Beckwith-Wiedemann syndrome (BWS) is a genetic overgrowth syndrome with multiple clinical manifestations, including hypoglycemia. Various genetic alterations leading to BWS have been described. Literature has also described the association between BWS and congenital diabetes, but little is known about the association with type 1 diabetes (T1D). We report a 4-year-old female patient with co-occurring BWS and T1D. The patient presented with 2.4-kilogram weight loss in 3 months accompanied by headache, polyuria, and polydipsia. Initial workup showed blood glucose of 681 mg/dL (37.8 mmol/L). Additional workup revealed marked elevation of the glutamic acid decarboxylase 65 and insulin antibodies, confirming the diagnosis of T1D. The patient's initial genetic test results revealed BWS caused by hypomethylation of the imprinting center 2 (IC2) found on maternal chromosome 11. Concurrence of BWS and T1D is rare and there are cases previously described where BWS has co-occurred with congenital diabetes but not T1D. Although the etiology of acquired autoimmunity is unclear, the answer may lie in genetic analysis or autoimmunity secondary to preceding viral illness. Regardless of the etiology, this case emphasizes further exploration of the association between BWS and T1D.

BACKGROUND:Environmental enteric dysfunction (EED), a chronic inflammatory condition of the small intestine, is an important driver of childhood malnutrition globally. Quantifying intestinal morphology in EED allows for exploration of its association with functional and disease outcomes. OBJECTIVES:We sought to define morphometric characteristics of childhood EED and determine whether morphology features were associated with disease pathophysiology. METHODS:Morphometric measurements and histology were assessed on duodenal biopsy slides for this cross-sectional study from children with EED in Bangladesh, Pakistan, and Zambia (n = 69), and those with no pathologic abnormality (NPA; n = 8) or celiac disease (n = 18) in North America. Immunohistochemistry was also conducted on 46, 8, and 18 biopsy slides, respectively. Linear mixed-effects regression models were used to reveal morphometric differences between EED compared with NPA or celiac disease and identify associations between morphometry and histology or immunohistochemistry among children with EED. RESULTS:In duodenal biopsies, median EED villus height (248 μm), crypt depth (299 μm), and villus:crypt (V:C) ratio (0.9) values ranged between those of NPA (396 μm villus height; 246 μm crypt depth; 1.6 V:C ratio) and celiac disease (208 μm villus height; 365 μm crypt depth; 0.5 V:C ratio). Among EED biopsy slides, morphometric assessments were not associated with histologic parameters or immunohistochemical markers, other than pathologist-determined subjective semiquantitative villus architecture. CONCLUSIONS:Morphometric analysis of duodenal biopsy slides across geographies identified morphologic features of EED, specifically short villi, elongated crypts, and a smaller V:C ratio relative to NPA slides, although not as severe as in celiac slides. Morphometry did not explain other EED features, suggesting that EED histopathologic processes may be operating independently of morphology. Although acknowledging the challenges with obtaining relevant tissue, these data form the basis for further assessments of the role of morphometry in EED.

Environmental enteric dysfunction (EED) is a subclinical enteropathy challenging to diagnose due to an overlap of tissue features with other inflammatory enteropathies. EED subjects (n=52) from Pakistan, controls (n=25), and a validation EED cohort (n=30) from Zambia were used to develop a machine-learning-based image analysis classification model. We extracted histologic feature representations from the Pakistan EED model and correlated them to transcriptomics and clinical biomarkers. In-silico metabolic network modeling was used to characterize alterations in metabolic flux between EED and controls and validated using untargeted lipidomics. Genes encoding beta-ureidopropionase, CYP4F3, and epoxide hydrolase 1 correlated to numerous tissue feature representations. Fatty acid and glycerophospholipid metabolism-related reactions showed altered flux. Increased phosphatidylcholine, lysophosphatidylcholine (LPC), and ether-linked LPCs, and decreased ester-linked LPCs were observed in the duodenal lipidome of Pakistan EED subjects, while plasma levels of glycine-conjugated bile acids were significantly increased. Together, these findings elucidate a multi-omic signature of EED.

