Article Text

Download PDFPDF

1480 Spatially resolved T-cell microenvironment in mantle cell lymphoma using combined image analysis and spatial omics
  1. Lavanya Lokhande,
  2. Daniel Nilsson,
  3. Joana Rodrigues,
  4. May Hassan,
  5. Lina Olsson,
  6. Anna Porwit,
  7. Anna S Gerdtsson,
  8. Mats Jerkeman and
  9. Sara Ek
  1. Lund University, Lund, Skane, Sweden
  • Journal for ImmunoTherapy of Cancer (JITC) preprint. The copyright holder for this preprint are the authors/funders, who have granted JITC permission to display the preprint. All rights reserved. No reuse allowed without permission.


Background Deciphering the tumor-immune microenvironment (TIME) and the impact of cell-to-cell interaction for immunotherapy response is pivotal for developing biomarker signatures for patient stratification and novel therapies. Spatial phenotyping and functional deconvolution of the TIME can be achieved by using spatial omics technologies. Furthermore, high-throughput image segmentation algorithms can provide deeper insights into histology and composition of the tissue. Thus, in this study we aim to decipher the nature of infiltrating T-cell subtypes in mantle cell lymphoma (MCL) and assess how variation in T-cell subtype frequencies impact the local T-cell- and tumor-associated proteome and transcriptome.

Methods GeoMxTM(Nanostring Inc.) was used for spatial profiling (69 proteins and 1811 mRNA) of T-cell subsets and combined with metrics derived from CellPose cell segmentation of the multiplexed immunofluorescence (mIF) images. Tissue microarrays were constructed from 102 MCL patients (BLISS cohort) and omics expression data was collected across three tissue sections (figure 1). Cell-specific protein data was extracted for CD20+ tumor cells, TC,57- (cytotoxic, CD3+ CD8+ CD57-), TH,57- (helper, CD3+ CD8- CD57-), TC,57+ (matured cytotoxic, CD3+ CD8+ CD57+) and TH,57+ (matured helper, CD3+ CD8- CD57+). Transcriptome data was collected from CD20+ tumor cells, TC (cytotoxic T-cells) and TH (helper T-cells). For image analysis, two Cellpose models were simultaneously retrained, a ‘nuclei’ model based on Syto13 staining and a ‘cyto’ model using the CD3, CD8 and CD57 staining. Cells are classified into the four T-cell subtypes based on overlapping the centroid of the nuclei segmentation mask with the marker-specific cell segmentation masks. The metrics per cell and image were extracted and used to correlate T-cell frequency to omics expression.

Results We show that variation in T-cell content, not only impacts the composition of the cellular MCL microenvironment but is associated with distinct phenotypic and transcriptional profiles of malignant and infiltrating immune cells. For example, increasing expression of IDO1, a known mediator of immunosuppression, was correlated with increasing T-cell frequency. In contrast, GITR was negatively correlated with IDO1 expression and predominantly abundant in regions with sparse T-cell content. Ongoing investigations on integrated models aim to identify novel therapeutic targets and methods of patient stratification.

Conclusions This study demonstrates how deep-learning cell segmentations strategies, combined with spatially guided proteomics and transcriptomics data reveal variation in T-cell composition and functionality in MCL. Deconvoluting the local tumor immune microenvironment contributes to improved biological insight, essential for understanding response to immunomodulatory treatment.

Abstract 1480 Figure 1

Overview of the study. (A) Spatial omics data collection performed across three TMA sections. Section 1 was stained for CD20 for tumor compartment. Section 2 was stained for CD3 (bulk T-cells), CD8 (differentiating between cytotoxic and helper subtypes), and CD57 differentiating between active and matured T-cell subtypes (pink and yellow represent positive and negative for cell markers, respectively). Section 1 and 2 was used to collect cell.specìfic expression data. Transcriptome data was collected from section 3, for tumor, TC and TH compartment. (B) Overview of image analysis performed on mIF images collected from slice 2 stained for four T-cell subsets. Two Cellpose models were retrained, a nuclei model based on Syto13 staining and a cyto model based on cell membrane markers CD3, CD8 and CD57 which all together generated four segmentation masks. A cell was classified into the four cell subtypes if the centroid of the nuclei segmentation mask overlapped with the respective cytoplasmic masks. Image derived metrics were used in conjunction with expression data for the current analysis

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.