An Immunogram for the Cancer-Immunity Cycle: Towards Personalized Immunotherapy of Lung Cancer

Accepted Manuscript An immunogram for the cancer-immunity cycle: towards personalized immunotherapy of lung cancer Takahiro Karasaki, Kazuhiro Nagayama, Hideki Kuwano, Jun-ichi Nitadori, Masaaki Sato, Masaki Anraku, Akihiro Hosoi, Hirokazu Matsushita, Yasuyuki Morishita, Kosuke Kashiwabara, Masaki Takazawa, Osamu Ohara, Kazuhiro Kakimi, Jun Nakajima PII:

S1556-0864(17)30008-4

DOI:

10.1016/j.jtho.2017.01.005

Reference:

JTHO 475

To appear in:

Journal of Thoracic Oncology

Received Date: 31 August 2016 Revised Date:

21 November 2016

Accepted Date: 4 January 2017

Please cite this article as: Karasaki T, Nagayama K, Kuwano H, Nitadori J-i, Sato M, Anraku M, Hosoi A, Matsushita H, Morishita Y, Kashiwabara K, Takazawa M, Ohara O, Kakimi K, Nakajima J, An immunogram for the cancer-immunity cycle: towards personalized immunotherapy of lung cancer, Journal of Thoracic Oncology (2017), doi: 10.1016/j.jtho.2017.01.005. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

An immunogram for the cancer-immunity cycle: towards personalized immunotherapy of lung cancer

RI PT

Takahiro Karasaki1, Kazuhiro Nagayama1, Hideki Kuwano1, Jun-ichi Nitadori1, Masaaki Sato1, Masaki Anraku1, Akihiro Hosoi2,6, Hirokazu Matsushita2, Yasuyuki Morishita3, Kosuke Kashiwabara4, Masaki Takazawa5, Osamu Ohara5, Kazuhiro Kakimi2, Jun Nakajima1 1

Department of Thoracic Surgery, 2Department of Immunotherapeutics and 3Department of Molecular Pathology, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan 4 Department of Biostatistics, School of Public Health, The University of Tokyo, Tokyo, Japan 6

Department of Technology Development, Kazusa DNA Research Institute, Kisarazu, Japan Medinet Co. Ltd., Yokohama, Japan

SC

5

M AN U

Running title: Immunogram for personalized immunotherapy Correspondence to: Kazuhiro Kakimi Department of Immunotherapeutics,

TE D

Graduate School of Medicine, The University of Tokyo 7-3-1 Hongo, Bunkyo-Ku, Tokyo 113-8655, Japan Tel.: 81-35805-3161, Fax: 81-35805-3164, E-mail: [email protected] Keywords: lung cancer, cancer-immunity cycle, neoantigen, transcriptome, immunogram

AC C

EP

Financial support: This study was supported in part by JSPS KAKENHI (Grants-in-Aid for Scientific Research) Grant Number 16H04708 (Kazuhiro Kakimi), 15H04942 (Jun Nakajima), 16K07162 (Hirokazu Matsushita), and 26462124 (Kazuhiro Nagayama). Conflict of interest: Department of Immunotherapeutics, Graduate School of Medicine, The University of Tokyo is endowed by Medinet Co. Ltd. Dr. Kazuhiro Kakimi received research support from Medinet Co. Ltd.. The study sponsors had no involvement in study design; collection, analysis, and interpretation of data; writing the report; and the decision to submit the report for publication. All other authors have declared there is no financial conflicts of interest related to this work. Word count: 4248 (main text) Total number of Figures and Tables: 7 Supplemental Data: 2 Methods, 4 Tables, 4 Figures

1

ACCEPTED MANUSCRIPT

Abstract

Introduction: The interaction of immune cells and cancer cells shapes the immunosuppressive tumor microenvironment. For successful cancer immunotherapy, comprehensive knowledge of

RI PT

anti-tumor immunity as a dynamic spacio-temporal process is required for each individual patient. To this end, we developed an immunogram for the cancer-immunity cycle using next-generation sequencing.

SC

Methods: Whole-exome sequencing and RNA-Seq was performed in 20 non-small cell lung cancer patients (12 adenocarcinoma, 7 squamous cell carcinoma, and 1 large cell neuroendocrine

M AN U

carcinoma). Mutated neoantigens and cancer-germline antigens expressed in the tumor were assessed for predicted binding to patients’ HLA molecules. The expression of genes related to cancer immunity was assessed and normalized to construct a radar chart composed of 8 axes reflecting 7 steps in the cancer-immunity cycle.

TE D

Results: Three immunogram patterns were observed in lung cancer patients: T-cell-rich, T-cell-poor and intermediate. The T cell-rich pattern was characterized by gene signatures of abundant T cells, Tregs and MDSCs, checkpoint molecules and immune-inhibitory molecules in the tumor, suggesting

EP

the presence of anti-tumor immunity dampened by an immunosuppressive microenvironment. The T cell-poor phenotype reflected lack of anti-tumor immunity, inadequate DC activation, and

AC C

insufficient antigen presentation in the tumor. Immunograms for both the adenocarcinoma patients and the non-adenocarcinoma patients included both T cell-rich and T cell-poor phenotypes, suggesting that histology does not necessarily reflect the cancer-immunity status of the tumor. Conclusions: The patient-specific landscape of the tumor microenvironment can be appreciated using immunogram as integrated biomarkers, which may thus become a valuable resource for optimal personalized immunotherapy.

2

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

(246 words)

3

ACCEPTED MANUSCRIPT

Introduction

The development of immune checkpoint blockade therapy achieving robust and durable responses of refractory malignancies, including lung cancers, has opened new avenues for cancer treatment.1-5

RI PT

However, response rates are still unsatisfactory with clinical responses usually achieved in only a fraction of patients. PD-L1 expression in the tumor, or its neoantigen load, has been reported to be associated with response to checkpoint blockade therapy and prognosis to some extent,4-6 but this

SC

information is insufficient for distinguishing between those patients who will respond to therapy and those who should be offered alternative treatments.7, 8 Therefore, efficient and reliable biomarkers are

M AN U

urgently required.

The evaluation of anti-tumor immune responses in each individual patient is viewed as a crucial dynamic spacio-temporal process proposed by Chen and Mellman to be a “cancer-immunity cycle”

TE D

(CIC) consisting of 7 steps: 1) Release of cancer antigens, 2) Cancer antigen presentation, 3) Priming and activation, 4) Trafficking of T cells to tumors, 5) Infiltration of T cells into tumors, 6) Recognition of cancer cells by T cells and 7) Killing of cancer cells.

