Debaditya Chakraborty, Elizabeth Gutierrez-Chakraborty, Cristian Rodriguez-Aguayo, Hakan Başağaoğlu, Gabriel Lopez-Berestein, Paola Amero
High-Grade Serous Ovarian Cancer (HGSC) is the most prevalent and lethal form of gynecologic malignancies [1], accounting for 70%-80% of ovarian cancer fatalities. Despite decades of research, the overall survival rate for HGSC has remained largely unchanged [2], and patients with advanced stages of the disease have only a 41% chance of surviving beyond five years [3]. Investigating the genomic and immune profiles of long-term HGSC survivors could offer valuable insights into the underlying tumor biology and inform potential therapeutic strategies [4]. This study advances upon prior research by employing an innovative eXplainable Artificial Intelligence (XAI) integrated with a hypothesis-driven probabilistic methodology to dissect the intricate genetic underpinnings linked to HGSC's survival outcomes in a cohort of 407 patients. The objective of this article was to uncover the most critical prognostic biomarkers from a pool of 655 potential targets through our distinctive data-driven approach and determine the impacts of potentially modulating the identified biomarkers on HGSC outcomes.
Recent studies indicate that AI models are often referred to as “black boxes” because their decision-making process lacks transparency [5]. The consensus is that the lack of inherent explainability is problematic as this produces biases, creates difficulties in detecting false positives and negatives, and conceals potential insights that may be derived from AI [6]. In this study, we provide evidence demonstrating how XAI can enhance biological explainability by revealing novel insights from the underlying data (Supplementary Materials and Methods). Our XAI approach distinctively predicts patient outcomes and survival duration based on genetic signatures (predictive AI aspect of the models) and discovers and helps visualize critical biomarkers (biological explainability aspect of the models) in HGSC. To ensure the viability of explanations generated by our XAI, we subsequently validated the most prominent HGSC-promoting biomarker identified by XAI using in vivo murine tumor models (Supplementary Figure S1). The XAI approach outlined in this study is a proof-of-concept that is not only intended to generate high predictive accuracy but also infer the cause-effect relations behind the predictions, identify counterfactuals that are useful for optimizing interventional therapies, and assess the resultant improvements in patients.
We report that our models predicted the ≥5-year overall survival probability based on the genetic features of patients (n = 407) with 97.52% accuracy, 100% precision, and 94.74% recall on the testing data that comprised 25% of the total samples that were hidden from the models during the training phase. Insights derived through XAI prioritized the biomarkers that are of utmost importance in determining prognosis for patients with HGSC, which we r
{"title":"Discovering genetic biomarkers for targeted cancer therapeutics with eXplainable Artificial Intelligence","authors":"Debaditya Chakraborty, Elizabeth Gutierrez-Chakraborty, Cristian Rodriguez-Aguayo, Hakan Başağaoğlu, Gabriel Lopez-Berestein, Paola Amero","doi":"10.1002/cac2.12530","DOIUrl":"10.1002/cac2.12530","url":null,"abstract":"<p>High-Grade Serous Ovarian Cancer (HGSC) is the most prevalent and lethal form of gynecologic malignancies [<span>1</span>], accounting for 70%-80% of ovarian cancer fatalities. Despite decades of research, the overall survival rate for HGSC has remained largely unchanged [<span>2</span>], and patients with advanced stages of the disease have only a 41% chance of surviving beyond five years [<span>3</span>]. Investigating the genomic and immune profiles of long-term HGSC survivors could offer valuable insights into the underlying tumor biology and inform potential therapeutic strategies [<span>4</span>]. This study advances upon prior research by employing an innovative eXplainable Artificial Intelligence (XAI) integrated with a hypothesis-driven probabilistic methodology to dissect the intricate genetic underpinnings linked to HGSC's survival outcomes in a cohort of 407 patients. The objective of this article was to uncover the most critical prognostic biomarkers from a pool of 655 potential targets through our distinctive data-driven approach and determine the impacts of potentially modulating the identified biomarkers on HGSC outcomes.</p><p>Recent studies indicate that AI models are often referred to as “black boxes” because their decision-making process lacks transparency [<span>5</span>]. The consensus is that the lack of inherent explainability is problematic as this produces biases, creates difficulties in detecting false positives and negatives, and conceals potential insights that may be derived from AI [<span>6</span>]. In this study, we provide evidence demonstrating how XAI can enhance biological explainability by revealing novel insights from the underlying data (Supplementary Materials and Methods). Our XAI approach distinctively predicts patient outcomes and survival duration based on genetic signatures (predictive AI aspect of the models) and discovers and helps visualize critical biomarkers (biological explainability aspect of the models) in HGSC. To ensure the viability of explanations generated by our XAI, we subsequently validated the most prominent HGSC-promoting biomarker identified by XAI using in vivo murine tumor models (Supplementary Figure S1). The XAI approach outlined in this study is a proof-of-concept that is not only intended to generate high predictive accuracy but also infer the cause-effect relations behind the predictions, identify counterfactuals that are useful for optimizing interventional therapies, and assess the resultant improvements in patients.