Recent 3D analyses reported an abundance of small β-cell-rich endocrine objects (EOs) in the human pancreas. Here, we used archival, immunolabelled 2D pancreas sections to assess morphological EO parameters in donors with or without type 1 diabetes (T1D), varying in age and disease duration. We confirm that abundant small, insulin-positive EOs are present in donors without diabetes and comprise most of the pancreatic endocrine area in early life. Small EOs are virtually absent in individuals with T1D, and this effect is most pronounced in children diagnosed with T1D in their earliest years. We conclude that extra-islet β-cells are impacted in T1D development, and their early loss is a characteristic feature. This finding has important implications that will inform future screening and treatment strategies for T1D.

Small things matter: Lack of extra-islet β-cells in Type 1 diabetes

Dryad DOI: https://doi.org/10.5061/dryad.6q573n6b5

*** Manuscript information ***

Manuscript title: Small things matter: Lack of extra-islet β-cells in Type 1 diabetes

Authors: Kathryn Murrall, Teifion Luckett et al.

Corresponding author: Sarah Richardson

Corresponding author email. s.richardson@exeter.ac.uk

README created on: 2025-08-14

*** Abbreviations ***

EO = endocrine object: comprised of a single or contiguous clusters of endocrine cells (labelled with either Chromogranin A or Insulin/Glucagon)

CgA = Chromogranin A

Ins/Gluc = Insulin Glucagon

T1D = Type 1 Diabetes

*** Research scope ***

Restricted tissue availability, particularly from donors in the first decade of life, has limited our understanding of the human pancreas and how it develops. The early changes that may underpin conditions such as type 1 diabetes, an autoimmune disease that targets the insulin-producing β-cells, are even less clear. Therefore, we sought to investigate how the endocrine architecture of the pancreas develops post-birth and whether this is impacted in individuals with T1D. We also investigated whether there are age-related differences in pancreas histology related to age at onset.

*** Project objective ***

Define the endocrine cell topography in 2D in 458 pancreas sections from 250 donors across the life course, held in multiple biobanks, including individuals with, or at-risk of developing, type 1 diabetes.

*** Methodology ***

458 whole slide images of pancreas tissue sections were scanned. HALO-AI densenet-v2 classifiers were trained to demarcate the pancreas tissue area, acinar and endocrine compartments, whilst filtering out artefacts. For tissue immunolabelled with insulin and glucagon, the HALO 3.6 area quantification algorithm was applied to calculate the insulin and glucagon positively labelled area within each EO. Individual EOs were attributed a bin size (EO bin = log2(EO area mm2 /170mm2) and rounded down to the nearest integer. Summary statistics were generated for each EO bin size at the individual whole slide image level and at the donor level.

#================================ File descriptions ==============================#

Note: all EO tabular data are wide.

* DONOR-LEVEL DATA *

1. insgluc_donor_level.csv

EO bin summary data at the donor level for pancreas sections immunolabelled with insulin & glucagon

2. insgluc_donor_level_byEOcontent.csv

EO bin summary data at the donor level (grouped by endocrine content) for pancreas sections immunolabelled with insulin & glucagon.

3. insgluc_donor_level_hormone_area.csv

Donor level data for % hormone (insulin or glucagon) positive area as a proportion of the total tissue area within EO bin sizes.

4. cga_EO_data.csv

EO bin summary data for each individual pancreas section (n=1 per donor) immunolabelled with Chromogranin A & CK19

* IMAGE-LEVEL DATA *

5. insgluc_image_level.csv

EO bin summary data for each individual pancreas section immunolabelled with insulin & glucagon

6. insgluc_image_level_byEOcontent.csv

EO bin summary data (grouped by endocrine content) for each individual pancreas section immunolabelled with insulin & glucagon.

7. insgluc_image_level_hormone_area.csv

Image-level data for each individual pancreas section. % hormone (insulin or glucagon) positive area as a proportion of the total tissue area within EO bin sizes.

* SPATIAL PLOTS *

8. spatial_plot_colour_size.zip

A compressed folder containing spatial plots for each pancreas section immunolabelled with insulin & glucagon. Points are coloured and sized according to EO bin size.

9. spatial_plot_endocrine_content.zip

A compressed folder containing spatial plots for each pancreas section immunolabelled with insulin & glucagon. Points are coloured by endocrine content and sized according to EO bin size.

CLINICAL METADATA

10. donor_clinical_data.csv

Descriptions of relevant clinical information for the donors studied.

