Survival prediction for patients with lung adenocarcinoma: A prognostic risk model based on gene mutations

Abstract

BACKGROUND:

Lung adenocarcinoma is the most common type of lung cancer, and it is one of the most aggressive and rapidly fatal tumor types.

OBJECTIVE:

To identify a signature mutation genes for prognostic prediction of lung adenocarcinoma.

METHODS:

Four hundred and sixty-two lung adenocarcinoma cases were screened out and downloaded from TCGA database. Mutation data of 18 targeted genes were detected by MuTect. LASSO-COX model was used to screen gene loci, and then a prognosis model was established. Afterwards, 40 clinical patients of lung adenocarcinoma were collected to verify the mutation features and the predictive function of the above prognostic model. The mutations of above 18 genes were sequenced with targeted next generation sequencing (NGS) and analyzed with GATK and MuTect.

RESULTS:

TP53 (282, 32.38%), NF1 (82, 9.41%) and EGFR (80, 9.18%) were the top 3 most frequent mutation genes. A total of 7 variables were screened out after lasso-COX analysis (tumor stage, age, diagnostic type, SMARCA4, GNAS, PTCH2, TSC2). SMARCA4, GNAS and TSC2 were a gene mutation signature to predict a poor prognosis.

CONCLUSIONS:

We established a prognostic model for lung adenocarcinoma, and further concluded that SMARCA4, GNAS and TSC2 were a gene signature which plays a prognostic role.

Keywords

Lung adenocarcinoma gene mutation signature prognosis a risk model

1. Introduction

Lung cancer is “the double most cancer” with the highest morbidity and the leading cancer-associated mortality in the worldwide. It is estimated that there are 228,150 new cases of lung cancer and 142,670 deaths by the American Cancer Society in 2019 [1]. Lung adenocarcinoma is the most common type of lung cancer, comprising around 40% of all lung cancer cases, and it is one of the most aggressive and rapidly fatal tumor types with overall survival less than 5 years [2]. Outcome of patients with lung adenocarcinoma treated with chemotherapy is usually poor, and it is improved with the development of targeted therapy and immunotherapy [3, 4]. However, the prognosis of patients with lung adenocarcinoma is still depressing with a 5-year survival of 5–16% [5]. Predicting prognosis is helpful to formulate reasonable follow-up treatment plan and reduce ineffective or over-treatment. The gene mutations have been studied as the prognosis prediction markers for lung adenocarcinoma. In Asia-Pacific region, EGFR exon 18/21 mutation is the most common type of oncogenic mutation in lung adenocarcinoma, with incidence rate of 55% [6]. KRAS and TP53 have also been shown to be a negative prognostic marker for lung adenocarcinoma [7]. PIK3CA gene mutation associated with poor prognosis of lung adenocarcinoma [8]. Nevertheless, most gene mutations have a more clear predictive effect on target therapy, and have a poor predictive effect, or sometimes wrong, on chemotherapy, radiotherapy and various combination therapies [9, 10]. There is still a lack of effective prognostic judgement and evaluation index of chemotherapy efficacy in clinic. Additionally, due to individual differences and the cost of gene testing, mutation genes that are really useful for clinical use are very limited. Therefore, it is necessary and meaningful to find more oncogenic mutations and their role in predicting the prognosis.

In this article, the prognostic correlation of 18 genes related to lung cancer was analyzed, namely, TP53, EGFR, SMARCA4, HGF, ERBB4, PDGFRB, NF1, GNAS, C11orf30, BRCA2, PTCH2, DDR2, MAPK3, TSC2, FLT4, APC, SMO and ABL1. The aim was to identify a signature mutation genes for prognostic prediction of lung adenocarcinoma on the basis of further understanding of its molecular genetic characteristics.

