Single-cell spatial landscapes of the lung tumour immune microenvironment.

Single-cell technologies have revealed the complexity of the tumour immune microenvironment with unparalleled resolution1-9. Most clinical strategies rely on histopathological stratification of tumour subtypes, yet the spatial context of single-cell phenotypes within these stratified subgroups is poorly understood. Here we apply imaging mass cytometry to characterize the tumour and immunological landscape of samples from 416 patients with lung adenocarcinoma across five histological patterns. We resolve more than 1.6 million cells, enabling spatial analysis of immune lineages and activation states with distinct clinical correlates, including survival. Using deep learning, we can predict with high accuracy those patients who will progress after surgery using a single 1-mm2 tumour core, which could be informative for clinical management following surgical resection. Our dataset represents a valuable resource for the non-small cell lung cancer research community and exemplifies the utility of spatial resolution within single-cell analyses. This study also highlights how artificial intelligence can improve our understanding of microenvironmental features that underlie cancer progression and may influence future clinical practice.

Multi-institutional prognostic modeling of survival outcomes in NSCLC patients treated with first-line immunotherapy using radiomics.

BACKGROUND: Immune checkpoint inhibitors (ICIs) have emerged as one of the most promising first-line therapeutics in the management of non-small cell lung cancer (NSCLC). However, only a subset of these patients responds to ICIs, highlighting the clinical need to develop better predictive and prognostic biomarkers. This study will leverage pre-treatment imaging profiles to develop survival risk models for NSCLC patients treated with first-line immunotherapy. METHODS: Advanced NSCLC patients (n = 149) were retrospectively identified from two institutions who were treated with first-line ICIs. Radiomics features extracted from pretreatment imaging scans were used to build the predictive models for progression-free survival (PFS) and overall survival (OS). A compendium of five feature selection methods and seven machine learning approaches were utilized to build the survival risk models. The concordance index (C-index) was used to evaluate model performance. RESULTS: From our results, we found several combinations of machine learning algorithms and feature selection methods to achieve similar performance. K-nearest neighbourhood (KNN) with ReliefF (RL) feature selection was the best-performing model to predict PFS (C-index = 0.61 and 0.604 in discovery and validation cohorts), while XGBoost with Mutual Information (MI) feature selection was the best-performing model for OS (C-index = 0.7 and 0.655 in discovery and validation cohorts). CONCLUSION: The results of this study highlight the importance of implementing an appropriate feature selection method coupled with a machine learning strategy to develop robust survival models. With further validation of these models on external cohorts when available, this can have the potential to improve clinical decisions by systematically analyzing routine medical images.

Advances in personalized radiotherapy.

Radiotherapy is a mainstay of cancer treatment. The clinical response to radiotherapy is heterogeneous, from a complete response to early progression. Recent studies have explored the importance of patient characteristics in response to radiotherapy. In this editorial, we invite contributions for a BMC Cancer collection of articles titled 'Advances in personalized radiotherapy' towards the improvement of treatment response.

Gene expression signature predicts radiation sensitivity in cell lines using the integral of dose-response curve.

BACKGROUND: Although substantial efforts have been made to build molecular biomarkers to predict radiation sensitivity, the ability to accurately stratify the patients is still limited. In this study, we aim to leverage large-scale radiogenomics datasets to build genomic predictors of radiation response using the integral of the radiation dose-response curve. METHODS: Two radiogenomics datasets consisting of 511 and 60 cancer cell lines were utilized to develop genomic predictors of radiation sensitivity. The intrinsic radiation sensitivity, defined as the integral of the dose-response curve (AUC) was used as the radioresponse variable. The biological determinants driving AUC and SF2 were compared using pathway analysis. To build the predictive model, the largest and smallest datasets consisting of 511 and 60 cancer cell lines were used as the discovery and validation cohorts, respectively, with AUC as the response variable. RESULTS: Utilizing a compendium of three pathway databases, we illustrated that integral of the radiobiological model provides a more comprehensive characterization of molecular processes underpinning radioresponse compared to SF2. Furthermore, more pathways were found to be unique to AUC than SF2-30, 288 and 38 in KEGG, REACTOME and WIKIPATHWAYS, respectively. Also, the leading-edge genes driving the biological pathways using AUC were unique and different compared to SF2. With regards to radiation sensitivity gene signature, we obtained a concordance index of 0.65 and 0.61 on the discovery and validation cohorts, respectively. CONCLUSION: We developed an integrated framework that quantifies the impact of physical radiation dose and the biological effect of radiation therapy in interventional pre-clinical model systems. With the availability of more data in the future, the clinical potential of this signature can be assessed, which will eventually provide a framework to integrate genomics into biologically-driven precision radiation oncology.

