Lung Cancer Cohort Research Articles

Prognostic model incorporating immune checkpoint genes to predict the immunotherapy efficacy for lung adenocarcinoma: a cohort study integrating machine learning algorithms.

This study aimed to develop and validate a nomogram based on immune checkpoint genes (ICGs) for predicting prognosis and immune checkpoint blockade (ICB) efficacy in lung adenocarcinoma (LUAD) patients. A total of 385 LUAD patients from the TCGA database and 269 LUAD patients in the combined dataset (GSE41272 + GSE50081) were divided into training and validation cohorts, respectively. Three different machine learning algorithms including random forest (RF), least absolute shrinkage and selection operator (LASSO) logistic regression analysis, and support vector machine (SVM) were employed to select the predictive markers from 82 ICGs to construct the prognostic nomogram. The X-tile software was used to stratify patients into high- and low-risk subgroups based on the nomogram-derived risk scores. Differences in functional enrichment and immune infiltration between the two subgroups were assessed using gene set variation analysis (GSVA) and various algorithms. Additionally, three lung cancer cohorts receiving ICB therapy were utilized to evaluate the ability of the model to predict ICB efficacy in the real world. Five ICGs were identified as predictive markers across all three machine learning algorithms, leading to the construction of a nomogram with strong potential for prognosis prediction in both the training and validation cohorts (all AUC values close to 0.800). The patients were divided into high- (risk score ≥ 185.0) and low-risk subgroups (risk score < 185.0). Compared to the high-risk subgroup, the low-risk subgroup exhibited enrichment in immune activation pathways and increased infiltration of activated immune cells, such as CD8 + T cells and M1 macrophages (P < 0.05). Furthermore, the low-risk subgroup had a greater likelihood of benefiting from ICB therapy and longer progression-free survival (PFS) than did the high-risk subgroup (P < 0.05) in the two cohorts receiving ICB therapy. A nomogram based on ICGs was constructed and validated to aid in predicting prognosis and ICB treatment efficacy in LUAD patients.

Systematic dissection of tumor-normal single-cell ecosystems across a thousand tumors of 30 cancer types

The complexity of the tumor microenvironment poses significant challenges in cancer therapy. Here, to comprehensively investigate the tumor-normal ecosystems, we perform an integrative analysis of 4.9 million single-cell transcriptomes from 1070 tumor and 493 normal samples in combination with pan-cancer 137 spatial transcriptomics, 8887 TCGA, and 1261 checkpoint inhibitor-treated bulk tumors. We define a myriad of cell states constituting the tumor-normal ecosystems and also identify hallmark gene signatures across different cell types and organs. Our atlas characterizes distinctions between inflammatory fibroblasts marked by AKR1C1 or WNT5A in terms of cellular interactions and spatial co-localization patterns. Co-occurrence analysis reveals interferon-enriched community states including tertiary lymphoid structure (TLS) components, which exhibit differential rewiring between tumor, adjacent normal, and healthy normal tissues. The favorable response of interferon-enriched community states to immunotherapy is validated using immunotherapy-treated cancers (n = 1261) including our lung cancer cohort (n = 497). Deconvolution of spatial transcriptomes discriminates TLS-enriched from non-enriched cell types among immunotherapy-favorable components. Our systematic dissection of tumor-normal ecosystems provides a deeper understanding of inter- and intra-tumoral heterogeneity.

Multicenter PET image harmonization using generative adversarial networks

PurposeTo improve reproducibility and predictive performance of PET radiomic features in multicentric studies by cycle-consistent generative adversarial network (GAN) harmonization approaches.MethodsGAN-harmonization was developed to harmonize whole-body PET scans to perform image style and texture translation between different centers and scanners. GAN-harmonization was evaluated by application to two retrospectively collected open datasets and different tasks. First, GAN-harmonization was performed on a dual-center lung cancer cohort (127 female, 138 male) where the reproducibility of radiomic features in healthy liver tissue was evaluated. Second, GAN-harmonization was applied to a head and neck cancer cohort (43 female, 154 male) acquired from three centers. Here, the clinical impact of GAN-harmonization was analyzed by predicting the development of distant metastases using a logistic regression model incorporating first-order statistics and texture features from baseline 18F-FDG PET before and after harmonization.ResultsImage quality remained high (structural similarity: left kidney ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 0.800, right kidney ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 0.806, liver ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 0.780, lung ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 0.838, spleen ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 0.793, whole-body ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 0.832) after image harmonization across all utilized datasets. Using GAN-harmonization, inter-site reproducibility of radiomic features in healthy liver tissue increased at least by ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 5 ± 14% (first-order), ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 16 ± 7% (GLCM), ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 19 ± 5% (GLRLM), ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 16 ± 8% (GLSZM), ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 17 ± 6% (GLDM), and ≥\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ge$$\\end{document} 23 ± 14% (NGTDM). In the head and neck cancer cohort, the outcome prediction improved from AUC 0.68 (95% CI 0.66–0.71) to AUC 0.73 (0.71–0.75) by application of GAN-harmonization.ConclusionsGANs are capable of performing image harmonization and increase reproducibility and predictive performance of radiomic features derived from different centers and scanners.

Correlation of Systemic Inflammation Parameters and Serum SLFN11 in Small Cell Lung Cancer-A Prospective Pilot Study.