9

To develop efficient and

EP

reliable methods for evaluating anti-tumor immune responses, several strategies including flow cytometry, immunohistochemistry and transcriptome analysis using microarrays or next-generation

AC C

sequencing (NGS) can be applied. One example using immunohistochemistry is

an

“immunoscore”.10-12 In the case of colorectal cancer, an “immunoscore” based on the assessment of the tumor-infiltrating lymphocytes (TIL) is closely associated with better prognosis than current TNM classification.10-12

Large-scale genomics datasets of different tumors are now available for

extensive analysis of the genetic landscape of tumors. 13-19 Prognostic immune cells and genes have been reported by extensive review of gene profile studies and computational analyses of large scale genomic datasets.13, 19 It has been reported that increased expression of genes related to cytotoxic 4

ACCEPTED MANUSCRIPT activity in the local tumor microenvironment is also associated with increased expression of those related to immunosuppression.17,

18

Comprehensive analysis of various functional pathways and

molecular networks contribute to uncover the integrated mechanisms of tumor-immune interactions.

RI PT

These advances in sophisticated analytical methods now allow “omics” data to be generated for each individual patient. In practice, omics data themselves are still a challenge for clinicians to deal with, such that it may be said that there is a compelling need for the development of a “translator” for

SC

converting cumbersome omics data into easily comprehensible information for clinical use.

M AN U

Recently, Blank et al. proposed the concept of the “cancer immunogram” to visualize general and local cancer immunity status in each patient.20 In their Perspectives article, the concept was applied theoretically to patients, but not tested in practice. To accomplish this, here we have developed an immunogram reflecting the cancer-immunity cycle and have applied it to real lung cancer patients.

AC C

EP

individual patient.

TE D

We discuss its potential application for tailoring effective personalized cancer immunotherapy to the

5

ACCEPTED MANUSCRIPT

Materials and Methods

Patient selection and characteristics Twenty patients with non-small cell lung cancer (NSCLC) who underwent lung resection were

RI PT

included in the study after written informed consent had been obtained. Twelve patients had adenocarcinoma (AD), 7 squamous cell carcinoma (SQ), and 1 large cell neuroendocrine carcinoma (LCNEC) (Table 1). None of the patients had received any preoperative treatment. Among 12 AD

SC

patients, EGFR mutations were present in 7 and no ALK rearrangements were detected. The study was conducted with the approval of the Human Genome and Gene Analysis Research Ethics

M AN U

Committee of the Faculty of Medicine and Graduate School of Medicine of The University of Tokyo and The University of Tokyo Hospital (G3545).

Application of whole-exome sequencing and RNA-Seq

TE D

Sequencing libraries of DNA and RNA from tumor and normal tissue samples were prepared as described in Supplemental Data 1 (Supplemental Method 1), and the enriched sequencing libraries were analyzed by massively parallel sequencing on a HiSeq 1500 (Illumina, San Diego, CA).

EP

Sequence data were processed and utilized for calling somatic missense mutations and expression analyses. Expression values were calculated as fragments per kilobase of transcript per million

AC C

fragments mapped (FPKM).

Identification of candidate neoantigens and cancer-germline antigen expression Candidate neoantigens were identified as previously described.21 In brief, 4-digit human leukocyte antigen (HLA) class I alleles of each patient were identified from whole-exome sequencing data of normal lung tissue or PBMC, using Omixon Target HLA Typing (Omixon, Cambridge, MA).22 The Immune Epitope Database (IEDB) analysis resource NetMHCpan (v2.8) tool was used to predict 6

ACCEPTED MANUSCRIPT binding affinities of 8- to 11-mer mutant peptides to the patients’ HLA-A, HLA-B, and HLA-C alleles.23, 24 Peptides derived from somatic mutations with predicted IC50 values <500 nM were considered candidate neoantigens. In this study, we also used RNA-Seq expression data (FPKM>1)

RI PT

to filter neoantigens accompanying tumor expression.

A list of cancer-germline (CG) antigens including 276 genes was obtained from the Cancer Immunity CT Gene Database (http://www.cta.lncc.br/) in May 2016.25 A total of 103 genes which had FPKM=0

SC

in RNA-Seq data of all 6 normal lung tissue samples was included in the analysis. CG antigens with FPKM >1 were analyzed for their binding capacity to HLA molecules using the NetMHCpan tool.

M AN U

CG antigens which generate peptides with predicted IC50 values <500 nM were considered immunogenic CG (imCG) antigens; the number of these was counted in each patient.

Gene expression analysis with RNA-Seq data

TE D

Gene Set Enrichment Analysis (GSEA) was used to estimate if immune cells were over-represented in the tumor microenvironment, as previously described.26, 27 In brief, the expression levels of each gene (FPKM) were z-score normalized across all samples (20 tumor samples and 6 normal lung

EP

samples). For each tumor, all genes were then ranked in descending order according to their z-scores. GSEA PreRanked was performed with default parameters and the association with each immune cell

AC C

gene set was represented by a normalized enrichment score (NES). The immune metagene list provided by Angelova et al. was used.27 For activated DC, a gene set provided by Newman et al. (LM22) was used.28 Incorporated gene sets are listed in Supplemental Data 2 (Supplemental Table 1). An immune cell type was considered significantly enriched or poorer in a tumor when the false discovery rate (q-value) was ≤10%.

7

ACCEPTED MANUSCRIPT When an appropriately validated gene set for GSEA was unavailable, a panel of immune-related genes reflecting each step of the CIC was created.9 RNA-Seq data expression profiles for each gene (FPKM) were transformed into log2 fold-changes (FC) using mean expression of 6 normal samples

RI PT

as the control.

Immunohistochemistry

Immunohistochemistry (IHC) analysis was performed on formalin-fixed paraffin-embedded tissue

SC

sections for assessment of T cell and Treg infiltration using polyclonal rabbit anti-human CD3 antibody (code A0452: Dako, Carpinteria, CA), monoclonal mouse anti-human CD8 (clone

M AN U

C8/144B: Dako), and monoclonal mouse anti-human forkhead box P3 (FOXP3) antibody (clone 236A/E7: AbCam, Cambridge, UK). The slides were counterstained with hematoxylin. Images were captured using BIOREVO-9000 (Keyence, Osaka, Japan); the positively stained area was quantified

TE D

digitally by image analysis software BZ-II Analyzer (Keyence).

