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

Multiparametric MRI Scoring System of the Pancreas for the Diagnosis of Chronic Pancreatitis

BACKGROUND

Ductal features alone may not offer high diagnostic sensitivity or most accurate disease severity of chronic pancreatitis (CP).

PURPOSE

Diagnose CP based on multiparametric MRI and MRCP features.

STUDY TYPE

Prospective.

POPULATION

Between February 2019 and May 2021, 46 control (23 males, 49.3 ± 14.1 years), 45 suspected (20 males, 48.7 ± 12.5 years), and 46 definite (20 males, 53.7 ± 14.6 years) CP patients were enrolled at seven hospitals enrolled in the MINIMAP study. CP classification was based on imaging findings and clinical presentation.

FIELD STRENGTH AND SEQUENCES

1.5 T. T1-weighted (T1W) spoiled gradient echo, T1 map with variable flip angle, dual-echo Dixon, secretin-enhanced MRCP before and after secretin infusion.

ASSESSMENT

Dual-echo fat fraction (FF), T1 relaxation time, extracellular volume (ECV), T1 signal intensity ratio of the pancreas to the spleen (T1 score), arterial-to-venous enhancement ratio (AVR), pancreatic tail diameter (PTD), pancreas volume, late gadolinium enhancement, pancreatic ductal elasticity (PDE), and duodenal filling grade of secretin-enhanced MRCP were measured.

STATISTICAL TESTS

Logistic regression analysis generated CP-MRI and secretin-enhanced CP-SMRI scores. Receiver operating characteristics analysis was used to differentiate definite CP from control. Interobserver agreement was assessed using Lin's concordance correlation coefficient.

RESULTS

Compared to control, definite CP cohort showed significantly higher dual-echo FF (7% vs. 11%), lower AVR (1.35 vs. 0.85), smaller PTD (2.5 cm vs. 1.95 cm), higher ECV (28% vs. 38%), and higher incidence of PDE loss (6.5% vs. 50%). With the cut-off of >2.5 CP-MRI score (dual-echo FF, AVR, and PTD) and CP-SMRI score (dual-echo FF, AVR, PTD, and PDE) had cross-validated area under the curves of 0.84 (sensitivity 87%, specificity 68%) and 0.86 (sensitivity 89%, specificity 67%), respectively. Interobserver agreement for both CP-MRI and CP-SMRI scores was 0.74.

CONCLUSION

The CP-MRI and CP-SMRI scores yielded acceptable performance and interobserver agreement for the diagnosis of CP.

EVIDENCE LEVEL

1 TECHNICAL EFFICACY: Stage 2.

Advances in MRI of Chronic Pancreatitis

MRI and MRCP play an essential role in diagnosing CP by imaging pancreatic parenchyma and ducts. Quantitative and semi-quantitative MR imaging offers potential advantages over conventional MR imaging, including simplicity of analysis, quantitative and population-based comparisons, and more direct interpretation of disease progression or response to drug therapy. Using parenchymal imaging techniques may provide quantitative metrics for determining the presence and severity of acinar cell loss and aid in diagnosing CP. Given that the parenchymal changes of CP precede the ductal involvement, there would be a significant benefit from developing a new MRI/MRCP based, more robust diagnostic criteria combining ductal and parenchymal findings.

Impaired islet function and normal exocrine enzyme secretion occur with low inter-regional variation in type 1 diabetes

Histopathological heterogeneity in the human pancreas is well documented; however, functional evidence at the tissue level is scarce. Herein, we investigate in situ glucose-stimulated islet and carbachol-stimulated acinar cell secretion across the pancreas head (PH), body (PB), and tail (PT) regions in donors without diabetes (ND; n = 15), positive for one islet autoantibody (1AAb+; n = 7), and with type 1 diabetes (T1D; <14 months duration, n = 5). Insulin, glucagon, pancreatic amylase, lipase, and trypsinogen secretion along with 3D tissue morphometrical features are comparable across regions in ND. In T1D, insulin secretion and beta-cell volume are significantly reduced within all regions, while glucagon and enzymes are unaltered. Beta-cell volume is lower despite normal insulin secretion in 1AAb+, resulting in increased volume-adjusted insulin secretion versus ND. Islet and acinar cell secretion in 1AAb+ are consistent across the PH, PB, and PT. This study supports low inter-regional variation in pancreas slice function and, potentially, increased metabolic demand in 1AAb+.

