Image-Based Machine Learning Algorithms for Disease Characterization in the Human Type 1 Diabetes Pancreas.

Emerging data suggest that type 1 diabetes affects not only the ß-cell-containing islets of Langerhans, but also the surrounding exocrine compartment. Using digital pathology, machine learning algorithms were applied to high-resolution, whole-slide images of human pancreata to determine whether the tissue composition in individuals with or at risk for type 1 diabetes differs from those without diabetes. Transplant-grade pancreata from organ donors were evaluated from 16 nondiabetic autoantibody-negative controls, 8 nondiabetic autoantibody-positive subjects with increased type 1 diabetes risk, and 19 persons with type 1 diabetes (0 to 12 years' duration). HALO image analysis algorithms were implemented to compare architecture of the main pancreatic duct as well as cell size, density, and area of acinar, endocrine, ductal, and other nonendocrine, nonexocrine tissues. Type 1 diabetes was found to affect exocrine area, acinar cell density, and size, whereas the type of difference correlated with the presence or absence of insulin-positive cells remaining in the pancreas. These changes were not observed before disease onset, as indicated by modeling cross-sectional data from pancreata of autoantibody-positive subjects and those diagnosed with type 1 diabetes. These data provide novel insights into anatomic differences in type 1 diabetes pancreata and demonstrate that machine learning can be adapted for the evaluation of disease processes from cross-sectional data sets.

Hospital time prior to death and pancreas histopathology: implications for future studies.

AIMS/HYPOTHESIS: Diabetes research studies routinely rely upon the use of tissue samples from human organ donors. It remains unclear whether the length of hospital stay prior to organ donation affects the presence of cells infiltrating the pancreas or the frequency of replicating beta cells. METHODS: To address this, 39 organ donors without diabetes were matched for age, sex, BMI and ethnicity in groups of three. Within each group, donors varied by length of hospital stay immediately prior to organ donation (<3 days, 3 to <6 days, or ≥6 days). Serial sections from tissue blocks in the pancreas head, body and tail regions were immunohistochemically double stained for insulin and CD45, CD68, or Ki67. Slides were electronically scanned and quantitatively analysed for cell positivity. RESULTS: No differences in CD45+, CD68+, insulin+, Ki67+ or Ki67+/insulin+ cell frequencies were found when donors were grouped according to duration of hospital stay. Likewise, no interactions were observed between hospitalisation group and pancreas region, age, or both; however, with Ki67 staining, cell frequencies were greater in the body vs the tail region of the pancreas (∆ 0.65 [unadjusted 95% CI 0.25, 1.04]; p = 0.002) from donors <12 year of age. Interestingly, frequencies were less in the body vs tail region of the pancreas for both CD45+ cells (∆ -0.91 [95% CI -1.71, -0.10]; p = 0.024) and insulin+ cells (∆ -0.72 [95% CI -1.10, -0.34]; p < 0.001). CONCLUSIONS/INTERPRETATION: This study suggests that immune or replicating beta cell frequencies are not affected by the length of hospital stay prior to donor death in pancreases used for research. DATA AVAILABILITY: All referenced macros (adopted and developed), calculations, programming code and numerical dataset files (including individual-level donor data) are freely available on GitHub through Zenodo at https://doi.org/10.5281/zenodo.1034422.

The influence of type 1 diabetes on pancreatic weight.

