK3, and WEE1, which have not been reported in the Cancer Gene Census (CGC) database. Further experimental validation is required to elucidate their biological mechanisms in LUAD.

To assess the prognostic significance of LUAD-related genomic alterations, including CCRGs, driver genes, and gene mutations in oncogene pathways, we developed an MPGM risk model using a multivariable Cox regression algorithm in a publicly available Chinese LUAD cohort (EAS cohort). The risk model successfully stratified patients into two groups: MPGM-High and MPGM-Low. Patients in the MPGM-High group exhibited inferior OS compared to those in the MPGM-Low group, validated in an independent cohort (MSK-LUAD cohort). Subgroup analysis shows the robust predicting performance of the MPGM model to stratify the LUAD patients. The MPGM risk model could effectively distinguish LUAD patients of the EAS cohort into MPGM-High and MPGM-Low groups, irrespective of EGFR mutation status, treatment type (chemotherapy or targeted therapy), and disease stage (early or advanced). Remarkable survival differences were observed between the two risk groups, demonstrating the model's clinically solid applicability and stable prognostic performance.

Although statistical significance was not achieved in the DPH cohort, an analysis of 54 patients with complete PFS follow-up information revealed that those in the MPGM-Low group exhibited better PFS than those in the MPGM-High group. Additionally, several characteristics were observed in the MPGM-High group, including higher ki67 expression, smoking history, lower differentiation grade, high tumor mutational burden (TMB), and specific gene mutations in the RTK-RAS pathway. Significant differences were also observed in targetable gene mutation sites for targeted therapy. The MPGM-High group showed a higher prevalence mutation of ERBB2_20ins, KRAS_G12/13X, and BRAF_V600E mutations, while the MPGM-Low group had a higher occurrence of EGFR_L858R, EGFR_19del, EGFR_20ins, and EGFR_T790M, EGFR_S768I, EGFR_L861Q, and EGFR_G719A mutations. Interestingly, it was observed that all patients harboring detected oncogenic fusion genes, including RET, ROS1, ALK, and NRG1 (Ettinger, Wood, Aisner et al. 2023), were consistently classified into the MPGM-High group. These findings provide compelling evidence that patients classified as MPGH-High risk exhibit a higher risk hazard, both in clinical and molecular characteristics. Moreover, these results emphasize the model's substantial clinical utility and robust prognostic capacity in real-world settings. Due to the favorable clinical application of the MPGM risk model and improved prognosis assessment performance when combined with the clinicopathology feature, we establish a nomogram based on the Cox regression model, which is pretty valuable for the clinics to precisely predict the survival status using the clinical information at baseline and MPGM risk model according to the molecular biological characteristics, including three CCRGs (BRCA2, ALK, PDGFRA) and two driver genes (BRAF, EGFR). These five genes all play essential roles in LUAD.

To get the reason for the different survival outcomes between MPGM-High and MPGM-Low groups, we analyzed immune infiltration differences using the RNA-seq data. Patients in the MPGM-High group had higher tumor purity and a lower ESTIMATEScore, suggesting more infiltrated tumor cells and less enriched immune cells (Yoshihara, Shahmoradgoli, Martínez et al. 2013). Patients in the MPGM-Low group showed significant enrichment of dendritic cells (DC) and monocytic lineage cells, and a higher MicroenvironmentScore and ImmuneScore, demonstrating robust immune activation and anti-tumor capabilities. A previous study has confirmed that tumor-associated DC is related to increased survival outcomes in lung cancer due to an increased anti-tumor T-cell response (Broz, Binnewies, Boldajipour et al. 2014; Guilliams, Dutertre, Scott et al. 2016). The monocytic lineage cells could differentiate into macrophage cells, affecting tumor growth and survival outcomes (Ugel, Canè, De Sanctis et al. 2021). The monocyte lineage and macrophage cells demonstrate a significant interaction in LUAD, correlating with unfavorable survival outcomes and high-grade tumor subtypes (Sorin, Rezanejad, Karimi et al. 2023). But this disadvantage may be negated by heightened immune activity in the MPGM-Low group.