Development and validation of genomic predictors of radiation sensitivity using preclinical data.

BACKGROUND: Radiation therapy is among the most effective and commonly used therapeutic modalities of cancer treatments in current clinical practice. The fundamental paradigm that has guided radiotherapeutic regimens are 'one-size-fits-all', which are not in line with the dogma of precision medicine. While there were efforts to build radioresponse signatures using OMICS data, their ability to accurately predict in patients is still limited. METHODS: We proposed to integrate two large-scale radiogenomics datasets consisting of 511 with 23 tissues and 60 cancer cell lines with 9 tissues to build and validate radiation response biomarkers. We used intrinsic radiation sensitivity, i.e., surviving fraction of cells (SF2) as the radiation response indicator. Gene set enrichment analysis was used to examine the biological determinants driving SF2. Using SF2 as a continuous variable, we used five different approaches, univariate, rank gene ensemble, rank gene multivariate, mRMR and elasticNet to build genomic predictors of radiation response through a cross-validation framework. RESULTS: Through the pathway analysis, we found 159 pathways to be statistically significant, out of which 54 and 105 were positively and negatively enriched with SF2. More importantly, we found cell cycle and repair pathways to be enriched with SF2, which are inline with the fundamental aspects of radiation biology. With regards to the radiation response gene signature, we found that all multivariate models outperformed the univariate model with a ranking based approach performing well compared to other models, indicating complex biological processes underpinning radiation response. CONCLUSION: To summarize, we found biological processes underpinning SF2 and systematically compared different machine learning approaches to develop and validate predictors of radiation response. With more patient data available in the future, the clinical value of these biomarkers can be assessed that would allow for personalization of radiotherapy.

Transcriptome-wide association study reveals candidate causal genes for lung cancer.

We have recently completed the largest GWAS on lung cancer including 29,266 cases and 56,450 controls of European descent. The goal of our study has been to integrate the complete GWAS results with a large-scale expression quantitative trait loci (eQTL) mapping study in human lung tissues (n = 1,038) to identify candidate causal genes for lung cancer. We performed transcriptome-wide association study (TWAS) for lung cancer overall, by histology (adenocarcinoma, squamous cell carcinoma and small cell lung cancer) and smoking subgroups (never- and ever-smokers). We performed replication analysis using lung data from the Genotype-Tissue Expression (GTEx) project. DNA damage assays were performed in human lung fibroblasts for selected TWAS genes. As expected, the main TWAS signal for all histological subtypes and ever-smokers was on chromosome 15q25. The gene most strongly associated with lung cancer at this locus using the TWAS approach was IREB2 (pTWAS = 1.09E-99), where lower predicted expression increased lung cancer risk. A new lung adenocarcinoma susceptibility locus was revealed on 9p13.3 and associated with higher predicted expression of AQP3 (pTWAS = 3.72E-6). Among the 45 previously described lung cancer GWAS loci, we mapped candidate target gene for 17 of them. The association AQP3-adenocarcinoma on 9p13.3 was replicated using GTEx (pTWAS = 6.55E-5). Consistent with the effect of risk alleles on gene expression levels, IREB2 knockdown and AQP3 overproduction promote endogenous DNA damage. These findings indicate genes whose expression in lung tissue directly influences lung cancer risk.

Transcriptomic data helps refining classification of pulmonary carcinoid tumors with increased mitotic counts.