Statistical analysis

Spearman’s correlation test was used to compare NES obtained by GSEA and IHC-stained areas.

EP

Neoantigen load and imCG antigen load were analyzed with two-tailed Welch’s t-tests on log(x + 1) transformed data to accommodate data skewness and zero values. P values <0.05 were considered significant.

All

AC C

statistically

statistical

analyses

(http://www.r-project.org).

8

were

performed

using

R

3.3.0

ACCEPTED MANUSCRIPT

Results

An immunogram for the cancer-immunity cycle To evaluate the dynamic process of anti-tumor immunity in an individual patient and formulate

RI PT

strategies to manipulate anti-tumor immunity for cancer treatment, we developed an immunogram for the cancer-immunity cycle (Fig. 1). Here, 7 steps of the CIC are depicted by the 8 axes of Immunogram Scores (IGS) as follows: IGS1, existence of T cell immunity in the tumor; IGS2, tumor

SC

antigenicity; IGS3, priming and activation; IGS4, trafficking and infiltration; IGS5, recognition of tumor antigens; IGS6-8, suppressive factors preventing killing of cancer cells. Because CIC depicts

M AN U

dynamic process for the development of anti-tumor T cell immunity, assessment of T cell immunity was set as the first axis of the “Immunogram for the Cancer-Immunity Cycle”.

IGS1: anti-tumor T cell immunity

TE D

First, we investigated the presence of T cells in the tumor by Gene Set Enrichment Analysis (Fig. 2).26 A normalized enrichment score (NES) of 86 genes27 comprising a T cell signature was calculated by GSEA in 20 patients. Examples of enrichment plots for patient LK071 obtained by

EP

GSEA are shown in Supplemental Data 3 (Supplemental Figure 1). In all, 9 patients had a T cell-inflamed (rich) phenotype with significant enrichment of T cells (q-value <0.1) (Fig. 2). Detailed

AC C

results of TIL enrichment analysis by GSEA are shown in Supplemental Data 4 (Supplemental Table 2). Of the remaining 11 patients, classified as having a non-T cell-inflamed phenotype, 6 showed significantly poorer T cell infiltration (q<0.1) and the other 5 were classified as intermediate. NES as an indicator for anti-cancer T cell presence in the tumor was validated by IHC analysis. Tumor sections were stained for CD3+ and CD8+ T cells, and positively stained areas (µm2/mm2) were quantified digitally. The NES of T cells was correlated significantly with the CD3-positive area (Spearman’s R = 0.83, p<0.0001) and with the CD8-positive area (Spearman’s R = 0.80, p<0.0001). 9

ACCEPTED MANUSCRIPT All these data are shown in Supplemental Data 5 (Supplemental Table 3). IHC staining of representative patients either with enriched or poor T cell infiltration are shown in Supplemental Data 6 (Supplemental Fig. 2). For plotting the immunogram, the NES of T cells in each patient was converted into a z-score, and

RI PT

then converted into the “Immunogram Score” (IGS). If M represents the mean NES and SD represents the standard deviation of the NES, the z-score (Z) of each patient is calculated by Z = (NES - M)/SD

SC

We then defined the score of the first axis of the immunogram (IGS1) in each patient as IGS1 = 3 + 1.5×Z

M AN U

This formula was applied for all axes of the patient´s immunogram. The lower and upper limits of the IGS were set at 1 and 5. By definition, IGS = 3 represents an NES equivalent to the mean NES, and IGS = 4.5 (1.5) represents an NES equivalent to the mean+1SD (-1SD). The IGS was defined in this way so that patients would be well-distributed over the range from 1 to 5. The IGS calculation

IGS2: tumor antigenicity

TE D

for patient LK071 is shown as an example in Supplemental Data 7 (Supplemental Method 2).

EP

Tumor antigenicity was evaluated by the number of candidate neoantigens and imCG antigens (Fig. 3A). Numbers of candidate neoantigens and imCG antigens are shown in Supplemental Data 8

AC C

(Supplemental Table 4). Median neoantigen load was 23 (range 3-183) in T cell-inflamed and 71 (range 31-301) in non-T cell-inflamed tumor (P = 0.053, Welch’s t-test on log-transformed data). Median imCG antigen load was 1 (range 0-11) in T cell-inflamed and 4 (range 1-29) in non-T cell-inflamed tumor (P = 0.14). The number of neoantigens and imCG was summed to give the number of putative cancer antigens to be plotted on the second axis of the immunogram (IGS2) after the z-score was calculated, and converted to IGS by the same formula as for the first axis (Fig. 3A).

10

ACCEPTED MANUSCRIPT IGS3: T cell priming and activation Antigen presentation and T cell priming was evaluated by analysis of gene signatures for activated dendritic cells (DCs). The enrichment analysis of activated DC was performed using 53 genes28 and the NES was determined (Fig. 3B).26, 27 Detailed results of activated DC enrichment analysis by

RI PT

GSEA are shown in Supplemental Data 4 (Supplemental Table S2). Activated DC were significantly enriched in 11 patients (q<0.1), all of whose tumors were of the T cell-rich (LK001, 013, 047, 051, 053, 066, 070, 071) or intermediate phenotype (LK029, 059, 075) (Fig. 3B). The IGS of the third

M AN U

IGS4: trafficking and infiltration of T cells into tumors

SC

axis (IGS3) was calculated in the same manner as the first axis.

Because there is no established gene set which is suitable for GSEA of T cell trafficking and infiltration, expression variance was analyzed for each of the genes thought to be positively associated with T cell trafficking and infiltration. Based on a previous report,9 CXCL9, CXCL10,

TE D

CCL5, CX3CL1, LFA1, ICAM1 and SELE were selected as being informative for trafficking and infiltration (Fig. 4A). All expression data are shown in Supplemental Data 4 (Supplemental Table 2). The number of genes the expression of which was up-regulated with log2FC>1 was counted, and

EP

converted to a z-score, and the IGS4 calculated.

AC C

IGS5: recognition of cancer cells by T cells The recognition of cancer cells by T cells requires that the antigen processing and presentation machinery of cancer cells is intact. Therefore, a gene signature for antigen processing and presentation machinery was used for the evaluation of the fifth axis of the immunogram (IGS5). The gene set consisted of HLA-A, HLA-B, HLA-C, B2M, TAP1, and TAP2 (Fig. 4B). These genes are in general expressed in normal cells; however, some tumors lack or down-regulate their expression,

11

ACCEPTED MANUSCRIPT making them invisible to T cells. Therefore, the number of down-regulated genes with log2FC<0 was counted, and converted to a z-score (Z). The IGS was calculated as IGS5 = 3 - 1.5×Z Note that a high IGS5 implies that the number of down-regulated genes is low and antigen processing

RI PT

and presentation machinery is intact.