AIMS/HYPOTHESIS: Previous studies of pancreases obtained at autopsy or by radiography note reduced pancreas weight (PW) and size, respectively, in type 1 diabetes; this finding is widely considered to be the result of chronic insulinopenia. This literature is, however, limited with respect to the influence of age, sex, anthropometric factors and disease duration on these observations. Moreover, data are sparse for young children, a group of particular interest for type 1 diabetes. We hypothesised that the pancreas-to-body weight ratio would normalise confounding inter-subject factors, thereby permitting better characterisation of PW in type 1 diabetes. METHODS: Transplant-grade pancreases were recovered from 216 organ donors with type 1 diabetes (n = 90), type 2 diabetes (n = 40) and no diabetes (n = 86). Whole-organ and head, body and tail weights were determined. The relative PW (RPW; PW [g] / body weight [kg]) was calculated and tested for normalisation of potential differences due to age, sex and BMI. RESULTS: PW significantly correlated with body weight in control donors (R (2) = 0.76, p < 0.001) while RPW (1.03 ± 0.36, mean ± SD) did not significantly differ across ages (0-58 years). Donors with type 1 diabetes (0.57 ± 0.18, p < 0.001), but not those with type 2 diabetes (0.93 ± 0.30), had significantly lower RPW. The relative weights of each pancreatic region from donors with type 1 diabetes were significantly smaller than those of regions from control donors and donors with type 2 diabetes (p < 0.001). Perhaps most interestingly, the RPW was not significantly associated with duration of type 1 diabetes or type 2 diabetes. CONCLUSIONS/INTERPRETATION: RPW allows for comparisons across a wide range of donor ages by eliminating confounding variables. These data validate an interesting feature of the type 1 diabetes pancreas and underscore the need for additional studies to identify the mechanistic basis for this finding, including those beyond the chronic loss of endogenous insulin secretion.

Islet cell hyperexpression of HLA class I antigens: a defining feature in type 1 diabetes.

AIMS/HYPOTHESIS: Human pancreatic beta cells may be complicit in their own demise in type 1 diabetes, but how this occurs remains unclear. One potentially contributing factor is hyperexpression of HLA class I antigens. This was first described approximately 30 years ago, but has never been fully characterised and was recently challenged as artefactual. Therefore, we investigated HLA class I expression at the protein and RNA levels in pancreases from three cohorts of patients with type 1 diabetes. The principal aims were to consider whether HLA class I hyperexpression is artefactual and, if not, to determine the factors driving it. METHODS: Pancreas samples from type 1 diabetes patients with residual insulin-containing islets (n = 26) from the Network for Pancreatic Organ donors with Diabetes (nPOD), Diabetes Virus Detection study (DiViD) and UK recent-onset type 1 diabetes collections were immunostained for HLA class I isoforms, signal transducer and activator of transcription 1 (STAT1), NLR family CARD domain containing 5 (NLRC5) and islet hormones. RNA was extracted from islets isolated by laser-capture microdissection from nPOD and DiViD samples and analysed using gene-expression arrays. RESULTS: Hyperexpression of HLA class I was observed in the insulin-containing islets of type 1 diabetes patients from all three tissue collections, and was confirmed at both the RNA and protein levels. The expression of ß2-microglobulin (a second component required for the generation of functional HLA class I complexes) was also elevated. Both 'classical' HLA class I isoforms (i.e. HLA-ABC) as well as a 'non-classical' HLA molecule, HLA-F, were hyperexpressed in insulin-containing islets. This hyperexpression did not correlate with detectable upregulation of the transcriptional regulator NLRC5. However, it was strongly associated with increased STAT1 expression in all three cohorts. Islet hyperexpression of HLA class I molecules occurred in the insulin-containing islets of patients with recent-onset type 1 diabetes and was also detectable in many patients with disease duration of up to 11 years, declining thereafter. CONCLUSIONS/INTERPRETATION: Islet cell HLA class I hyperexpression is not an artefact, but is a hallmark in the immunopathogenesis of type 1 diabetes. The response is closely associated with elevated expression of STAT1 and, together, these occur uniquely in patients with type 1 diabetes, thereby contributing to their selective susceptibility to autoimmune-mediated destruction.