In summary, the significantly higher anti-tumor capability and better survival of the MPGM-Low group can mainly be attributed to the infiltration of immune cells and DCs. In contrast, the MPGM-High group exhibits extensive invasion and infiltration of tumor cells, resulting in a markedly suppressive tumor microenvironment and compromised survival prognosis. These findings reinforce the reliable and robust predictive performance of our model, highlighting its credibility and resilience.

We will apply the MPGH risk model to our clinical practice. For these 55 LUAD patients who keep a close follow-up in the DPH cohort, patients in the MPGM-High group will be paid extra attention to the disease progression and gene alterations during the subsequent follow-up. In contrast, patients in the MPGM-Low group will undergo the standard therapy method and avoid overtreatment.

There are still some limitations in our study. First, most LUAD patients are at an early stage, resulting in a loss of follow-up in the DPH cohort. Finally, only 55 patients keep regular follow-up visits. Second, the MPGM risk model doesn’t show a statistically significant difference between MPGM-High and MPGH-Low groups in the DPH cohort, mainly because the follow-up period was relatively short, and only a minority of patients experienced disease progression. However, we will persistently conduct follow-ups for this specific group of individuals.

In addition to exploring therapeutic targets for targeted therapy opportunities, it is essential to consider patient stratification based on prognosis. Beyond utilizing clinical and pathological features like TNM staging and histological subtyping, leveraging patient-specific molecular alterations is crucial for precise risk stratification of heterogeneous patients, enabling personalized diagnostic and therapeutic assessments. In this study, we comprehensively investigated the molecular characteristics of LUAD and identified several novel putative driver genes and genes associated with clinical pathological characteristics. These findings provide new insight for studying molecular biological features and drug development for patients with LUAD. By assessing the risk level of LUAD patients using the MPGM risk model, we can make informed decisions regarding treatment selection. For patients classified into the MPGM-Low group, a thorough evaluation will determine if they can be spared from excessive treatments, reducing unnecessary suffering. Conversely, the MPGM-High group LUAD patients in advanced stages may benefit from more comprehensive evaluations, allowing for the exploration of more effective treatment options, such as immunotherapy. By utilizing patient-specific genetic alteration data, we can not only identify actionable targets for targeted therapy but also predict their survival prognosis. Ultimately, this risk model has the potential to improve patient outcomes by providing personalized treatment strategies.

AUC: area under the ROC curve;

CI: Confidence interval;

HR: Hazard ratio;

ICIs: immune checkpoint inhibitors;

KEGG: Kyoto Encyclopedia of Genes and Genomes;

LUAD: Lung adenocarcinoma;

NSCLC: non-small cell lung cancers;

NGS: next-generation sequencing;

PD-1: programmed cell death protein-1;

PD-L1: programmed cell death protein-1 ligand;

TMB: tumor mutational burden;

TME: tumor microenvironment;

TCGA: The Cancer Genome Atlas;

GEO: Gene Expression Omnibus

GO: Gene Ontology

GSEA: Gene set enrichment analysis

HPA: Human Protein Atlas

KEGG: Kyoto Encyclopedia of Genes and Genomes

PCA: Principal component analysis

ROC: Receiver Operating Characteristic

Supplementary materials

Supplementary Figures and Tables.

Funding

This work was supported by grants from the Natural Science Foundation of Chongqing (cstc2020jcyj-msxmX1060), Dianjiang County Science and Technology Bureau of Chongqing (no. djkjxm2021shmskjcxyw002).

Data Availability

The analysis results data generated during this study are in Supplementary Tables.

Ethical approval

This study was approved by the ethics committee of Dianjiang People's Hospital of Chongqing (Number 20220906).

Author contributions

Conception and design of the study: TL, JL, YF, and PH. Acquisition of data: SH, DW, and JC. Data analysis: TL and YZ. Visualization: JL and SH. Review & editing: YF and PH. Original draft: TL and JL. Supervision of the study: YF and PH. All authors contributed to the article and approved the submitted version.

