Integrating genome-scale metabolic modelling and transfer learning for human gene regulatory network reconstruction.

MOTIVATION: Gene regulation is responsible for controlling numerous physiological functions and dynamically responding to environmental fluctuations. Reconstructing the human network of gene regulatory interactions is thus paramount to understanding the cell functional organization across cell types, as well as to elucidating pathogenic processes and identifying molecular drug targets. Although significant effort has been devoted towards this direction, existing computational methods mainly rely on gene expression levels, possibly ignoring the information conveyed by mechanistic biochemical knowledge. Moreover, except for a few recent attempts, most of the existing approaches only consider the information of the organism under analysis, without exploiting the information of related model organisms. RESULTS: We propose a novel method for the reconstruction of the human gene regulatory network, based on a transfer learning strategy that synergically exploits information from human and mouse, conveyed by gene-related metabolic features generated in silico from gene expression data. Specifically, we learn a predictive model from metabolic activity inferred via tissue-specific metabolic modelling of artificial gene knockouts. Our experiments show that the combination of our transfer learning approach with the constructed metabolic features provides a significant advantage in terms of reconstruction accuracy, as well as additional clues on the contribution of each constructed metabolic feature. AVAILABILITY AND IMPLEMENTATION: The method, the datasets and all the results obtained in this study are available at: https://doi.org/10.6084/m9.figshare.c.5237687. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Multimodal regularized linear models with flux balance analysis for mechanistic integration of omics data.

MOTIVATION: High-throughput biological data, thanks to technological advances, have become cheaper to collect, leading to the availability of vast amounts of omic data of different types. In parallel, the in silico reconstruction and modeling of metabolic systems is now acknowledged as a key tool to complement experimental data on a large scale. The integration of these model- and data-driven information is therefore emerging as a new challenge in systems biology, with no clear guidance on how to better take advantage of the inherent multisource and multiomic nature of these data types while preserving mechanistic interpretation. RESULTS: Here, we investigate different regularization techniques for high-dimensional data derived from the integration of gene expression profiles with metabolic flux data, extracted from strain-specific metabolic models, to improve cellular growth rate predictions. To this end, we propose ad-hoc extensions of previous regularization frameworks including group, view-specific and principal component regularization and experimentally compare them using data from 1143 Saccharomyces cerevisiae strains. We observe a divergence between methods in terms of regression accuracy and integration effectiveness based on the type of regularization employed. In multiomic regression tasks, when learning from experimental and model-generated omic data, our results demonstrate the competitiveness and ease of interpretation of multimodal regularized linear models compared to data-hungry methods based on neural networks. AVAILABILITY AND IMPLEMENTATION: All data, models and code produced in this work are available on GitHub at https://github.com/Angione-Lab/HybridGroupIPFLasso_pc2Lasso. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Functional constipation masked as irritable bowel syndrome.

BACKGROUND: Rome IV criteria for functional gastrointestinal disorders state that children suspected of having Irritable Bowel Syndrome (IBS) with Constipation (IBS-C) should be preliminarily treated for constipation. We aimed at verifying if functional constipation may indeed lead to an erroneous diagnosis of IBS with diarrhea (IBS-D) or IBS with mixed pattern of diarrhea and constipation (IBS-M). METHODS: We prospectively enrolled in an unblinded fashion 10 and 16 consecutive children referred to our center who met Rome IV criteria for a diagnosis of IBS-D and IBS-M, respectively. Patients who fulfilled criteria for suspect "occult constipation" were then given a bowel cleaning regimen with Polyethylene glycol 3350, re-evaluated at 2 months and followed up for at least 6 months. Sixteen additional patients with IBS with Constipation (IBS-C) referred in the same period served as control. The endpoints were: 1) a decrease of more than 50% in abdominal pain intensity and frequency scores; and 2) for patients with IBS-D and IBS-M: resolution of diarrhea. RESULTS: The endpoints were met by 8 (80%) and 14 (87%) of the patients with IBS-D and IBS-M, respectively, with decrease of abdominal pain and resolution of "diarrhea". The response was not significantly different from that observed in 15 (93%) of the IBS-C control group. CONCLUSION: Acknowledging the limitations of the small number of patients and of the uncontrolled nature of the study, we suggest that a possibly large number of patients labeled as IBS-D or IBS-M may actually simply present functional constipation and should be managed as such.