The Juvenile Diabetes Research Foundation Network for Pancreatic Organ Donors with Diabetes (nPOD) Program: goals, operational model and emerging findings.

nPOD actively promotes a multidisciplinary and unbiased approach toward a better understanding of T1D and identify novel therapeutic targets, through its focus on the study of human samples. Unique to this effort is the coordination of collaborative efforts and real-time data sharing. Studies supported by nPOD are providing direct evidence that human T1D isa complex and heterogeneous disease, in which a multitude of pathogenic factors may be operational and may contribute to the onset of the disease. Importantly, the concept that beta cell destruction is almost completed and that the autoimmune process is almost extinguished soon after diagnosis is being challenged. nPOD investigators are exploring the hypothesis that beta cell dysfunction may also be a significant cause of hyperglycemia, at least around the time of diagnosis, and are uncovering novel molecules and pathways that are linked to the pathogenesis and etiology of human T1D. The validation of therapeutic targets is also a key component of this effort, with recent and future findings providing new strategic direction for clinical trials.

Methylglyoxal Adducts Are Prognostic Biomarkers for Diabetic Kidney Disease in Patients With Type 1 Diabetes.

More than 30% of patients with type 1 diabetes develop diabetic kidney disease (DKD), which significantly increases mortality risk. The Diabetes Control and Complications Trial (DCCT) and follow-up study, Epidemiology of Diabetes Interventions and Complications (EDIC), established that glycemic control measured by HbA1c predicts DKD risk. However, the continued high incidence of DKD reinforces the urgent need for additional biomarkers to supplement HbA1c. Here, we assessed biomarkers induced by methylglyoxal (MG), a metabolic by-product that forms covalent adducts on DNA, RNA, and proteins, called MG adducts. Urinary MG adducts were measured in samples from patients with type 1 diabetes enrolled in DCCT/EDIC who did (case patients; n = 90) or did not (control patients; n = 117) develop DKD. Univariate and multivariable analyses revealed that measurements of MG adducts independently predict DKD before established DKD biomarkers such as glomerular filtration rate and albumin excretion rate. Elevated levels of MG adducts bestowed the greatest risk of developing DKD in a multivariable model that included HbA1c and other clinical covariates. Our work establishes a novel class of biomarkers to predict DKD risk and suggests that inclusion of MG adducts may be a valuable tool to improve existing predictors of complications like DKD prior to overt disease, and to aid in identifying at-risk individuals and personalized risk management.

Untangling the genetics of beta cell dysfunction and death in type 1 diabetes.

BACKGROUND: Type 1 diabetes (T1D) is a complex multi-system disease which arises from both environmental and genetic factors, resulting in the destruction of insulin-producing pancreatic beta cells. Over the past two decades, human genetic studies have provided new insight into the etiology of T1D, including an appreciation for the role of beta cells in their own demise. SCOPE OF REVIEW: Here, we outline models supported by human genetic data for the role of beta cell dysfunction and death in T1D. We highlight the importance of strong evidence linking T1D genetic associations to bona fide candidate genes for mechanistic and therapeutic consideration. To guide rigorous interpretation of genetic associations, we describe molecular profiling approaches, genomic resources, and disease models that may be used to construct variant-to-gene links and to investigate candidate genes and their role in T1D. MAJOR CONCLUSIONS: We profile advances in understanding the genetic causes of beta cell dysfunction and death at individual T1D risk loci. We discuss how genetic risk prediction models can be used to address disease heterogeneity. Further, we present areas where investment will be critical for the future use of genetics to address open questions in the development of new treatment and prevention strategies for T1D.

A multicenter study to standardize reporting and analyses of fluorescence-activated cell-sorted murine intestinal epithelial cells.