IGS6-8: killing of cancer cells

SC

The effector function of T cells and killing of cancer cells are influenced by the complex interaction of inhibitory cells and molecules. We divided this process into 3 axes for the immunogram: IGS6 for

the absence of other inhibitory molecules.

M AN U

the absence of inhibitory cells, IGS7 for the absence of checkpoint molecule expression and IGS8 for

As shown in Figure 3C, the infiltration of inhibitory immune cells, such as MDSC and Tregs, was

TE D

evaluated using GSEA.26 Gene signatures consisted of 26 genes and 83 genes27 to calculate the NES for Treg and MDSC, respectively. Detailed results of enrichment analysis of MDSC and Tregs are shown in Supplemental Data 4 (Supplemental Table 2). The NES for Tregs was validated by IHC

EP

staining, showing that it correlated well with FOXP3 staining (Spearman’s R = 0.75, p = 0.0002). Examples of IHC staining for FOXP3 are shown in Supplemental Data 6 (Supplemental Fig. 2) and

AC C

all such data are shown in Supplemental Data 5 (Supplemental Table 3). The sum of the NES for Treg and MDSC was calculated and then converted to a z-score (Z). Enrichment of these cells suggests an immunosuppressive microenvironment in the tumor. Thus, the sixth axis of the immunogram (IGS6) was calculated as IGS6 = 3 - 1.5×Z Low IGS6 suggests that anti-tumor immunity is suppressed by the infiltration of inhibitory immune cells; absence of inhibitory immune cells results in high IGS6. 12

ACCEPTED MANUSCRIPT

IGS7: absence of checkpoint molecule expression It is known that anti-tumor T cell responses are inhibited by arrays of immune checkpoint molecules in the tumor microenvironment. For the seventh axis of the immunogram (IGS7), we selected a set of

RI PT

genes for immune checkpoints comprising PD-1, BTLA, TIM-3, LAG3, CTLA-4, PD-L1, and VISTA, which are expressed either on T cells, tumor cells or antigen presenting cells (Fig. 4C).9 Genes with log2FC>1 were considered to be up-regulated; they were counted and the result

SC

converted to a z-score (Z). Thus, the seventh axis of the immunogram (IGS7) was calculated as IGS7 = 3 - 1.5 × Z

M AN U

High expression of checkpoint molecules results in low IGS7.

IGS8: absence of other inhibitory molecules.

Tumor cells and infiltrating cells express a variety of inhibitory molecules that impair the effector

TE D

function of T cells. Among these, IL10, IDO1, TGFB1, ARG1, INOS, and CTNNB1 were selected for the gene set for IGS8, because therapeutic inhibition of these molecules may enhance T cell

likewise (Fig. 4D).

EP

immunity.9 Expression profiles of these inhibitory molecules were evaluated and IGS8 was calculated

AC C

Interpretation of Immunograms for the Cancer-Immunity Cycle As described above, each step of the cancer-immunity cycle was assessed, normalized, and scored onto the 8 axes of the immunogram (IGS1-8), generated for each individual patient by integration onto a radar chart (Fig. 5A-C). Immunograms of patients with T cell-rich, -intermediate, or -poor phenotypes displayed distinct features for each IGS (Fig. 5D). In patients with a T cell-rich phenotype, high IGS1 (existence of putative T cell immunity in the tumor) was accompanied by a high IGS3 (activation of DCs), high IGS4 (trafficking and infiltration), and high IGS5 (likely intact 13

ACCEPTED MANUSCRIPT recognition of tumor antigens by T cells), whereas IGS6, IGS7 and IGS8 were all low, suggesting increased recruitment of Tregs and MDSCs, and up-regulated expression of checkpoint molecules and immunoinhibitory molecules in the tumor. In contrast, in patients with a T cell-poor phenotype, IGS1, IGS3, IGS4 and IGS5 were quite low, while IGS2, IGS6, IGS7 and IGS8 were maintained.

RI PT

Activation of DC was impaired, even if the tumors had relatively high antigenicity. These tumors had recruited few immunosuppressive cells and lacked the expression of immune checkpoint molecules and immunoinhibitory molecules. Each axis of the immunogram from patients with an intermediate

SC

phenotype largely showed values lying between the two extremes above. Interestingly, immunograms for AD patients could display either a T cell-rich or T cell-poor phenotype,

M AN U

irrespective of EGFR mutational status. Similar results were obtained in non-AD patients (Fig 5E). These results suggest that histology does not necessarily reflect the cancer-immunity status of the

AC C

EP

TE D

tumor.

14

ACCEPTED MANUSCRIPT

Discussion

Since the anti-PD-1 monoclonal antibodies nivolumab and pembrolizumab were approved for the treatment of NSCLC, robust and durable responses have been observed, but only 20~30% of patients

RI PT

responded to the therapy.3-5 Although the expression of PD-L1 or the number of tumor neoantigens was reported to correlate with treatment outcome,4-6 better predictive biomarkers for selecting patients who would or would not respond to therapy is warranted. Anti-cancer immunity is a

SC

dynamic process described as a cancer-immunity cycle; different steps in the cycle by which tumors escape immunosurveillance are likely to be different patient by patient. Therefore, we propose to

M AN U

construct an immunogram for each patient in order to better understand the individual patient´s CIC and to clarify the steps where the anti-cancer response is blocked. In the present study, we defined the “Immunogram for the Cancer-Immunity Cycle” using NGS data and were able to visualize the status of potential anti-tumor immune responses within the tumor. Utilizing this immunogram, the

TE D

landscape of the tumor microenvironment in each patient can be appreciated and the compromised steps of the cancer-immunity cycle can be easily visualized.

EP

Immunograms of lung cancer patients have several characteristics in common, reflecting their gene signatures relevant to T cell responses (IGS1). In patients with a T cell-rich phenotype, high scores

AC C

for IGS3, IGS4 and IGS5 and low scores for IGS6, IGS7 and IGS8 were found to be typical (Fig. 5A and D). This suggests that the tumor microenvironment is dominated by inhibitory cell infiltration and the expression of checkpoint and immunoinhibitory molecules which counter-regulate anti-tumor immune responses. For example, the immunogram of patient LK071 with a T cell-rich phenotype demonstrated that T cell immunity potentially existed in the tumor, but was suppressed by inhibitory cells and molecules including checkpoint molecules (Fig. 5A). Reviewing the transcriptomic data for this patient, we hypothesize that IDO1 may be one of the key molecules in 15

ACCEPTED MANUSCRIPT this respect (Fig. 4D). Therefore, strategies to unleash T cell responses by depleting immunosuppressive cells, immune checkpoint blockade or enzymatic inhibition of the immunosuppressive molecules might be recommended for this patient (Fig. 6).

RI PT

In contrast, low scores for IGS3, IGS4, IGS5 and high scores for IGS6, IGS7, IGS8 were common features of the immunogram in patients with a T cell-poor phenotype (Fig. 5C and D). Lack of induction of anti-tumor immunity in these patients might have been due to inadequate DC activation

SC

or insufficient antigen processing and presentation in the tumor. The lack of recruitment of immunosuppressive cells and absence of expression of immune checkpoint and immunoinhibitory

M AN U

molecules also reflects the lack of induction of counter-regulatory immunosuppression in the tumor. Inspecting the immunograms, it would be predicted that immune checkpoint blockade would be unlikely to be effective for these T cell-poor patients. For example, the immunogram of patient LK073 with a T cell-poor phenotype demonstrated that although a certain amount of cancer antigens

TE D

was detected, DC activation was significantly impaired (Fig. 5C). For this patient, intervention to enhance T cell trafficking into tumors such as via the induction of immunologic cell death by chemotherapy or radiotherapy, or the direct activation of DCs by neoantigen-targeted vaccines might

EP

be a preferred therapeutic option29, 30 (Fig. 6).

AC C

The hierarchical clustering using each z-scores for IGS also displayed two clusters of patients: a cluster which contained all T cell-rich patients, and another cluster which contained all T cell-poor patients (Supplemental Data 9 Supplemental Fig. 3, the result of hierarchical clustering using z-scores for IGS). In addition, IGS axes were clustered into two groups: a cluster associated with immunological status of T cells, and another cluster regarding antigens and antigen presentation of the tumor.

16

ACCEPTED MANUSCRIPT In the present study, IGS2 was generally low in tumors with a T cell-rich phenotype, especially due to their lower neoantigen load. This may be partly explained by cancer immunoediting, that subclones of tumor cells expressing immunogenic antigens are more likely to be eliminated by the T cell immunity in these pateints.12, 16 This may also suggest that T cells can nonetheless be recruited by

RI PT

only a few effective neoantigens, and also implies that T cells are not be recruited even by abundant neoantigens if certain other steps of the CIC are compromised. Hence, non-T cell-inflamed tumors with abundant neoantigens might be suitable targets for neoantigen-based vaccine therapy. In the

SC

case of patients LK070, 001, and 044, reduced expression of antigen-presenting machinery was observed in the presence of abundant tumor-infiltrating T cells, suggesting that T cells might not

M AN U

adequately recognize tumor cells (Fig. 4B, 5A). Interventions to upregulate MHC expressions such as epigenetic therapy, or alternative interventions, such as chimeric antigen receptor T cell therapy, that does not depend on MHC class I antigen presentation for recognition of antigens, might be

TE D

required in these cases.31, 32

Interestingly, when we overlay the immunograms for adenocarcinoma with or without EGFR mutation and non-adenocarcinoma, no typical pattern for each tumor type emerged (Fig. 5E). Both T

EP

cell-inflamed and non-T cell-inflamed phenotypes were present in each histology. These results are consistent with previous studies showing that clinical responses on checkpoint blockade were not

AC C

easily predicted by the histology of the tumor.3-5 When we overlay the immunograms by age, sex and smoking status, similar pattern with relatively low antigenicity (IGS2) was observed in female and never-smoking patients (Supplemental Data 10, Supplemental Figure 4). Generally, EGFR mutation is related to female sex and never-smoking status.33 Never-smoking status is related to lower mutational load.34 Thus, patients with EGFR mutation positive adenocarcinoma, female sex, and never-smoking status are overlapped, resulting in similar immunogram pattern of lower mutational

17

ACCEPTED MANUSCRIPT burden. These results support that immunogram of each patient successfully reflected the patient’s immunological status and background.

We propose that the application of “Immunograms for the Cancer-Immunity Cycle” in each patient

RI PT

will be beneficial for tailoring personalized immunotherapy. One of the advantages of integrative analysis of multiple genes is that it may increase the reliability of gene-expression data with low levels of expression. However, there are several limitations to the present study. First, immunogram

SC

axes and each gene list are not absolutes, and should be altered along with advances in knowledge and technologies. Although the presented immunogram depicts the landscape of cancer immunity in

M AN U

each patient, there are many other factors related to T cell dysfunction, such as abnormal tumor metabolism, that may also play a role. Inclusion of these factors may strengthen prediction of the effectiveness of cancer immunotherapy. Second, not only neoantigens and CG antigens act as tumor antigens. There are still many other types of cancer antigens that cannot be found from exome data or

TE D

classical expression analysis, for example, viral antigens from oncoviruses, or abnormal expression of human endogenous retroviral antigens.16, 18 Third, although the immunogram may precisely depict a complicated immune status, it is a snapshot of only that moment in time. Thus, for example, if T

EP

cell immunity could be induced in the T cell-poor tumor by an appropriate treatment (e.g. induction of immunogenic cell death by chemoradiotherapy; neoantigen vaccines; anti-CTLA4 antibody; IFNα

AC C

agent, CD40-agonist, and manipulation of microbiota for enhancement of DC activation),9, 35, 36 then counter-regulatory checkpoint molecule expression, and induction or recruitment of inhibitory cells and inhibitory molecules may follow. Re-evaluation of the patient´s immunological status would be required to monitor this. Combinations of several modalities of therapy might then be recommended according to up-dated immunograms (Fig. 6). Finally, because of the small number of patients included in the present study, we could not offer the complete panorama of immune landscape in lung cancer patients. Further accumulation of patient data including treatment outcome is crucial for 18

ACCEPTED MANUSCRIPT making an optimal immunogram that could appropriately determine the best treatment combinations and strategies for every lung cancer patient.

Despite these limitations, the scoring and visualization of the cancer-immunity status of each patient

RI PT

using immunogram may yield a better understanding of cancer-immunity interactions than current data management relying on restricted information with IHC and limited amounts of gene-expression data. Immunogram for the Cancer-Immunity Cycle may thus represent a promising tool for

SC

translating cumbersome omics data into easily comprehensible “report cards” for clinicians to use to

M AN U

tailor optimal immunotherapy for each patient.