#================ Calculations performed to generate donor level data ==============#

1. Mean EO summary statistics were generated for each pancreas location where multiple whole slide images exist.

2. Mean EO summary statistics of all pancreas locations present to generate donor-level EO summary statistics.

NOTE: donor level summary data was used for data visualisation in the manuscript.

#============================ Column heading descriptions ==========================#

unique_scan_id = all 413 scans in the insgluc dataset are designated a unique ID from 1 to 413.

caseID = ID used to identify donors.

pancreas_location = Pancreas tissue from the head (PH), body (PB), or tail (PT), or unknown location (PO)

pancreas_tissue_area_mm2 = measured pancreas tissue area from whole slide image analysis.

total_EO_count = number of EOs in each pancreas tissue section

total_endocrine_area_mm2 = sum of all EO area (mm2)

percentage_endocrine_area = % of the tissue area made up by EOs

absolute_EO_density = EO density for all objects across the whole pancreas tissue section.

bin0 - bin9 = EO's designated bin based on the formula and rounded down to the nearest integer. 170µm² represents the average cell size of an EO (see manuscript for reference).

log2(EO area mm2/ 170 mm2)

endocrine_content = If the insulin or glucagon area that exceeds the detection threshold is greater than 40µm², then the EO will be designated one of three classes: Ins+Gluc-, Ins+Gluc+, Ins-Gluc+. This describes the endocrine cells present within an EO.

bin0:9_count = Number of EOs belonging to that bin size.

bin0:9_proportion_of_EO_count = (number of EO in that bin size / total EO count) * 100

bin0:9_proportion_of_endocrine_area = (EO area made up by that bin size / total endocrine area of the pancreas tissue section) * 100

bin0:9_density_mm2 = number of EOs belonging to that bin size / pancreas tissue area mm²

bin0:9_circularity = mean circularity of EOs from that bin size (calculated in QuPath)

bin0:9_solidity = mean solidity of EO's from that bin size (calculated in QuPath)

hormone = refers to the hormone selected for positive area quantification.

bin0:9_hormone_area_mm2 = sum of hormone-positive area within all EOs of that bin size

bin0:9_percentage_hormone_area = (hormone positive area within all EOs of that bin size / pancreas tissue area) * 100

Human subjects data

This repository contains de-identified data from human subjects. caseIDs for each donor were generated by the respective biobanks referenced in this dataset. No more than 3 indirect donors identified were used publicised to preserve anonymity. Further information on study participant details is as follows:

Pancreas tissue from human organ donors with or without T1D was procured for the Network for Pancreatic Organ Donors with Diabetes (nPOD) program at the University of Florida (RRID: SCR_014641, https://www.jdrfnpod.org) and processed in accordance with the established standard operating procedures of the nPOD/OPPC as approved by the University of Florida Institutional Review Board (IRB201600029) with strict adherence to the guidelines of the United Network for Organ Sharing (UNOS) and federal regulations, with informed consent obtained from each donor’s legal representative.

The Exeter Archival Diabetes Biobank (EADB) is held with ethical permission from the West of Scotland Research Ethics Committee (ref: 25/WS/0017; IRAS Project ID: 354341).

Human fetal tissues were obtained from the Human Developmental Biology Resource (HDBR). These samples were collected with appropriate maternal written informed consent and approval from the Newcastle and North Tyneside (Newcastle University) and London—Fulham (UCL) Research NHS Health Authority Joint Ethics Committees (Ref: 23/NE/0135; IRAS Project ID: 330783). HDBR is regulated by the UK Human Tissue Authority (HTA; www.hta.gov.uk) and operates in accordance with the relevant HTA Codes of Practice.

The DiViD study was approved by The Norwegian Governments Regional Ethics Committee (IRB 0000 1870). Written informed consent was obtained from all cases after oral and written information from the diabetologist and the surgeon separately. Three women and three men, three to nine weeks after diagnosis, (aged 24-35 years) participated.

Organs for the MRC QUOD Whole Pancreas Biobank were retrieved after informed and written donor family consent in compliance with the UK Human Tissue Act of 2004 under specific ethical approvals by the UK Human Research Authority (05/MRE09/48 and 16NE0230). EUnPOD: the studies involving human participants were reviewed and approved by local ethics committee of the University of Pisa (Italy). Pancreata not suitable for organ transplantation were obtained with informed written consent by organ donors’ next-of-kin and processed with the approval of the local ethics committee.

Small things matter: Lack of extra-islet β-cells in Type 1 diabetes

Data files

Abstract

Small things matter: Lack of extra-islet β-cells in Type 1 diabetes

Human subjects data