2. Materials and methods

2.1 TCGA data screening

Tissue mutation data were downloaded from The Cancer Genome Atlas (TCGA), and complete clinical data sets are available at the TCGA website (https://tcga-data.nci.nih.gov/docs/publications/tcga/). The screening criteria: a. The disease is lung cancer and the pathological type is adenocarcinoma; b. The samples were tissues, including cancer tissues and adjacent tissues; c. Mutation data with all target gene regions were available (18 targeted genes: TP53, EGFR, SMARCA4, HGF, ERBB4, PDGFRB, NF1, GNAS, C11orf30, BRCA2, PTCH2, DDR2, MAPK3, TSC2, FLT4, APC, SMO and ABL1.); d. Patients had corresponding clinical information such as age, sex, stage, location of tumors and survival time. Thus, a total of 462 lung adenocarcinoma cases were screened out.

2.2 Patients and samples

Forty patients diagnosed as lung adenocarcinoma were collected from February 2013 to April 2019 in Tianjin Chest Hospital. The tumor tissues were sequenced with targeted next generation sequencing (NGS) of 18 lung cancer gene panel, and the paracancerous tissues or leukocyte were as controls.

2.3 Sequencing (targeted NGS)

Fresh tissues samples were used for DNA extraction with Genomic DNA extraction kit (Qiagen, Valencia, CA) strictly following the manufacturer’ s instructions. The DNA samples were purified by Agencourt AMPure XP beads (Agentcourt Biosciences, Beverly, MA). Afterwards, the adaptor library was amplified and linked, and the total library was accurately quantified by Qubit DNA HS Assay Kit (Invitrogen, USA, CA). A library hybridization kit, SeqCap EZ MedExome Enrichment kits (Roche, Basel, CH), was used to capture target sequences and bead capture and elution hybridization libraries with Roche’s customized 18 targeted gene probes (Roche, Basel, CH). After amplifying the captured library by PCR, the constructed library was sequenced by Illumina MiniSeq sequencer. The average sequencing depth of tissue samples was 500 X. It could detect mutations with very low frequency to 0.1%.

2.4 Mutation analysis

Using SOAPnuke to filter the sequencing data, removing the sequence containing adapter and the low quality data, and the quality of the original data after filtering meets Q30 85%. BWA was used to compare data to the Human Reference Genome (hg19.fa); Genome Analysis Toolkit (GATK) version 2.3.9 was adapted to re-compare the reads in the interval, calibrate and rearrange the alkali matrix quality values, and to count the sequence depth and coverage. Insertion/deletion (Indel) and single nucleotide polymorphism (SNP) were detected by MuTect version 2.0. SNP and Indel with more than 1% allele mutation frequencies were filtered out. Furthermore, the functions of mutations were annotated with PolyPhen version 2, a software to predict the possible impact of an amino acid substitution on the structure and function of a human protein. The higher the score, the greater the hazard. The classification criteria: probably damaging 0.90–1.0, possibly damaging 0.45–0.90, benign 0–0.45.

Table 1
Clinical characteristics of TCGA cases

Indicators	Type	N	Percentage
Gender	Female	257	55.60%
	Male	205	44.40%
Race	American Indian or Alaska Native	1	0.20%
	Asian	7	2.00%
	Black or African American	48	10%
	White	367	79%
	Not reported	39	8.40%
Diagnosis	Adenocarcinoma	280	60.60%
	Adenocarcinoma with mixed subtypes	99	21.40%
	Acinar cell carcinoma	22	4.80%
	Papillary adenocarcinoma	20	4.30%
	Bronchiolo-alveolar carcinoma, non-mucinous	14	3.00%
	Mucinous adenocarcinoma	11	2.40%
	Solid carcinoma	5	1.10%
	Bronchio-alveolar carcinoma, mucinous	4	1%
	Bronchiolo-alveolar adenocarcinoma	3	0.60%
	Micropapillary carcinoma	2	0.40%
	Clear cell adenocarcinoma	1	0.20%
	Signet ring cell carcinoma	1	0.20%
Tumor stage	I	257	55.60%
	II	106	22.90%
	III	76	16.50%
	IV	23	5%
Location of onset	Upper lobe, lung	279	60%
	Lower lobe, lung	158	34%
	Middle lobe, lung	19	4.10%
	Overlapping lesion of lung	4	0.90%
	Main bronchus	2	0.40%

TCGA, The Cancer Genome Atlas.