Diagnostic Yield of 2 Strategies for Adult Celiac Disease Identification in Primary Care.

GOALS: To compare the diagnostic yield and cost-consequences of 2 strategies, screening regardless of symptoms versus case finding (CF), using a point-of-care test (POCT), for the detection of celiac disease (CD) in primary care, to bridge the diagnostic gap of CD in adults. MATERIALS AND METHODS: All subjects under 75 years of age who consecutively went to their general practitioners' offices were offered POCT for anti-transglutaminase immunoglobulin A antibodies. The POCT was performed on all subjects who agreed, and then a systematic search for symptoms or conditions associated with higher risk for CD was performed, immediately after the test but before knowing the test results. The 2 resulting groups were: (a) POCT positive and (b) symptomatic subject at CF. Subjects were defined as symptomatic at CF in the presence of 1 or more symptoms. All POCT-positive or symptomatic subjects at CF were referred to the CD Centers for confirmation of CD. Data on resource consumption were gathered from patients' charts. Cost of examinations, and diagnostic and laboratory tests were estimated with regional outpatient tariffs (Sicily), and a price of &OV0556;2.5 was used for each POCT. RESULTS: Of a total of 2197 subjects who agreed to participate in the study, 36 (1.6%) and 671 (30.5%) were POCT positive and symptomatic at CF, respectively. The yield from the screening and CF was 5 new celiac patients. The total cost and mean cost for each new CD case were &OV0556;7497.35 and &OV0556;1499.47 for the POCT screening strategy, and &OV0556;9855.14 and &OV0556;1971.03 for the CF strategy, respectively. Assuming consecutive use of both strategies, performing POCT only in symptomatic subjects at CF, the calculated yield would be 4 new diagnoses with a total cost of &OV0556;2345.84 and a mean cost of &OV0556;586.46 for each newly diagnosed patient. Only 1 patient was celiac despite a negative POCT. CONCLUSIONS: Testing symptomatic subjects at CF only by POCT seems the most cost-effective strategy to bridge the diagnostic gap of adult CD in primary care.

Case report on pathogenetic link between gluten and IgA nephropathy.

BACKGROUND: A relationship between IgA nephropathy (IgAN) and celiac disease (CD) has been reported. We show the pathogenetic link for the first time. CASE PRESENTATION: A 39-year-old man with cystic fibrosis (CF) and CF-related diabetes started to present gross hematuria, back pain and headache. At admission, laboratory analysis showed increase in serum creatinine of 1.5 mg/dl, together with hematuria and mild proteinuria (1 g/24 h). He underwent a renal biopsy to investigate the cause of hematuria and renal failure. Biopsy was consistent with IgAN. In view of patient reported dyspepsia, an upper gastrointestinal endoscopy with duodenal biopsies was undertaken and was normal. We looked for mucosal deposits of tTG-2 in the duodenum and the renal mesangium. tTG-2 deposits were found both in the duodenum and in renal biopsies, where they topographically replicated mesangial IgA deposits. After one year on a continued gluten containing diet, the patient developed a Marsh 2 type duodenal pathology. CONCLUSIONS: Our findings suggest a connection between CD and IgAN in terms of an immune-mediated gluten-induced pathogenesis even in the absence of villous atrophy and serum celiac autoantibodies.

C3 glomerulopathy in cystic fibrosis: a case report.