In conclusion, comprehensive understanding of cancer-immunity interactions is essential for providing effective cancer immunotherapy. Immunogram for the Cancer-Immunity Cycle can be used as an integrated biomarker, providing us with a clearer view of the landscape of

personalized immunotherapy.

EP

Acknowledgements

TE D

cancer-immunity interaction status in each patient, and may thus become a valuable resource for

AC C

The authors thank Mr. Makoto Ikeda (Kazusa DNA Research Institute) for data analyses.

References

19

ACCEPTED MANUSCRIPT 1.

Hodi FS, O'Day SJ, McDermott DF, et al. Improved survival with ipilimumab in patients

with metastatic melanoma. N Engl J Med 2010;363:711-723. 2.

Herbst RS, Soria JC, Kowanetz M, et al. Predictive correlates of response to the anti-PD-L1

antibody MPDL3280A in cancer patients. Nature 2014;515:563-567. Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus Docetaxel in Advanced

RI PT

3.

Squamous-Cell Non-Small-Cell Lung Cancer. N Engl J Med 2015;373:123-135. 4.

Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus Docetaxel in Advanced

5.

SC

Nonsquamous Non-Small-Cell Lung Cancer. N Engl J Med 2015;373:1627-1639.

Herbst RS, Baas P, Kim DW, et al. Pembrolizumab versus docetaxel for previously treated,

trial. Lancet 2016;387:1540-1550. 6.

M AN U

PD-L1-positive, advanced non-small-cell lung cancer (KEYNOTE-010): a randomised controlled

Rizvi NA, Hellmann MD, Snyder A, et al. Cancer immunology. Mutational landscape

determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 2015;348:124-128. Whiteside TL, Demaria S, Rodriguez-Ruiz ME, et al. Emerging Opportunities and

TE D

7.

Challenges in Cancer Immunotherapy. Clin Cancer Res 2016;22:1845-1855. 8.

Topalian SL, Taube JM, Anders RA, et al. Mechanism-driven biomarkers to guide immune

9.

Chen DS, Mellman I. Oncology meets immunology: the cancer-immunity cycle. Immunity

AC C

2013;39:1-10. 10.

EP

checkpoint blockade in cancer therapy. Nat Rev Cancer 2016;16:275-287.

Galon J, Costes A, Sanchez-Cabo F, et al. Type, density, and location of immune cells within

human colorectal tumors predict clinical outcome. Science 2006;313:1960-1964. 11.

Galon J, Angell HK, Bedognetti D, et al. The continuum of cancer immunosurveillance:

prognostic, predictive, and mechanistic signatures. Immunity 2013;39:11-26.

20

ACCEPTED MANUSCRIPT 12.

Mlecnik B, Bindea G, Angell HK, et al. Integrative Analyses of Colorectal Cancer Show

Immunoscore Is a Stronger Predictor of Patient Survival Than Microsatellite Instability. Immunity 2016;44:698-711. 13.

Suzuki K, Kachala SS, Kadota K, et al. Prognostic immune markers in non-small cell lung

14.

RI PT

cancer. Clin Cancer Res 2011;17:5247-5256.

The Cancer Genome Atlas Research Network. Comprehensive genomic characterization of

squamous cell lung cancers. Nature 2012;489:519-525.

The Cancer Genome Atlas Research Network. Comprehensive molecular profiling of lung

SC

15.

adenocarcinoma. Nature 2014;511:543-550.

Rooney MS, Shukla SA, Wu CJ, et al. Molecular and genetic properties of tumors

M AN U

16.

associated with local immune cytolytic activity. Cell 2015;160:48-61. 17.

McGranahan N, Furness AJ, Rosenthal R, et al. Clonal neoantigens elicit T cell

immunoreactivity and sensitivity to immune checkpoint blockade. Science 2016;351:1463-1469. Ock CY, Keam B, Kim S, et al. Pan-cancer immunogenomic perspective on the tumor

TE D

18.

microenvironment based on PD-L1 and CD8 T cell infiltration. Clin Cancer Res 2016;22:2261-2270. 19.

Gentles AJ, Newman AM, Liu CL, et al. The prognostic landscape of genes and infiltrating

20.

EP

immune cells across human cancers. Nat Med 2015;21:938-945. Blank CU, Haanen JB, Ribas A, et al. CANCER IMMUNOLOGY. The "cancer

21.

AC C

immunogram". Science 2016;352:658-660. Karasaki T, Nagayama K, Kawashima M, et al. Identification of Individual Cancer-Specific

Somatic Mutations for Neoantigen-Based Immunotherapy of Lung Cancer. J Thorac Oncol 2016;11:324-333. 22.

Major E, Rigó K, Hague T, et al. HLA Typing from 1000 Genomes Whole Genome and

Whole Exome Illumina Data. PLoS ONE 2013;8:e78410.

21

ACCEPTED MANUSCRIPT 23.

Nielsen M, Lundegaard C, Blicher T, et al. NetMHCpan, a method for quantitative

predictions of peptide binding to any HLA-A and -B locus protein of known sequence. PLoS One 2007;2:e796. 24.

Hoof I, Peters B, Sidney J, et al. NetMHCpan, a method for MHC class I binding prediction

25.

RI PT

beyond humans. Immunogenetics 2009;61:1-13.

Almeida LG, Sakabe NJ, deOliveira AR, et al. CTdatabase: a knowledge-base of

high-throughput and curated data on cancer-testis antigens. Nucleic Acids Res 2009;37:D816-819. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a

SC

26.

S A 2005;102:15545-15550. 27.

M AN U

knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U

Angelova M, Charoentong P, Hackl H, et al. Characterization of the immunophenotypes and

antigenomes of colorectal cancers reveals distinct tumor escape mechanisms and novel targets for immunotherapy. Genome Biol 2015;16:64.

Newman AM, Liu CL, Green MR, et al. Robust enumeration of cell subsets from tissue

TE D

28.

expression profiles. Nat Methods 2015;12:453-457. Schumacher TN, Schreiber RD. Neoantigens in cancer immunotherapy. Science

2015;348:69-74. 30.

EP

29.

Carreno BM, Magrini V, Becker-Hapak M, et al. Cancer immunotherapy. A dendritic cell

AC C

vaccine increases the breadth and diversity of melanoma neoantigen-specific T cells. Science 2015;348:803-808. 31.