Abstract Introduction Beckwith-Wiedemann Syndrome (BWS) is the most common overgrowth syndrome with up to 50% of infants found to have hypoglycemia which has been believed to be related to hyperinsulinemia. However, our case report highlights a patient with the unusual presentation of concurrent BWS and T1DM along with the discussion of genetic factors that may contribute to the pathogenesis. Clinical Case We present the case of a 4-year-old African American female with a past medical history of BWS, twin gestation, prematurity, omphalocele, and patent ductus arteriosus. Her BWS had been diagnosed at birth via genetic testing which found a loss of methylation on the maternal chromosome at imprinting center 2 (IC2) on chromosome 11. During her newborn period, she did not have hypoglycemia. She presented to her primary care physician with a 2.4-kilogram weight loss, new-onset headaches, polyuria, and polydipsia for three months. She also had a viral upper respiratory tract infection a month prior to presentation. Her twin sister did not have similar complaints. Her physical exam including vital signs was reassuring. With the concern for diabetes, further work-up was done which showed an elevated serum glucose level of 681 mg/dL, and hemoglobin A1C of 12.4%. She was admitted for new-onset diabetes. Further labs on admission also showed elevated transglutaminase IgG and positive glutamic acid decarboxylase 65 (GAD65) antibody (level: 444 nmol/L, normal: ≤0.02 nmol/L). Notably, no acidosis was detected in the initial investigations and, unfortunately, due to laboratory error the diabetes antibody panel was cancelled. She was started on a basal-bolus insulin regimen with the diagnosis of new-onset T1DM without ketoacidosis. BWS does not have T1DM as a known characteristic. Hypoglycemia secondary to hyperinsulinemia rather than the opposite, such as our case, is associated with BWS. Due to this, our case is rare, and we found a case report from Europe describing a similar phenomenon. Our patient had a GAD65 antibody being positive which explains having T1DM although there are no studies showing the association of BWS with a positive GAD65 antibody. Our patient also had a viral infection prior to the diagnosis of T1DM which may have triggered autoimmunity. Further, IC2 loss of methylation present in our patient regulates the expression of CDKN1C which has been previously reported to be associated with diabetes. Conclusion Further studies will be required to identify the possible case of T1DM in a patient with BWS. We plan on obtaining a diabetes antibody panel to further assess the genetics of our patient.

Crohn's disease (CD) is a chronic inflammatory disease of the gastrointestinal tract. A clear gap in our existing CD diagnostics and current disease management approaches is the lack of highly specific biomarkers that can be used to streamline or personalize disease management. Comprehensive profiling of metabolites holds promise; however, these high-dimensional profiles need to be reduced to have relevance in the context of CD. Machine learning approaches are optimally suited to bridge this gap in knowledge by contextualizing the metabolic alterations in CD using genome-scale metabolic network reconstructions. Our work presents a framework for studying altered metabolic reactions between patients with CD and controls using publicly available transcriptomic data and existing gene-driven metabolic network reconstructions. Additionally, we apply the same methods to patient-derived ileal enteroids to explore the utility of using this experimental in vitro platform for studying CD. Furthermore, we have piloted an untargeted metabolomics approach as a proof-of-concept validation strategy in human ileal mucosal tissue. These findings suggest that in silico metabolic modeling can potentially identify pathways of clinical relevance in CD, paving the way for the future discovery of novel diagnostic biomarkers and therapeutic targets.