2.5 Statistical analysis

The statistical analysis was performed using IBM SPSS statistics 21 and Rstudio R3.4.4 software. The Chi-square test was used to determine the correlation between gene mutations and clinical indicators, and the fisher test was used when the sample size was too small. Establishing a prognostic model using the Lasso model and COX model, and the proportional hazards (PH) assumption in the COX model is assumed to use the Schoenfeld residual test. Besides, the prognosis model was evaluated and validated by K-M survival analysis, ROC curve and clinical validation data.

3. Results

3.1 Clinical characteristics

The clinical characteristics of 462 cases were listed in Table 1. In the cohort, 55.6% were female 44.4% male, and 79% were white race, and 78.5% were I–II stage.

3.2 Mutation features and Clinical indicator correlation

In the 462 cases, a total of 871 mutations were screened out, namely, TP53 (282, 32.38%), NF1 (82, 9.41%), EGFR (80, 9.18%), ERBB4 (63, 7.23%), HGF (58, 6.66%), SMARCA4 (50, 5.74%), BRCA2 (39, 4.48%), APC (35, 4.02%), FLT4 (30, 3.44%), DDR2 (27, 3.10%), GNAS (25, 2.87%), C11orf30 (21, 2.41%), PDGFRB (19, 2.18%), ABL1 (17, 1.95%), TSC2 (16, 1.84%), SMO (15, 1.72%), MAPK3 (7, 0.80%) and PTCH2 (5, 0.57%). The mutation features of them were showed in Fig. 1A–C. Among them, 768 were SNPs (88.2%), 84 deletions (9.6%), and 19 inserts (2.2%) (Fig. 1A); In the 768 SNPs, 24.74% were C-A ( $n=$ 190), and 17.32% were G-T ( $n=$ 133) (Fig. 1C). Furthermore, the location and possible functional significance of mutations were exhibited in Fig. 1B. Most mutations occurred in the coding region.

Figure 1.

The mutation features of targeted genes. A. The overall catastrophe type distribution; B. The location and possible functional significance of mutations, None indicated the functional significance of mutations were unknown or unclear in Polyphen; C. The mutation forms of bases of SNPs.

Figure 2.

The prognostic model. A. COX variable selection cyclic voltammograms based on LASSO method; B. The schoenfeld Residual graph of the 6 screened variables, When the $p$ value $>$ 0.05, the variable passed the PH hypothesis test, and did not change with time; the global schoenfeld test $p$ value $=$ 0.3792; C. The kaplan-meier curve of overall survival; D. The ROC of 3- and 5-year overall survival.

Figure 3.

The verification results in clinical cases. A. Comparison of the mutation frequencies of all 18 targeted genes between TCGA and clinical cases; B. Comparison of the mutation frequencies of 18 targeted genes between TCGA and clinical cases in subgroups of stage I–II and stage III–IV; C. The number of mutations in each gene in 40 clinical cases, the size of dots indicates the number of mutations, and the color of dots represents different genes; D. The overall survival curves of low-risk and high-risk groups based on above prognosis model.

As for the possible impact on the structure and function of a human protein, 121 were benign, 62 possibly damaging, and 316 probably damaging. Additionally, all of them, with relatively clear functions, were missense mutations.