Pulmonary neuroendocrine neoplasms are classified by WHO as either typical or atypical carcinoids, large cell (LCNEC) or small cell (SCLC) neuroendocrine carcinoma based on mitotic count, morphology, and necrosis assessment. LCNEC with low mitotic count and sharing morphologic features with carcinoids are in a gray zone for classification and their rare prevalence and the paucity of studies precludes proper validation of the current grading system. In this study, we aim to investigate their clinicopathological and transcriptomic profiles. Lung resection specimens obtained from 18 patients diagnosed with carcinoids or LCNEC were selected. Four of them were characterized as borderline tumors based on a mitotic rate ranging between 10 and 30 mitoses per 2 mm2. Comprehensive morphological and immunohistochemical (IHC) evaluation was performed and tumor-based transcriptomic profiles were analyzed through unsupervised clustering. Clustering analysis revealed two distinct molecular groups characterized by low (C1) and high (C2) proliferation. C1 was comprised of seven carcinoids and three borderline tumors, while C2 was comprised of seven LCNEC and one borderline tumor. Furthermore, patients in cluster C1 had a better recurrence-free survival compared with patients in cluster C2 (20% vs 75%). Histological features, IHC profile, and molecular analysis showed that three out of four borderline tumors showed features consistent with carcinoids. Therefore, our findings convey that the current diagnostic guidelines are suboptimal for classification of pulmonary neuroendocrine tumors with increased proliferative index and carcinoid-like morphology. These results support the emerging concept that neuroendocrine tumors with carcinoid-like features and mitotic count of <20 mitoses per 2 mm2 should be regarded as pulmonary carcinoids instead of LCNEC.

Predictive approaches for drug combination discovery in cancer.

Drug combinations have been proposed as a promising therapeutic strategy to overcome drug resistance and improve efficacy of monotherapy regimens in cancer. This strategy aims at targeting multiple components of this complex disease. Despite the increasing number of drug combinations in use, many of them were empirically found in the clinic, and the molecular mechanisms underlying these drug combinations are often unclear. These challenges call for rational, systematic approaches for drug combination discovery. Although high-throughput screening of single-agent therapeutics has been successfully implemented, it is not feasible to test all possible drug combinations, even for a reduced subset of anticancer drugs. Hence, in vitro and in vivo screening of a large number of drug combinations are not practical. Therefore, devising computational methods to efficiently explore the space of drug combinations and to discover efficacious combinations has attracted a lot of attention from the scientific community in the past few years. Nevertheless, in the absence of consensus regarding the computational approaches used to predict efficacious drug combinations, a plethora of methods, techniques and hypotheses have been developed to date, while the research field lacks an elaborate categorization of the existing computational methods and the available data sources. In this manuscript, we review and categorize the state-of-the-art computational approaches for drug combination prediction, and elaborate on the limitations of these methods and the existing challenges. We also discuss about the recent pan-cancer drug combination data sets and their importance in revising the available methods or developing more performant approaches.

SIGN: similarity identification in gene expression.

MOTIVATION: High-throughput molecular profiles of human cells have been used in predictive computational approaches for stratification of healthy and malignant phenotypes and identification of their biological states. In this regard, pathway activities have been used as biological features in unsupervised and supervised learning schemes. RESULTS: We developed SIGN (Similarity Identification in Gene expressioN), a flexible open-source R package facilitating the use of pathway activities and their expression patterns to identify similarities between biological samples. We defined a new measure, the transcriptional similarity coefficient, which captures similarity of gene expression patterns, instead of quantifying overall activity, in biological pathways between the samples. To demonstrate the utility of SIGN in biomedical research, we establish that SIGN discriminates subtypes of breast tumors and patients with good or poor overall survival. SIGN outperforms the best models in DREAM challenge in predicting survival of breast cancer patients using the data from the Molecular Taxonomy of Breast Cancer International Consortium. In summary, SIGN can be used as a new tool for interrogating pathway activity and gene expression patterns in unsupervised and supervised learning schemes to improve prognostic risk estimation for cancer patients by the biomedical research community. AVAILABILITY AND IMPLEMENTATION: An open-source R package is available (https://cran.r-project.org/web/packages/SIGN/).

Network science in clinical trials: A patient-centered approach.

There has been a paradigm shift in translational oncology with the advent of novel molecular diagnostic tools in the clinic. However, several challenges are associated with the integration of these sophisticated tools into clinical oncology and daily practice. High-throughput profiling at the DNA, RNA and protein levels (omics) generate a massive amount of data. The analysis and interpretation of these is non-trivial but will allow a more thorough understanding of cancer. Linear modelling of the data as it is often used today is likely to limit our understanding of cancer as a complex disease, and at times under-performs to capture a phenotype of interest. Network science and systems biology-based approaches, using machine learning and network science principles, that integrate multiple data sources, can uncover complex changes in a biological system. This approach will integrate a large number of potential biomarkers in preclinical studies to better inform therapeutic decisions and ultimately make substantial progress towards precision medicine. It will however require development of a new generation of clinical trials. Beyond discussing the challenges of high-throughput technologies, this review will develop a framework on how to implement a network science approach in new clinical trial designs in order to advance cancer care.

Modeling age-dependent radiation-induced second cancer risks and estimation of mutation rate: an evolutionary approach.

Although the survival rate of cancer patients has significantly increased due to advances in anti-cancer therapeutics, one of the major side effects of these therapies, particularly radiotherapy, is the potential manifestation of radiation-induced secondary malignancies. In this work, a novel evolutionary stochastic model is introduced that couples short-term formalism (during radiotherapy) and long-term formalism (post-treatment). This framework is used to estimate the risks of second cancer as a function of spontaneous background and radiation-induced mutation rates of normal and pre-malignant cells. By fitting the model to available clinical data for spontaneous background risk together with data of Hodgkin's lymphoma survivors (for various organs), the second cancer mutation rate is estimated. The model predicts a significant increase in mutation rate for some cancer types, which may be a sign of genomic instability. Finally, it is shown that the model results are in agreement with the measured results for excess relative risk (ERR) as a function of exposure age and that the model predicts a negative correlation of ERR with increase in attained age. This novel approach can be used to analyze several radiotherapy protocols in current clinical practice and to forecast the second cancer risks over time for individual patients.

CXCL13 Positive Cells Localization Predict Response to Anti-PD-1/PD-L1 in Pulmonary Non-Small Cell Carcinoma.

Background: Immune checkpoint inhibitors (ICIs) have revolutionized non-small cell lung cancers (NSCLCs) treatment, but only 20-30% of patients benefit from these treatments. Currently, PD-L1 expression in tumor cells is the only clinically approved predictor of ICI response in lung cancer, but concerns arise due to its low negative and positive predictive value. Recent studies suggest that CXCL13+ T cells in the tumor microenvironment (TME) may be a good predictor of response. We aimed to assess if CXCL13+ cell localization within the TME can predict ICI response in advanced NSCLC patients. Methods: This retrospective study included 65 advanced NSCLC patients treated with Nivolumab/Pembrolizumab at IUCPQ or CHUM and for whom a pretreatment surgical specimen was available. Good responders were defined as having a complete radiologic response at 1 year, and bad responders were defined as showing cancer progression at 1 year. IHC staining for CXCL13 was carried out on a representative slide from a resection specimen, and CXCL13+ cell density was evaluated in tumor (T), invasive margin (IM), non-tumor (NT), and tertiary lymphoid structure (TLS) compartments. Cox models were used to analyze progression-free survival (PFS) and overall survival (OS) probability, while the Mann-Whitney test was used to compare CXCL13+ cell density between responders and non-responders. Results: We showed that CXCL13+ cell density localization within the TME is associated with ICI efficacy. An increased density of CXCL13+ cells across all compartments was associated with a poorer prognostic (OS; HR = 1.22; 95%CI = 1.04-1.42; p = 0.01, PFS; HR = 1.16; p = 0.02), or a better prognostic when colocalized within TLSs (PFS; HR = 0.84, p = 0.03). Conclusion: Our results support the role of CXCL13+ cells in advanced NSCLC patients, with favorable prognosis when localized within TLSs and unfavorable prognosis when present elsewhere. The concomitant proximity of CXCL13+ and CD20+ cells within TLSs may favor antigen presentation to T cells, thus enhancing the effect of PD-1/PD-L1 axis inhibition. Further validation is warranted to confirm the potential relevance of this biomarker in a clinical setting.

The Cross-Scale Association between Pathomics and Radiomics Features in Immunotherapy-Treated NSCLC Patients: A Preliminary Study.