Introduction:Technical burdens and time-intensive review processes limit the practical utility of video capsule endoscopy (VCE). Artificial intelligence (AI) is poised to address these limitations, but the intersection of AI and VCE reveals challenges that must first be overcome. We identified five challenges to address. Challenge #1: VCE data are stochastic and contains significant artifact. Challenge #2: VCE interpretation is cost-intensive. Challenge #3: VCE data are inherently imbalanced. Challenge #4: Existing VCE AIMLT are computationally cumbersome. Challenge #5: Clinicians are hesitant to accept AIMLT that cannot explain their process.Methods:An anatomic landmark detection model was used to test the application of convolutional neural networks (CNNs) to the task of classifying VCE data. We also created a tool that assists in expert annotation of VCE data. We then created more elaborate models using different approaches including a multi-frame approach, a CNN based on graph representation, and a few-shot approach based on meta-learning.Results:When used on full-length VCE footage, CNNs accurately identified anatomic landmarks (99.1%), with gradient weighted-class activation mapping showing the parts of each frame that the CNN used to make its decision. The graph CNN with weakly supervised learning (accuracy 89.9%, sensitivity of 91.1%), the few-shot model (accuracy 90.8%, precision 91.4%, sensitivity 90.9%), and the multi-frame model (accuracy 97.5%, precision 91.5%, sensitivity 94.8%) performed well.Discussion:Each of these five challenges is addressed, in part, by one of our AI-based models. Our goal of producing high performance using lightweight models that aim to improve clinician confidence was achieved.

ABSTRACT. Environmental enteric dysfunction (EED) is a subclinical enteropathy prevalent in resource-limited settings, hypothesized to be a consequence of chronic exposure to environmental enteropathogens, resulting in malnutrition, growth failure, neurocognitive delays, and oral vaccine failure. This study explored the duodenal and colonic tissues of children with EED, celiac disease, and other enteropathies using quantitative mucosal morphometry, histopathologic scoring indices, and machine learning–based image analysis from archival and prospective cohorts of children from Pakistan and the United States. We observed villus blunting as being more prominent in celiac disease than in EED, as shorter lengths of villi were observed in patients with celiac disease from Pakistan than in those from the United States, with median (interquartile range) lengths of 81 (73, 127) µm and 209 (188, 266) µm, respectively. Additionally, per the Marsh scoring method, celiac disease histologic severity was increased in the cohorts from Pakistan. Goblet cell depletion and increased intraepithelial lymphocytes were features of EED and celiac disease. Interestingly, the rectal tissue from cases with EED showed increased mononuclear inflammatory cells and intraepithelial lymphocytes in the crypts compared with controls. Increased neutrophils in the rectal crypt epithelium were also significantly associated with increased EED histologic severity scores in duodenal tissue. We observed an overlap between diseased and healthy duodenal tissue upon leveraging machine learning image analysis. We conclude that EED comprises a spectrum of inflammation in the duodenum, as previously described, and the rectal mucosa, warranting the examination of both anatomic regions in our efforts to understand and manage EED.

It is now standard of care to offer genetic testing to patients at risk of hereditary breast cancer and make management decisions based on these results. Although great strides have been made in ensuring access to genetic testing and genetic counseling by establishing hereditary breast cancer clinics in well-resourced countries, these are essentially non-existent in low-middle income countries like Pakistan. We established a hereditary breast cancer clinic involving a multidisciplinary team, including a medical geneticist and a genetic counselor. Our efforts were based on consensus guidelines and included educating medical providers about the importance of genetic testing in breast cancer care and the mandatory presence of a genetics team member at the weekly Breast Tumor Board meeting. This resulted in an increase in the number of referrals of breast cancer patients for genetic testing. In this report, we describe the challenges we faced in setting up such a system in Pakistan and the measures to overcome them. There is a need to establish such hereditary breast cancer clinics, which can also be replicated at other centers in low-resource settings, to improve standardized assessment and management of the patients with hereditary breast cancer according to consensus guidelines.