3.3 Establishment of the prognostic model

We used the lasso model to select meaningful variables. The lasso model introduced 23 variables, including gender, tumor stage, tumor location, age, diagnosis type, and 18 gene mutation information. After cross-validation, the best lambda value, lambdamin $=$ 0.03335, was selected. The relationship between lambda and residual is exhibited in Fig. 2A. The residual variables were tumor stage, age, diagnostic type, SMARCA4, GNAS, PTCH2, TSC2, a total of 7 variables. When COX regression was established for the 7 variables, the number of PTCH2 gene mutation was small, leading to regression non-convergence, so this variable was excluded. Schoenfeld residual test chart was used to test whether the variables in the model conform to the proportional hazard (PH) assumption. As shown in the Fig. 2B, the $p$ value of the 6 variables was greater than 0.05, and all of which passed the PH hypothesis test, that was, all variables in the equation did not change with time. The prognostic risk score of each group was calculated by multivariate COX regression analysis. Through COX regression analysis, the risk index formula was obtained as follows: The Risk Index $=$ 0.561505 (Tumor Stage) $-$ 0.600823 $\times$ (Diagnostic Type) $+$ 0.021123 $\times$ (Year) $+$ 0.310566 $\times$ (SMARCA4) $+$ 0.795032 $\times$ (GNAS) $+$ 0.411406 $\times$ (TSC2). Stage: indicated I $=$ 1, stage II $=$ 2, stage III $=$ 3, stage IV $=$ 4; Diagnostic type: Adenocarcinoma $=$ 1, Adenocarcinoma with mixed subtypes $=$ 1, Other types $=$ 2; SMARCA4, GNAS, TSC2: mutation $=$ 1, no mutation $=$ 0; Age: Age of diagnosis. According to the risk index, the patients were divided into two groups by median (median $=$ 1.5794), the high-risk group and the low-risk group. The log rank test revealed that there was a significant difference in the survival time between the two groups; the low-risk group had a higher survival rate than the high-risk group (Fig. 2C), as well as a lower risk of death ( $p<$ 0.0001). At the same time, we made the receiver operating characteristic (ROC) curves of the 3-year and 5-year overall survival (OS) rates. The results are as follows: the area under curve (AUC) of the 3-year OS rate was 0.7205, and the AUC of the 5-year OS rate was 0.7028 (Fig. 2D).

3.4 Clinical verification

Forty clinical patients of lung adenocarcinoma were collected to verify the mutation features and the predictive function of the above prognostic model with age of 63.75 $\pm$ 10.5969. Among them, 10 cases were I–II stages and 30 cases III–IV stages; 14 cases underwent surgery, 2 cases underwent surgery and chemotherapy, 2 cases treated with surgery combined with chemotherapy and targeted therapy, 10 cases received chemotherapy, 2 cases received chemotherapy and Bevacizumab, 2 cases underwent chemotherapy and radiotherapy, 2 cases were not treated, and 6 cases did not record therapeutic method. In the 40 cases, a total of 285 mutations were detected in the 18 targeted genes. It was obvious from Fig. 3A that the mutation frequencies of all 18 targeted genes were higher in clinical cases than in TCGA cases. Since the therapeutic strategies were different from each stage in the clinical setting, we have further analyzed the mutation frequencies of TCGA cases and clinical cases in subgroup of stage I–II and stage III–IV. Besides TP53 in both two subgroups and HGF in stage I–II subgroup, the mutation frequencies of other genes in clinical cases were higher compared with TCGA cases (Fig. 3B). We speculated that the mutation frequencies might be related to the race, sample size, stage and individual differences between the two groups. Simultaneously, we summarized the number of mutations in each gene in clinical data as exhibited in the Fig. 3C. Among them, 16 EGFR mutations were found in No. 33 patient (P33). Clinical cases were with more advanced cases than those in TCGA, and fewer cases were lower than the median TCGA risk index of 1.5794. So we used the median (3.0712) calculated from clinical cases as the critical value, and divided them into high-risk group and low-risk group based on the risk index formula. Afterwards, the survival curve revealed that the survival time of low-risk group was higher than that of high-risk group, and there was significant difference through the log rank test ( $p=$ 0.033) (Fig. 3D).

4. Discussion

The combined application of multiple marker genes can improve the accuracy of lung cancer screening and diagnosis. The combined application of multiple marker genes and multiple targets can reduce the probability of false positive or false negative, thereby improving the diagnostic efficiency of gene detection [11]. The current study used multiple mutation genes to predict the prognosis of patients with lung adenocarcinoma. LASSO was first proposed by Robert Tibshirani in 1996, which added a Lp norm as a penalty constraint in the calculation of really simple syndication (RSS) minimization [12]. When the lambda component is large, the Lp norm can reduce the coefficients of some parameters with smaller coefficients to 0. By choosing the absolute value of compression coefficient and adjusting the parameter lambda, LASSO can realize the continuous stable process of variable selection and estimation at the same time, so it is widely used. The LASSO method can remove some meaningless or insignificant variable coefficients by adding a constraint condition to the absolute value of the coefficients, so that a better model can be obtained. COX proportional risk regression model is a model with survival outcome and survival time as dependent variables, which is mostly used in the correlation analysis of patient survival data. Combining the characteristics of the two models, we analyzed the prognostic risk of 462 TCGA patients by LASSO-COX model. Subsequently, SMARCA4, GNAS and TSC2 were excavated to risk factors with poor outcome (Fig. 2). In addition, it was further validation in 40 clinical cases (Fig. 3D).