Activation of ductal progenitor-like cells from adult human pancreas requires extracellular matrix protein signaling

Ductal progenitor-like cells are a sub-population of ductal cells in the adult human pancreas that have the potential to contribute to regenerative medicine. However, the microenvironmental cues that regulate their activation are poorly understood. Here, we establish a 3-dimensional suspension culture system containing six defined soluble factors in which primary human ductal progenitor-like and ductal non-progenitor cells survive but do not proliferate. Expansion and polarization occur when suspension cells are provided with a low concentration (5% v/v) of Matrigel, a sarcoma cell product enriched in many extracellular matrix (ECM) proteins. Screening of ECM proteins identified that collagen IV can partially recapitulate the effects of Matrigel. Inhibition of integrin α1β1, a major collagen IV receptor, negates collagen IV- and Matrigel-stimulated effects. These results demonstrate that collagen IV is a key ECM protein that stimulates the expansion and polarization of human ductal progenitor-like and ductal non-progenitor cells via integrin α1β1 receptor signaling.

Impaired islet function with normal exocrine enzyme secretion is consistent across the head, body, and tail pancreas regions in type 1 diabetes

Histopathological heterogeneity in human pancreas has been well documented; however, functional evidence at the tissue level is scarce. Herein we investigated in situ glucose-stimulated islet and carbachol-stimulated acinar cell secretion across the pancreas head (PH), body (PB), and tail (PT) regions in no diabetes (ND, n=15), single islet autoantibody-positive (1AAb+, n=7), and type 1 diabetes donors (T1D, <14 months duration, n=5). Insulin, glucagon, pancreatic amylase, lipase, and trypsinogen secretion along with 3D tissue morphometrical features were comparable across the regions in ND. In T1D, insulin secretion and beta-cell volume were significantly reduced within all regions, while glucagon and enzymes were unaltered. Beta-cell volume was lower despite normal insulin secretion in 1AAb+, resulting in increased volume-adjusted insulin secretion versus ND. Islet and acinar cell secretion in 1AAb+ were consistent across PH, PB and PT. This study supports low inter-regional variation in pancreas slice function and potentially, increased metabolic demand in 1AAb+.

Machine learning application in autoimmune diseases: State of art and future prospectives

Autoimmune diseases are a group of disorders resulting from an alteration of immune tolerance, characterized by the formation of autoantibodies and the consequent development of heterogeneous clinical manifestations. Diagnosing autoimmune diseases is often complicated, and the available prognostic tools are limited. Machine learning allows us to analyze large amounts of data and carry out complex calculations quickly and with minimal effort. In this work, we examine the literature focusing on the use of machine learning in the field of the main systemic (systemic lupus erythematosus and rheumatoid arthritis) and organ-specific autoimmune diseases (type 1 diabetes mellitus, autoimmune thyroid, gastrointestinal, and skin diseases). From our analysis, interesting applications of machine learning emerged for developing algorithms useful in the early diagnosis of disease or prognostic models (risk of complications, therapeutic response). Subsequent studies and the creation of increasingly rich databases to be supplied to the algorithms will eventually guide the clinician in the diagnosis, allowing intervention when the pathology is still in an early stage and immediately directing towards a correct therapeutic approach.

Diagnosis of chronic pancreatitis using semi-quantitative MRI features of the pancreatic parenchyma: results from the multi-institutional MINIMAP study

PURPOSE

To determine the diagnostic performance of parenchymal MRI features differentiating CP from controls.