BACKGROUND: Recent advances in cancer biomarker development have led to a surge of distinct data modalities, such as medical imaging and histopathology. To develop predictive immunotherapy biomarkers, these modalities are leveraged independently, despite their orthogonality. This study aims to explore the cross-scale association between radiological scans and digitalized pathology images for immunotherapy-treated non-small cell lung cancer (NSCLC) patients. METHODS: This study involves 36 NSCLC patients who were treated with immunotherapy and for whom both radiology and pathology images were available. A total of 851 and 260 features were extracted from CT scans and cell density maps of histology images at different resolutions. We investigated the radiopathomics relationship and their association with clinical and biological endpoints. We used the Kolmogorov-Smirnov (KS) method to test the differences between the distributions of correlation coefficients with the two imaging modality features. Unsupervised clustering was done to identify which imaging modality captures poor and good survival patients. RESULTS: Our results demonstrated a significant correlation between cell density pathomics and radiomics features. Furthermore, we also found a varying distribution of correlation values between imaging-derived features and clinical endpoints. The KS test revealed that the two imaging feature distributions were different for PFS and CD8 counts, while similar for OS. In addition, clustering analysis resulted in significant differences in the two clusters generated from the radiomics and pathomics features with respect to patient survival and CD8 counts. CONCLUSION: The results of this study suggest a cross-scale association between CT scans and pathology H&E slides among ICI-treated patients. These relationships can be further explored to develop multimodal immunotherapy biomarkers to advance personalized lung cancer care.

EGFR Pathway Expression Persists in Recurrent Glioblastoma Independent of Amplification Status.

BACKGROUND: Glioblastoma mortality is driven by tumour progression or recurrence despite administering a therapeutic arsenal consisting of surgical resection, radiation, and alkylating chemotherapy. The genetic changes underlying tumour progression and chemotherapy resistance are poorly understood. METHODS: In this study, we sought to define the relationship between EGFR amplification status, EGFR mRNA expression, and EGFR pathway activity. We compared RNA-sequencing data from matched primary and recurrent tumour samples (n = 40 patients, 20 with EGFR amplification). RESULTS: In the setting of glioblastoma recurrence, the EGFR pathway was overexpressed regardless of EGFR-amplification status, suggesting a common genomic endpoint in recurrent glioblastoma, although EGFR amplification did associate with higher EGFR mRNA expression. Three of forty patients in the study cohort had EGFR-amplified tumours and received targeted EGFR therapy. Their molecular subtypes and clinical outcomes did not significantly differ from patients who received conventional chemotherapy. CONCLUSION: Our findings suggest that while the EGFR amplification may confer a unique molecular profile in primary glioblastoma, pathway analysis reveals upregulation of the EGFR pathway in recurrence, regardless of amplification status. As such, the EGFR pathway may be a key mediator of glioblastoma progression.

Radiomics approaches to predict PD-L1 and PFS in advanced non-small cell lung patients treated with immunotherapy: a multi-institutional study.

With the increasing use of immune checkpoint inhibitors (ICIs), there is an urgent need to identify biomarkers to stratify responders and non-responders using programmed death-ligand (PD-L1) expression, and to predict patient-specific outcomes such as progression free survival (PFS). The current study is aimed to determine the feasibility of building imaging-based predictive biomarkers for PD-L1 and PFS through systematically evaluating a combination of several machine learning algorithms with different feature selection methods. A retrospective, multicenter study of 385 advanced NSCLC patients amenable to ICIs was undertaken in two academic centers. Radiomic features extracted from pretreatment CT scans were used to build predictive models for PD-L1 and PFS (short-term vs. long-term survivors). We first employed the LASSO methodology followed by five feature selection methods and seven machine learning approaches to build the predictors. From our analyses, we found several combinations of feature selection methods and machine learning algorithms to achieve a similar performance. Logistic regression with ReliefF feature selection (AUC = 0.64, 0.59 in discovery and validation cohorts) and SVM with Anova F-test feature selection (AUC = 0.64, 0.63 in discovery and validation datasets) were the best-performing models to predict PD-L1 and PFS. This study elucidates the application of suitable feature selection approaches and machine learning algorithms to predict clinical endpoints using radiomics features. Through this study, we identified a subset of algorithms that should be considered in future investigations for building robust and clinically relevant predictive models.

Imaging-Based Biomarkers Predict Programmed Death-Ligand 1 and Survival Outcomes in Advanced NSCLC Treated With Nivolumab and Pembrolizumab: A Multi-Institutional Study.