</p><p>We report that our models predicted the ≥5-year overall survival probability based on the genetic features of patients (<i>n</i> = 407) with 97.52% accuracy, 100% precision, and 94.74% recall on the testing data that comprised 25% of the total samples that were hidden from the models during the training phase. Insights derived through XAI prioritized the biomarkers that are of utmost importance in determining prognosis for patients with HGSC, which we r","PeriodicalId":9495,"journal":{"name":"Cancer Communications","volume":null,"pages":null},"PeriodicalIF":16.2,"publicationDate":"2024-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/cac2.12530","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140560445","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Tumors can be classified into distinct immunophenotypes based on the presence and arrangement of cytotoxic immune cells within the tumor microenvironment (TME). Hot tumors, characterized by heightened immune activity and responsiveness to immune checkpoint inhibitors (ICIs), stand in stark contrast to cold tumors, which lack immune infiltration and remain resistant to therapy. To overcome immune evasion mechanisms employed by tumor cells, novel immunologic modulators have emerged, particularly ICIs targeting cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) and programmed cell death protein 1/programmed death-ligand 1(PD-1/PD-L1). These agents disrupt inhibitory signals and reactivate the immune system, transforming cold tumors into hot ones and promoting effective antitumor responses. However, challenges persist, including primary resistance to immunotherapy, autoimmune side effects, and tumor response heterogeneity. Addressing these challenges requires innovative strategies, deeper mechanistic insights, and a combination of immune interventions to enhance the effectiveness of immunotherapies. In the landscape of cancer medicine, where immune cold tumors represent a formidable hurdle, understanding the TME and harnessing its potential to reprogram the immune response is paramount. This review sheds light on current advancements and future directions in the quest for more effective and safer cancer treatment strategies, offering hope for patients with immune-resistant tumors.
{"title":"Immunologic tumor microenvironment modulators for turning cold tumors hot","authors":"Gholam-Reza Khosravi, Samaneh Mostafavi, Sanaz Bastan, Narges Ebrahimi, Roya Safari Gharibvand, Nahid Eskandari","doi":"10.1002/cac2.12539","DOIUrl":"10.1002/cac2.12539","url":null,"abstract":"<p>Tumors can be classified into distinct immunophenotypes based on the presence and arrangement of cytotoxic immune cells within the tumor microenvironment (TME). Hot tumors, characterized by heightened immune activity and responsiveness to immune checkpoint inhibitors (ICIs), stand in stark contrast to cold tumors, which lack immune infiltration and remain resistant to therapy. To overcome immune evasion mechanisms employed by tumor cells, novel immunologic modulators have emerged, particularly ICIs targeting cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) and programmed cell death protein 1/programmed death-ligand 1(PD-1/PD-L1). These agents disrupt inhibitory signals and reactivate the immune system, transforming cold tumors into hot ones and promoting effective antitumor responses. However, challenges persist, including primary resistance to immunotherapy, autoimmune side effects, and tumor response heterogeneity. Addressing these challenges requires innovative strategies, deeper mechanistic insights, and a combination of immune interventions to enhance the effectiveness of immunotherapies. In the landscape of cancer medicine, where immune cold tumors represent a formidable hurdle, understanding the TME and harnessing its potential to reprogram the immune response is paramount. This review sheds light on current advancements and future directions in the quest for more effective and safer cancer treatment strategies, offering hope for patients with immune-resistant tumors.</p>","PeriodicalId":9495,"journal":{"name":"Cancer Communications","volume":null,"pages":null},"PeriodicalIF":16.2,"publicationDate":"2024-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/cac2.12539","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140326366","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Haiyang Zhou, Jiahui Yin, Anqi Wang, Xiaomao Yin, Taojun Jin, Kai Xu, Lin Zhu, Jiexuan Wang, Wenqiang Wang, Wei Zhang, Xinxiang Li, Zhiqian Hu, Xinxing Li
The presence of malignant ascites in colorectal cancer (CRC) patients is associated with a poor prognosis, a high risk of recurrence, and resistance to chemotherapy and immune therapy [1-3]. Understanding the complex interactions among different kinds of cells and the ecosystem of peritoneal metastasized colorectal cancer (pmCRC) ascites may provide insights into effective treatment strategies.
We profiled the single-cell transcriptomes of 96,065 cells from ascites samples of 12 treatment-naïve patients with pmCRC using the 10× single-cell RNA-sequencing (scRNA-seq) (Supplementary Figure S1A, Supplementary Table S1). Eleven major cell types were identified by characteristic canonical cell markers, including epithelial cells, endothelial cells, fibroblasts, T cells, B cells, monocytes, macrophages, plasma cells, natural killer (NK) cells, dendritic cells (DCs), and mast cells (Figure 1A-B). The main cellular components of pmCRC ascites are T cells (40,095; 41.7%), macrophages (28,487; 29.7%), and fibroblasts (5,932; 6.2%). Compared with primary CRC, which showed 14.8% epithelial cells [4], only 0.3% (291) epithelial cells were found in the ascites. The low percentage of epithelial cells in pmCRC ascites was consistent with the scRNA-seq studies of another tumor ascites [5-7].
We classified the 12 patients into 2 groups according to their treatment response as follows: 8 patients (P02, P03, P04, P07, P08, P09, P11, and P12) had stable disease (SD), while 4 (P01, P05, P06, and P10) had progressive disease (PD). Single-cell transcriptomic analyses have revealed high heterogeneity of cell composition in 12 patients. The SD group exhibited a higher proportion of fibroblasts and epithelial cells (Figure 1B). Remarkably, fibroblasts had significantly different expression characteristics between the 2 groups (Figure 1C), and the top five upregulated/downregulated genes were visualized in 11 cell types (Figure 1D). We also found a significant increase in the frequency of macrophages in pmCRC ascites compared with the primary tumors [4] (Figure 1E). It hinted that significant inter-patient variability in the composition and functional programs of pmCRC ascites cells under different disease states.
To comprehensively study the cellular interactions within the pmCRC ascites ecosystem, we predicted cell-cell communication networks using CellChat. Overall, we identified 44 significant ligand-receptor pair interactions. Although T cells were the most abundant cell population (41.7%) in pmCRC ascites, fibroblasts and macrophages were the core of the cellular interaction network (Figure 1F), suggesting their important roles in recruiting and cross-talking with diverse cells in the pmCRC ascites ecosystem.