Competing interests

All authors declare that the research was conducted without any commercial or financial relationships that could be construed as a potential conflict of interest.

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Consent to publish

Not applicable.

Acknowledgements

Not applicable.

Alvarado-Ortiz E, de la Cruz-López KG, Becerril-Rico J, et al. (2020) Mutant p53 Gain-of-Function: Role in Cancer Development, Progression, and Therapeutic Approaches. Frontiers In Cell and Developmental Biology 8: 607670. doi: 10.3389/fcell.2020.607670
Aran D, Hu Z, Butte AJ (2017) xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biology 18: 220. doi: 10.1186/s13059-017-1349-1
Becht E, Giraldo NA, Lacroix L, et al. (2016) Estimating the population abundance of tissue-infiltrating immune and stromal cell populations using gene expression. Genome Biology 17: 218.
Broz ML, Binnewies M, Boldajipour B, et al. (2014) Dissecting the tumor myeloid compartment reveals rare activating antigen-presenting cells critical for T cell immunity. Cancer Cell 26: 638-652. doi: 10.1016/j.ccell.2014.09.007
Chen B, Khodadoust MS, Liu CL, et al. (2018) Profiling Tumor Infiltrating Immune Cells with CIBERSORT. Methods In Molecular Biology (Clifton, N.J.) 1711: 243-259. doi: 10.1007/978-1-4939-7493-1_12
Chen H, Chong W, Wu Q, et al. (2019) Association of LRP1B Mutation With Tumor Mutation Burden and Outcomes in Melanoma and Non-small Cell Lung Cancer Patients Treated With Immune Check-Point Blockades. Frontiers In Immunology 10: 1113. doi: 10.3389/fimmu.2019.01113
Chen J, Yang H, Teo ASM, et al. (2020) Genomic landscape of lung adenocarcinoma in East Asians. Nature Genetics 52: 177-186. doi: 10.1038/s41588-019-0569-6
Chen P, Liu Y, Wen Y, et al. (2022) Non-small cell lung cancer in China. Cancer Communications (London, England) 42: 937-970. doi: 10.1002/cac2.12359
Dai C-X, Gao Q, Qiu S-J, et al. (2009) Hypoxia-inducible factor-1 alpha, in association with inflammation, angiogenesis and MYC, is a critical prognostic factor in patients with HCC after surgery. BMC Cancer 9: 418. doi: 10.1186/1471-2407-9-418
Detterbeck FC, Boffa DJ, Kim AW, et al. (2017) The eighth edition lung cancer stage classification. Chest 151: 193-203.
Di Federico A, De Giglio A, Parisi C, et al. (2021) STK11/LKB1 and KEAP1 mutations in non-small cell lung cancer: Prognostic rather than predictive? European Journal of Cancer (Oxford, England : 1990) 157: 108-113. doi: 10.1016/j.ejca.2021.08.011
Ettinger DS, Wood DE, Aisner DL, et al. (2023) NCCN Guidelines® Insights: Non–Small Cell Lung Cancer, Version 2.2023: Featured Updates to the NCCN Guidelines. Journal of the National Comprehensive Cancer Network 21: 340-350.
Finotello F, Mayer C, Plattner C, et al. (2017) quanTIseq: quantifying immune contexture of human tumors. BioRxiv: 223180.
Forde PM, Spicer J, Lu S, et al. (2022) Neoadjuvant Nivolumab plus Chemotherapy in Resectable Lung Cancer. The New England Journal of Medicine 386: 1973-1985. doi: 10.1056/NEJMoa2202170
Guerreiro T, Forjaz G, Antunes L, et al. (2021) Lung cancer survival and sex-specific patterns in Portugal: A population-based analysis. Pulmonology. doi: 10.1016/j.pulmoe.2021.09.001
Guilliams M, Dutertre C-A, Scott CL, et al. (2016) Unsupervised High-Dimensional Analysis Aligns Dendritic Cells across Tissues and Species. Immunity 45: 669-684. doi: 10.1016/j.immuni.2016.08.015
