Patient characteristics of training and validation set. Patient selection and inclusion overview can be found in Supp Data 9. Training set (103)

Similar documents
Supplementary. properties of. network types. randomly sampled. subsets (75%

Journal: Nature Methods

Cumulative metformin use and its impact on survival in gastric cancer patients after INDEX

Nature Medicine: doi: /nm.3967

Ratio of maximum standardized uptake value to primary tumor size is a prognostic factor in patients with advanced non-small cell lung cancer

Nature Immunology: doi: /ni Supplementary Figure 1. RNA-Seq analysis of CD8 + TILs and N-TILs.

SUPPLEMENTARY MATERIAL

Edge attributes. Node attributes

The 16th KJC Bioinformatics Symposium Integrative analysis identifies potential DNA methylation biomarkers for pan-cancer diagnosis and prognosis

Temporal Trends in Demographics and Overall Survival of Non Small-Cell Lung Cancer Patients at Moffitt Cancer Center From 1986 to 2008

Gene expression profiling predicts clinical outcome of prostate cancer. Gennadi V. Glinsky, Anna B. Glinskii, Andrew J. Stephenson, Robert M.

Gene expression correlates of clinical prostate cancer behavior

Supplementary Figure S1 Expression of mir-181b in EOC (A) Kaplan-Meier

The Current Status of Immune Checkpoint Inhibitors: Arvin Yang, MD PhD Oncology Global Clinical Research Bristol-Myers Squibb

Evaluation of AJCC, UICC, and Brigham and Women's Hospital Tumor Staging for Cutaneous Squamous Cell Carcinoma

Supplementary webappendix

Expert-guided Visual Exploration (EVE) for patient stratification. Hamid Bolouri, Lue-Ping Zhao, Eric C. Holland

GSK VACCINES: BUILDING A THERAPEUTIC PORTFOLIO

GENE EXPRESSION ANALYSIS OF PROSTATE TUMORS FROM AFRICAN-AMERICAN AND EUROPEAN-AMERICAN MEN

EGFR inhibitors in NSCLC

2.28E E AKT/CAT P-3-4:

Supplementary information for: Community detection for networks with. unipartite and bipartite structure. Chang Chang 1, 2, Chao Tang 2

Locoregional treatment Session Oral Abstract Presentation Saulo Brito Silva

Colon cancer subtypes from gene expression data

Correlation of pretreatment surgical staging and PET SUV(max) with outcomes in NSCLC. Giancarlo Moscol, MD PGY-5 Hematology-Oncology UTSW

Supplemental Figure S1. RANK expression on human lung cancer cells.

T. R. Golub, D. K. Slonim & Others 1999

a) List of KMTs targeted in the shrna screen. The official symbol, KMT designation,

Antiangiogenic Agents in NSCLC Where are we? Which biomarkers? VEGF Is the Only Angiogenic Factor Present Throughout the Tumor Life Cycle

Supplementary Materials

Salvage prostatectomy for post-radiation adenocarcinoma with treatment effect: Pathological and oncological outcomes

ARTICLE RESEARCH. Macmillan Publishers Limited. All rights reserved

Geisinger Clinic Annual Progress Report: 2011 Nonformula Grant

Nature Genetics: doi: /ng Supplementary Figure 1. Phenotypic characterization of MES- and ADRN-type cells.

Survival Analysis in Clinical Trials: The Need to Implement Improved Methodology

Radiation Technology, Hyogo Ion Beam Medical Center, Tatsuno, Hyogo, JAPAN

SSM signature genes are highly expressed in residual scar tissues after preoperative radiotherapy of rectal cancer.