BACKGROUND: C3 glomerulonephritis is a rare glomerulopathy characterized at renal biopsy by C3 deposition, alone or with scanty immunoglobulins, as well as by an electron-dense material in mesangium, subendothelial and subepithelial space. An abnormal systemic activation of the alternative pathway of the complement cascade is responsible for the development of the disease if triggered by several possible environmental conditions. We report the first case in literature of a patient affected by cystic fibrosis and C3GN. CASE PRESENTATION: Our case involves a young woman with cystic fibrosis, who had persistent microscopic hematuria, proteinuria and hypocomplementemia C3 for over three months. Renal biopsy confirmed the diagnosis of C3 glomerulopathy. Complement system dysregulation was tested and resulted in a strong terminal pathway activation proved by high levels of sC5b-9 complex, amounting to 1588 ng/ml (normal value < 400 ng/ml). Next generation sequencing (NGS) showed polymorphism in CFH (p.V62I in SCR1) and THBD (p.A473V), already known as pathogenic for C3GN, as well as a mutation in C3 (p.R102G) associated only with age-related macular degeneration (AMD) so far. Treatment was based on ACE inhibitors and kidney function is currently stable (GFR 50 ml/min, serum creatinine 1.7). CONCLUSIONS: The co-existence of C3 glomerulopathy in a patient with CF, which is characterized by chronic infection/inflammation, makes this case an interesting model of chronic altered systemic activation of the alternative pathway of the complement cascade.

Endoscopic and Histologic Healing in Children With Inflammatory Bowel Diseases Treated With Thalidomide.

BACKGROUND & AIMS: Mucosal healing, determined by endoscopic evaluation, is one of the most important prognostic markers for patients with inflammatory bowel diseases. Findings from histologic evaluation, however, could complement findings from endoscopy in assessing mucosal responses to treatment. We analyzed long-term results of children treated with thalidomide to determine the association between clinical response and histology and endoscopy findings. METHODS: We collected data from 2 multicenter trials of 70 children with refractory Crohn's disease (CD) or ulcerative colitis (UC) (2-18 years old; ileocolonic or colonic disease) given thalidomide or placebo (NCT00720538). Clinical remission and clinical response at 8 weeks were defined as a pediatric CD activity index scores 10 points or lower and a decrease of at least 50% from baseline, respectively, for patients with CD; and as a pediatric UC activity index score below 10 and a decrease of at least 20 points from baseline, respectively, for patients with UC. Patients with a clinical response to 8 weeks' treatment with thalidomide underwent endoscopic examination with biopsy collection at study weeks 12 and 52. Severity of inflammation in patients with UC was assessed by Mayo score and in patients with CD by 4-grade system. Biopsies were assessed for signs of active inflammation, erosion or ulceration, and crypt abscesses and assigned a histologic score. RESULTS: Clinical remission was observed in 42 patients (60.0%) and clinical response in 45 patients (64.2%) at Week 8. At Week 52, a total of 38 patients (54.3%) were still in clinical remission or still had a clinical response; 29 patients (41.4%) had mucosal healing, defined as complete healing of erosions or ulcerations, and 20 patients (27.7%) had histologic healing, defined as complete absence of markers of inflammation. Of patients with clinical remission or clinical response, 75.3% also had mucosal healing and 52.6% also had histologic healing. The probability of achieving mucosal healing decreased significantly with increasing values of erythrocyte sedimentation rate (adjusted odds ratio, 0.96; 95% CI, 0.93-0.98; P = .006). CONCLUSIONS: In a long-term analysis of data from 2 clinical trials of pediatric patients with CD or UC, 52 weeks' treatment with thalidomide led to clinical remission in 54.3% of patients with ileocolonic or colonic disease; of these patients, 75.3% had mucosal healing and 52.6% also had histologic healing. Further studies are needed to determine how thalidomide therapy affects long-term progression of inflammatory bowel diseases. (ClinicalTrials.gov number NCT00720538).

Abdominal Pain-Associated Functional Gastrointestinal Disorder Prevalence in Children and Adolescents with Celiac Disease on Gluten-Free Diet: A Multinational Study.