The objective of this research was to analyze the correlation of the neutrophil-to-lymphocyte ratio (NLR), C-reactive protein (CRP), soluble programmed cell death ligand 1 (sPD-L1), and Schlafen 11 (SLFN11) with the response to first-line chemotherapy in a cohort of small cell lung cancer (SCLC) patients, and to determine their potential as predictive serum biomarkers. A total of 60 SCLC patients were included. Blood samples were taken to determine CRP, sPD-L1, and SLFN11 levels. The first sampling was performed before the start of chemotherapy, the second after two cycles, and the third after four cycles of chemotherapy. The patients who died earlier during the study had NLR and SLFN11 concentrations significantly higher compared to the survivor group. In the group of survivors, after two cycles of chemotherapy, the NLR ratio decreased significantly (p < 0.01), but after four cycles, the NLR ratio increased (p < 0.05). Their serum SLFN11 concentration increased significantly (p < 0.001) after two cycles of chemotherapy, but after four cycles, the level of SLFN11 fell significantly (p < 0.01). CRP, NLR, and SLFN11 were significant predictors of patient survival according to Kaplan-Meier analysis. The combination of inflammatory parameters and SLFN11 with a cutoff value above the 75th percentile of the predicted probability was associated with significantly lower overall survival in SCLC patients (average survival of 3.6 months vs. 4.8 months). The combination of inflammatory markers and the levels of two specific proteins (sPD-L1, SLFN11) could potentially serve as a non-invasive biomarker for predicting responses to DNA-damaging therapeutic agents in SCLC.

Development and evaluation of two open-source nnU-Net models for automatic segmentation of lung tumors on PET and CT images with and without respiratory motion compensation

ObjectivesIn lung cancer, one of the main limitations for the optimal integration of the biological and anatomical information derived from Positron Emission Tomography (PET) and Computed Tomography (CT) is the time and expertise required for the evaluation of the different respiratory phases. In this study, we present two open-source models able to automatically segment lung tumors on PET and CT, with and without motion compensation.Materials and methodsThis study involved time-bin gated (4D) and non-gated (3D) PET/CT images from two prospective lung cancer cohorts (Trials 108237 and 108472) and one retrospective. For model construction, the ground truth (GT) was defined by consensus of two experts, and the nnU-Net with 5-fold cross-validation was applied to 560 4D-images for PET and 100 3D-images for CT. The test sets included 270 4D- images and 19 3D-images for PET and 80 4D-images and 27 3D-images for CT, recruited at 10 different centres.ResultsIn the performance evaluation with the multicentre test sets, the Dice Similarity Coefficients (DSC) obtained for our PET model were DSC(4D-PET) = 0.74 ± 0.06, improving 19% relative to the DSC between experts and DSC(3D-PET) = 0.82 ± 0.11. The performance for CT was DSC(4D-CT) = 0.61 ± 0.28 and DSC(3D-CT) = 0.63 ± 0.34, improving 4% and 15% relative to DSC between experts.ConclusionsPerformance evaluation demonstrated that the automatic segmentation models have the potential to achieve accuracy comparable to manual segmentation and thus hold promise for clinical application. The resulting models can be freely downloaded and employed to support the integration of 3D- or 4D- PET/CT and to facilitate the evaluation of its impact on lung cancer clinical practice.Clinical relevance statementWe provide two open-source nnU-Net models for the automatic segmentation of lung tumors on PET/CT to facilitate the optimal integration of biological and anatomical information in clinical practice. The models have superior performance compared to the variability observed in manual segmentations by the different experts for images with and without motion compensation, allowing to take advantage in the clinical practice of the more accurate and robust 4D-quantification.Key PointsLung tumor segmentation on PET/CT imaging is limited by respiratory motion and manual delineation is time consuming and suffer from inter- and intra-variability.Our segmentation models had superior performance compared to the manual segmentations by different experts.Automating PET image segmentation allows for easier clinical implementation of biological information.

Identifying social determinants of health from clinical narratives: A study of performance, documentation ratio, and potential bias

ObjectiveTo develop a natural language processing (NLP) package to extract social determinants of health (SDoH) from clinical narratives, examine the bias among race and gender groups, test the generalizability of extracting SDoH for different disease groups, and examine population-level extraction ratio. MethodsWe developed SDoH corpora using clinical notes identified at the University of Florida (UF) Health. We systematically compared 7 transformer-based large language models (LLMs) and developed an open-source package – SODA (i.e., SOcial DeterminAnts) to facilitate SDoH extraction from clinical narratives. We examined the performance and potential bias of SODA for different race and gender groups, tested the generalizability of SODA using two disease domains including cancer and opioid use, and explored strategies for improvement. We applied SODA to extract 19 categories of SDoH from the breast (n = 7,971), lung (n = 11,804), and colorectal cancer (n = 6,240) cohorts to assess patient-level extraction ratio and examine the differences among race and gender groups. ResultsWe developed an SDoH corpus using 629 clinical notes of cancer patients with annotations of 13,193 SDoH concepts/attributes from 19 categories of SDoH, and another cross-disease validation corpus using 200 notes from opioid use patients with 4,342 SDoH concepts/attributes. We compared 7 transformer models and the GatorTron model achieved the best mean average strict/lenient F1 scores of 0.9122 and 0.9367 for SDoH concept extraction and 0.9584 and 0.9593 for linking attributes to SDoH concepts. There is a small performance gap (∼4%) between Males and Females, but a large performance gap (>16 %) among race groups. The performance dropped when we applied the cancer SDoH model to the opioid cohort; fine-tuning using a smaller opioid SDoH corpus improved the performance. The extraction ratio varied in the three cancer cohorts, in which 10 SDoH could be extracted from over 70 % of cancer patients, but 9 SDoH could be extracted from less than 70 % of cancer patients. Individuals from the White and Black groups have a higher extraction ratio than other minority race groups. ConclusionsOur SODA package achieved good performance in extracting 19 categories of SDoH from clinical narratives. The SODA package with pre-trained transformer models is available at https://github.com/uf-hobi-informatics-lab/SODA_Docker.

A Multicenter Study on the Challenges and Real-World Utilization of Immune Checkpoint Inhibitors in Resource-Constrained Settings: Insights and Implications from India

Using immune checkpoint inhibitors (ICIs) has revolutionized cancer treatment, but access and affordability remain significant challenges, particularly in resource-constrained settings. This multicenter study evaluated the utilization, outcomes, and challenges associated with ICIs in India. Data from multiple centers involving patients treated between January 2018 and December 2021 were retrospectively collected. Patient demographics, treatment indications, biomarker testing, financial coverage, toxicity, treatment discontinuation, clinical benefit, progression-free survival (PFS), and overall survival (OS) were analyzed. Ninety-one patients were analyzed; lung cancer (39.6%) and renal cancer (11%) were the main indications for ICI use. Programmed death ligand 1 expression was tested in 40.7% and tumor mutational burden in 3.3%. Financial constraints influenced 41.8% of patients with out-of-pocket expenses. Treatment discontinuation due to financial constraints occurred in 17.6%, with 50% showing ongoing responses. The median number of cycles was 4; the median PFS was 4.6 months, and the median OS was 15.4 months. The lung cancer cohort had a median PFS of 5.7 months and a 1-year OS of 57.6%. Limited biomarker testing and 6.6% grade ¾ toxicities were observed. This study revealed challenges in ICI utilization in resource-constrained settings driven by financial constraints. Compared with prior studies, improved outcomes reflect better patient selection and evolving understanding of ICI use. However, in the absence of biosimilars, cost remains a significant barrier. Solutions to increase access include using lower doses, which may be as effective.

PBMCs as Tool for Identification of Novel Immunotherapy Biomarkers in Lung Cancer.

Lung cancer (LC), including both non-small (NSCLC) and small (SCLC) subtypes, is currently treated with a combination of chemo- and immunotherapy. However, predictive biomarkers to identify high-risk patients are needed. Here, we explore the role of peripheral blood mononuclear cells (PBMCs) as a tool for novel biomarkers searching. We analyzed the expression of the cGAS-STING pathway, a key DNA sensor that activates during chemotherapy, in PBMCs from LC patients divided into best responders (BR), responders (R) and non-responders (NR). The PBMCs were whole exome sequenced (WES). PBMCs from BR and R patients of LC cohorts showed the highest levels of STING (p < 0.0001) and CXCL10 (p < 0.0001). From WES, each subject had at least 1 germline/somatic alteration in a DDR gene and the presence of more DDR gene mutations correlated with clinical responses, suggesting novel biomarker implications. Thus, we tested the effect of the pharmacological DDR inhibitor (DDRi) in PBMCs and in three-dimensional spheroid co-culture of PBMCs and LC cell lines; we found that DDRi strongly increased cGAS-STING expression and tumor infiltration ability of immune cells in NR and R patients. Furthermore, we performed FACS analysis of PBMCs derived from LC patients from the BR, R and NR cohorts and we found that cytotoxic T cell subpopulations displayed the highest STING expression. cGAS-STING signaling activation in PBMCs may be a novel potential predictive biomarker for the response to immunotherapy and high levels are correlated with a better response to treatment along with an overall increased antitumor immune injury.

Safety net hospital risk model demonstrates stronger, population-specific applicability in characterizing lung cancer risk.

Determining lung cancer (LC) risk using personalized risk stratification may improve screening effectiveness. While the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) is a well-established stratification model for LC screening, it was derived from a predominantly Caucasian population and its effectiveness in a safety net hospital (SNH) population is unknown. We have developed a model more tailored to the SNH population and compared its performance to the PLCO model in a SNH setting. Retrospective dataset was compiled from patients screened for LC at SNH from 2015 to 2019. Descriptive statistics were calculated using the following variables: age, sex, race, education, body mass index (BMI), smoking history, personal cancer history, family LC history, chronic obstructive pulmonary disease (COPD), and emphysema. Variables distribution was compared using t- and chi-square tests. LC risk scores were calculated using SNH and PLCO models and categorized as low (scores <0.65%), moderate (0.65-1.49%), and high (>1.5%). Linear regression was applied to evaluate the relationship between models and covariates. Of 896 individuals, 38 were diagnosed with LC. Data reflected the SNH patient demographics, which predominantly were African American (53.5%), current smokers (69.9%), and with emphysema (70.1%). Among the non-LC cohort, SNH model most frequently categorized patients as low risk, while PLCO model most frequently classified patients as moderate risk. Among the LC cohort, there was no significant difference between mean scores or risk stratification. SNH model showed 92.1% sensitivity and 96.8% specificity while PLCO model showed 89.4% sensitivity and 26.1% specificity. Emphysema demonstrated a strong association in SNH model (P<0.001) while race showed no relation. SNH model demonstrated greater specificity for characterizing LC risk in a SNH population. The results demonstrated the importance of study sample representation when identifying risk factors in a stratification model.

34 Predictive value of PD-L1 expression IHC on histology vs. cytology-specimen in non-small cell lung cancer (NSCLC) cohort in Leicester

34 Predictive value of PD-L1 expression IHC on histology vs. cytology-specimen in non-small cell lung cancer (NSCLC) cohort in Leicester

Carrying on with life as a lung cancer survivor: a qualitative study of Australian survivors' employment, finances, relationships, and healthcare experiences.

With novel therapies, more individuals are living longer with lung cancer (LC). This study aimed to understand the impacts of LC on life domains such as employment, finances, relationships, and healthcare needs. Individuals 18+, diagnosed with LC, 6-24 months post-treatment were recruited through an Australian LC cohort study (Embedding Research and Evidence in Cancer Healthcare-EnRICH). Demographic, clinical, quality-of-life and distress data were obtained through the EnRICH study database. Participants completed telephone interviews. Qualitative data were analysed via Framework methods. Twenty interviews (10 females) were conducted. Most participants were diagnosed with advanced LC (Stage III =8, Stage IV =6), and were on average 17 (range, 10-24) months post-diagnosis. Four themes related to "carrying on with life" as a LC survivor were identified: (I) the winding path back to work: those working pre-diagnosis discussed challenges of maintaining/returning to employment, and the meaning and satisfaction derived from work. (II) Vulnerability versus protection: managing the financial impacts of LC: wide variations in financial impacts, some described lost income and high healthcare expenses, others felt financially protected. (III) Connection and loneliness: navigating relationships as a survivor: some experienced lost friendships due to their diagnosis, others noted more meaningful connections. (IV) Still under the umbrella: healthcare during survivorship: participants noted the importance of ongoing oncology team connection and the vital role of cancer care coordinators. Many individuals living with LC want to "carry on" with life. Participants spoke of challenges and opportunities across life domains of relationships, work, and finances, and noted the importance of continued specialist healthcare throughout survivorship.

Wemics: A Single-Base Resolution Methylation Quantification Method for Enhanced Prediction of Epigenetic Regulation.

DNA methylation, an epigenetic mechanism that alters gene expression without changing DNA sequence, is essential for organism development and key biological processes like genomic imprinting and X-chromosome inactivation. Despite tremendous efforts in DNA methylation research, accurate quantification of cytosine methylation remains a challenge. Here, a single-base methylation quantification approach is introduced by weighting methylation of consecutive CpG sites (Wemics) in genomic regions. Wemics quantification of DNA methylation better predicts its regulatory impact on gene transcription and identifies differentially methylated regions (DMRs) with more biological relevance. Most Wemics-quantified DMRs in lung cancer are epigenetically conserved and recurrently occurred in other primary cancers from The Cancer Genome Atlas (TCGA), and their aberrant alterations can serve as promising pan-cancer diagnostic markers. It is further revealed that these detected DMRs are enriched in transcription factor (TF) binding motifs, and methylation of these TF binding motifs and TF expression synergistically regulate target gene expression. Using Wemics on epigenomic-transcriptomic data from the large lung cancer cohort, a dozen novel genes with oncogenic potential are discovered that are upregulated by hypomethylation but overlooked by other quantification methods. These findings increase the understanding of the epigenetic mechanism by which DNA methylation regulates gene expression.

Abstract 2490: Quantification of tumor fraction and outcomes association in a real-world non-small cell lung cancer (NSCLC) cohort using a tissue agnostic epigenomic circulating tumor DNA (ctDNA) assay

Abstract Background: The validity of ctDNA assays for evaluating molecular response to therapy by measuring changes in variant allele fraction (VAF) while on treatment is well established, but they are prone to limitations such as low ctDNA levels and interference from copy number variation and clonal hematopoeisis (CH), which may be overcome by methylation-based quantification. ctDNA levels may be monitored throughout a patient's (pts) journey to indicate when relapse or progression is present, often sooner than current methods (RECIST). Here we describe the performance of a methylation-based ctDNA assay to quantify ctDNA levels and correlate changes with outcomes in a real-world NSCLC cohort. Methods: RADIOHEAD is a pan-tumor observational study of 1200 patients on standard of care ICI treatment regimens with plasma samples collected prospectively for retrospective analysis. We randomly selected 241 NSCLC pts that have a combined total of 761 samples (236 baseline, 191 C3D1, and 128 at 6 months post first dose). Samples were analyzed using a commercial NGS assay, which quantifies circulating tumor fraction (cTF) via interrogating thousands of methylation sites from ctDNA, validated with LoB, LoD, LoQ, and linearity studies. Cox proportional hazards (CPH) were used for comparison of real-world progression free survival (rwPFS). Gender, age, disease stage, and blood-assessed tumor mutational burden (bTMB) were included as covariates. Median rwPFS was calculated using Kaplan Meier analysis. rwPFS ordinal groups were defined as PFS event within 3mo, 3-6mo, 6-12mo, and greater than 12mo, and association with early ctDNA changes was assessed with a chi-squared test. Results: In this cohort, 4, 11, 83 and 142 pts were stage I to IV, respectively (1 unknown). Methylation based detection of ctDNA at baseline or C3D1 was associated with shorter rwPFS (baseline: mPFS 12.7 mo [9.3-19.5] vs NR [16.4-NR]; HR=2.2 [1.3-3.7] p&amp;lt;0.005, C3D1: mPFS 12.4 mo [9.3-16.1] vs NR [19.5-NR] HR= 2.2 [1.3-3.5], p&amp;lt;0.005) independent of stage, gender, or age. Longer rwPFS was associated with &amp;gt;95% reduction in cTF from baseline to C3D1 or low ctDNA at both timepoints (rwPFS &amp;lt;3mo = 12.5%, 3-6 mo = 30.8%, 6-12 = 37.1%, ≥12 mo = 55.3%; N=153, p=0.015). For pts with ctDNA not detected at C3D1, ctDNA detection at 6 months post first dose was strongly associated with shorter rwPFS (mPFS 13.5 mo [9.5-NR] vs. NR [19.5-NR]; HR= 6.2 [1.9-20.1], p&amp;lt;0.005). Conclusions: These data demonstrate a significant association of ctDNA detection via methylation-based cTF and on-treatment changes in cTF with rwPFS. Furthermore, subsequent cTF detection 6-months on-treatment, without cTF detected early on-treatment was associated with worse outcomes, suggesting the value of serial monitoring to improve identification of patients likely to progress. Citation Format: Sara Wienke, Sean Gordon, Samantha Liang, Jing Wang, Reagan Barnett, Kyle Chang, Shile Zhang, Carin R. Espenschied, Katie Quinn, Kimberly Banks. Quantification of tumor fraction and outcomes association in a real-world non-small cell lung cancer (NSCLC) cohort using a tissue agnostic epigenomic circulating tumor DNA (ctDNA) assay [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2490.

Abstract 3634: Copy number alterations in chromosome 19 genes associate with reduced immune infiltrate and adverse outcomes in lung cancer

Abstract Deleterious mutations in STK11/LKB1, KEAP1, and SMARCA4 have been shown to be associated with adverse response to immunotherapy in several lung cancer cohorts. However, the biological and clinical implications of deletions of these genes, which all reside on chromosome 19p (chr19p), has not been extensively described. We sought to first characterize the association between immune infiltration and these copy number changes in chr19p. To do this, we analyzed mutation, copy number, and gene expression data from non-small cell lung cancer cohorts in the Cancer Genome Atlas (TCGA). Chr19p is frequently deleted across cancers in the TCGA dataset, with higher rates in lung adenocarcinoma. In both lung adenocarcinoma and squamous cell carcinoma, nearly half of TCGA tumors have deletion of chr19p including STK11, KEAP1, and SMARCA4, and these tumors are enriched for STK11 mutation. Immune infiltrate (estimated by methylation patterns) was not significantly associated with STK11 mutation but was significantly reduced in tumors with chr19 deletion when controlling for tumor type and overall copy number load. In addition, low-level deletion of any of these genes correlated with decreased expression of immune signaling pathways. Based on these findings, we next wanted to characterize the association between chr19p copy number and immunotherapy response. We previously reported that low-level deletion of STK11 correlated with an immunosuppressive, treatment-refractory phenotype in samples collected from non-small cell lung cancer (NSCLC) patients with resectable disease in a phase II study of neoadjuvant atezolizumab + chemotherapy. To confirm these findings in a larger cohort, we examined the AACR GENIE NSCLC biopharma collective (BPC) dataset. A subset of this cohort included patients treated with immunotherapy (atezolizumab, nivolumab, and/or pembrolizumab) and with copy number data available or GISTIC analysis of individual genes (n = 204). Here, deletion of any of these three genes correlated with worse overall survival, and this was statistically significant for STK11 deletion. Overall, our data suggests that a broader set of patients, defined by copy number changes in these genes, may experience this adverse immunobiology. Further study to elucidate the mechanistic implications of this association is warranted. Citation Format: Brian S. Henick, Yohanna Georgis, Benjamin O. Herzberg, Carla P. Concepcion-Crisol, Alison M. Taylor. Copy number alterations in chromosome 19 genes associate with reduced immune infiltrate and adverse outcomes in lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3634.

Abstract 1169: Increased immunotherapy sensitivity associated with damaging IFN-gamma pathway mutations in NSCLC

Abstract Introduction: In vivo CRISPR screens across multiple murine models of malignancy revealed that loss of interferon gamma (IFN-γ)-mediated sensing predisposes to immune elimination. We assessed the association between damaging mutations in the IFN-γ pathway and clinical outcomes after immune checkpoint inhibition in the Stand Up to Cancer-Mark Foundation non-small cell lung cancer (NSCLC) cohort. Methods: The sensitizing IFN-γ pathway genes were defined as IFNGR1, IFNGR2, JAK1, JAK2, STAT1, IRF1 &amp; IRF9. Damaging mutations (including missense mutations) in any of these genes was considered as loss of pathway activity. We used the best observed response (measured by RECIST v1.1 criteria across time points) to define outcome. Complete and partial response were categorized as responses (R), while stable and progressive disease were categorized as non-responses (NR). The percentage change in target lesions (CTL) was also assessed. Response data was analyzed using additive logistic (LogR) and linear regression (LinR). Overall survival (OS) and progression-free survival (PFS) were analyzed using Cox proportional-hazards models. Three predictor variable groups were established; the final models comprised IFN-γ pathway loss, tumor mutational burden (TMB), programmed death-ligand 1 (PD-L1) tumor proportion score (TPS), prior platinum chemotherapy, age, sex, histological subtype, initial tumor stage, and smoking pack years. Results: 309 patients had evaluable whole-exome sequencing data, treatment with PD-(L)1 inhibitor regimens, and matched radiological responses (121 R [39%], 188 NR [61%]); 306 had matched OS data; 297 had matched PFS data; and 227 had matched CTL data. 27 (7%) patients had damaging mutations in the IFN-γ pathway [IFNGR1 (3); IFNGR2 (1); JAK1 (6); JAK2 (8); STAT1 (3); IRF1 (1); IRF9 (3); and JAK2 &amp; IRF9 (1)]. Significant relationships were observed in univariable models between IFN-γ pathway loss and both response (LogR: OR 0.17, p=2 × 10−4*; LinR: OR 0.70, p=4.8 × 10−4*) and survival (PFS: HR 0.45, p=0.0067*; OS: HR 0.51, p=0.037*). IFN-γ pathway loss is significantly associated with TMB (p=9.7 × 10−8*, Wilcoxon signed rank test). Multivariable models incorporating TMB and PD-L1 TPS confirmed the significance of IFN-γ pathway loss in both response (LogR: OR 0.29, p=0.050*; LinR: OR 0.71, p=0.0052*) and survival (PFS: HR 0.37, p=0.036*; OS: HR 0.27, p=0.030*). These relationships persisted in the final multivariable models (LogR: OR 0.32, p=0.11; LinR: OR 0.69, p=0.012*; PFS: HR 0.39, p=0.068; OS: HR 0.15, p=0.0038*). Random gene sets, of similar total protein coding sequence length, significantly associated with TMB but not with multivariable models of response or survival. Conclusion: In this NSCLC cohort, damaging IFN-γ pathway mutations associate with increased immunotherapy sensitivity, independently of TMB and other co-variates, as evidenced by multiple clinical outcomes. Citation Format: Alexander Azizi, Arvind Ravi, Natalie Vokes, Stephen-John Sammut, Justin Gainor, Gad Getz. Increased immunotherapy sensitivity associated with damaging IFN-gamma pathway mutations in NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1169.

Abstract 1200: A novel algorithm for deconvolving cancer allele-specific clone copy number and copy number evolution

Abstract Both single nucleotide variants (SNVs) and somatic copy number alterations (SCNAs) accumulate in cancer cells during tumor development, fuelling clonal evolution. However, accurate inference of their coevolution from bulk DNA sequencing is challenging. We present ALPACA (ALlele-specific Phylogenetic Analysis of clone Copy number Alterations), a novel algorithm to allow the practical inference of SNV and SCNA coevolution. ALPACA is framed as an optimisation problem, that takes as input bulk tumor sample copy number mixtures and the tumor phylogenetic tree constructed from SNVs, and deconvolves the optimal, allele-specific integer copy number profiles for every distinct clone present in the tumor. To circumvent the challenge of joint SNV/SCNA inference, we postulate that in high mutation burden tumors all relevant SCNA clones are identifiable by their unique SNVs, hence ALPACA leverages phylogenetic trees constructed from bulk tumor sequencing data using SNVs. Secondly, to restrict the clone-specific copy number search space, ALPACA leverages constraints imposed by multisample sequencing, specifically in cases where clones are present across multiple samples or clones from different samples are phylogenetically related. Finally, ALPACA relies on parsimony and biologically informed constraints to further guide the deconvolution process. We demonstrate that ALPACA infers the copy number evolution of complex simulated tumors with higher accuracy than current state-of-the-art methods. We apply ALPACA to the large, multisample non-small cell lung cancer (NSCLC) cohort from the recent longitudinal, prospective TRACERx421 study, and show that ALPACA uncovers loss of heterozygosity and amplification events in minor subclones that were previously missed using standard approaches. ALPACA's assignment of SCNA driver events to branches of the phylogenetic tree reveals the evolutionary ordering of SNVs and SCNAs during NSCLC tumor evolution, such as late focal amplifications affecting the TERT gene locus occurring in subclones which expand and dominate a tumor sample. ALPACA enables an in-depth analysis of the coevolution of mutations and SCNAs in the tumor’s evolutionary history, revealing distinct patterns of copy number evolution. Notably, ALPACA identifies common patterns of copy number changes across the genome characterizing metastatic seeding clones, revealing that they harbor an increased number of SCNAs compared to clones that do not metastasize. Additionally, ALPACA uncovers subclonal enrichment for CCND1 amplification in primary tumor subclones that seed metastasis, and an overall increase of SCNA events occurring in both tumors with polyclonal seeding and extrathoracic metastases. Clone-level results obtained with ALPACA can offer new clinical insights and enable new types of analysis, e.g. copy number signature analysis including temporal order of SCNA acquisition. Citation Format: Piotr Pawlik, Kristiana K. Grigoriadis, Abigail Bunkum, Simone Zaccaria, Nicholas McGranahan, Charles Swanton. A novel algorithm for deconvolving cancer allele-specific clone copy number and copy number evolution [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1200.

Abstract 6095: Lung cancer risk assessment with the INTEGRAL protein panel: Preliminary results from development and validation in the Lung Cancer Cohort Consortium

Abstract Background: In the NCI-funded Integrative Analysis of Lung Cancer Etiology and Risk (INTEGRAL) program, the Lung Cancer Cohort Consortium (LC3) recently identified 36 proteins associated with lung cancer risk in pre-diagnostic blood samples after evaluating 1,161 proteins in 731 case-control pairs (Nature Communications, 2023). Based on these data, we designed the INTEGRAL panel, which measures absolute concentrations of 21 proteins, to optimize lung cancer risk assessment for screening beyond prediction models (e.g. PLCOm2012) or categorical screening criteria (e.g. USPSTF-2021). Methods: To evaluate the performance of the INTEGRAL panel, we assayed blood from participants selected as a representative case-cohort sample from 14 LC3 cohorts, divided into development and validation sets (Annals of Epidemiology, 2023). Here, we present preliminary results from the 7 cohorts in the development set, including 807 lung cancer cases diagnosed within 3 years of blood draw and 1,144 sub-cohort representatives, who were weighted to represent all current and former smokers in their corresponding cohorts. We first fit a flexible parametric survival model in 4 cohorts (n=478 cases) including a subset of proteins and age, smoking duration, and smoking intensity. The model was subsequently evaluated in the remaining 3 cohorts (n=329 cases) and benchmarked against the PLCOm2012 risk score and USPSTF2021 screening criteria. Results: The preliminary model includes 4 proteins (CEACAM5, MMP12, SCF, LPL). Compared with the PLCOm2012 score, the model improved discrimination of future lung cancer for cases occurring over 3 years (AUC=0.81 vs. 0.76, p&amp;lt;0.0001) and over 1 year (AUC=0.86 vs. 0.78, p&amp;lt;0.0001). At the USPSTF-2021 specificity of 76%, the protein model increased sensitivity by an absolute difference of 14% over 3 years and 25% over 1 year compared to USPSTF-2021, and by 9% over 3 years and 17% over 1 year compared to the PLCOm2012 model. When screening the same number of participants as PLCOm2012, the protein model identified more future cases (223 vs 193), and the cases identified only by the protein model were predicted to gain more life-years from 3 LDCT screens compared with the PLCOm2012-identified cases (4.2 vs. 3.6 life-years). Conclusions: Preliminary results suggest that the INTEGRAL protein panel can improve risk discrimination beyond questionnaire-based risk prediction models, as well as identify future cases who have more life-years to gain from screening. Final validation is planned for completion in early 2024, by analysis of fully independent data from 7 additional LC3 cohorts (n=560 cases and 1,126 sub-cohort representatives). Citation Format: Hana Zahed, Karine Alcala, David C. Muller, Rayjean Hung, Mattias Johansson, Hilary A. Robbins, The Lung Cancer Cohort Consortium. Lung cancer risk assessment with the INTEGRAL protein panel: Preliminary results from development and validation in the Lung Cancer Cohort Consortium [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6095.

Abstract 5260: Combination activity of eribulin liposomal formulation and anti-PD-1 antibody after tumor regrowth during anti-PD-1 antibody treatment in mice

Abstract Eribulin (ERI) has been reported a microtubule dynamics inhibitor with unique tumor microenvironment modulations such as vascular remodeling activity. In the previous meeting (AACR2022, 2023), we reported that ERI and liposomal formulation of ERI (ERI-LF) also have immunomodulatory activity that induces CD8+ T cells via its vascular remodeling activity and ERI-LF showed more potent immune modulation and combination activities with anti-PD-1 antibody (PD-1 Ab) than ERI in syngeneic mice models and humanized mice models. Although immune-checkpoint inhibitors (ICI) are using as a key drug for various types of cancers in clinical setting, several clinical questions regarding suitable therapies after ICI therapy and efficacy of ICI rechallenge are remaining. In this study, we evaluated anti-tumor activities of combination of ERI-LF and PD-1 Ab in two syngeneic transplantation models using mouse non-small lung cancer cells, P-glycoprotein-knockout (Pgp-KO) KLN205 cells and Pgp-KO LL2 cells. As a result, combination of 1mg/kg ERI-LF (Q7D×2) with PD-1 Ab (200 ug/head, twice a week ×2 weeks) showed significant stronger antitumor activity compared with each monotherapy in only Pgp-KO KLN205 model, in which ERI-LF increased intratumoral microvascular density as a vascular remodeling activity, not in Pgp-KO LL2 model. We next conducted continuous treatment of PD-1 Ab in Pgp-KO KLN205 model to investigate tumor regrowth during treatment of PD-1 Ab. In this model, tumor growth was inhibited by PD-1 Ab until about two weeks after initiation of the treatment and then this inhibitory activity disappeared after two weeks even by continuous treatment of PD-1 Ab. We performed flow cytometry analysis (FCM) for tumor-infiltrating immune cells and RNAseq analysis using regrowing tumors compared with non-treatment tumors. Expressions of mRNA of several immune inhibitory receptors were upregulated in tumors with regrowth, although CD4+ cells and CD8+ cells were also increased in FCM using tumor samples. Furthermore, we investigated anti-tumor activity of combination of ERI-LF and PD-1 Ab against tumors which changed from inhibitory phase to regrowth phase during PD-1 Ab treatment in Pgp-KO KLN205 model. The combination of ERI-LF at 1mg/kg (Q7D×3) and PD-1 Ab (200 ug/head, twice a week ×3 weeks) also showed potent anti-tumor activity against tumors with regrowth by prior PD-1 Ab treatment as well as PD-1 Ab-naïve tumor, suggesting that this combination therapy might be effective after prior ICI therapy. Currently, Phase 1b/2 clinical trial of ERI-LF plus nivolumab in patients with selected solid cancers (NCT04078295) is underway. Patients with or without prior ICI therapy were enrolled in a small cell lung cancer cohort of this clinical study. Our preclinical results might provide a scientific rationale for a uniqueness of ERI-LF as a post ICI therapy. Citation Format: Moe Tamura, Yuki Niwa, Taro Semba. Combination activity of eribulin liposomal formulation and anti-PD-1 antibody after tumor regrowth during anti-PD-1 antibody treatment in mice [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 5260.

Abstract 4966: Machine learning and large language model approach to pancancer data elements

Abstract Introductory Statement: The goal is to use machine learning (ML) and large language model (LLM) to augment the manual curation of cancer data elements. Introduction: Memorial Sloan Kettering Cancer Center (MSKCC) has ~100,000 cancer patients and counting with genomic testing. Clinicians use genomic data for research but lack clinical data to analyze together. We use a vendor, VASTA Global to hire curators to manually curate cancer patient’s core clinical data elements (CCDE) within unstructured/paragraph text in electronic medical record (EMR) notes. CCDE encompasses 122 data elements that include a patient’s full cancer history that can take up to 1 working day to curate. We collaborated with the Realyze Intelligence Healthcare Solutions vendor to use their AI pipeline to generate the manual curated dataset. Realyze generated the CCDE data elements such as histology, pathology site, MMR, TNM staging, ECOG, and KPS for a pilot lung cancer cohort of 150 patients. We manually validated the generated data for 74 out of 150 patients. Methods:The Realyze platform uses a combination of LLMs, ML algorithms and standard terminologies to create a cancer patient model. These models are flexible enough to address the unique needs and challenges of a pan-cancer oncology model. By using standardized FHIR export, results were delivered to a data lake solution and written into a REDCap database to enable human review. Summary:We manually assessed 74 patients. The NLP gave concordant values for MMR, KPS and TNM staging for 100% of the instances. For MMR these were all null values with false negative (FN) of 100% accuracy. Pathology site had 92.15% accuracy while histology has 97.5% accuracy. Conclusion:Will work on refining pathology site and histology’s ICDO3 list to increase the percentage of accuracy. Once Realyze refines their model for these data elements we will re-run it on a larger cohort of cancer patients and calculate the accuracy. Accuracy Results Clinical data elements 74 patients assessed: Accuracy % ECOG 98.6 KPS 100 T (path) 100 T (clinical) 100 N (path) 100 N (clinical) 100 M (path) 100 M(clinical) 100 MMR 100 Histology (path) 97.5 Path site 92.15 Citation Format: Andrew Niederhausern, Nadia S. Bahadur, Gary Wallace, Gilan E. Saadawi, John Philip. Machine learning and large language model approach to pancancer data elements [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4966.

DeepLION2: deep multi-instance contrastive learning framework enhancing the prediction of cancer-associated T cell receptors by attention strategy on motifs.

T cell receptor (TCR) repertoires provide valuable insights into complex human diseases, including cancers. Recent advancements in immune sequencing technology have significantly improved our understanding of TCR repertoire. Some computational methods have been devised to identify cancer-associated TCRs and enable cancer detection using TCR sequencing data. However, the existing methods are often limited by their inadequate consideration of the correlations among TCRs within a repertoire, hindering the identification of crucial TCRs. Additionally, the sparsity of cancer-associated TCR distribution presents a challenge in accurate prediction. To address these issues, we presented DeepLION2, an innovative deep multi-instance contrastive learning framework specifically designed to enhance cancer-associated TCR prediction. DeepLION2 leveraged content-based sparse self-attention, focusing on the top k related TCRs for each TCR, to effectively model inter-TCR correlations. Furthermore, it adopted a contrastive learning strategy for bootstrapping parameter updates of the attention matrix, preventing the model from fixating on non-cancer-associated TCRs. Extensive experimentation on diverse patient cohorts, encompassing over ten cancer types, demonstrated that DeepLION2 significantly outperformed current state-of-the-art methods in terms of accuracy, sensitivity, specificity, Matthews correlation coefficient, and area under the curve (AUC). Notably, DeepLION2 achieved impressive AUC values of 0.933, 0.880, and 0.763 on thyroid, lung, and gastrointestinal cancer cohorts, respectively. Furthermore, it effectively identified cancer-associated TCRs along with their key motifs, highlighting the amino acids that play a crucial role in TCR-peptide binding. These compelling results underscore DeepLION2's potential for enhancing cancer detection and facilitating personalized cancer immunotherapy. DeepLION2 is publicly available on GitHub, at https://github.com/Bioinformatics7181/DeepLION2, for academic use only.