Pulmonary resection remains the most effective. Survival in Synchronous vs Single Lung Cancer. Upstaging Better Reflects Prognosis

Bin Liu, Lei Yang, Binfang Huang, Mei Cheng, Hui Wang, Yinyan Li, Dongsheng Huang, Jian Zheng,

The BCL11A-XL expression predicts relapse in squamous cell carcinoma and large cell carcinoma

The role of cytoreductive. nephrectomy in elderly patients. with metastatic renal cell. carcinoma in an era of targeted. therapy

Survival Prediction Models for Estimating the Benefit of Post-Operative Radiation Therapy for Gallbladder Cancer and Lung Cancer

Comparison of Triple Negative Breast Cancer between Asian and Western Data Sets

MOLECULAR AND CLINICAL ONCOLOGY 4: , 2016

Refining Prognosis of Early Stage Lung Cancer by Molecular Features (Part 2): Early Steps in Molecularly Defined Prognosis

National Surgical Adjuvant Breast and Bowel Project (NSABP) Foundation Annual Progress Report: 2009 Formula Grant

Nivolumab: esperienze italiane nel carcinoma polmonare avanzato

PROSPERO International prospective register of systematic reviews

The Impact of Adjuvant Chemotherapy in Pulmonary Large Cell Neuroendocrine Carcinoma (LCNC)

EGFR Tyrosine Kinase Inhibitors Prolong Overall Survival in EGFR Mutated Non-Small-Cell Lung Cancer Patients with Postsurgical Recurrence

Nature Getetics: doi: /ng.3471

Integrative analysis of survival-associated gene sets in breast cancer

Changing demographics of smoking and its effects during therapy

SUPPLEMENTARY INFORMATION

Conflicts of Interest

SUPPLEMENTARY INFORMATION

Expanded View Figures

IRESSA (Gefitinib) The Journey. Anne De Bock Portfolio Leader, Oncology/Infection European Regulatory Affairs AstraZeneca

Supplementary Table 1. Baseline Characteristics by Quintiles of Systolic and Diastolic Blood Pressures

Supplementary Online Content

Supplementary data Table S3. GO terms, pathways and networks enriched among the significantly correlating genes using Tox-Profiler

Supplementary Online Content

Gene Expression-Based Approaches in Differentiation of Metastases and Second Primary Tumour

Supporting Information Identification of Amino Acids with Sensitive Nanoporous MoS 2 : Towards Machine Learning-Based Prediction

Supplementary Figure 1. A. Bar graph representing the expression levels of the 19 indicated genes in the microarrays analyses comparing human lung

2 nd line Therapy and Beyond NSCLC. Alan Sandler, M.D. Oregon Health & Science University

gliomas. Fetal brain expected who each low-

Safety of Retreatment With Immunotherapy After Immune-Related Toxicity in Patients With Lung Cancers Treated With Anti-PD-(L)1 Therapy

Supplementary Appendix to manuscript submitted by Trappe, R.U. et al:

Pleural lavage cytology as an independent prognostic factor in non-small cell lung cancer patients with stage I disease and adenocarcinoma

8/1/2017. Imaging and Molecular Biomarkers of Lung Cancer Prognosis. Disclosures. The Era of Precision Oncology

This online (electronic) survey contains twenty four simple questions. Those marked with an asterisk (*) must be answered.

DISCLOSURE SLIDE. ARGOS: research funding, scientific advisory board

Technology appraisal guidance Published: 28 July 2010 nice.org.uk/guidance/ta192

Tumor Growth Rate (TGR) A New Indicator of Antitumor Activity in NETS? IPSEN NET Masterclass Athens, 12 th November 2016

Predicting prognosis of post-chemotherapy patients with resected IIIA non-small cell lung cancer

Thoracic and head/neck oncology new developments

Correlation between expression and significance of δ-catenin, CD31, and VEGF of non-small cell lung cancer

Supplementary Online Content

Supplementary Information

Recursive Partitioning Method on Survival Outcomes for Personalized Medicine

Supplementary Figure 1. ALVAC-protein vaccines and macaque immunization. (A) Maximum likelihood

Neuroendocrine Lung Tumors Myers

Joachim Aerts Erasmus MC Rotterdam, Netherlands. Drawing the map: molecular characterization of NSCLC

Evaluation of public cancer datasets and signatures identifies TP53 mutant signatures with robust prognostic and predictive value

Classification. Methods Course: Gene Expression Data Analysis -Day Five. Rainer Spang

Nature Immunology: doi: /ni Supplementary Figure 1. Transcriptional program of the TE and MP CD8 + T cell subsets.