Background: Although the immune checkpoint inhibitors, nivolumab and pembrolizumab, were found to be promising in patients with advanced NSCLC, some of them either do not respond or have recurrence after an initial response. It is still unclear who will benefit from these therapies, and, hence, there is an unmet clinical need to build robust biomarkers. Methods: Patients with advanced NSCLC (N = 323) who were treated with pembrolizumab or nivolumab were retrospectively identified from two institutions. Radiomics features extracted from baseline pretreatment computed tomography scans along with the clinical variables were used to build the predictive models for overall survival (OS), progression-free survival (PFS), and programmed death-ligand 1 (PD-L1). To develop the imaging and integrative clinical-imaging predictive models, we used the XGBoost learning algorithm with ReliefF feature selection method and validated them in an independent cohort. The concordance index for OS, PFS, and area under the curve for PD-L1 was used to evaluate model performance. Results: We developed radiomics and the ensemble radiomics-clinical predictive models for OS, PFS, and PD-L1 expression. The concordance indices of the radiomics model were 0.60 and 0.61 for predicting OS and PFS and area under the curve was 0.61 for predicting PD-L1 in the validation cohort, respectively. The combined radiomics-clinical model resulted in higher performance with 0.65, 0.63, and 0.68 to predict OS, PFS, and PD-L1 in the validation cohort, respectively. Conclusions: We found that pretreatment computed tomography imaging along with clinical data can aid as predictive biomarkers for PD-L1 and survival end points. These imaging-driven approaches may prove useful to expand the therapeutic options for nonresponders and improve the selection of patients who would benefit from immune checkpoint inhibitors.

A Radiomics-Clinical Model Predicts Overall Survival of Non-Small Cell Lung Cancer Patients Treated with Immunotherapy: A Multicenter Study.

BACKGROUND: Immune checkpoint inhibitors (ICIs) are a great breakthrough in cancer treatments and provide improved long-term survival in a subset of non-small cell lung cancer (NSCLC) patients. However, prognostic and predictive biomarkers of immunotherapy still remain an unmet clinical need. In this work, we aim to leverage imaging data and clinical variables to develop survival risk models among advanced NSCLC patients treated with immunotherapy. METHODS: This retrospective study includes a total of 385 patients from two institutions who were treated with ICIs. Radiomics features extracted from pretreatment CT scans were used to build predictive models. The objectives were to predict overall survival (OS) along with building a classifier for short- and long-term survival groups. We employed the XGBoost learning method to build radiomics and integrated clinical-radiomics predictive models. Feature selection and model building were developed and validated on a multicenter cohort. RESULTS: We developed parsimonious models that were associated with OS and a classifier for short- and long-term survivor groups. The concordance indices (C-index) of the radiomics model were 0.61 and 0.57 to predict OS in the discovery and validation cohorts, respectively. While the area under the curve (AUC) values of the radiomic models for short- and long-term groups were found to be 0.65 and 0.58 in the discovery and validation cohorts. The accuracy of the combined radiomics-clinical model resulted in 0.63 and 0.62 to predict OS and in 0.77 and 0.62 to classify the survival groups in the discovery and validation cohorts, respectively. CONCLUSIONS: We developed and validated novel radiomics and integrated radiomics-clinical survival models among NSCLC patients treated with ICIs. This model has important translational implications, which can be used to identify a subset of patients who are not likely to benefit from immunotherapy. The developed imaging biomarkers may allow early prediction of low-group survivors, though additional validation of these radiomics models is warranted.

Single-cell spatial landscape of immunotherapy response reveals mechanisms of CXCL13 enhanced antitumor immunity.

BACKGROUND: Immunotherapy has revolutionized clinical outcomes for patients suffering from lung cancer, yet relatively few patients sustain long-term durable responses. Recent studies have demonstrated that the tumor immune microenvironment fosters tumorous heterogeneity and mediates both disease progression and response to immune checkpoint inhibitors (ICI). As such, there is an unmet need to elucidate the spatially defined single-cell landscape of the lung cancer microenvironment to understand the mechanisms of disease progression and identify biomarkers of response to ICI. METHODS: Here, in this study, we applied imaging mass cytometry to characterize the tumor and immunological landscape of immunotherapy response in non-small cell lung cancer by describing activated cell states, cellular interactions and neighborhoods associated with improved efficacy. We functionally validated our findings using preclinical mouse models of cancer treated with anti-programmed cell death protein-1 (PD-1) immune checkpoint blockade. RESULTS: We resolved 114,524 single cells in 27 patients treated with ICI, enabling spatial resolution of immune lineages and activation states with distinct clinical outcomes. We demonstrated that CXCL13 expression is associated with ICI efficacy in patients, and that recombinant CXCL13 potentiates anti-PD-1 response in vivo in association with increased antigen experienced T cell subsets and reduced CCR2+ monocytes. DISCUSSION: Our results provide a high-resolution molecular resource and illustrate the importance of major immune lineages as well as their functional substates in understanding the role of the tumor immune microenvironment in response to ICIs.