The encoded protein of SMARCA4 is part of the large ATP-dependent chromatin remodeling complex SNF/SWI, which is required for transcriptional activation of genes normally repressed by chromatin [13]. In addition, this protein can bind BRCA1, as well as regulate the expression of the tumorigenic protein CD44 [14]. Some studies have indicated that the high expression of SMARCA4 was associated with the dismal prognosis of lung adenocarcinomas [15, 16, 17]. The present study demonstrated from the risk index formula that the mutation of SMARCA4 was a poor outcome risk for lung adenocarcinomas, which was similar with previous studies. We have not studied the specific mechanism of action, and the relevant report holds that BRG1/SMARCA4 inactivation promotes NSCLC aggressiveness by altering chromatin organization [18]. GNAS, also known as GNAS1, mainly encodes a stimulatory alpha subunit of guanine nucleotide binding protein (G protein), which activates adenylate cyclase in G protein-coupled receptor signaling pathway, leading to an increase in the level of cAMP, and then participates in the regulation of cell growth and cell division [19]. At present, GNAS1 point mutation T393C is widely studied, which is a predictor of recurrence and survival of resectable NSCLC, particularly the advanced NSCLC, and a predictor of chemosensitivity and survival rate of advanced NSCLC patients treated with gemcitabine combined with platinum [20, 21, 22]. Our study found that GANS mutation is a poor prognostic factor in lung adenocarcinoma patients, consistent with previous studies. TSC2 associates with hamartin in a cytosolic complex, possibly acting as a chaperone for hamartin, and the protein Alternative splicing results in multiple transcript variants encoding different isoforms [23, 24]. Mutations of TSC2 lead to tuberous sclerosis complex and its gene product is believed to be a tumor suppressor and is able to stimulate specific GTPases [25, 26]. There are few studies on TSC2 in the prognosis of lung adenocarcinoma. Our results suggested for the first time that TSC2 mutation might be a poor prognostic factor of lung adenocarcinoma. However, this result needs further confirmation. Concurrently, we speculated that SMARCA4, GNAS and TSC2 were a gene mutation signature to predict a poor prognosis. In the future, we would further study the accuracy and specificity of the prognosis of three genome combinations and one or two genes alone, and further explore the relevant molecular mechanisms.

In summary, we established a prognostic model for lung adenocarcinoma, and further concluded that SMARCA4, GNAS and TSC2 were a gene signature which plays a prognostic role.

Footnotes

Acknowledgments

This work was supported by the 2011 Key Project of Medical Science Research of Health Commision of Hebei Province (No. 20110444).

Conflict of interest

The authors declare no conflicts of interest.

References

Siegel

R.L.

Miller

K.D.

and Jemal

, Cancer statistics, 2019, CA: A Cancer Journal for Clinicians 69 (2019), 7–34.

Denisenko

T.V.

Budkevich

I.N.

and Zhivotovsky

, Cell death-based treatment of lung adenocarcinoma, Cell Death & Disease 9 (2018), 117.

Huang

C.Y.

Chen

B.H.

Chou

W.C.

Yang

C.T.

and Chang

J.W.

, Factors associated with the prognosis and long-term survival of patients with metastatic lung adenocarcinoma: a retrospective analysis, Journal of Thoracic Disease 10 (2018), 2070–2078.

Aguiar

P.N.

Mello

R.A.D.

Hall

Tadokoro

and Lima Lopes

, PD-L1 expression as a predictive biomarker in advanced non-small-cell lung cancer: Updated survival data, Immunotherapy 9 (2017), 499–506.

Ruiz-Patiño

Castro

C.D.

Ricaurte

L.M.

Cardona

A.F.

Rojas

Zatarain-Barrón

Z.L.

Wills

Reguart

Carranza

Vargas

Otero

Corrales

Martín

Archila

Rodriguez

Avila

Bravo

Pino

L.E.

Rosell

Arrieta

and Latin-American Consortium for the Investigation of Lung Cancer (CLICaP), EGFR amplification and sensitizing mutations correlate with survival in lung adenocarcinoma patients treated with erlotinib (MutP-CLICaP), Targeted Oncology 13 (2018), 621–629.

Hsu

K.H.

C.C.

Hsia

T.C.

Tseng

J.S.

K.Y.

M.F.

Chiu

K.L.

Yang

T.Y.

Chen

K.C.

Ooi

T.C.

Chen

H.J.

Chen

H.Y.

Chang

C.S.

Hsu

C.P.

Hsia

J.Y.

Chuang

C.Y.

Lin

C.H.

Chen

J.J.

Chen

K.Y.

Liao

W.Y.

Shih

J.Y.

S.L.

C.J.

Yang

P.C.

and Chang

G.C.

, Identification of five driver gene mutations in patients with treatment-na? Lung adenocarcinoma in Taiwan, PLOS ONE 10 (2015), e0120852.

Fukuyama

Mitsudomi

Sugio

Ishida

Akazawa

and Sugimachi

, K-ras and p53 mutations are an independent unfavourable prognostic indicator in patients with non-small-cell lung cancer, Br J Cancer 78 (1997), 1125–1130.

Zhang

Shi

Zhao

Wang

and Yue

, PIK3CA gene mutation associated with poor prognosis of lung adenocarcinoma, Oncotargets & Therapy 6 (2013), 497–502.

Silini

E.M.

Bosi

Solcia

Volpato

Romano

Nazari

Tinelli

Ranzani

G.N.

Solcia

and Fiocca

, K-rasgene mutations: An unfavorable prognostic marker in stage I lung adenocarcinoma, Virchows Archiv An International Journal of Patholog 424 (1994), 367–373.

10.

M.L.

Park

J.Y.

and Kim

D.S.

, Wif1hypermethylation as unfavorable prognosis of non-small cell lung cancers with EGFR mutation, Molecules & Cell 36 (2013), 69–73.

11.

G.D.C.

Domínguez

J.F.

Bolton

R.D.

Fernández Pérez

Cabeza Martínez

García García-Esquinas

and Carreras Delgado

J.L.

, PET-CT in presurgical lymph node staging in non-small cell lung cancer: the importance of false-negative and false-positive findings, Radiologia 59 (2017), 147.

12.

Tibshirani

, Regression shrinkage and selection via the lasso: A retrospective, Journal of the Royal Statistical Society 73 (2011), 267–288.

13.

Lissanu

D.Y.

Yuting

Christopher

Khan

Martinez-Ledesma

Gay

Gao

Mullinax

R.A.

Khor

Feng

Lin

Y.H.

C.C.

Reyes

Peng

Robinson

Inoue

Kochat

Liu

C.G.

Asara

J.M.

Moran

Muller

Wang

Fang

Papadimitrakopoulou

Wistuba

I.I.

Rai

Marszalek

and Futreal

P.A.

, Author correction: Mutations in the SWI/SNF complex induce a targetable dependence on oxidative phosphorylation in lung cancer, Nature Medicine 24 (2018), 1047–1057.

14.

Hoffman

J.A.

Trotter

K.W.

Ward

J.M.

and Archer

T.K.

, BRG1 governs glucocorticoid receptor interactions with chromatin and pioneer factors across the genome, Elife 7 (2018), e35073.

15.

Wilkerson

M.D.

Yin

Walter

Zhao

Cabanski

C.R.

Hayward

M.C.

Miller

C.R.

Socinski

M.A.

Parsons

A.M.

Thorne

L.B.

Haithcock

B.E.

Veeramachaneni

N.K.

Funkhouser

W.K.

Randell

S.H.

Bernard

P.S.

Perou

C.M.

and Hayes

D.N.