Porter DL, Levine BL, Kalos M, et al. Chimeric antigen receptor-modified T cells in chronic

lymphoid leukemia. N Engl J Med 2011;365:725-733. 32.

Maude SL, Frey N, Shaw PA, et al. Chimeric antigen receptor T cells for sustained

remissions in leukemia. N Engl J Med 2014;371:1507-1517.

22

ACCEPTED MANUSCRIPT 33.

Shi Y, Au JS, Thongprasert S, et al. A prospective, molecular epidemiology study of EGFR

mutations in Asian patients with advanced non-small-cell lung cancer of adenocarcinoma histology (PIONEER). J Thorac Oncol 2014;9:154-162. Govindan R, Ding L, Griffith M, et al. Genomic landscape of non-small cell lung cancer in

smokers and never-smokers. Cell 2012;150:1121-1134. 35.

RI PT

34.

Sivan A, Corrales L, Hubert N, et al. Commensal Bifidobacterium promotes antitumor

immunity and facilitates anti-PD-L1 efficacy. Science 2015;350:1084-1089.

Vetizou M, Pitt JM, Daillere R, et al. Anticancer immunotherapy by CTLA-4 blockade relies

AC C

EP

TE D

M AN U

on the gut microbiota. Science 2015;350:1079-1084.

SC

36.

23

ACCEPTED MANUSCRIPT

Figure Legends

Figure 1. Immunogram for the Cancer-Immunity Cycle. Immunological status of an individual patient in terms of the cancer-immunity cycle can be depicted

RI PT

by the 8 axes of immunogram scores IGS1, existence of T cell immunity in the tumor; IGS2, tumor antigenicity; IGS3, priming and activation; IGS4, trafficking and infiltration; IGS5, recognition of

SC

tumor antigens; IGS6-8, suppressive factors inhibiting killing of cancer cells.

CG antigen, cancer-germline antigen; aDC, activated dendritic cell; Treg, regulatory T cell; MDSC,

M AN U

myeloid-derived suppressor cell

Figure 2. T-cell enrichment analysis for the 1st axis of the immunogram (IGS1). Fragments per kilobase of transcript per million fragments mapped (FPKM) from RNA-Seq of lung

TE D

cancers were compared with mean FPKM values of the corresponding genes in 6 normal lung tissues and displayed as fold-change (FC). Heat-map for log2FC of 86 T cell signature genes27 is shown. Independently, a normalized enrichment score (NES) of 86 T cell signature genes was calculated by 27

First, expression levels of each gene (FPKM) obtained by

EP

Gene Set Enrichment Analysis.26,

RNA-Seq were z-score normalized across all samples (20 tumor samples and 6 normal lung

AC C

samples); for each tumor, all genes were then ranked in descending order according to their z-scores and an NES of the T cell signature was calculated for each patient. Again, the NES was z-score normalized across 20 patients and IGS1 was calculated by the formula: IGS1 = 3 + 1.5×Z, where Z is the z-score of NES. Nine patients had a T cell-inflamed (rich) phenotype with significant enrichment of T cells (q-value <0.1). Of the remaining 11 patients, classified as non-T cell-inflamed phenotype, 6 showed significantly poor T cell infiltration (q<0.1) and the rest were classified as of intermediate phenotype. 24

ACCEPTED MANUSCRIPT

NES, normalized enrichment score; FC, fold change

and activation for IGS3, and inhibitory cells for IGS6

RI PT

Figure 3. Assessment of tumor antigenicity for the 2nd axis of the immunogram (IGS2), priming

(A) Tumor antigenicity was evaluated by the total number of putative cancer antigens (i.e. sum of candidate neoantigens and immunogenic cancer-germline antigens). Z-scores of number of cancer

SC

antigens were calculated, then converted to IGS2. (B) In addition to the heat-map showing the level of gene expression as fold-change compared to normal tissues, priming and activation of T cells was

M AN U

assessed using Gene Set Enrichment Analysis (GSEA) of activated dendritic cells (aDC). Normalized enrichment scores (NES) for aDC signatures (53 genes)28 were calculated by GSEA. Z-scores of NES were calculated and converted to IGS3. (C) Infiltration of Tregs and MDSCs into the tumor was assessed using GSEA of Treg (26 genes) and MDSC (83 genes).26, 27 Z-scores of the

TE D

sum of NES in each patient were calculated and then converted to IGS6. Heat-maps of Treg and MDSC signatures are also shown.

EP

NES, normalized enrichment score; FC, fold change; imCG antigen, immunogenic cancer-germline

cell

AC C

antigen, aDC, activated dendritic cell; Treg, regulatory T cell; MDSC, myeloid-derived suppressor

Figure 4. Assessment of T cell trafficking and infiltration for the 4th axis of the immunogram (IGS4), recognition of tumor cells for IGS5, immune checkpoint expression for IGS7, and other inhibitory molecules for IGS8. (A) Expression profiles of genes related to trafficking and infiltration are shown.9 FPKM values from RNA-Seq were compared with the mean FPKM value of 6 normal lung tissues and displayed as 25

ACCEPTED MANUSCRIPT fold-change (FC). Z-scores of the number of up-regulated genes with log2FC >1 were calculated, and then converted to IGS4. (B) Z-scores of the number of down-regulated (log2FC<0) genes related to antigen presentation machinery were calculated, and then converted to IGS5. Z-scores of the number of up-regulated (log2FC>1) genes related to immune checkpoint and other inhibitory molecules were

RI PT

calculated,9 and then converted to IGS7 (C) and IGS8 (D).

SC

FC, fold-change

Figure 5. Immunogram for the Cancer-Immunity Cycle of 20 non-small cell lung cancer

M AN U

patients.

Each axis represents: 1 (IGS1), putative existence of T cell immunity in the tumor; 2 (IGS2), tumor antigenicity; 3 (IGS3), priming and activation of T cells; 4 (IGS4), trafficking and infiltration; 5 (IGS5), recognition of tumor antigens; 6 (IGS6), absence of inhibitory immune cells; 7 (IGS7),

TE D

absence of immune checkpoint molecules; 8 (IGS8), absence of other inhibitory molecules. Immunograms of an individual patient with a T cell-rich phenotype (A), T cell-intermediate phenotype (B) or T cell-poor phenotype (C). (D) Overlaid immunograms of 9 patients with a T cell-

Overlaid

EP

rich phenotype, 5 with a T cell-intermediate phenotype and 6 with a T cell-poor phenotype. (E) immunograms

of

adenocarcinomas

with

and

without

EGFR

mutation,

and

AC C

non-adenocarcinoma tumors. All scoring data used for immunogram construction are shown in Figures 2-4 and Supplemental Data 4 (Supplemental Table 2).