Han YB, Kim H, Mino-Kenudson M, et al. (2021) Tumor spread through air spaces (STAS): prognostic significance of grading in non-small cell lung cancer. Modern Pathology : an Official Journal of the United States and Canadian Academy of Pathology, Inc 34: 549-561. doi: 10.1038/s41379-020-00709-2
Herbst RS, Morgensztern D, Boshoff C (2018) The biology and management of non-small cell lung cancer. Nature 553: 446-454. doi: 10.1038/nature25183
Kadara H, Choi M, Zhang J, et al. (2017) Whole-exome sequencing and immune profiling of early-stage lung adenocarcinoma with fully annotated clinical follow-up. Annals of Oncology : Official Journal of the European Society For Medical Oncology 28: 75-82. doi: 10.1093/annonc/mdw436
Kichenadasse G, Miners JO, Mangoni AA, et al. (2020) Association Between Body Mass Index and Overall Survival With Immune Checkpoint Inhibitor Therapy for Advanced Non-Small Cell Lung Cancer. JAMA Oncology 6: 512-518. doi: 10.1001/jamaoncol.2019.5241
Koboldt DC, Zhang Q, Larson DE, et al. (2012) VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Research 22: 568-576. doi: 10.1101/gr.129684.111
Lawrence MS, Stojanov P, Polak P, et al. (2013) Mutational heterogeneity in cancer and the search for new cancer-associated genes. Nature 499: 214-218. doi: 10.1038/nature12213
Le Calvez F, Mukeria A, Hunt JD, et al. (2005) TP53 and KRAS mutation load and types in lung cancers in relation to tobacco smoke: distinct patterns in never, former, and current smokers. Cancer Research 65: 5076-5083.
Lee GY, Chung J-H, Cho S, et al. (2022) Impact of Preoperative Diagnostic Biopsy Procedure on Spread Through Airspaces and Related Outcomes in Resected Stage I Non-Small Cell Lung Cancer. Chest 162: 1199-1212. doi: 10.1016/j.chest.2022.05.002
Li Q, Wang R, Yang Z, et al. (2022) Molecular profiling of human non-small cell lung cancer by single-cell RNA-seq. Genome Medicine 14: 87. doi: 10.1186/s13073-022-01089-9
Li T, Fan J, Wang B, et al. (2017) TIMER: A Web Server for Comprehensive Analysis of Tumor-Infiltrating Immune Cells. Cancer Research 77: e108-e110. doi: 10.1158/0008-5472.CAN-17-0307
Martínez-Jiménez F, Muiños F, Sentís I, et al. (2020) A compendium of mutational cancer driver genes. Nature Reviews. Cancer 20: 555-572. doi: 10.1038/s41568-020-0290-x
McKenna A, Hanna M, Banks E, et al. (2010) The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Research 20: 1297-1303. doi: 10.1101/gr.107524.110
Mitsudomi T, Tan D, Yang JC-H, et al. (2023) Expert Consensus Recommendations on Biomarker Testing in Metastatic and Nonmetastatic NSCLC in Asia. Journal of Thoracic Oncology : Official Publication of the International Association For the Study of Lung Cancer 18: 436-446. doi: 10.1016/j.jtho.2022.10.021
Network CGAR (2014) Comprehensive molecular profiling of lung adenocarcinoma. Nature 511: 543.
Nguyen B, Fong C, Luthra A, et al. (2022) Genomic characterization of metastatic patterns from prospective clinical sequencing of 25,000 patients. Cell 185. doi: 10.1016/j.cell.2022.01.003
Sanchez-Vega F, Mina M, Armenia J, et al. (2018) Oncogenic Signaling Pathways in The Cancer Genome Atlas. Cell 173. doi: 10.1016/j.cell.2018.03.035
Sondka Z, Bamford S, Cole CG, et al. (2018) The COSMIC Cancer Gene Census: describing genetic dysfunction across all human cancers. Nature Reviews Cancer 18: 696-705.
Sorin M, Rezanejad M, Karimi E, et al. (2023) Single-cell spatial landscapes of the lung tumour immune microenvironment. Nature 614: 548-554. doi: 10.1038/s41586-022-05672-3