, Differential pathogenesis of lung adenocarcinoma subtypes involving sequence mutations, copy number, chromosomal instability, and methylation, PLoS One 7 (2012), e36530.

16.

Tomida

Takeuchi

Shimada

Arima

Matsuo

Mitsudomi

Yatabe

and Takahashi

, Relapse-related molecular signature in lung adenocarcinomas identifies patients with dismal prognosis, J Clin Oncol 27 (2009), 2793–2799.

17.

Raponi

Zhang

Chen

Lee

Taylor

J.M.

Macdonald

Thomas

Moskaluk

Wang

and Beer

D.G.

, Gene expression signatures for predicting prognosis of squamous cell and adenocarcinomas of the lung, Cancer Res 66 (2006), 7466–7472.

18.

Tess

Austin

Vonn

Song

Simon

Parker

Wilskerson

M.D.

Desai

Major

M.B.

Hayes

D.N.

Davis

I.J.

and Weissman

, BRG1/SMARCA4 inactivation promotes non-small cell lung cancer aggressiveness by altering chromatin organization, Cancer Research 74 (2014), 6486–6498.

19.

Podyma

Sun

Wilson

E.A.

Carlson

Pritikin

Gavrilova

Weinstein

L.S.

and Chen

, The stimulatory G protein Gsα is required in melanocortin 4 receptor-expressing cells for normal energy balance, thermogenesis and glucose metabolism, Journal of Biological Chemistry 293 (2018), 10993–11005.

20.

Uzunoglu

F.G.

Heumann

Musici

Kutup

Koenig

Roch

Thomssen

Dohrmann

Tsui

T.Y.

Mann

Izbicki

J.R.

and Vashist

Y.K.

, The T393C polymorphism of GNAS1 is a predictor for relapse and survival in resectable non-small cell lung cancer, Lung Cancer 79 (2013), 151–155.

21.

Gong

H.Y.

W.G.

Wang

X.L.

Zhu

and Song

Q.B.

, TT genotype of GNAS1 T393C polymorphism predicts better outcome of advanced non-small cell lung cancer patients, World Journal of Gastrointestinal Oncology 6 (2014), 444–449.

22.

Xie

F.J.

Zhao

Kou

J.Y.

Hong

Gao

and Zhang

Y.P.

, The T393C polymorphism of GNAS1 as a predictor for chemotherapy sensitivity and survival in advanced non-small-cell lung cancer patients treated with gemcitabine plus platinum, Cancer Chemotherapy & Pharmacology 69 (2012), 1443–1448.

23.

Renzo

Samuela

Federica

P.M.

Sartori

Nosadini

Labriola

Canevini

Vignoli

and Toldo

, Genetic and imaging features of cerebellar abnormalities in tuberous sclerosis complex: more insights into their pathogenesis, Developmental Medicine & Child Neurology 60 (2018), 724–725.

24.

Pal

Xiong

and Sardiello

, Abnormal glycogen storage in tuberous sclerosis complex caused by impairment of mTORC1 – dependent and – independent signaling pathways, Proc Natl Acad Sci U S A 116 (2019), 2977–2986.

25.

Wang

Xiong

Chopra

and Wu

, A novel TSC2 missense variant associated with a variable phenotype of tuberous sclerosis complex: Case report of a Chinese family, BMC Med Genet 19 (2018), 90.

26.

Alsidawi

and Kasi

P.M.

, Exceptional response to everolimus in a novel tuberous sclerosis complex-2 mutation-associated metastatic renal-cell carcinoma, Molecular Case Studies 4 (2018), a002220.

Survival prediction for patients with lung adenocarcinoma: A prognostic risk model based on gene mutations

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Materials and methods

2.1 TCGA data screening

2.2 Patients and samples

2.3 Sequencing (targeted NGS)

2.4 Mutation analysis

Table 1 Clinical characteristics of TCGA cases

3. Results

3.1 Clinical characteristics

3.2 Mutation features and Clinical indicator correlation

3.4 Clinical verification

4. Discussion

Footnotes

Acknowledgments

Conflict of interest

References

Table 1
Clinical characteristics of TCGA cases