Figure 6. Immunogram towards personalized immunotherapy. For effective T cell immunotherapy against cancer, induction of T cells is essential in patients whose tumors have a T cell-poor phenotype (e.g. LK073). Once T cell immunity is successfully induced, the microenvironment of the tumor might convert to a T cell-rich phenotype (e.g. LK071). If so, 26

ACCEPTED MANUSCRIPT counter-regulatory mechanisms blocking effector T cells by checkpoint molecule expression and inhibitory cell recruitment, as well as production of other inhibitory molecules may develop and overcome immunosurveillance. Re-evaluation of immunological status might be required over the

recommended according to the up-dated immunogram.

RI PT

course of treatment and a combination of several therapeutic modalities may need to be

1. T cell immunity, 2. Tumor Antigenicity, 3. Priming & activation, 4. Trafficking & infiltration, 5.

SC

Recognition of tumor cells, 6. Absence of inhibitory cells, 7. Absence of checkpoint expressions, 8. Absence of inhibitory molecules

M AN U

CAR-T cell, chimeric antigen receptor T cell; Treg, regulatory T cell; MDSC, myeloid-derived

AC C

EP

TE D

suppressor cell

27

ACCEPTED MANUSCRIPT Supplemental Fig. 1 (Supplemental Data 3). Gene Set Enrichment Analysis. Enrichment plot for T cell signature (A), activated dendritic cell (aDC) signature (B), regulatory T cell (Treg) signature (C), myeloid-derived suppressor cell (MDSC) signature (D) for patient LK071.

RI PT

NES, Normalized enrichment score

Supplemental Fig. 2 (Supplemental Data 6). Immunohistochemical analysis.

SC

Infiltration of T cells into the tumor was evaluated for CD3+ T cells (A, B), CD8+ T cells (C, D) and FOXP3+ regulatory T cells (E, F). Staining patterns of representative patients with a T cell-rich

M AN U

phenotype (A, C, E; LK071) or T cell-poor phenotype (B, D, F; LK073) are shown. Images were captured by BIOREVO-9000 (Keyence, Osaka, Japan) at a magnification of x20 (objective lens).

TE D

NES, normalized enrichment score

Supplemental Fig. 3 (Supplemental Data 9). Hierarchical clustering using z-scores for Immunogram Score (IGS) calculation.

EP

Z-scores for the calculation of IGS were used in hierarchical clustering (Euclidean distance). Patients made two clusters: a cluster which contained all T cell-rich patients, and another cluster which

AC C

contained all T cell-poor patients. Z-scores for IGS also made two clusters: a cluster of IGS reflecting the cancer-immunity status of T cell side, and another cluster of IGS featuring the tumor side.

Supplemental Fig. 4 (Supplemental Data 10). Immunograms overlaid with age, sex and smoking status.

28

ACCEPTED MANUSCRIPT Each axis represents: 1 (IGS1), putative existence of T cell immunity in the tumor; 2 (IGS2), tumor antigenicity; 3 (IGS3), priming and activation of T cells; 4 (IGS4), trafficking and infiltration; 5 (IGS5), recognition of tumor antigens; 6 (IGS6), absence of inhibitory immune cells; 7 (IGS7), absence of immune checkpoint molecules; 8 (IGS8), absence of other inhibitory molecules. 75 years, age >75 years, male sex, female sex,

RI PT

Immunogram were overlaid by patients with age

ever-smoking status, and never-smoking status. Overlaid immunogram of female sex and

AC C

EP

TE D

M AN U

SC

never-smoking status showed relatively reduced tumor antigenicity (IGS2).

29

ACCEPTED MANUSCRIPT

List of supplementary materials

AC C

EP

TE D

M AN U

Supplemental Data 9. (Supplemental Figure 3) tiff Supplemental Data 10. (Supplemental Figure 4) tiff

SC

Supplemental Data 5. (Supplemental Table 3) xlsx Supplemental Data 6. (Supplemental Figure 2) tiff Supplemental Data 7. (Supplemental Method 2) docx Supplemental Data 8. (Supplemental Table 4) xlsx

RI PT

Supplemental Data 1. (Supplemental Method 1) docx Supplemental Data 2. (Supplemental Table 1) xlsx Supplemental Data 3. (Supplemental Figure 1) tiff Supplemental Data 4. (Supplemental Table 2) xlsx

30

ACCEPTED MANUSCRIPT TABLE 1. Patients’ characteristics

Sex

Pack-Year

Histology

EGFR mutation

P-Stage

LK001

78

M

116

AD

(-)

T1bN1M0-IIA

LK004

80

F

0

AD

exon19del

LK005

76

M

114

LCNEC

(-)

LK010

71

F

0

AD

E709G,L858R

LK012

66

M

46

SQ

(-)

T3N0M0-IIB

LK013

34

F

0

AD

exon19del

T1bN0M0-IA

LK029

78

M

60

LK044

73

F

0

LK047

41

M

21

LK049

65

M

111

LK051

70

M

LK053

59

F

LK056

81

M

LK059

75

LK060

65

RI PT

Age

T2aN2M0-IIIA T2bN1M0-IIB

M AN U

SC

T1bN2M0-IIIA

(-)

T1bN0M0-IA

AD

exon19del

T2aN0M0-IB

AD

L858R

T1bN2M0-IIIA

AD

(-)

T2aN1M0-IIA

112

AD

E709V,L858R

T2aN2M0-IIIA

0

AD

(-)

T2bN2M0-IIIA

56

AD

(-)

T2bNxM1-IV

M

28

AD

(-)

T2bN0M0-IIA

M

62

SQ

(-)

T2aN1M0-IIA

AC C

EP

TE D

SQ

LK066

81

M

80

SQ

(-)

T2bN0M0-IIA

LK070

67

M

71

SQ

(-)

T3N0M0-IIB

LK071

79

F

10

SQ

(-)

T1aN2M0-IIIA

LK073

67

M

40

AD

Q701L,L858R

T2aN0M0-IB

LK075

81

M

70

SQ

(-)

T1aN0M0-IA

M, male; F, female; AD, adenocarcinoma; LCNEC, large cell neuroendocrine carcinoma; SQ, squamous cell carcinoma

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT