Article Text

Download PDFPDF

1465 Multiplex imaging identifies unique immunophenotypic and spatial characteristics associated with response to immune checkpoint inhibitors (ICIs) in metastatic urothelial cancer (mUC)
  1. Jonathan Anker1,
  2. John-William Sidhom1,
  3. Guray Akturk2,
  4. Sudeh Izadmehr3,
  5. Justin David4,
  6. Saurabh Gupta5,
  7. Seunghee Kim-Schulze3,
  8. Padmanee Sharma6,
  9. Sacha Gnjatic3 and
  10. Matthew Galsky7
  1. 1Mount Sinai Hospital, New York, NY, USA
  2. 2Merck, Demarest, NJ, USA
  3. 3Icahn School of Medicine at Mount Sinai, New York, NY, USA
  4. 4Bristol Myers Squibb, Mt. Laurel, NJ, USA
  5. 5Bristol Myers Squibb, Princeton, NJ, USA
  6. 6The University of Texas MD Anderson Cancer Center, Houston, TX, USA
  7. 7Tisch Cancer Institute At Mount Sinai Medical Center, New York, NY, USA
  • 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 ICIs increase survival in mUC1 but only a subset (~15–25%) of patients experience durable disease control.2 Differences in the tumor microenvironment (TME) might underlie such differential responses. However, the complex network of cellular interactions within the TME that associate with response and resistance to ICIs remains underexplored.

Methods Multiplex Immunohistochemical Consecutive Staining on a Single Slide (MICSSS)3 was performed on UC specimens (N=40) from CheckMate 2754 prior to treatment with nivolumab. 9 immunohistochemical stains (PD-L1, CD8, CD3, pan-cytokeratin, fibronectin, CD68, FAP, DC-LAMP,5 CD11b) were sequentially performed on a single slide per patient. Image processing, whole slide annotation (median 464,554 cells/slide), and intra- and extra-tumoral compartment training, were performed using QuPath (figure 1).6 Responders (CR, PR) and non-responders (SD, PD) were defined per RECIST v1.1. Immunophenotypic designations of ‘inflamed’, ‘excluded’, and ‘desert’ were defined via tumor margin CD8 analysis.7 Lymphoid aggregates were identified morphologically with dense CD3 positivity. Single-cell spatial analysis was performed defining neighborhoods as the 25 nearest neighboring cells.

Results TME characterization demonstrated inter-tumoral heterogeneity, both in the intra- and extra-tumoral compartments (figure 2). Responders contained >2-fold increased intra-tumoral CD8 cells, though no cell types were significantly altered in comparison to non-responders. In contrast, extra-tumoral CD3, CD8, CD3CD8-, DC-LAMP (PD-L1- and PD-L1+), and PD-L1+ CD11b cells were significantly enriched in responders (figure 3). Inflamed tumors were more prevalent and excluded/desert tumors less prevalent in responders, with inflamed tumors containing increased intra-tumoral T cell and DC-LAMP infiltration. There were no significant differences in infiltrate composition between inflamed responders and inflamed non-responders, while excluded/desert responders demonstrated enrichment for extra-tumoral DC-LAMP cells (PD-L1- and PD-L1+) and intra-tumoral PD-L1- DC-LAMP cells as compared to excluded/desert non-responders (figure 4). Responder tumors also contained an increased density of lymphoid aggregates, which were found in closer proximity to tumor regions, were associated with increased survival, and were comprised of a greater degree of DC-LAMP cells (figure 5). Spatial analysis of extra-tumoral cells identified a unique immune and tumor-enriched PD-L1+ neighborhood (cluster 0) predominant in responders, and a distinct CD11b and tumor-based neighborhood devoid of PD-L1 and other immune cells (cluster 2) predominant in non-responders (figure 6).

Conclusions Multiplex immunohistochemistry identified unique immunophenotypic and spatial TME features specific to mUC responders to ICI. Both increased infiltration and the geographic arrangement of T cells, dendritic cells, and PD-L1 positivity, particularly in the extra-tumoral compartment, may prove key in identifying responders to ICI.

Trial Registration This is a secondary translational analysis from NCT02387996 (CheckMate 275, CA209–275)


  1. Bellmunt J, de Wit R, Vaughn DJ, Fradet Y, Lee J-L, Fong L, et al. Pembrolizumab as Second-Line Therapy for Advanced Urothelial Carcinoma. New England Journal of Medicine. 2017;376:1015–26.

  2. Galsky MD, Arija JÁA, Bamias A, Davis ID, Santis MD, Kikuchi E, et al. Atezolizumab with or without chemotherapy in metastatic urothelial cancer (IMvigor130): a multicentre, randomised, placebo-controlled phase 3 trial. The Lancet. 2020;395:1547–57.

  3. Akturk G, Sweeney R, Remark R, Merad M, Gnjatic S. Multiplexed Immunohistochemical Consecutive Staining on Single Slide (MICSSS): Multiplexed Chromogenic IHC Assay for High-Dimensional Tissue Analysis. Methods Mol Biol. 2020;2055:497–519.

  4. Sharma P, Retz M, Siefker-Radtke A, Baron A, Necchi A, Bedke J, et al. Nivolumab in metastatic urothelial carcinoma after platinum therapy (CheckMate 275): a multicentre, single-arm, phase 2 trial. The Lancet Oncology. 2017;18:312–22.

  5. Maier B, Leader AM, Chen ST, Tung N, Chang C, LeBerichel J, et al. A conserved dendritic-cell regulatory program limits antitumour immunity. Nature. 2020;580:257–62.

  6. Bankhead P, Loughrey MB, Fernández JA, Dombrowski Y, McArt DG, Dunne PD, et al. QuPath: Open source software for digital pathology image analysis. Sci Rep. 2017;7:16878.

  7. Braun DA, Hou Y, Bakouny Z, Ficial M, Sant’ Angelo M, Forman J, et al. Interplay of somatic alterations and immune infiltration modulates response to PD-1 blockade in advanced clear cell renal cell carcinoma. Nat Med. 2020;26:909–18.

Ethics Approval The initial clinical trial NCT02387996 obtained appropriate ethics approval, and was conducted in accordance with Good Clinical Practice guidelines defined by the International Conference on Harmonisation. All participants provided informed consent before taking part in the study based on the principles of the Declaration of Helsinki. Approval was granted from local institutional review boards or ethics committees at each center (as published).

Abstract 1465 Figure 1

MICSSS analysis pipeline. 9 sequential immunohistocliemical stains were performed on a single slide tot each patient. Images were then co-registered to the single cell level, confirmed at each quadrant of the specimen. Tissue annotation arid cell segmentation were performed on whole tissue specimens (median 464,554, range 134,796–1,255,791 cells per patient). Stain positivity was confirmed at multiple regions for each stain and patient. Intra- and extra-tumoral compartment training was performed on multiple regions

Abstract 1465 Figure 2

Whole cohort TME characterization. (A) Waterfall plot demonstrating the TME composition breakdown from each cell type, each column repesents an individual patient (B) Waterfall plots (top), scatter plots (bottom) demonstrating the percentage of each cell type within the intra-tumoral (left) or extra-tumoral (right) campartrnents

Abstract 1465 Figure 3

Compartment-specific TME infiltrate in responders and non-responders. Comparison of lymphoid, myeloid (including PD-L1 positive and negative subsets), and matrisome cell types between responders (blue) and non-responders (gray) in the (A) intra-tumoral and (B) extra-tumoral compartments. Bar plots (right) represent the ratio of the median cell concentration between responders and non-responders, dotted lines indicating ≥2-fold difference. * P<0.05, ** P<0.01, *** P<0.001

Abstract 1465 Figure 4

TME immunophenotypes in responders and non-responders. (A) Example images of inflamed, excluded, and desert TMEs, CD8 in pink, CK in brown (left) Breakdown of irnmunophenotype designation by CD8 analysis at the tumor outer margin and tumor inner margin (center), Breakdown of immunophenotype designation in responders and non-responders (right). Bar plots represent the ratio of the median cell concentration between (B) inflamed and desert/excluded patients, (C) inflamed responders and inflamed non-responders, and (D) excluded/desert responders and excluded/desert non-responders (right). Dotted lines indicating ≥2-fold difference. *P<0.05, **P<0.01, ***P<0.001, P>0.05 and <0.1 listed

Abstract 1465 Figure 5

Lymphoid aggregate analysis in responders and non-responders. (A) Density of lymphoid aggregates per patitent. (B) Distance from each lymphoid aggregate to the nearest tumor region. Black bar respresents the median (left). Distribution of distances (right). (C) Kaplan-Meier curve for overall survival of patitents separated by median density of lymphoid aggregates. (D) Bar plot represents the ratio of the median cell concentration within lymphoid aggregates between responders and non-responders, dotted line indicates ≥2-fold difference. * P<0.05, ** P<0.01, *** P<0.001, *** P<0.0001

Abstract 1465 Figure 6

Neighborhood Analysìs of Extra-Tumoral Tissue. (A) Neighborhood characterization was completed by applyrig K-Nearest Neighbors (k = 25) to single-cell expression dala. where the mean expression value over 25 nearest neighbors (including center cell) was averaged over each stain to create a mean expression vector for each neighborhood. UMAP dimensionality reduction was applied on neighborhoods. where the centre cell is classified as within the extra-tumoral tissue, to visualize distribution of neighborhoods along with corresponding antibody expressions. (B) K-means clustering was applied on neighborhood expression vectors (n-clusters = 4) and is visual-ized in UMAP space. (C) Composition of each sample by clusters was conducted and stratified between responders/non-responders by each cluster (** P<0.01). (D) Cluster charaterization visualized via clustermap where each row represents a cluster and columns represent average expression of given marker for each cluster (z-scores shown, red = high expression, blue = low expression)

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.