The result of cellular communications suggested that there was a complex interplay between various signaling molecule. Macrophage migration inhibitory factor (MIF), annexin, complement
{"title":"Single-cell landscape of malignant ascites from patients with metastatic colorectal cancer","authors":"Haiyang Zhou, Jiahui Yin, Anqi Wang, Xiaomao Yin, Taojun Jin, Kai Xu, Lin Zhu, Jiexuan Wang, Wenqiang Wang, Wei Zhang, Xinxiang Li, Zhiqian Hu, Xinxing Li","doi":"10.1002/cac2.12541","DOIUrl":"10.1002/cac2.12541","url":null,"abstract":"<p>The presence of malignant ascites in colorectal cancer (CRC) patients is associated with a poor prognosis, a high risk of recurrence, and resistance to chemotherapy and immune therapy [<span>1-3</span>]. Understanding the complex interactions among different kinds of cells and the ecosystem of peritoneal metastasized colorectal cancer (pmCRC) ascites may provide insights into effective treatment strategies.</p><p>We profiled the single-cell transcriptomes of 96,065 cells from ascites samples of 12 treatment-naïve patients with pmCRC using the 10× single-cell RNA-sequencing (scRNA-seq) (Supplementary Figure S1A, Supplementary Table S1). Eleven major cell types were identified by characteristic canonical cell markers, including epithelial cells, endothelial cells, fibroblasts, T cells, B cells, monocytes, macrophages, plasma cells, natural killer (NK) cells, dendritic cells (DCs), and mast cells (Figure 1A-B). The main cellular components of pmCRC ascites are T cells (40,095; 41.7%), macrophages (28,487; 29.7%), and fibroblasts (5,932; 6.2%). Compared with primary CRC, which showed 14.8% epithelial cells [<span>4</span>], only 0.3% (291) epithelial cells were found in the ascites. The low percentage of epithelial cells in pmCRC ascites was consistent with the scRNA-seq studies of another tumor ascites [<span>5-7</span>].</p><p>We classified the 12 patients into 2 groups according to their treatment response as follows: 8 patients (P02, P03, P04, P07, P08, P09, P11, and P12) had stable disease (SD), while 4 (P01, P05, P06, and P10) had progressive disease (PD). Single-cell transcriptomic analyses have revealed high heterogeneity of cell composition in 12 patients. The SD group exhibited a higher proportion of fibroblasts and epithelial cells (Figure 1B). Remarkably, fibroblasts had significantly different expression characteristics between the 2 groups (Figure 1C), and the top five upregulated/downregulated genes were visualized in 11 cell types (Figure 1D). We also found a significant increase in the frequency of macrophages in pmCRC ascites compared with the primary tumors [<span>4</span>] (Figure 1E). It hinted that significant inter-patient variability in the composition and functional programs of pmCRC ascites cells under different disease states.</p><p>To comprehensively study the cellular interactions within the pmCRC ascites ecosystem, we predicted cell-cell communication networks using CellChat. Overall, we identified 44 significant ligand-receptor pair interactions. Although T cells were the most abundant cell population (41.7%) in pmCRC ascites, fibroblasts and macrophages were the core of the cellular interaction network (Figure 1F), suggesting their important roles in recruiting and cross-talking with diverse cells in the pmCRC ascites ecosystem.</p><p>The result of cellular communications suggested that there was a complex interplay between various signaling molecule. Macrophage migration inhibitory factor (MIF), annexin, complement","PeriodicalId":9495,"journal":{"name":"Cancer Communications","volume":null,"pages":null},"PeriodicalIF":20.1,"publicationDate":"2024-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11260760/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140292889","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Infiltrating immune cells in the tumor microenvironment (TME) play critical roles in the initiation, progression, and metastasis of cancer [1]. Previous studies have reported that the infiltration levels of various immune cell types are significantly associated with patient prognosis in different cancers [2, 3]. Specifically, in non-small cell lung cancer (NSCLC) the prognostic associations of major immune cell types have been investigated [4-6], however, some of the reported associations are inconsistent and remain debated [7]. Limited by technical issues, most studies focused on a few immune cell lineages or relied on inferred immune cell levels from computational deconvolution. To investigate the prognostic effects of all major immune cell types unbiasedly, more systematic high-quality immune cell profiling data with matched patient survival information are needed.
Recently, Sorin et al. [8] used imaging mass cytometry (IMC) to characterize the immunological landscape of 416 distinct lung adenocarcinoma (LUAD) samples at single-cell resolution. The IMC images provide the counts and spatial distribution of 16 cell types with high precision in each sample. These cell types include cancer and endothelial cells, along with 14 immune cell types, including CD163+ and CD163− macrophages, CD8+, CD4+, regulatory, and other T cells, classical, non-classical, and intermediate monocytes, natural killer cells, dendritic cells, mast cells, neutrophils, and other immune cells. Additionally, the data provide patient survival and other clinical information. Using these data, we investigated the prognostic associations of the cell density (#cells/megapixel) and fractions of the 16 cell types as well as the fraction ratio between each pair of cell types (Supplementary Methods). Our results indicated that the relative abundance between cell types (fraction ratios) was more prognostic than cell fractions and densities.