Background Breast cancer is the most common malignancy in women, affecting over 1.5 million women every year, which accounts for the highest number of cancer-related deaths in women globally. Hereditary breast cancer (HBC), an important subset of breast cancer, accounts for 5–10% of total cases. However, in Low Middle-Income Countries (LMICs), the population-specific risk of HBC in different ethnicities and the correlation with certain clinical characteristics remain unexplored. Methods Retrospective chart review of patients who visited the HBC clinic and proceeded with multi-gene panel testing from May 2017 to April 2020. Descriptive and inferential statistics were used to analyze clinical characteristics of patients. Fisher’s exact, Pearson’s chi-squared tests and Logistic regression analysis were used for categorical variables and Wilcoxon rank-sum test were used for quantitative variables. For comparison between two independent groups, Mann-Whitney test was performed. Results were considered significant at a p value of < 0.05. Results Out of 273 patients, 22% tested positive, 37% had a VUS and 41% had a negative genetic test result. Fifty-five percent of the positive patients had pathogenic variants in either BRCA1 or BRCA2 , while the remaining positive results were attributed to other genes. Patients with a positive result had a younger age at diagnosis compared to those having a VUS and a negative result; median age 37.5 years, IQR (Interquartile range) (31.5–48). Additionally, patients with triple negative breast cancer (TNBC) were almost 3 times more likely to have a positive result (OR = 2.79, CI = 1.42–5.48 p = 0.003). Of all patients with positive results, 25% of patients had a negative family history of breast and/or related cancers. Conclusions In our HBC clinic, we observed that our rate of positive results is comparable, yet at the higher end of the range which is reported in other populations. The importance of expanded, multi-gene panel testing is highlighted by the fact that almost half of the patients had pathogenic or likely pathogenic variants in genes other than BRCA1/2 , and that our test positivity rate would have only been 12.8% if only BRCA1/2 testing was done. As the database expands and protocol-driven referrals are made across the country, our insight about the genetic architecture of HBC in our population will continue to increase.