METHODS

This prospective study performed abdominal MRI scans at seven institutions, using 1.5 T Siemens and GE scanners, in 50 control and 51 definite CP participants, from February 2019 to May 2021. MRI parameters included the T1-weighted signal intensity ratio of the pancreas (T1 score), arterial-to-venous enhancement ratio (AVR) during venous and delayed phases, pancreas volume, and diameter. We evaluated the diagnostic performance of these parameters individually and two semi-quantitative MRI scores derived using logistic regression: SQ-MRI Model A (T1 score, AVR venous, and tail diameter) and Model B (T1 score, AVR venous, and volume).

RESULTS

When compared to controls, CP participants showed a significantly lower mean T1 score (1.11 vs. 1.29), AVR venous (0.86 vs. 1.45), AVR delayed (1.07 vs. 1.57), volume (54.97 vs. 80.00 ml), and diameter of the head (2.05 vs. 2.39 cm), body (2.25 vs. 2.58 cm), and tail (1.98 vs. 2.51 cm) (p < 0.05 for all). AUCs for these individual MR parameters ranged from 0.66 to 0.79, while AUCs for the SQ-MRI scores were 0.82 and 0.81 for Model A (T1 score, AVR venous, and tail diameter) and Model B (T1 score, AVR venous, and volume), respectively. After propensity-matching adjustments for covariates, AUCs for Models A and B of the SQ-MRI scores increased to 0.92 and 0.93, respectively.

CONCLUSION

Semi-quantitative parameters of the pancreatic parenchyma, including T1 score, enhancement ratio, pancreas volume, diameter and multi-parametric models combining these parameters are helpful in diagnosis of CP. Longitudinal analyses including more extensive population are warranted to develop new diagnostic criteria for CP.

Longitudinal MRI Shows Progressive Decline in Pancreas Size and Altered Pancreas Shape in Type 1 Diabetes

CONTEXT

Individuals with type 1 diabetes (T1D) have a smaller pancreas, but longitudinal changes in pancreas size and shape are unclear.

OBJECTIVE

We monitored changes in pancreas size and shape after diagnosis with T1D.

DESIGN

We conducted a prospective cohort study at an academic medical center between 2014 and 2022.

PATIENTS AND HEALTHY CONTROLS

Individuals with T1D (n = 91) or controls (n = 90) underwent magnetic resonance imaging (MRI) of the pancreas, including longitudinal MRI in 53 individuals with new-onset T1D.

INTERVENTION

Interventions included MRI and continuous glucose monitoring (CGM).

MAIN OUTCOME MEASURES

Pancreas size and shape were measured from MRI. For participants who used CGM, measures of glycemic variability were calculated.

RESULTS

On longitudinal imaging, pancreas volume and pancreas volume index normalized for body weight declined during the first year after diagnosis. Pancreas volume index continued to decline through the fifth year after diagnosis. A cross-sectional study of individuals with diabetes duration up to 60 years demonstrated that pancreas size in adults negatively correlated with age and disease duration, whereas pancreas volume and pancreas volume index remained stable in controls. Pancreas volume index correlated inversely with low blood glucose index, a measure of risk for hypoglycemia. Pancreas shape was altered in individuals with T1D and further diverged from controls over the first 5 years after diagnosis. Pancreas size and shape are altered in nondiabetic individuals at genetic risk for T1D. Combined pancreas size and shape analysis better distinguished the pancreas of individuals with T1D from controls than size alone.

CONCLUSIONS

Pancreas size declines most rapidly near the clinical diagnosis of T1D and continues to decline throughout adulthood. Declines in pancreas size are accompanied by changes in pancreas shape.

The pathogenic "symphony" in type 1 diabetes: A disorder of the immune system, β cells, and exocrine pancreas

Type 1 diabetes (T1D) is widely considered to result from the autoimmune destruction of insulin-producing β cells. This concept has been a central tenet for decades of attempts seeking to decipher the disorder's pathogenesis and prevent/reverse the disease. Recently, this and many other disease-related notions have come under increasing question, particularly given knowledge gained from analyses of human T1D pancreas. Perhaps most crucial are findings suggesting that a collective of cellular constituents-immune, endocrine, and exocrine in origin-mechanistically coalesce to facilitate T1D. This review considers these emerging concepts, from basic science to clinical research, and identifies several key remaining knowledge voids.

