henhaobio
Sign in

bioRxivPathologyPreprint

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

Kathryn MurralliD, Teifion LuckettiD, Christiana LekkaiD, Christine S. FlaxmaniD, Rebecca C. WyattiD, Pouria AkhbariiD, Irina KusmartsevaiD, Stephanie L. HunteriD, Pia LeeteiD, Isabel BurniD, Elena Osokina, James ShawiD, Noel G. MorganiD, Sarah J. RichardsoniD

PDFDOIData
In-depth full-text digest available · Jump to digest ↓

Abstract

Recent 3D analyses reported abundant, small beta-cell-rich endocrine objects (EOs) in the human pancreas. Here, we assessed morphological parameters of >262,000 EOs in pancreas sections from 220 donors with or without type 1 diabetes (T1D), ranging in age and disease duration. We observe many insulin (Ins)+/glucagon (Gluc)-EOs in donors without diabetes. Their relative contribution to the total endocrine area is greatest in early life (0-2y) but reduces thereafter. Strikingly, we show the virtual absence of Ins+Gluc- EOs in individuals with T1D, where only the medium and large EOs retain beta cells. We also report a lower EO density in T1D, especially in individuals diagnosed in early life. These findings suggest that extra-islet beta cells are impacted in the development of T1D, and their early loss is a characteristic feature. This new understanding has important implications for defining beta-cell mass, which may inform future screening and treatment strategies in T1D. HighlightsO_LIThe present extensive 2D studies confirm and extend recent 3D analyses of human pancreata, demonstrating that 50% of endocrine objects (EOs) are much smaller than classical islets of Langerhans and consist predominantly of beta cells (Ins+). C_LIO_LISmall insulin+ EOs comprise the largest proportion of the total endocrine area in early childhood and persist throughout the life course in donors without diabetes. C_LIO_LIThere is a shift towards larger EO size with increasing age, with the most pronounced changes in size occurring in the first few years of life. C_LIO_LISmall Ins+ EOs are virtually absent in individuals with type 1 diabetes, while the persisting EOs with beta cells are larger, suggesting a selective destruction of beta cells in small EOs. C_LIO_LIDevelopment of type 1 diabetes, particularly at an early age, is associated with fewer larger EOs in adulthood. This implies that the lack or early destruction of small Ins+ EOs may be detrimental to the generation of larger EOs. C_LI

In-depth summary

Claude Opus 5.5 · xhigh effort · 2 Nov 2026 · from the full text of version 1 (17 Apr 2025) · CC BY 4.0

Image analysis of 262,832 pancreatic endocrine objects across 220 donors reveals that small beta-cell-only clusters are abundant in early childhood but virtually absent in type 1 diabetes.

Histological and isolated-islet studies of human pancreas development and type 1 diabetes (T1D) have focused on medium and large islets because single endocrine cells and small clusters (<50 µm diameter) are lost during islet isolation. Profiling endocrine objects across the human life course (0-72 years) and across diabetes stages shows how extra-islet beta-cell clusters shift during postnatal growth and are selectively depleted in T1D, especially in childhood-onset disease.

Key results

  1. ShownSmall insulin-positive, glucagon-negative (Ins+Gluc-) endocrine objects (EOs) are the most frequent EO class in non-diabetic (ND) pancreata but are virtually absent in T1D, where residual beta cells are confined to medium and large EOs (bins 4-9).

    Across 114 T1D donors (88,617 EOs) versus 62 ND donors (109,936 EOs), Ins+Gluc- EOs represent 5.8±2.5% vs 54±3.3% of total EO count (p<0.001) and 1.6±0.9% vs 17.2±2.5% of total EO area (p<0.001), whereas Ins-Gluc+ EOs rise from 8.9±1.6% to 81.6±2.1% of EO count (p<0.001). · Hormone immunolabelling-based class splitting of…; Beta cells as single cells or in small clusters…

  2. ShownIn healthy donors, EO density and proportional endocrine area decline over the first 10 years of life, while the endocrine profile shifts from small EOs toward larger islets with age.

    Across 45 CgA/CK19-labelled ND sections (42,793 EOs) and 62 Ins/Gluc-labelled ND donors aged 0-72 years (243,326 EOs), small EOs in bins 0-3 (<50 µm minimum Feret diameter, lost during standard islet isolation) comprise 76.3-98.6% (73-87% in the 0-72y cohort) of total EOs and 15.9-36.7% (19-49%) of endocrine area. · Endocrine object density and area decrease…

  3. ShownYounger age at T1D clinical onset (<13 years, endotype 1) is associated with more severe loss of insulin-containing EOs and permanently lower density of medium and large islets compared to later onset (>=13 years, endotype 2).