Fluorescence-activated cell sorting (FACS) is an essential tool for studies requiring isolation of distinct intestinal epithelial cell populations. Inconsistent or lack of reporting of the critical parameters associated with FACS methodologies has complicated interpretation, comparison, and reproduction of important findings. To address this problem a comprehensive multicenter study was designed to develop guidelines that limit experimental and data reporting variability and provide a foundation for accurate comparison of data between studies. Common methodologies and data reporting protocols for tissue dissociation, cell yield, cell viability, FACS, and postsort purity were established. Seven centers tested the standardized methods by FACS-isolating a specific crypt-based epithelial population (EpCAM+/CD44+) from murine small intestine. Genetic biomarkers for stem/progenitor (Lgr5 and Atoh 1) and differentiated cell lineages (lysozyme, mucin2, chromogranin A, and sucrase isomaltase) were interrogated in target and control populations to assess intra- and intercenter variability. Wilcoxon's rank sum test on gene expression levels showed limited intracenter variability between biological replicates. Principal component analysis demonstrated significant intercenter reproducibility among four centers. Analysis of data collected by standardized cell isolation methods and data reporting requirements readily identified methodological problems, indicating that standard reporting parameters facilitate post hoc error identification. These results indicate that the complexity of FACS isolation of target intestinal epithelial populations can be highly reproducible between biological replicates and different institutions by adherence to common cell isolation methods and FACS gating strategies. This study can be considered a foundation for continued method development and a starting point for investigators that are developing cell isolation expertise to study physiology and pathophysiology of the intestinal epithelium.

Low risk for diabetic complications in type 1 diabetes patients carrying a protective insulin gene variant.

Type 1 diabetes patients carrying a 'protective' insulin gene (INS) variant present a disease endotype with reduced insulin antibody titers, preserved beta cell function and improved glycemic control. We tested whether this protective INS variant associated with lowered risk for development of proliferative diabetic retinopathy (PDR) and diabetic kidney disease (DKD) as long-term diabetic complications. Insulin gene polymorphisms were evaluated in 1,363 type 1 diabetes patients participating in the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) study that compared intensive versus conventional insulin therapy in relation with development of PDR and DKD with a follow-up of over two decades. PDR and DKD were absent in type 1 diabetes patients carrying the protective INS variant and receiving intensive insulin therapy (the current standard of clinical care) 1-5 years from their diagnosis (n = 67; mean post-diagnosis follow up of 20.4 ± 1.6 years), versus 11 of 258 patients (4.3%) lacking this variant (20.4 ± 1.8 years follow up). In the secondary intervention group of the intensive therapy arm (1-15 years of disease), PDR was significantly less frequent in carriers of the protective INS variant than those without it (4 of 83 [4.8%] vs. 31 of 260 [11.9%]; p = 0.032; 26.1 ± 3.9 and 26.3 ± 4.1 years follow-up, respectively), whereas DKD frequencies were no different between those with or without this variant (5 of 83 [6.0%] vs. 11 of 260 [4.2%]). Carrying a copy of this protective INS variant further reduces the risk of diabetic complications achieved by intensive insulin therapy and marks a disease endotype with superior glycemic control, increased and extended beta cell function, and prevention of DKD and PDR.

A genomic data archive from the Network for Pancreatic Organ donors with Diabetes.

The Network for Pancreatic Organ donors with Diabetes (nPOD) is the largest biorepository of human pancreata and associated immune organs from donors with type 1 diabetes (T1D), maturity-onset diabetes of the young (MODY), cystic fibrosis-related diabetes (CFRD), type 2 diabetes (T2D), gestational diabetes, islet autoantibody positivity (AAb+), and without diabetes. nPOD recovers, processes, analyzes, and distributes high-quality biospecimens, collected using optimized standard operating procedures, and associated de-identified data/metadata to researchers around the world. Herein describes the release of high-parameter genotyping data from this collection. 372 donors were genotyped using a custom precision medicine single nucleotide polymorphism (SNP) microarray. Data were technically validated using published algorithms to evaluate donor relatedness, ancestry, imputed HLA, and T1D genetic risk score. Additionally, 207 donors were assessed for rare known and novel coding region variants via whole exome sequencing (WES). These data are publicly-available to enable genotype-specific sample requests and the study of novel genotype:phenotype associations, aiding in the mission of nPOD to enhance understanding of diabetes pathogenesis to promote the development of novel therapies.

