Luis Munoz-Erazo ^1,2 Saem Park ^1,2 Shelly Lin ^1,2 Chun-Jen J Chen ^1,2 Lisa Y Y Zhou ^1,2 Janet L Rhodes ^2,3 Taesung Jeon ^4,5 Sonya Fenton ⁶ John L Mccall ⁶ Roslyn Kemp ^2,3 Rod Dunbar ^2,7*

• ¹ School of Biological Sciences, Faculty of Science, The University of Auckland, Auckland, Auckland, New Zealand
• ² Maurice Wilkins Centre for Molecular Biodiscovery, Faculty of Science, The University of Auckland, Auckland, Auckland, New Zealand
• ³ Department of Microbiology and Immunology, School of Biomedical Sciences, University of Otago, Dunedin, Otago, New Zealand
• ⁴ Department of Pathology, College of Medicine, Korea University, Seoul, Republic of Korea
• ⁵ Korea University Guro Hospital, Seoul, Republic of Korea
• ⁶ Department of Surgical Sciences, Otago Medical School, University of Otago, Dunedin, Otago, New Zealand
• ⁷ The University of Auckland, Auckland, New Zealand

Tertiary Lymphoid Structures (TLS) in cancer tissue are potential sites for the organisation of immune responses to cancer, and correlate positively with improved clinical outcomes for patients including in colorectal cancer (CRC). However it has proven challenging to standardise assessment of TLS due to the highly variable appearances of circumscribed domains of TLS within tissue sections. A recent three-dimensional reconstruction of TLS in CRC tissue showed that TLS are often large, multi-lobular structures, suggesting that assessing TLS across whole sections may be necessary to provide an accurate view of TLS activity in a patient's tumour. In a pilot study we therefore used whole slide scans of multiplexed immunofluorescence images to characterise TLS from 22 subjects with CRC. Multiplexed staining for CD20, CD3, CD8, FoxP3 and Ki-67 enabled us to identify B-cells, CD8+ T cells, FoxP3-CD4 T-cells, and Foxp3+ CD4 T cells in all sections, and quantify both the presence of these cell subsets in lymphocytic clusters and their degree of proliferation within those clusters. In total we identified 524 lymphocytic clusters with morphology consistent with TLS. TLS domains varied substantially between samples in size, morphology, cellular constituents, count (from 4 to 100), and proportion of total section area they occupied (0.2%-7.8%). We quantified proliferation of B-cells and T-cell subsets within TLS domains across entire sections and compared data to the canonical approach of counting and phenotyping individual TLS domains. The whole-slide approach proved far simpler, generating digital summaries that identified patients with strikingly different levels of immune activity within their TLS. Strong correlations were observed between the proliferation of Bcells and T-cell subsets. The presence of non-proliferating Foxp3+ CD4 T cells within TLS showed no correlation with the level of proliferation of other lymphocyte subsets. In summary, whole-section digital quantification of immune cell activity within TLS has major advantages over canonical approaches, and should accelerate research into correlations between TLS status and clinical outcomes, with potential to enable a standardised assay for clinical use.