Generalization optimizing machine learning to improve CT scan radiomics and assess immune checkpoint inhibitors' response in non-small cell lung cancer: a multicenter cohort study.

Background: Recent developments in artificial intelligence suggest that radiomics may represent a promising non-invasive biomarker to predict response to immune checkpoint inhibitors (ICIs). Nevertheless, validation of radiomics algorithms in independent cohorts remains a challenge due to variations in image acquisition and reconstruction. Using radiomics, we investigated the importance of scan normalization as part of a broader machine learning framework to enable model external generalizability to predict ICI response in non-small cell lung cancer (NSCLC) patients across different centers. Methods: Radiomics features were extracted and compared from 642 advanced NSCLC patients on pre-ICI scans using established open-source PyRadiomics and a proprietary DeepRadiomics deep learning technology. The population was separated into two groups: a discovery cohort of 512 NSCLC patients from three academic centers and a validation cohort that included 130 NSCLC patients from a fourth center. We harmonized images to account for variations in reconstruction kernel, slice thicknesses, and device manufacturers. Multivariable models, evaluated using cross-validation, were used to estimate the predictive value of clinical variables, PD-L1 expression, and PyRadiomics or DeepRadiomics for progression-free survival at 6 months (PFS-6). Results: The best prognostic factor for PFS-6, excluding radiomics features, was obtained with the combination of Clinical + PD-L1 expression (AUC = 0.66 in the discovery and 0.62 in the validation cohort). Without image harmonization, combining Clinical + PyRadiomics or DeepRadiomics delivered an AUC = 0.69 and 0.69, respectively, in the discovery cohort, but dropped to 0.57 and 0.52, in the validation cohort. This lack of generalizability was consistent with observations in principal component analysis clustered by CT scan parameters. Subsequently, image harmonization eliminated these clusters. The combination of Clinical + DeepRadiomics reached an AUC = 0.67 and 0.63 in the discovery and validation cohort, respectively. Conversely, the combination of Clinical + PyRadiomics failed generalizability validations, with AUC = 0.66 and 0.59. Conclusion: We demonstrated that a risk prediction model combining Clinical + DeepRadiomics was generalizable following CT scan harmonization and machine learning generalization methods. These results had similar performances to routine oncology practice using Clinical + PD-L1. This study supports the strong potential of radiomics as a future non-invasive strategy to predict ICI response in advanced NSCLC.

Unravelling actionable biology using transcriptomic data to integrate mitotic index and Ki-67 in the management of lung neuroendocrine tumors.

Pulmonary neuroendocrine tumors (NETs) are a heterogeneous family of malignancies whose classification relies on morphology and mitotic rate, unlike extrapulmonary neuroendocrine tumors that require both mitotic rate and Ki-67. As mitotic count is proportional to Ki-67, it is crucial to understand if Ki-67 can complement the existing diagnostic guidelines, as well as discover the benefit of these two markers to unravel the biological heterogeneity. In this study, we investigated the association of mitotic rate and Ki-67 at gene- and pathway-level using transcriptomic data in lung NET malignancies. Lung resection tumor specimens obtained from 28 patients diagnosed with NETs were selected. Mitotic rate, Ki-67 and transcriptomic data were obtained for all samples. The concordance between mitotic rate and Ki-67 was evaluated at gene-level and pathway-level using gene expression data. Our analysis revealed a strong association between mitotic rate and Ki-67 across all samples and cell cycle genes were found to be differentially ranked between them. Pathway analysis indicated that a greater number of pathways overlapped between these markers. Analyses based on lung NET subtypes revealed that mitotic rate in carcinoids and Ki-67 in large cell neuroendocrine carcinomas provided comprehensive characterization of pathways among these malignancies. Among the two subtypes, we found distinct leading-edge gene sets that drive the enrichment signal of commonly enriched pathways between mitotic index and Ki-67. Overall, our findings delineated the degree of benefit of the two proliferation markers, and offers new layer to predict the biological behavior and identify high-risk patients using a more comprehensive diagnostic workup.