Table S1. Comparison of the characteristics between the PLCO trial and HLCS study

Table S2. List of 220 selected genes in the endosome-related gene-set

Table S3. Associations of the first 10 principal components and overall survival of NSCLC in the PLCO trial

Table S4. Stratified analyses for associations between the unfavorable genotypes and 10-year survival of NSCLC in the PLCO trial^a

Table S5. Function prediction of the three SNPs and other SNPs in high LD (r² > =0.8)

Figure S1. The distribution of the imputation information score of the present study.

Figure S2. Manhattan plot of associations between genotype data and overall survival of patients with NSCLC from the discovery and validation datasets in the PLCO trial and HLCS study, respectively. (a) The statistical values across the autosomes of associations between 44,112 SNPs and overall survival in PLCO are plotted as −log10 p values. (b) The statistical values across the autosomes of associations between 821 SNPs and overall survival in HLCS are plotted as −log10 p values. The blue horizontal line indicates p = 0.050 and the red line indicates BFDP = 0.80. Abbreviations: NSCLC, nonsmall cell lung cancer; SNPs, single nucleotide polymorphisms; PLCO, Prostate, Lung, Colorectal and Ovarian Cancer Screening trial; OS, overall survival; BFDP, Bayesian false discovery probability.

Figure S3. Regional association plots for the three independent SNPs in the endosome pathway genes in the 1000 Genomes Project. Single nucleotide polymorphisms (SNPs) in the region of 50kb up or downstream of (a) KIF16B rs1555195 C>T, (b) NEDD4L rs11660748 A>G and (c) NEDD4L rs73440898 A>G. Data points are colored according to the level of linkage disequilibrium of each pair of SNPs based on the hg19/1000 Genomes European population. The left-hand y-axis shows the association p-values of individual SNPs in the discovery dataset, which is plotted as −log10 (p) against chromosomal base-pair position. The right-hand y-axis shows the recombination rate estimated from HapMap Data Rel 22/phase II European population.

Figure S4. Five-year NSCLC survival prediction of the combined three SNPs by ROC curve in the PLCO dataset. (a) Time-dependent AUC estimation based on age, sex, smoking status, histology, tumor stage, chemotherapy, surgery, principal component and the risk genotypes of the four genes. (b) Five-year NSCLC OS prediction by ROC curve, (c) five-year NSCLC DSS prediction by ROC curve. Abbreviations: NSCLC, nonsmall cell lung cancer; SNPs, single nucleotide polymorphism; ROC, receiver operating characteristic curve; AUC, area under the curve; PLCO, Prostate, Lung, Colorectal and Ovarian Cancer Screening trial; OS, overall survival; DSS, disease-specific survival;

Figure S5. Correlations of rs1555195, rs11660748 and rs73440898 genotypes with their corresponding mRNA expression levels. The eQTL for KIF16B rs1555195 (a), NEDD4L rs11660748 ( b) and rs73440898 (c) in 373 Europeans from the 1000 Genomes Project. Abbreviation: eQTL, expression quantitative trait loci.

Figure S6. Correlations of rs11660748 and rs73440898 genotypes with their corresponding mRNA expression levels. The eQTL for NEDD4L rs11660748 in normal lung tissue (a) and in whole blood; (b) NEDD4L rs73440898 in normal lung tissue; (c) and in whole blood; (d) in GTEx Project. Abbreviation: eQTL, expression quantitative trait loci.

Figure S7. Expanded view of the ENCODE data for the rs1555195, rs11660748 and rs73440898. The H3K27Ac, H3K4Me1 and H3K4Me3 tracks show the genome-wide levels of enrichment of acetylation of lysine 27, the mono-methylation of lysine 4, and tri-methylation of lysine 4 of the H3 histone protein, as determined by the ChIP-seq assays. These levels are thought to be associated with the promoter and enhancer regions. DNase clusters track showed Dnase hypersensitivity areas. Tnx factor track show regions of transcription factor binding of DNA, as assayed by ChIP-seq experiments. Transcription shows target genes of transcription factors from transcription factor binding site profiles. Abbreviation: ChIP, chromatin immunoprecipitation sequencing.

Figure S8. Differential mRNA expression analysis and overall survival analysis of the KIF16B gene in the TCGA database. (a) The expression levels of KIF16B were not different in the 109 paired NSCLC tissues; (b) Higher expression levels of KIF16B were found in the normal tissue of the 58 paired LUAD tissues, compared to the tumor tissue; and (c) lower expression levels of KIF16B were found in the normal tissue of the 51 paired LUSC, compared to the tumor tissues. Higher KIF16B expression was associated with a better overall survival; (d) in NSCLC; (e) in LUAD patients, while with a worse overall survival; and (f) in LUSC patients. Abbreviations: TCGA, The Cancer Genome Atlas; NSCLC, nonsmall cell lung cancer; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma.

Figure S9. Differential mRNA expression analysis and overall survival analysis of the NEDD4L gene in the TCGA database. Higher expression levels of NEDD4L were found (a) in the 109 paired NSCLC tissues; (b) in the 58 paired LUAD tissues; and (c) in the 51 paired LUSC tissues. Higher expression levels of NEDD4L was associated with a better overall survival in (d) NSCLC and (e) LUAD patients, while with a worse survival (f) in LUSC patients. Abbreviations: TCGA, The Cancer Genome Atlas; NSCLC, nonsmall cell lung cancer; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma.