Tissue specimens taken from primary tumors or metastases contain important information for diagnosis and treatment of cancer patients. Multispectral imaging allows in situ visualization of heterogeneous cell subsets, such as lymphocytes, in tissue samples. Many image processing pipelines first segment cell boundaries and then measure marker expression to assign cell phenotypes. In dense tissue environments such as solid tumors, segmentation-based phenotyping can be inaccurate due to segmentation errors or overlapping cell boundaries. Here we introduce a machine learning pipeline design called ImmuNet that directly identifies the positions and phenotypes of immune cells without determining their exact boundaries. ImmuNet is easy to train: human annotators only need to click on immune cells and rank their expression of each marker; full annotation of tissue regions is not necessary. We demonstrate that ImmuNet is a suitable approach for immune cell detection and phenotyping in multiplex immunohistochemistry: it compares favourably to segmentation-based methods, especially in dense tissues, and we externally validate ImmuNet results by comparing them to flow cytometric measurements from the same tissue. In summary, ImmuNet performs well on diverse tissue specimens, takes relatively little effort to train and implement, and is a simpler alternative to segmentation-based approaches when only cell positions and phenotypes, but not their shapes are required for downstream analyses. We hope that ImmuNet will help cancer researchers to analyze multichannel tissue images more easily and accurately.

Checkpoint inhibitors targeting PD-(L)1 induce objective responses in 20% of patients with metastatic urothelial cancer (UC). CD8+ T cell infiltration has been proposed as a putative biomarker for response to checkpoint inhibitors. Nevertheless, data on spatial and temporal heterogeneity of tumor-infiltrating lymphocytes in advanced UC are lacking. The major aims of this study were to explore spatial heterogeneity for lymphocyte infiltration and to investigate how the immune landscape changes during the disease course. We performed multiplex immunohistochemistry to assess the density of intratumoral and stromal CD3+, CD8+, FoxP3+ and CD20+ immune cells in longitudinally collected samples of 49 UC patients. Within these samples, spatial heterogeneity for lymphocyte infiltration was observed. Regions the size of a 0.6 tissue microarray core (0.28 mm2) provided a representative sample in 60.6 to 71.6% of cases, depending on the cell type of interest. Regions of 3.30 mm2, the median tumor surface area in our biopsies, were representative in 58.8 to 73.8% of cases. Immune cell densities did not significantly differ between untreated primary tumors and metachronous distant metastases. Interestingly, CD3+, CD8+ and FoxP3+ T cell densities decreased during chemotherapy in two small cohorts of patients treated with neoadjuvant or palliative platinum-based chemotherapy. In conclusion, spatial heterogeneity in advanced UC challenges the use of immune cell infiltration in biopsies as biomarker for response prediction. Our data also suggests a decrease in tumor-infiltrating T cells during platinum-based chemotherapy.

Background Whereas neoadjuvant cisplatin-based chemotherapy (NAC) followed by a radical cystectomy remains the standard treatment for patients with muscle-invasive bladder cancer (MIBC), increasing evidence suggests that checkpoint inhibitors, either alone or in combination with chemotherapy, are effective in the (neo)adjuvant setting. The major aim of this study was to improve our understanding of the immune-modulating effects of NAC in MIBC. Methods Tumor tissue of 81 patients was used, including 60 patients treated with NAC and 21 patients undergoing upfront cystectomy. Multiplex immunohistochemistry was performed to assess CD3+, CD3+CD8+, CD3+CD8−FoxP3−, CD3+FoxP3+, and CD20+ cells. Patients were classified into a favorable or unfavorable outcome group based on the development of a recurrence within a year. Results The density of intratumoral CD3+ T cells decreased following NAC in patients with a recurrence at one year, while it remained stable in patients without a recurrence (median fold change 0.6 [95CI 0.3; 1.0] versus 1.0 [95CI 0.6; 2.2]). This decrease was mainly attributable to a decrease in CD3+CD8−FoxP3− and CD3+FoxP3+ T cells and was not observed in patients with an early recurrence after upfront cystectomy. Additionally, in cystectomy tissue of patients treated with NAC, median CD3+ and CD3+CD8+ T cell densities were significantly lower in patients with versus patients without a recurrence (CD3: 261. cells/mm2 [95CI 22.4; 467.2]; CD8: 189.6 cells/mm2 [95CI 2.0;462.0]). Conclusion T cell density decreases following NAC in MIBC patients with poor clinical outcome. Further research is needed to investigate whether this decrease in T cell density affects the efficacy of subsequent checkpoint inhibitors. Précis The major aim of this study was to improve our understanding of the immune-modulating effects of NAC in patients with MIBC. We reveal a decline in intratumoral CD3+ T cell density following NAC in patients with an early recurrence.

Homologous recombination repair deficiency (HRD) is observed in 10% of patients with castrate-resistant prostate cancer (PCa). Preliminary data suggest that HRD-PCa might be more responsive to immune checkpoint inhibitors (ICIs). In this study, we compare the tumor immune landscape and peripheral T cell receptor (TCR) repertoire of patients with and without HRD-PCa to gain further insight into the immunogenicity of HRD-PCa. Immunohistochemistry was performed on tumor tissue of 81 patients, including 15 patients with HRD-PCa. Peripheral TCR sequencing was performed in a partially overlapping cohort of 48 patients, including 16 patients with HRD-PCa. HRD patients more frequently had intratumoral CD3 + , CD3 + CD8 − FoxP3 − or Foxp3 + TILs above median compared to patients without DNA damage repair alterations (DDRwt; CD3 + and Foxp3 + : 77% vs 35%, p = .013; CD3 + CD8 − FoxP3 − : 80% vs 44%, p = .031). No significant difference in CD8 + TILs or PD-L1 expression was observed. In peripheral blood, HRD patients displayed a more diverse TCR repertoire compared to DDRwt patients (p = .014). Additionally, HRD patients shared TCR clusters with low generation probability, suggesting patient-overlapping T cell responses. A pooled analysis of clinical data from 227 patients with molecularly characterized PCa suggested increased efficacy of ICIs in HRD-PCa. In conclusion, patients with HRD-PCa display increased TIL density and an altered peripheral TCR repertoire. Further research into the efficacy of ICIs and the presence of shared neoantigens in HRD-PCa is warranted.