Integrated Physiology of the Exocrine and Endocrine Compartments in Pancreatic Diseases: Workshop Proceedings

The Integrated Physiology of the Exocrine and Endocrine Compartments in Pancreatic Diseases workshop was a 1.5-day scientific conference at the National Institutes of Health (Bethesda, MD) that engaged clinical and basic science investigators interested in diseases of the pancreas. This report provides a summary of the proceedings from the workshop. The goals of the workshop were to forge connections and identify gaps in knowledge that could guide future research directions. Presentations were segregated into six major theme areas, including 1) pancreas anatomy and physiology, 2) diabetes in the setting of exocrine disease, 3) metabolic influences on the exocrine pancreas, 4) genetic drivers of pancreatic diseases, 5) tools for integrated pancreatic analysis, and 6) implications of exocrine-endocrine cross talk. For each theme, multiple presentations were followed by panel discussions on specific topics relevant to each area of research; these are summarized here. Significantly, the discussions resulted in the identification of research gaps and opportunities for the field to address. In general, it was concluded that as a pancreas research community, we must more thoughtfully integrate our current knowledge of normal physiology as well as the disease mechanisms that underlie endocrine and exocrine disorders so that there is a better understanding of the interplay between these compartments.

Development and validation of a risk score for diabetic kidney disease prediction in type 2 diabetes patients: a machine learning approach

PURPOSE

This study aims to develop and validate a risk score to predict the occurrence of DKD in individuals with type 2 diabetes using a machine learning (ML) approach.

METHODS

By implementing Recursive Feature Elimination with Cross-Validation (RFECV) and RFE on the Diabetes Clinic of Imam Khomeini Hospital Complex (IKHC) dataset, the most critical features were identified. These features were used in the multivariate logistic regression (LR) analysis, and the discrimination and calibration of the model were evaluated. Finally, external validation of the model was assessed.

RESULTS

The development dataset included 1907 type 2 diabetic patients, 763 of whom developed DKD over 5 years. The predictive model performed well in the development dataset by implementing RFECV with the RF algorithm and considering six features (AUC: 79%). Using these features, the LR-based risk score indicated appropriate discrimination (AUC: 75.5%, 95% CI 73-78%) and acceptable calibration ([Formula: see text]= 7.44; p value = 0.49). This risk score was then used for 1543 diabetic patients in the validation dataset, including 633 patients with DKD over 5 years. The results showed sufficient discrimination (AUC: 75.8%, 95% CI 73-78%) of the risk score in the validation dataset.

CONCLUSIONS

We developed and validated a new risk score for predicting DKD via ML approach, which used common features in the periodic screening of type 2 diabetic patients that are readily available. In addition, a web-based online tool that is readily available to the public was developed to calculate the DKD risk score.

Quantitative MRI of chronic pancreatitis: results from a multi-institutional prospective study, magnetic resonance imaging as a non-invasive method for assessment of pancreatic fibrosis (MINIMAP)

PURPOSE

To determine if quantitative MRI techniques can be helpful to evaluate chronic pancreatitis (CP) in a setting of multi-institutional study.

METHODS

This study included a subgroup of participants (n = 101) enrolled in the Prospective Evaluation of Chronic Pancreatitis for Epidemiologic and Translational Studies (PROCEED) study (NCT03099850) from February 2019 to May 2021. MRI was performed on 1.5 T using Siemens and GE scanners at seven clinical centers across the USA. Quantitative MRI parameters of the pancreas included T1 relaxation time, extracellular volume (ECV) fraction, apparent diffusion coefficient (ADC), and fat signal fraction. We report the diagnostic performance and mean values within the control (n = 50) and CP (n = 51) groups. The T1, ECV and fat signal fraction were combined to generate the quantitative MRI score (Q-MRI).

RESULTS

There was significantly higher T1 relaxation time; mean 669 ms (± 171) vs. 593 ms (± 82) (p = 0.006), ECV fraction; 40.2% (± 14.7) vs. 30.3% (± 11.9) (p < 0.001), and pancreatic fat signal fraction; 12.2% (± 5.5) vs. 8.2% (± 4.4) (p < 0.001) in the CP group compared to controls. The ADC was similar between groups (p = 0.45). The AUCs for the T1, ECV, and pancreatic fat signal fraction were 0.62, 0.72, and 0.73, respectively. The composite Q-MRI score improved the diagnostic performance (cross-validated AUC: 0.76).

CONCLUSION

Quantitative MR parameters evaluating the pancreatic parenchyma (T1, ECV fraction, and fat signal fraction) are helpful in the diagnosis of CP. A Q-MRI score that combines these three MR parameters improves diagnostic performance. Further studies are warranted with larger study populations including patients with acute and recurrent acute pancreatitis and longitudinal follow-ups.

Precision medicine in type 1 diabetes

First envisioned by early diabetes clinicians, a person-centred approach to care was an aspirational goal that aimed to match insulin therapy to each individual's unique requirements. In the 100 years since the discovery of insulin, this goal has evolved to include personalised approaches to type 1 diabetes diagnosis, treatment, prevention and prediction. These advances have been facilitated by the recognition of type 1 diabetes as an autoimmune disease and by advances in our understanding of diabetes pathophysiology, genetics and natural history, which have occurred in parallel with advancements in insulin delivery, glucose monitoring and tools for self-management. In this review, we discuss how these personalised approaches have improved diabetes care and how improved understanding of pathogenesis and human biology might inform precision medicine in the future.

T1 signal intensity ratio of the pancreas as an imaging biomarker for the staging of chronic pancreatitis

PURPOSE

Our purpose was to validate the T1 SIR (T1 score) as an imaging biomarker for the staging of CP in a large, multi-institutional, prospective study.

METHODS

The prospective study population included 820 participants enrolled in the PROCEED study from nine clinical centers between June 2017 and December 2021. A radiologist at each institution used a standardized method to measure the T1 signal intensity of the pancreas and the reference organs (spleen, paraspinal muscle, liver), which was used to derive respective T1 scores. Participants were stratified according to the seven mechanistic stages of chronic pancreatitis (MSCP 0-6) based on their clinical history, MRCP, and CT findings.

RESULTS

The mean pancreas-to-spleen T1 score was 1.30 in participants with chronic abdominal pain, 1.22 in those with acute or recurrent acute pancreatitis, and 1.03 in definite CP. After adjusting for covariates, we observed a linear, progressive decline in the pancreas-to-spleen T1 score with increasing MSCP from 0 to 6. The mean pancreas-to-spleen T1 scores were 1.34 (MSCP 0), 1.27 (MSCP 1), 1.21 (MSCP 2), 1.16 (MSCP 3), 1.18 (MSCP 4), 1.12 (MSCP 5), and 1.05 (MSCP 6) (p < 0.0001). The pancreas-to-liver and pancreas-to-muscle T1 scores showed less linear trends and wider confidence intervals.

CONCLUSION

The T1 score calculated by SIR of the pancreas-to-spleen shows a negative linear correlation with the progression of chronic pancreatitis. It holds promise as a practical imaging biomarker in evaluating disease severity in clinical research and practice.

Histopathologic correlation of pancreatic fibrosis with pancreatic magnetic resonance imaging quantitative metrics and Cambridge classification

PURPOSE

To determine the correlation of the T1-weighted signal intensity ratio (T1 SIR, or T1 Score) and arterial-to-delayed venous enhancement ratio (ADV ratio) of the pancreas with pancreatic fibrosis on histopathology.