Generation of human islet cell type-specific identity genesets.

Generation of surrogate cells with stable functional identities is crucial for developing cell-based therapies. Efforts to produce insulin-secreting replacement cells to treat diabetes require reliable tools to assess islet cellular identity. Here, we conduct a thorough single-cell transcriptomics meta-analysis to identify robustly expressed markers used to build genesets describing the identity of human α-, ß-, γ- and δ-cells. These genesets define islet cellular identities better than previously published genesets. We show their efficacy to outline cell identity changes and unravel some of their underlying genetic mechanisms, whether during embryonic pancreas development or in experimental setups aiming at developing glucose-responsive insulin-secreting cells, such as pluripotent stem-cell differentiation or in adult islet cell reprogramming protocols. These islet cell type-specific genesets represent valuable tools that accurately benchmark gain and loss in islet cell identity traits.

Tolerogenic dendritic cells pulsed with islet antigen induce long-term reduction in T-cell autoreactivity in type 1 diabetes patients.

Introduction: Restoration of immune tolerance may halt progression of autoimmune diseases. Tolerogenic dendritic cells (tolDC) inhibit antigen-specific proinflammatory T-cells, generate antigen-specific regulatory T-cells and promote IL-10 production in-vitro, providing an appealing immunotherapy to intervene in autoimmune disease progression. Methods: A placebo-controlled, dose escalation phase 1 clinical trial in nine adult patients with long-standing type 1 diabetes (T1D) demonstrated the safety and feasibility of two (prime-boost) vaccinations with tolDC pulsed with a proinsulin peptide. Immunoregulatory effects were monitored by antigen-specific T-cell assays and flow and mass cytometry. Results: The tolDC vaccine induced a profound and durable decline in pre-existing autoimmune responses to the vaccine peptide up to 3 years after therapy and temporary decline in CD4 and CD8+ T-cell responses to other islet autoantigens. While major leukocyte subsets remained stable, ICOS+CCR4+TIGIT+ Tregs and CD103+ tissue-resident and CCR6+ effector memory CD4+ T-cells increased in response to the first tolDC injection, the latter declining thereafter below baseline levels. Discussion: Our data identify immune correlates of mechanistic efficacy of intradermally injected tolDC reducing proinsulin autoimmunity in T1D.

Chronic marijuana usage by human pancreas donors is associated with impaired islet function.

We investigated the effect of chronic marijuana use, defined as 4 times weekly for more than 3 years, on human pancreatic islets. Pancreata from deceased donors who chronically used marijuana were compared to those from age, sex and ethnicity matched non-users. The islets from marijuana-users displayed reduced insulin secretion as compared to islets from non-users upon stimulation with high glucose (AUC, 3.41 ± 0.62 versus 5.14 ±0.47, p<0.05) and high glucose plus KCl (AUC, 4.48 ± 0.41 versus 7.69 ± 0.58, p<0.001). When human islets from chronic marijuana-users were transplanted into diabetic mice, the mean reversal rate of diabetes was 35% versus 77% in animals receiving islets from non-users (p<0.01). Immunofluorescent staining for cannabinoid receptor type 1 (CB1R) was shown to be colocalized with insulin and enhanced significantly in beta cells from marijuana-users vs. non-users (CB1R intensity/islet area, 14.95 ± 2.71 vs. 3.23 ± 0.87, p<0.001). In contrast, CB1R expression was not co-localized with glucagon or somatostatin. Furthermore, isolated islets from chronic marijuana-users appeared hypertrophic. In conclusion, excessive marijuana use affects islet endocrine phenotype and function in vitro and in vivo. Given the increasing use of marijuana, our results underline the importance of including lifestyle when evaluating human islets for transplantation or research.

DNA methylation mediates development of HbA1c-associated complications in type 1 diabetes.

