Tumour Infiltrating Lymphocytes in Follicular Lymphoma
The composition of the tumour microenvironment in follicular lymphoma (FL) is a relevant factor in determining disease progression and treatment response. This dataset is a collection of 349 FL diagnostic tissue micro-array (TMA) cores from 130 patients, stained using multi-plex immunofluorescence for: • T helper cells (CD4+) • Cytotoxic T cells (CD8+) • T regulatory cells (Tregs [CD4+FOXP3+]) • Macrophages (CD68+) • PD1+ lymphocytes • Mature B cells/follicular dendritic cells (CD21+) • DAPI (4′,6-diamidino-2-phenylindole) nuclear counterstain Cohort -------- FL patients according to the WHO 2008 classification were identified from the archives of The Christie NHS Foundation Trust, Manchester, UK. The study was conducted with approval by the North-West Multi-centre Ethics Committee (03/08/016) and according to the Declaration of Helsinki. Examination of the records of 350 FL patients in a random order identified 262 patients meeting the inclusion criteria: adult patients with previously untreated FL; diagnosed from incisional or excisional biopsy; and treated at first presentation with radiotherapy, watchful waiting or a combination of chemotherapy and rituximab immunotherapy. Pre-treatment biopsies were requested for 262 patients, of which 130 had sufficient tissue for analysis. A histological diagnosis of FL was confirmed by an expert haemato-pathologist (R.B). Image format and software compatibility --------------------------------------------- Each image represents a single TMA core as a 32-bit depth multi-page .tif file. Each image is 2420x2420 pixels and contains 8 channels in this order: • DAPI (nuclear) • 650 fluorophore signal for CD68 marker (membrane) • 570 fluorophore signal for CD21 marker (membrane) • 540 fluorophore signal for CD8 marker (membrane) • 690 fluorophore signal for PD1 marker (membrane) • 620 fluorophore signal for CD4 marker (membrane) • 520 fluorophore signal for FOXP3 (nuclear) • Auto-fluorescence Each channel contains the intensity signal of a fluorophore, in units normalised by exposure. These intensities are non-negative float numbers. There are several software compatible with this format, such as the free open-source QuPath 0.2.0-m4 and commercial inForm 2.4. Annotations ------------- Annotations are provided for nuclear segmentation. In a set of 41 small patches the outlines have been drawn manually for 69780 nuclei. A patch from the DAPI channel and corresponding label are given as .tif images. Labels are integer mask arrays where all pixels belonging to each nucleus are randomly assigned an integer number. Some patches contain no stained nuclei, as negative controls. Data structure ---------------- • FL_0-129 zip files: Multi-plex spectrally unmixed images for 130 patients, each zip file a single patient • nuclear_segmentation_annotations.zip: The nuclear segmentation annotations split into training/ validation and test set.
Steps to reproduce
Immunofluorescence method --------------------------------- Regions of interest were identified by a pathologist (R.B) and cores were extracted in triplicate from the formalin-fixed, paraffin embedded (FFPE) blocks to construct five TMA blocks. The multi-plex immunofluorescence protocol is publicly available in dx.doi.org/10.17504/protocols.io.49ygz7w. Briefly, 4μm tissue sections underwent 6 repeated cycles of indirect immunofluorescence with tyramide signal amplification using the Ultra Discovery (Roche, Basel, Switzerland) auto-stainer. The OpalTM fluorophores (520, 540, 570, 620, 650, 690 [Akoya Biosciences, Menlo Park, CA, USA]) were used for signal detection. After each cycle the heat retrieval step that followed stripped the primary and secondary antibodies, while the fluorophores remained on the epitopes. DAPI was used as a counterstain to observe nucleated cells. Scanning and spectral unmixing ------------------------------------ Sections were scanned with a Vectra 3.5 microscope (Akoya Biosciences) multispectrally at 20x magnification (0.49μm/ pixel). Phenochart 1.0.8 software (Akoya Biosciences) was used to recognise the TMA grid and enable scanning each core individually. After manual inspection some cores were excluded because of poor quality and artefacts, and 349 cores remained. Linear spectral unmixing to separate the 6 fluorophore and DAPI spectra was carried out in inForm 2.4 software (Akoya Biosciences). For this purpose a spectral library containing the individual fluorophore spectra was pre-built using single-plex assays of positive control samples. An unstained control to obtain the auto-fluorescence spectrum is also included.