Studies of Squamous Cell Carcinoma of the Tongue (TSCC), with Focus on Histological Factors

Research Article Prognostic Factors in Advanced Non-Small-Cell Lung Cancer Patients: Patient Characteristics and Type of Chemotherapy

LOGO. Statistical Modeling of Breast and Lung Cancers. Cancer Research Team. Department of Mathematics and Statistics University of South Florida

Layered-IHC (L-IHC): A novel and robust approach to multiplexed immunohistochemistry So many markers and so little tissue

levels of genes were separated by their expression levels; 2,000 high, medium, and low

Long-term prognostic value of N-Terminal Pro-Brain Natriuretic Peptide (NT-proBNP) changes within one year in patients with coronary heart disease

ASSESSING LONG-TERM BENEFITS OF IMMUNOTHERAPY BASED ON EARLY TUMOR ASSESSMENT DATA

Title:Identification of a novel microrna signature associated with intrahepatic cholangiocarcinoma (ICC) patient prognosis

Long term survival study of de-novo metastatic breast cancers with or without primary tumor resection

Take home message. Emilio Bria. II SESSIONE: Immunoterapia nel tumore del polmone

Dynamic prediction using joint models for recurrent and terminal events: Evolution after a breast cancer

Supplemental Information. A Highly Sensitive and Robust Method. for Genome-wide 5hmC Profiling. of Rare Cell Populations

Classification of cancer profiles. ABDBM Ron Shamir

Transcription:

Roepman P, et al. An immune response enriched 72-gene prognostic profile for early stage Non-Small- Supplementary Data 1. Patient characteristics of training and validation set. Patient selection and inclusion overview can be found in Supp Data 9. Training set (103) Validation set (69) Gender (%) (%) male 77 74.8 51 73.9 female 26 25.2 18 26.1 Age at diagnosis median range Hospital 62 41-77 67 22-79 NKI 30 29.1 6 8.7 Heidelberg 18 17.5 14 20.3 Bialystok 12 11.7 1 1.4 Gdansk 32 31.1 27 39.1 Vumc 11 10.7 21 30.4 Smoking current smoker 45 43.7 30 43.5 former smoker 44 42.7 28 40.6 non-smoker 3 2.9 3 4.3 unknown 11 10.7 8 11.6 Histology large cell carcinoma 8 7.8 2 2.9 squamous cell carcinoma 57 55.3 35 50.7 adenocarcinoma 33 32.0 23 33.3 other 5 4.9 9 13.0 Stage I 72 69.9 45 65.2 II 31 30.1 24 34.8 Follow-up period (months) median range Status 46 4-156 24 0.5-111 alive / censored 59 57.3 33 47.8 death lung cancer 35 34.0 16 23.2 death other 9 8.7 20 29.0 Relapse-free survival time (months) median range Overall survival time (months) 43 2-156 22 0.5-111 median range Treatment before surgery 46 4.3-156 24 0.5-111 yes 5 4.9 2 2.9 no 96 93.2 58 84.1 unknown 2 1.9 9 13.0

Supplementary Data 2 Hierarchical clustering (Euclidian distance, complete clustering) of NSCLC samples based on expression of all genes Hospital Histology Stage Survival Hospital Histology Stage Survival NKI-AvL Heidelberg Bialystok Gdansk VUmc squamous cell carcinoma adenocarcinoma other non-small cell stage I stage II alive or censored death by lung cancer