We calculated the densities of the 16 cell types in each patient's IMC image and applied Cox regression analysis to examine their associations with progression-free survival (PFS) after adjusting for established clinical factors including age, sex, smoking status, and tumor stage. At the significance level of P < 0.05, only the density of non-classical monocytes was found to have a significant association with worse prognosis (hazard ratio [HR] = 1.004, P = 0.040, Figure 1A). After multiple testing corrections, none of the cell types was significant (false discovery rate [FDR] > 0.05). Similar results were obtained when cell fractions among all cells were used for prognostic association analysis (Figure 1B). In addition, we conducted prognostic analysis on 14 immune cell types, focusing on their proportions among immune cells (excluding cancer and endothelial cells), yielding similar results. It has b
{"title":"Immune cell pair ratio captured by imaging mass cytometry has superior predictive value for prognosis of non-small cell lung cancer than cell fraction and density","authors":"Jian-Rong Li, Chao Cheng","doi":"10.1002/cac2.12540","DOIUrl":"10.1002/cac2.12540","url":null,"abstract":"<p>Infiltrating immune cells in the tumor microenvironment (TME) play critical roles in the initiation, progression, and metastasis of cancer [<span>1</span>]. Previous studies have reported that the infiltration levels of various immune cell types are significantly associated with patient prognosis in different cancers [<span>2, 3</span>]. Specifically, in non-small cell lung cancer (NSCLC) the prognostic associations of major immune cell types have been investigated [<span>4-6</span>], however, some of the reported associations are inconsistent and remain debated [<span>7</span>]. Limited by technical issues, most studies focused on a few immune cell lineages or relied on inferred immune cell levels from computational deconvolution. To investigate the prognostic effects of all major immune cell types unbiasedly, more systematic high-quality immune cell profiling data with matched patient survival information are needed.</p><p>Recently, Sorin <i>et al.</i> [<span>8</span>] used imaging mass cytometry (IMC) to characterize the immunological landscape of 416 distinct lung adenocarcinoma (LUAD) samples at single-cell resolution. The IMC images provide the counts and spatial distribution of 16 cell types with high precision in each sample. These cell types include cancer and endothelial cells, along with 14 immune cell types, including CD163<sup>+</sup> and CD163<sup>−</sup> macrophages, CD8<sup>+</sup>, CD4<sup>+</sup>, regulatory, and other T cells, classical, non-classical, and intermediate monocytes, natural killer cells, dendritic cells, mast cells, neutrophils, and other immune cells. Additionally, the data provide patient survival and other clinical information. Using these data, we investigated the prognostic associations of the cell density (#cells/megapixel) and fractions of the 16 cell types as well as the fraction ratio between each pair of cell types (Supplementary Methods). Our results indicated that the relative abundance between cell types (fraction ratios) was more prognostic than cell fractions and densities.</p><p>We calculated the densities of the 16 cell types in each patient's IMC image and applied Cox regression analysis to examine their associations with progression-free survival (PFS) after adjusting for established clinical factors including age, sex, smoking status, and tumor stage. At the significance level of <i>P</i> < 0.05, only the density of non-classical monocytes was found to have a significant association with worse prognosis (hazard ratio [HR] = 1.004, <i>P</i> = 0.040, Figure 1A). After multiple testing corrections, none of the cell types was significant (false discovery rate [FDR] > 0.05). Similar results were obtained when cell fractions among all cells were used for prognostic association analysis (Figure 1B). In addition, we conducted prognostic analysis on 14 immune cell types, focusing on their proportions among immune cells (excluding cancer and endothelial cells), yielding similar results. It has b","PeriodicalId":9495,"journal":{"name":"Cancer Communications","volume":null,"pages":null},"PeriodicalIF":16.2,"publicationDate":"2024-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://onlinelibrary.wiley.com/doi/epdf/10.1002/cac2.12540","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140292888","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The cover image is based on the Review Article Cell fate regulation governed by p53: Friends or reversible foes in cancer therapy by Bin Song et al., https://doi.org/10.1002/cac2.12520.