Ward, Caitlin

Abstract Motivation Multiplex immunofluorescence (MxIF) enables the quantification of multiple protein markers at a single-cell level while preserving spatial information, offering a powerful tool for studying tissue microenvironments. However, the flexibility in MxIF panel design poses challenges in standardizing cell phenotyping. Results We present SpaFlow, an efficient, customizable pipeline for unsupervised clustering and classification of MxIF data, implemented using Nextflow. SpaFlow performs quality control, clustering, and postclustering analysis on segmented and quantified MxIF data, facilitating reproducible and scalable analyses across various computing platforms. The SpaFlow pipeline integrates three clustering and classification packages—Seurat, SCIMAP, and CELESTA—each providing unique methodologies for identifying cell types based on phenotypic markers. A novel “meta-clustering” approach condenses clusters across multiple regions of interest into common meta-clusters, streamlining the cell-type identification process in large datasets. SpaFlow’s robust quality control steps, including signal summation and cell density filtering, mitigate artifacts that may impact clustering accuracy. We demonstrate the utility of SpaFlow in a case study involving 297 ovarian tumor cores, where SpaFlow successfully identified biologically meaningful cell populations, including tumor-infiltrating lymphocytes, efficiently and rapidly. Additionally, SpaFlow’s reproducibility is validated using serial tonsil sections, confirming its capability to consistently identify distinctive cell populations across matched ROIs. Availability and implementation SpaFlow is freely available with detailed documentation and examples at https://github.com/dimi-lab/SpaFlow.

CD36 expression in both immune and non-immune cells is known to be directly involved in cancer metastasis. Extracellular vesicles (EVs) secreted by malignant melanocytes play a vital role in developing tumor-promoting microenvironments, but it is unclear whether this is mediated through CD36. To understand the role of CD36 in melanoma, we first analyzed the SKCM dataset for clinical prognosis, evaluated the percentage of CD36 in lymphatic fluid-derived EVs (LEVs), and tested whether melanoma-derived EVs increase CD36 expression and induce M2-macrophage-like characteristics. Furthermore, we performed a multiplex immunofluorescence (MxIF) imaging analysis to evaluate the CD36 expression and its colocalization with various other cells in the lymph node (LN) of patients and control subjects. Our findings show that cutaneous melanoma patients have a worse clinical prognosis with high CD36 levels, and a higher percentage of CD36 in total LEVs were found at baseline in melanoma patients compared to control. We also found that monocytic and endothelial cells treated with melanoma EVs expressed more CD36 than untreated cells. Furthermore, melanoma-derived EVs can regulate immunosuppressive macrophage-like characteristics by upregulating CD36. The spatial imaging data show that cells in tumor-involved sentinel LNs exhibit a higher probability of CD36 expression than cells from control LNs, but this was not statistically significant. Conclusively, our findings demonstrated that CD36 plays a vital role in controlling the immunosuppressive microenvironment in the LN, which can promote the formation of a protumorigenic niche.

Abstract Background: Neuropilin 2 (NRP2) is a co-receptor that enhances signaling of growth factor and cytokine receptors and is expressed in tumor and microenvironment (TME) cells in melanoma and other tumors. Neuropilin 2 expression in tumor cells is associated with more aggressive tumors. Expression in tumor associated macrophages enhances efferocytosis and immunologically silent tumor cell death. Methods: Whole excisional lymph node biopsies were obtained from 10 treatment naïve patients with metastatic cutaneous melanoma who subsequently received anti-PD1 treatment. A single 5mm formalin-fixed paraffin embedded tissue section was used to assess a panel of 39 analytes by cyclic multiplex immunofluorescence (MxIF). Fields of view (~1mm2, n=288) were selected from pathologist-annotated regions of the tumor core, tumor-immune interface and adjacent lymphoid tissue. Pixel-based single cell segmentation and a supervised classifier approach was applied to resolve 12 distinct tumor, stromal and immune cell phenotypes and functional states (NRP2 positive or negative) in 1.5 million cells. Results: Tumor recurrence was observed in 6 patients (relapsed), whereas 4 patients were recurrence-free (non-relapsed) during follow up. Among the different cell types, NRP2 expression was more frequent in macrophages (44%), tumor cells (40%), dendritic cells (40%), followed by Tregs (31%) and T helper cells (29%), and less frequent in neutrophils (24%), cytotoxic T cells (23%) and B cells (15%). In colocalization permutation analysis, NRP2 expressing macrophages clustered away from tumor cells in non-relapsed patients while a random spatial distribution of NRP2 expressing macrophages with respect to tumor cells was observed in relapsed patients. In comparing the odds of tumor cell interactions with NRP2-expressing macrophages to those with NRP2 non-expressing macrophages, we found in the tumor:stromal interface (10% tumor cells) that cells from both non-relapsed and relapsed patients were more likely to interact with NRP2 non-expressing macrophages (OR = 0.73, 95% CI = [0.71, 0.75] and OR = 0.56, 95% CI = [0.55, 0.58] for non-relapsed and relapsed patients, respectively). However, in the tumor core (90% tumor cells), tumor cells from relapsed patients had significantly higher odds of interacting with NRP2 expressing macrophages, (OR = 1.15, 95% CI = [1.12, 1.18]), and this was not seen for non-relapsed patients (OR = 0.68, 95% CI = [0.66, 0.71]). In FOVs dominated by tumor cells (75% tumor), tumor cells from relapsed patients had 4.97 times the odds of expressing NRP2 compared to tumor cells from non-relapsed patients (95% CI = [2.23, 11.09], p = 0.0001). Conclusions: These spatial analytics suggest complex interactions among NRP2 expression in tumor and TME of melanoma determining risk for relapse with anti-PD1 treatment. Citation Format: Anastasios Dimou, Raymond M. Moore, Caitlin Ward, Alexey Leontovich, Ruifeng Guo, Chen Wang, Shankar Suman, Betty Dicke, Jill M. Schimke, Noah A. Stueven, Chathu A. Atherton, Wendy Nevala, Svetomir N. Markovic. Spatial analysis of neuropilin 2 expression in the microenvironment of melanoma [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 2905.

Introduction Survival prediction in glioblastoma remains challenging, and identification of robust imaging markers could help with this relevant clinical problem. We evaluated multiparametric magnetic resonance imaging–derived radiomics to assess prediction of overall survival (OS) and progression-free survival (PFS). Methodology A retrospective, institutional review board–approved study was performed. There were 93 eligible patients, of which 55 underwent gross tumor resection and chemoradiation (GTR-CR). Overall survival and PFS were assessed in the entire cohort and the GTR-CR cohort using multiple machine learning pipelines. A model based on multiple clinical variables was also developed. Survival prediction was assessed using the radiomics-only, clinical-only, and the radiomics and clinical combined models. Results For all patients combined, the clinical feature–derived model outperformed the best radiomics model for both OS (C-index, 0.706 vs 0.597; P < 0.0001) and PFS prediction (C-index, 0.675 vs 0.588; P < 0.001). Within the GTR-CR cohort, the radiomics model showed nonstatistically improved performance over the clinical model for predicting OS (C-index, 0.638 vs 0.588; P = 0.4). However, the radiomics model outperformed the clinical feature model for predicting PFS in GTR-CR cohort (C-index, 0.641 vs 0.550; P = 0.004). Combined clinical and radiomics model did not yield superior prediction when compared with the best model in each case. Conclusions When considering all patients, regardless of therapy, the radiomics-derived prediction of OS and PFS is inferior to that from a model derived from clinical features alone. However, in patients with GTR-CR, radiomics-only model outperforms clinical feature–derived model for predicting PFS.