Supplementary Data 3 Schematic overview of the multiple samples procedure that was used for development of a robust nearest mean classifier. A 10-fold cross validation loop was used to identify genes which expression ratios correlate with overall and recurrence free survival time. The initial 103 training samples were randomly split into a training set (n=93) and in a test set (n=10). The training set was used to identify which gene correlate best with overall (OS) and recurrence-free (RFS) survival (based on three statistics to reduce selection of genes based on noise as much as possible: Welsh t-test, logrank test and a Cox proportional hazard ratio). Subsequently the top 100 genes were used for building a nearest mean classifier and the performance was tested on the test set. Repeating this procedure for different training and test splits (multiple sampling) resulted in a multiple sets of most prognostic genes. The multiple gene rankings for OS and RFS were combined and the set of most prognostic genes (most often selected genes during the multiple sampling procedure) was selected via a top-down approach. The performance of the optimal set of genes (72 genes) was evaluated using a leave-one-out approach on all training samples to define the classifier performance using the optimal threshold. The NSCLC 72 gene classifier was finally validated on the independent set of validation samples (n=69). 10-fold cross validation loop ±42,000 gene probes 103 samples (>2y follow-up) >500x Randomly split samples 93 train 10 test Gene scoring Test on test samples Welsh t-test Logrank test Cox proportional hazard ratio Score sample outcome Combine multiple gene rankings Score gene ranking Combine ranking Select top 100 genes Nearest mean classifier Combine Multiple predictions Select set of best prognostic genes (72) Sample prediction robustness & accuracy LOO cross-validation on 103 training samples classifier perfomance using optimal threshold Determine low-risk and high-risk profile Independent validation 69 samples

Supplementary Data 4. Kaplan-Meier plot survival estimates of overall survival (OS) (A) and recurrence-free survival (RFS) (B) based on the multiple sampling outcomes of the test samples using the 10-fold cross validation procedure described in Supp. Data 3. These results indicated that the multiple samples approach is suitable for development of a nearest mean classifier that is unbiased towards the training samples. A B overall survival% 0.0 0.2 0.4 0.6 0.8 1.0 P= 0.001 low-risk profile high-risk profile recurrence-free survival% 0.0 0.2 0.4 0.6 0.8 1.0 P= 0.011 low-risk profile high-risk profile 0 20 40 60 80 100 120 140 months 0 20 40 60 80 100 120 140 months

Supplementary Data 5 NSCLC survival associated genes. The set of 72 prognostic classifier genes is ranked according to the gene expression association with recurrence-free survival.

Supplementary Data 6 Kaplan-Meier plot survival estimates of overall survival based on the 72-gene classifier within tumour stage (A) and based on the 72-gene classifier within tumour histology (B). A overall survival% 0.0 0.2 0.4 0.6 0.8 1.0 stage I & low-risk stage II & low-risk stage I & high-risk stage II & high-risk P<0.001 0 50 100 150 months B overall survival% 0.0 0.2 0.4 0.6 0.8 1.0 scc & low-risk scc & high-risk adeno & low-risk adeno & high-risk P<0.001 0 50 100 150 months

Supplementary Data 7 Functional category analysis of the classifier signature. Functional categories within the set of 72 prognostic genes were identified using gene ontology (GO) analysis. Significantly (P<0.01) overrepresented GO categories are colored red. Blue shaded area indicted categories associated with immune response, green with antigen binding, yellow with protein modification and degradation. antimicrobial humoral response humoral defense mechanism lymphocyte activation humoral immune response hemopoiesis immune cell activation protein modification response to pest, pathogenor parasite response to stress immune response response to other organism lymphocyte differentiation defense response response to biotic stimulus hemopoietic or lymphoid organ development development protein catabolism protein catabolism ubiquitin cycle protein metabolism proteolysis during cellular protein catabolism cellular protein catabolism protein catabolism macromolecule metabolism cellular macromolecule catabolism dent macromolecule catabolism lytic vacuole response to stimulus physiological process mrna binding nucleic acid binding lysosome vacuole cytoplasm organelle integral to plasma membrane membrane biological process gene ontology cellular component binding molecular function catalytic activity antigen binding translation factor activity, nucleic acid binding translation regulator activity carboxylic ester hydrolase activity intrinsic to plasma membrane hydrolase activity thiolester hydrolase activity

Supplementary Data 8 Comparison of different early stage NSCLC prognostic gene classifiers

Supplementary Data 9 Patient selection and inclusion overview.