METHODS

Sixty consecutive adult CP patients who had an MRI/MRCP study prior to pancreatic surgery were analyzed. Three blinded observers measured T1 SIR of pancreas to spleen (T1 SIR p/s), pancreas-to-paraspinal muscle (T1 SIR p/m), ADV ratio, and Cambridge grade. Histopathologic grades were given by a gastrointestinal pathologist using Ammann's fibrosis score. Statistical analysis included Spearman's correlation coefficient of the T1 SIR, ADV ratio, Cambridge grade with the fibrosis score, and weighted kappa for interobserver agreement.

RESULTS

The study population included 31 female and 29 male patients, with an average age of 52.1 (26-78 years). Correlations between fibrosis score and T1 SIR p/s, T1 SIR p/m, and ADV ratio were ρ = - 0.54 (p = 0.0001), ρ = - 0.19 (p = 0.19), and ρ = - 0.39 (p = 0.003), respectively. The correlation of Cambridge grade with fibrosis score was ρ = 0.26 (p = 0.07). There was substantial interobserver agreement (weighted kappa) for T1 SIR p/s (0.78), T1 SIR p/m (0.71), and ADV ratio (0.64). T1 SIR p/s of ≤ 1.20 provided a sensitivity of 74% and specificity of 50% (AUC: 0.74), while ADV ratio of ≤ 1.10 provided a sensitivity of 75% and specificity of 55% (AUC: 0.68) to detect a fibrosis score of ≥ 6.

CONCLUSION

There is a moderate negative correlation between the T1 Score (SIR p/s) and ADV ratio with pancreatic fibrosis and a substantial interobserver agreement. These parenchymal metrics show a higher correlation than the Cambridge grade.

Whole-Slide Image Analysis of Human Pancreas Samples to Elucidate the Immunopathogenesis of Type 1 Diabetes Using the QuPath Software

Type 1 diabetes is a chronic disease of the pancreas characterized by the loss of insulin-producing beta cells. Access to human pancreas samples for research purposes has been historically limited, restricting pathological analyses to animal models. However, intrinsic differences between animals and humans have made clinical translation very challenging. Recently, human pancreas samples have become available through several biobanks worldwide, and this has opened numerous opportunities for scientific discovery. In addition, the use of new imaging technologies has unraveled many mysteries of the human pancreas not merely in the presence of disease, but also in physiological conditions. Nowadays, multiplex immunofluorescence protocols as well as sophisticated image analysis tools can be employed. Here, we described the use of QuPath—an open-source platform for image analysis—for the investigation of human pancreas samples. We demonstrate that QuPath can be adequately used to analyze whole-slide images with the aim of identifying the islets of Langerhans and define their cellular composition as well as other basic morphological characteristics. In addition, we show that QuPath can identify immune cell populations in the exocrine tissue and islets of Langerhans, accurately localizing and quantifying immune infiltrates in the pancreas. Therefore, we present a tool and analysis pipeline that allows for the accurate characterization of the human pancreas, enabling the study of the anatomical and physiological changes underlying pancreatic diseases such as type 1 diabetes. The standardization and implementation of these analysis tools is of critical importance to understand disease pathogenesis, and may be informative for the design of new therapies aimed at preserving beta cell function and halting the inflammation caused by the immune attack.

State-of-the-art microscopy to understand islets of Langerhans: what to expect next?

The discovery of Langerhans and microscopic description of islets in the pancreas were crucial steps in the discovery of insulin. Over the past 150 years, many discoveries in islet biology and type 1 diabetes have been made using powerful microscopic techniques. In the past decade, combination of new probes, animal and tissue models, application of new biosensors and automation of light and electron microscopic methods and other (sub)cellular imaging modalities have proven their potential in understanding the beta cell under (patho)physiological conditions. The imaging evolution, from fluorescent jellyfish to real‐time intravital functional imaging, the revolution in automation and data handling and the increased resolving power of analytical imaging techniques are now converging. Here, we review innovative approaches that address islet biology from new angles by studying cells and molecules at high spatiotemporal resolution and in live models. Broad implementation of these cellular imaging techniques will shed new light on cause/consequence of (mal)function in islets of Langerhans in the years to come.