Sun L, Bleiberg B, Hwang W-T, et al. (2023) Association Between Duration of Immunotherapy and Overall Survival in Advanced Non-Small Cell Lung Cancer. JAMA Oncology. doi: 10.1001/jamaoncol.2023.1891
Sung H, Ferlay J, Siegel RL, et al. (2021) Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA: a Cancer Journal For Clinicians 71: 209-249. doi: 10.3322/caac.21660
Tamborero D, Rubio-Perez C, Deu-Pons J, et al. (2018) Cancer Genome Interpreter annotates the biological and clinical relevance of tumor alterations. Genome Medicine 10: 25. doi: 10.1186/s13073-018-0531-8
Tsuboi M, Herbst RS, John T, et al. (2023) Overall Survival with Osimertinib in Resected EGFR-Mutated NSCLC. The New England Journal of Medicine. doi: 10.1056/NEJMoa2304594
Tu H, Ye Y, Huang M, et al. (2022) Smoking, smoking cessation, and survival after cancer diagnosis in 128,423 patients across cancer types. Cancer Communications (London, England) 42: 1421-1424. doi: 10.1002/cac2.12357
Ugel S, Canè S, De Sanctis F, et al. (2021) Monocytes in the Tumor Microenvironment. Annual Review of Pathology 16. doi: 10.1146/annurev-pathmechdis-012418-013058
Wakelee H, Liberman M, Kato T, et al. (2023) Perioperative Pembrolizumab for Early-Stage Non-Small-Cell Lung Cancer. The New England Journal of Medicine. doi: 10.1056/NEJMoa2302983
Ye R, Yu Y, Zhao R, et al. (2023) Comprehensive molecular characterizations of stage I-III lung adenocarcinoma with tumor spread through air spaces. Frontiers In Genetics 14: 1101443. doi: 10.3389/fgene.2023.1101443
Yeo JG, Wasser M, Kumar P, et al. (2020) The Extended Polydimensional Immunome Characterization (EPIC) web-based reference and discovery tool for cytometry data. Nature Biotechnology 38: 679-684. doi: 10.1038/s41587-020-0532-1
Yoshihara K, Shahmoradgoli M, Martínez E, et al. (2013) Inferring tumour purity and stromal and immune cell admixture from expression data. Nature Communications 4: 2612. doi: 10.1038/ncomms3612
Zeng Q, Wang B, Li J, et al. (2020) Solid Nodule Appearance as a Predictor of Tumor Spread Through Air Spaces in Patients with Lung Adenocarcinoma: A Propensity Score Matching Study. Cancer Management and Research 12: 8197-8207. doi: 10.2147/CMAR.S266750
Zhang Z, Liu Z, Feng H, et al. (2020) Predictive value of radiological features on spread through air space in stage cIA lung adenocarcinoma. Journal of Thoracic Disease 12: 6494-6504. doi: 10.21037/jtd-20-1820

No competing interests reported.

Download PDF

Journal Publication

published 01 May, 2024

Read the published version in Cancer Medicine →

Version 1

posted

You are reading this latest preprint version

Clinical Relevance of Somatic Mutations in Chinese Lung Adenocarcinoma and Their Prognostic Implications for Survival

Status:

Journal Publication

Version 1

Abstract

Figures

1. Introduction

2. Material and Methods

2.1 Data

2.2 Targeted sequencing

2.3 Data processing

2.4 Genomic alterations analysis

2.5 Development of the risk model

2.6 Immune infiltrating analysis of risk model

2.7 Statistical analysis

3. Results

3.1 Clinical characters of DPH cohort.

3.2 Identification of a somatic mutation in LUAD of DPH cohort

3.4 Development and validation of Multiple Parameter Gene Mutation (MPGM) risk model

3.5 Prognosis stratification and immune infiltrates analysis for different LUAD subgroups using the MPGM risk model

4. Discussion

Conclusions

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1