Lynch Syndrome (LS), also known as hereditary non-polyposis colorectal cancer syndrome, is an autosomal dominant disorder caused by the presence of germline pathogenic (P) or likely pathogenic (LP) variants in DNA mismatch repair (MMR) genes, which include MLH1, MSH2, MSH6, PMS2, and EPCAM. Presence of a disease-causing variant in any of these MMR genes increases an individual's lifetime risk of developing colorectal cancer (CRC) (61%), endometrial (57%), ovarian (38%), renal pelvis or ureter (28%), prostate (24%), breast (19%), small bowel (11%), gastric (9%), brain (8%), hepatobiliary cancer (4%) and pancreatic cancer (2%). It is still unclear whether LS causes a predisposition to breast cancer, with current data suggesting a risk < 15%, which is close to the general population risk of 13%. Guidelines for high-risk surveillance and management of individuals with LS are thus not well-established for breast cancer, which at present is to be managed based on family history. Identification of germline disease-causing variants in MMR genes is thus pivotal for optimizing the treatment and surveillance of patients with LS and for the identification of at-risk family members, to reduce cancer-related morbidity and mortality. This is best done using Next Generation Sequencing (NGS) multi-gene panel testing, which increases the diagnostic yield, but also increases the chances of finding variants of uncertain significance (VUS). As the Pakistani population remains under-represented, the likelihood of finding a VUS is high, making it difficult to use these results for clinical decision-making. • To study the presence of disease-causing variants as well as VUSs in MMR genes causing LS in a series of patients diagnosed with breast cancer who underwent genetic testing using a multi-gene NGS panel • To study the clinical characteristics of patients with LS who presented with breast cancer • To use immunohistochemistry (IHC) on breast tumour tissue samples to clarify whether VUSs in MMR genes in breast cancer patients are associated with loss of staining. Retrospective chart review of patients who visited the hereditary cancer (HBC) clinic and proceeded with Hereditary Breast and Gynecological Cancer multi-gene NGS panel testing (at Prevention Genetic and Invitae Genetics, USA) from May 2017 to Oct 2021, at an Academic Medical Centre, Aga Khan University Hospital, Karachi, Pakistan. IHC was performed using standard antibodies against the protein products of MLH1, MSH2, MSH6 and PMS2. A total of 460 breast cancer patients qualified and proceeded with testing, considering their personal and/or family history of disease, based on NCCN criteria. 94 patients (20.4%) had a positive genetic test result, which included Pathogenic (P) and Likely Pathogenic (LP) variants. Of the remaining patients, 167 (36.3%) had one or more VUS[s], and 199 (43.3%) had a negative genetic test result. Five out the total 460 patients (1.1%) or five of the 94 patients with positive results (5.3%), tested positive for LS, with an average age at diagnosis of 40 years. One patient had a personal history of colon cancer at age 38 and had then presented with breast cancer at age 43; and had a family history of colon cancer. Another patient who presented with breast cancer only, harbored a pathogenic variant in MSH6 as well as BRCA1, having a positive family history of breast and uterine cancer. In the remaining three patients, personal or family history was not indicative of any possible established link with LS, and their diagnoses would have been missed if MMR genes were not included in the multi-gene panel. Thus, multi-gene testing including MMR genes increased the diagnostic yield by 4.3% (4/94), even after excluding the patient with a personal history of colon cancer. Details of these patients are mentioned in Table 1. Out of the total 460 patients who underwent testing, 28 (6.1%) harbored a VUS in MMR genes, namely PMS2 (n=12), MSH2 (n=7), MSH6 (n=8) and one patient had a VUS in both PMS2 and MSH2. No VUS in MLH1 was identified in breast cancer patients. IHC was done on 18/28 (64.3%) and no loss of staining was observed on any tissue sample, possibly indicating that the variants are not disease-causing. We also observed six recurrent VUSs in unrelated patients, which included, (variant.1) v.1: MSH2 NM_000251.2 (p.L135V) (Rs193096019 MAF in South Asians=0.05%), v.2: MSH6 NM_000179.2 (p.L1356Dfs*4), (rs775836476, MAF in South Asians=0.04%), v.3: MSH6, (p.R911Q), (Rs761622304,MAF in South Asians= 0.006%), v.4: PMS2 NM_000535.5 (p.R294W) (rs563433235, MAF in South Asians= 0.03%), v.5: PMS2 (p.L454S) (absent from population database), v.6: PMS2 (p.D784N) (unreliable region coverage).It is worth noting that, v. 2, 4 and 6 have been reported in diseased individuals, while v.5 is a novel variant, absent from the gnomAD, highlighting the need for further functional studies to understanding the role of these variants in tumorigenesis. In our experience with genetic testing in the HBC, a small proportion of patients were diagnosed with Lynch Syndrome, some presenting with breast cancer alone, without a personal or family history of LS associated tumors. This may justify the need to include MMR genes in multi-gene panels being offered to breast cancer patients. VUSs in these genes remain challenging, especially in underrepresented populations, and IHC may be a way to at least partially clarify their significance. More ethno-specific genomic studies, as well as better functional studies are required provide better clinical care.Tabled 1Patient IDVariant detailAge at diagnosisTumor IHCHistopathologySecondary cancer, ageFamily History1MLH1, NM_000249.3, Deletion (Exons 16-19)43ER+/PR+, HER2-IDC Grade IICa Colonfather, ca colon2MSH6, NM_000179.2, c.3261del (p. F1088Sfs*2)30Triple Negative DiseaseIDC Grade III-mother, maternal aunts, grandmother, ca breast and uterineBRCA1, NM_007294.3, Deletion (Exons 1-2)3MLH1, NM_000249.3, Exon 3, c.306G>T (p. E102D)39ER+/PR+, HER2-IDC Grade II-sister, ca breast4MLH1, NM_000249.3, Intron 16, c.1897-2A>G60Triple Negative DiseaseIDC (Grade not available)-sister, cousins, ca breast5MSH6, NM_000179.2, Exon 4 c.1222_1226del (p.P408Dfs*8),30ER-/PR-, HER2+DCIS-NegativeIHC= Immunohistochemistry, IDC= Invasive Ductal Carcinoma, DCIS= In-situ Ductal Carcinoma, ER= estrogen receptor, PR=progesterone receptor HER2= Human Epidermal Growth Factor Receptor 2 Open table in a new tab