    In short-duration T1D (<2y, n=70), Ins+Gluc+ EOs account for 31.6% of total endocrine area in donors diagnosed at 1-12y (vs 82% in ND) compared to 57.6% in those diagnosed at >=13y (vs 82.4% in ND); in longer-duration T1D (>=2y), medium-to-large EO density (bins 3-8) is significantly lower in <13y vs >=13y onset. · Age at diagnosis of type 1 diabetes correlates…; There is a lower density of large islets in…; Residual Ins+ area is greater in individuals who…

  4. SuggestedSingle-autoantibody-positive donors without diabetes have normal small EO densities, whereas donors with multiple autoantibodies including IA2A and histological features of progression show early shifts in small EO classes.

    Across 15 autoantibody-positive (AAb+) donors, single-AAb donors do not differ from age-matched ND controls, whereas 18-year-old donors with IA2A plus a second autoantibody (n=3) have significantly fewer Ins+Gluc- EOs in bins 0 and 2 and more Ins-Gluc+ EOs in bins 0-1 as a proportion of total EO count. · Endocrine object parameters are not impacted in…; Endocrine object parameters are impacted in…

Methods, data and reproducibility

Whole-slide immunohistochemistry scans of human pancreas sections stained for CgA/CK19, insulin/glucagon, or insulin/glucagon/CD45 were segmented with DenseNet V2 convolutional neural networks in HALO AI to quantify tissue compartments and individual endocrine objects (EOs). EO areas were divided by 170 µm^2 and log2-transformed into size bins, with morphology analyzed in QuPath and R.

  • Whole-slide immunohistochemistry scans of human pancreas sections across the life course (0-72y; ND, AAb+, and T1D) from EADB, HDBR, nPOD, Seattle, EUnPOD, and DiViD biobanks · human · n = 334 tissue sections from 220 unique donors (262,832 EOs) · PRJNA839725
  • CgA/CK19-labelled pancreas tissue sections from fetal and neonatal/paediatric ND donors (HDBR and EADB) · human · n = 45 donors (10 fetal, 35 neonatal/paediatric; 42,793 EOs)
  • Insulin/glucagon (n=106) and insulin/glucagon/CD45 (n=85) labelled pancreas sections from ND, T1D, and AAb+ donors · human · n = Ins/Gluc n=106, Ins/Gluc/CD45 n=85 (62 ND, 114 T1D, 15 AAb+)
Methods & tools
HALO / HALO AI 4.0 (DenseNet V2), QuPath v5.0 / v5.1, PhenoImager HT / 2.0, PhenoChart 2.0, inForm, Aperio CS2, R (v4.1 / v4.4.1), ggplot2 (v3.5.1), rstatix (v0.7.2), emmeans (v1.11.0), car (v3.1.2)
Reagents & environment
HALO 3.6 / HALO AI 4.0, QuPath v5.0/v5.1, R v4.1/v4.4.1 with ggplot2 v3.5.1, rstatix v0.7.2, emmeans v1.11.0, and car v3.1.2.
Reproducibility gaps
Spatial imaging data (.qptiff/.svs and .csv), trained HALO AI classifiers (.onnx), annotation files (.haloannotations), and R scripts are not publicly deposited and are available only from the lead author upon request; tissue samples require biobank approval.
Limitations (4)
  • The cross-sectional study design across donor biobanks precludes direct longitudinal inference of endocrine object loss or transition over time within individuals. · stated by the authors
  • Archival sections stained only for insulin and glucagon cannot rule out the presence of other endocrine cell subsets (beyond alpha and beta cells) in small EOs. · stated by the authors
  • Only 15 autoantibody-positive donors were available, 80% (12/15) of whom were aged >=18 years, and clinical glucose control measurements were unavailable for staging. · stated by the authors
  • Quantification was performed on 2D tissue sections sampled from a mixture of pancreatic anatomical regions across post-mortem, organ-donor, and live-donor biobanks, which may introduce regional variation. · stated by the authors

Read this if you study pancreatic beta-cell mass, islet architecture, or type 1 diabetes pathogenesis and want quantitative histological evidence on extra-islet endocrine cell clusters across human development.

Citations
11

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISSeiron '18Rippa '25Lehrstrand '25Huber '25Drotar '24Liang '26
81% semantic matchPathology
P. Seiron, A. Wiberg, L. Krogvold+4 more·bioRxiv·Nov 29, 2018