Metabolic memory, the persistent benefits of early glycaemic control on preventing and/or delaying the development of diabetic complications, has been observed in the Diabetes Control and Complications Trial (DCCT) and in the Epidemiology of Diabetes Interventions and Complications (EDIC) follow-up study, but the underlying mechanisms remain unclear. Here, we show the involvement of epigenetic DNA methylation (DNAme) in metabolic memory by examining its associations with preceding glycaemic history, and with subsequent development of complications over an 18-yr period in the blood DNA of 499 randomly selected DCCT participants with type 1 diabetes who are also followed up in EDIC. We demonstrate the associations between DNAme near the closeout of DCCT and mean HbA1c during DCCT (mean-DCCT HbA1c) at 186 cytosine-guanine dinucleotides (CpGs) (FDR < 15%, including 43 at FDR < 5%), many of which were located in genes related to complications. Exploration studies into biological function reveal that these CpGs are enriched in binding sites for the C/EBP transcription factor, as well as enhancer/transcription regions in blood cells and haematopoietic stem cells, and open chromatin states in myeloid cells. Mediation analyses show that, remarkably, several CpGs in combination explain 68-97% of the association of mean-DCCT HbA1c with the risk of complications during EDIC. In summary, DNAme at key CpGs appears to mediate the association between hyperglycaemia and complications in metabolic memory, through modifying enhancer activity at myeloid and other cells.

Human pancreatic islets and diabetes research.

Human islet research is crucial to understanding the cellular biology of the pancreas in developing therapeutic options for diabetes patients and in attempting to prevent the development of this disease. The national Islet Cell Resource Center Consortium provides human pancreatic islets for diabetes research while simultaneously addressing the need to improve islet isolation and transplantation technologies. Since its inception in 2001, the consortium has supplied 297.6 million islet equivalents to 151 national and international scientists for use in clinical and laboratory projects. Data on the volume, quality, and frequency of shipments substantiate the importance of human islets for diabetes research, as do the number of funded grants for beta-cell projects and publications produced as a direct result of islets supplied by this resource. Limitations in using human islets are discussed, along with the future of islet distribution centers. The information presented here is instructive to clinicians, basic science investigators, and policy makers who determine the availability of funding for such work. Organ procurement coordinators also may find the information useful in explaining to donor families why research consent is so valuable.

Factors That Influence the Quality of RNA From the Pancreas of Organ Donors.

OBJECTIVES: Attaining high-quality RNA from the tissues or organs of deceased donors used for research can be challenging due to physiological and logistical considerations. In this investigation, METHODS: RNA Integrity Number (RIN) was determined in pancreatic samples from 236 organ donors and used to define high (≥6.5) and low (≤4.5) quality RNAs. Logistic regression was used to evaluate the potential effects of novel or established organ and donor factors on RIN. RESULTS: Univariate analysis revealed donor cause of death (odds ratio [OR], 0.35; 95% confidence interval [CI], 0.15-0.77; P = 0.01), prolonged tissue storage before RNA extraction (OR, 0.65; 95% CI, 0.52-0.79; P < 0.01), pancreas region sampled (multiple comparisons, P < 0.01), and sample type (OR, 0.32; 95% CI, 0.15-0.67; P < 0.01) negatively influenced outcome. Conversely, duration of final hospitalization (OR, 3.95; 95% CI, 1.59-10.37; P < 0.01) and sample collection protocol (OR, 8.48; 95% CI, 3.96-19.30; P < 0.01) positively impacted outcome. Islet RNA obtained via laser capture microdissection improved RIN when compared with total pancreatic RNA from the same donor (ΔRIN = 1.3; 95% CI, 0.6-2.0; P < 0.01). CONCLUSIONS: A multivariable model demonstrates that autopsy-free and biopsy-free human pancreata received, processed, and preserved at a single center, using optimized procedures, from organ donors dying of anoxia with normal lipase levels increase the odds of obtaining high-quality RNA.