BACKGROUND:The underperformance of oral vaccines in children of low- and middle-income countries is partly attributable to underlying environmental enteric dysfunction (EED).METHODOLOGY:We conducted a longitudinal, community-based study to evaluate the association of oral rotavirus vaccine (Rotarix®) seroconversion with growth anthropometrics, EED biomarkers and intestinal enteropathogens in Pakistani infants. Children were enrolled between three to six months of their age based on their nutritional status. We measured serum anti-rotavirus immunoglobulin A (IgA) at enrollment and nine months of age with EED biomarkers and intestinal enteropathogens.RESULTS:A total of 391 infants received two doses of rotavirus (RV) vaccine. 331/391 provided paired blood samples. Of these 331 children, 45% seroconverted at 9 months of age, 35% did not seroconvert and 20% were seropositive at baseline. Non-seroconverted children were more likely to be stunted, wasted and underweight at enrollment. In univariate analysis, insulin-like growth factor (IGF) concentration at 6 months were higher in seroconverters, median (25th, 75th percentile): 26.3 (16.5, 43.5) ng/ml vs. 22.5 (13.6, 36.3) ng/ml for non-seroconverters, p-value = 0.024. At nine months, fecal myeloperoxidase (MPO) concentrations were significantly lower in seroconverters, 3050(1250, 7587) ng/ml vs. 4623.3 (2189, 11650) ng/ml in non-seroconverted children, p-value = 0.017. In multivariable logistic regression analysis, alpha-1 acid glycoprotein (AGP) and IGF-1 concentrations were positively associated with seroconversion at six months. The presence of sapovirus and rotavirus in fecal samples at the time of rotavirus administration, was associated with non-seroconversion and seroconversion, respectively.CONCLUSION:We detected high baseline RV seropositivity and impaired RV vaccine immunogenicity in this high-risk group of children. Healthy growth, serum IGF-1 and AGP, and fecal shedding of rotavirus were positively associated with RV IgA seroconversion following immunization, whereas the presence of sapovirus was more common in non-seroconverters.TRIAL REGISTRATION:Clinical Trials ID: NCT03588013.

The relationship between environmental factors and child health is not well understood in rural Pakistan. This study characterized the environmental factors related to the morbidity of acute respiratory infections (ARIs), diarrhea, and growth using geographical information systems (GIS) technology. Anthropometric, address and disease prevalence data were collected through the SEEM (Study of Environmental Enteropathy and Malnutrition) study in Matiari, Pakistan. Publicly available map data were used to compile coordinates of healthcare facilities. A Pearson correlation coefficient (r) was used to calculate the correlation between distance from healthcare facilities and participant growth and morbidity. Other continuous variables influencing these outcomes were analyzed using a random forest regression model. In this study of 416 children, we found that participants living closer to secondary hospitals had a lower prevalence of ARI (r = 0.154, p < 0.010) and diarrhea (r = 0.228, p < 0.001) as well as participants living closer to Maternal Health Centers (MHCs): ARI (r = 0.185, p < 0.002) and diarrhea (r = 0.223, p < 0.001) compared to those living near primary facilities. Our random forest model showed that distance has high variable importance in the context of disease prevalence. Our results indicated that participants closer to more basic healthcare facilities reported a higher prevalence of both diarrhea and ARI than those near more urban facilities, highlighting potential public policy gaps in ameliorating rural health.

AbstractObjectiveDevelop a deep learning-based methodology using the foundations of systems pathology to generate highly accurate predictive tools for complex gastrointestinal diseases, using celiac disease (CD) as a prototype.DesignTo predict the severity of CD, defined by Marsh–Oberhüber classification, we used deep learning to develop a model based on histopathologic features.ResultsThe study was based on a pediatric cohort of 124 patients identified with different classes of CD severity. The model predicted CD with an overall 88.7% accuracy with the highest for Marsh IIIc (91.0%; 95% sensitivity; 91% specificity). The model identified EECs as a defining feature of children with Marsh IIIc CD and endocrinopathies which was confirmed using immunohistochemistry.ConclusionThis deep learning image analysis platform has broad applications in disease treatment, management, and prognostication and paves the way for precision medicine.SummaryWhat is already known about this subject?–Deep Learning has the potential to generate predictive models for complex gastrointestinal diseases.What are the new findings?–Our deep learning-based model used the foundations of systems pathology to generate a highly accurate predictive tool for complex gastrointestinal diseases, using a celiac disease (CD) pediatric cohort as a prototype.–The model predicated CD severity with high accuracy and identified enteroendocrine cells as a defining feature of children with severe CD and endocrinopathies.How might it impact on clinical practice in the foreseeable future?–Assessment of histopathological markers at the time of diagnosis that can predict risk of severity or complications can have broad applications in disease treatment, management, and prognostication and pave the way for precision medicine.

Hematoxylin and Eosin (H&E) stained Whole Slide Images (WSIs) are utilized for biopsy visualization-based diagnostic and prognostic assessment of diseases. Variation in the H&E staining process across different lab sites can lead to significant variations in biopsy image appearance. These variations introduce an undesirable bias when the slides are examined by pathologists or used for training deep learning models. To reduce this bias, slides need to be translated to a common domain of stain appearance before analysis. We propose a Self-Attentive Adversarial Stain Normalization (SAASN) approach for the normalization of multiple stain appearances to a common domain. This unsupervised generative adversarial approach includes self-attention mechanism for synthesizing images with finer detail while preserving the structural consistency of the biopsy features during translation. SAASN demonstrates consistent and superior performance compared to other popular stain normalization techniques on H&E stained duodenal biopsy image data.

Candidate markers for Crohn’s Disease (CD) may be identified via gene expression-based construction of metabolic networks (MN). These can computationally describe gene-protein-reaction associations for entire tissues and also predict the flux of reactions (rate of turnover of specific molecules via a metabolic pathway). Recon3D is the most comprehensive human MN to date. We used publicly available CD transcriptomic data along with Recon3D to identify metabolites as potential diagnostic and prognostic biomarkers. Terminal ileal gene expression profiles (36,372 genes; 218 CD. 42 controls) from the RISK cohort (Risk Stratification and Identification of Immunogenetic and Microbial Markers of Rapid Disease Progression in Children with Crohn’s Disease) and their transcriptomic abundances were used. Recon3D was pruned to only include RISK dataset transcripts which determined metabolic reaction linkage with transcriptionally active genes. Flux balance analysis (FBA) was then run using RiPTiDe with context specific transcriptomic data to further constrain genes (Figure 1). RiPTiDe was independently run on transcriptomic data from both CD and controls. From the pruned and constricted MN obtained, reactions were extracted for further analysis. After applying the necessary constraints to modify Recon3D, 527 CD and 537 control reactions were obtained. Reaction comparison with a publicly available list of healthy small intestinal epithelial reactions (n=1282) showed an overlap of 80 CD and 84 control reactions. These were then further grouped based on their metabolic pathways. RiPTiDe identified context specific metabolic pathway activity without supervision and the percentage of forward, backward, and balanced reactions for each metabolic pathway (Figure 2). The metabolite concentrations in the small intestine was altered among CD patients. Notably, the citric acid cycle and malate-aspartate shuttle were affected, highlighting changes in mitochondrial metabolic pathways. This is illustrated by changes in the number of reactions at equilibrium between CD and control. The results are relevant as cytosolic acetyl-CoA is needed for fatty acid synthesis and is obtained by removing citrate from the citric acid cycle. An intermediate removal from the cycle has significant cataplerotic effects. The malate-aspartate shuttle also allows electrons to move across the impermeable membrane in the mitochondria (fatty acid synthesis location). These findings are reported by previously published studies where gene expression for fatty acid synthesis is altered in CD patients along with mitochondrial metabolic pathway changes, resulting in altered cell homeostasis. In-depth analysis is currently underway with our work supporting the utility of potential metabolic biomarkers for CD diagnosis, management and improved care.

Probe-based confocal laser endomicroscopy (pCLE) allows for real-time diagnosis of dysplasia and cancer in Barrett’s esophagus (BE) but is limited by low sensitivity. Even the gold standard of histopathology is hindered by poor agreement between pathologists. We deployed deep-learning-based image and video analysis in order to improve diagnostic accuracy of pCLE videos and biopsy images. Blinded experts categorized biopsies and pCLE videos as squamous, non-dysplastic BE, or dysplasia/cancer, and deep learning models were trained to classify the data into these three categories. Biopsy classification was conducted using two distinct approaches—a patch-level model and a whole-slide-image-level model. Gradient-weighted class activation maps (Grad-CAMs) were extracted from pCLE and biopsy models in order to determine tissue structures deemed relevant by the models. 1970 pCLE videos, 897,931 biopsy patches, and 387 whole-slide images were used to train, test, and validate the models. In pCLE analysis, models achieved a high sensitivity for dysplasia (71%) and an overall accuracy of 90% for all classes. For biopsies at the patch level, the model achieved a sensitivity of 72% for dysplasia and an overall accuracy of 90%. The whole-slide-image-level model achieved a sensitivity of 90% for dysplasia and 94% overall accuracy. Grad-CAMs for all models showed activation in medically relevant tissue regions. Our deep learning models achieved high diagnostic accuracy for both pCLE-based and histopathologic diagnosis of esophageal dysplasia and its precursors, similar to human accuracy in prior studies. These machine learning approaches may improve accuracy and efficiency of current screening protocols.

In recent years, the availability of digitized Whole Slide Images (WSIs) has enabled the use of deep learning-based computer vision techniques for automated disease diagnosis. However, WSIs present unique computational and algorithmic challenges. WSIs are gigapixel-sized ($\sim$100K pixels), making them infeasible to be used directly for training deep neural networks. Also, often only slide-level labels are available for training as detailed annotations are tedious and can be time-consuming for experts. Approaches using multiple-instance learning (MIL) frameworks have been shown to overcome these challenges. Current state-of-the-art approaches divide the learning framework into two decoupled parts: a convolutional neural network (CNN) for encoding the patches followed by an independent aggregation approach for slide-level prediction. In this approach, the aggregation step has no bearing on the representations learned by the CNN encoder. We have proposed an end-to-end framework that clusters the patches from a WSI into ${k}$-groups, samples ${k}'$ patches from each group for training, and uses an adaptive attention mechanism for slide level prediction; Cluster-to-Conquer (C2C). We have demonstrated that dividing a WSI into clusters can improve the model training by exposing it to diverse discriminative features extracted from the patches. We regularized the clustering mechanism by introducing a KL-divergence loss between the attention weights of patches in a cluster and the uniform distribution. The framework is optimized end-to-end on slide-level cross-entropy, patch-level cross-entropy, and KL-divergence loss (Implementation: https://github.com/YashSharma/C2C).

OBJECTIVES:Striking histopathological overlap between distinct but related conditions poses a disease diagnostic challenge. There is a major clinical need to develop computational methods enabling clinicians to translate heterogeneous biomedical images into accurate and quantitative diagnostics. This need is particularly salient with small bowel enteropathies; environmental enteropathy (EE) and celiac disease (CD). We built upon our preliminary analysis by developing an artificial intelligence (AI)-based image analysis platform utilizing deep learning convolutional neural networks (CNNs) for these enteropathies. METHODS:Data for the secondary analysis was obtained from three primary studies at different sites. The image analysis platform for EE and CD was developed using CNNs including one with multizoom architecture. Gradient-weighted class activation mappings (Grad-CAMs) were used to visualize the models' decision-making process for classifying each disease. A team of medical experts simultaneously reviewed the stain color normalized images done for bias reduction and Grad-CAMs to confirm structural preservation and biomedical relevance, respectively. RESULTS:Four hundred and sixty-one high-resolution biopsy images from 150 children were acquired. Median age (interquartile range) was 37.5 (19.0-121.5) months with a roughly equal sex distribution; 77 males (51.3%). ResNet50 and shallow CNN demonstrated 98% and 96% case-detection accuracy, respectively, which increased to 98.3% with an ensemble. Grad-CAMs demonstrated models' ability to learn different microscopic morphological features for EE, CD, and controls. CONCLUSIONS:Our AI-based image analysis platform demonstrated high classification accuracy for small bowel enteropathies which was capable of identifying biologically relevant microscopic features and emulating human pathologist decision-making process. Grad-CAMs illuminated the otherwise "black box" of deep learning in medicine, allowing for increased physician confidence in adopting these new technologies in clinical practice.