OBJECTIVE: To test the hypothesis that children with celiac disease (CD) on gluten-free diet are at increased risk of abdominal pain (AP) associated-functional gastrointestinal disorders (FGIDs). STUDY DESIGN: This was a multinational cross-sectional study performed from 2014 to 2015. Patients 4-18 years of age with CD on gluten-free diet for longer than 6 months were recruited from pediatric CD clinics in US and Italy. Control groups included siblings of children with CD (with normal tissue transglutaminase levels) and unrelated controls. Subjects or parents completed the Questionnaire on Pediatric Gastrointestinal Symptoms-Rome III. RESULTS: Children (n = 289) were recruited (55% US, 45% Italy): 96 children with CD, 96 sibling controls, and 97 unrelated controls. Chronic AP was present in 30 (30.9%) subjects with CD, 22 (22.7%) sibling controls, and 21 (21.6%) unrelated controls (P = .26 patients with CD vs siblings; P = .18 patients with CD vs unrelated; P = .96 siblings vs unrelated). AP-FGIDs were present in 8 (8.2%) subjects with CD, 8 (8.2%) sibling controls, and 2 (2.1%) unrelated controls (P = 1.00 subjects with CD vs sibling controls; P = .06 subjects with CD vs unrelated controls; P = .06 sibling controls vs unrelated controls). CONCLUSION: This multinational study evaluated the prevalence of chronic abdominal pain and AP-FGIDs in the pediatric population with CD. We found that subjects with CD and controls have a similar prevalence of chronic AP and AP-FGIDs. This suggests that not all types of gastrointestinal inflammation result in AP-FGIDs in children.

Diagnosis of celiac disease and applicability of ESPGHAN guidelines in Mediterranean countries: a real life prospective study.

BACKGROUND: We assessed how the diagnosis of Celiac Disease (CD) is made and how the new ESPGHAN guidelines can be applied in children from countries with different resources. METHODS: A real life prospective study was performed in 14 centres of 13 different Mediterranean countries. Participants were asked to apply the usual diagnostic work-up for CD according to their diagnostic facilities. RESULTS: There were 1974 patients enrolled in the study, mean age 4 years, 10 months; 865 male, 1109 female. CD was confirmed in 511 (25.9%) and was unconfirmed in 1391 (70.5%) patients; 14 patients were diagnosed as having CD according to the new ESPGHAN guidelines, 43 patients were classified as having potential CD. In all participating countries the diagnosis of CD relied on histology of duodenal biopsy; in 5 countries, HLA, and in one country endomysial antibodies (EMA) were not available. Symptoms did not add a significant increase to the pre-test probability of serological tests. The positive predictive value of tissue transglutaminase type 2 (tTG) antibodies performed with different kits but all corresponding to those recommended by ESPGHAN was 96.1% (95% CI 94-97.9%) in presence of tTG > 10xULN. In 135 patients with tTG >10xULN, HLA genotyping was performed and in all it was compatible with CD. CONCLUSIONS: The results of our study show that CD diagnosis still relies on intestinal biopsy in the Mediterranean area. New ESPGHAN criteria are not applicable in 5 countries due to lack of resources needed to perform HLA genotyping and, in one country, EMA assay. Further simplification of the new ESPGHAN guidelines might be made according to what preliminarily the present results suggest if confirmed by new prospective studies.

Cystic Fibrosis: A Risk Condition for Renal Disease.

OBJECTIVE: Cystic fibrosis (CF) is the most common autosomal recessive disease affecting the Caucasian population, with a birth incidence ranging between 1:2,500 and 1:1,800. It is caused by mutations in the CF transmembrane regulator gene which is localized on 7 chromosomes. Renal disease is reported as a relatively rare complication in adult patient with CF. We evaluated proteinuria and chronic renal failure (CRF) in a population of patients with CF. METHODS: A retrospective study was carried out in a referral center for CF at University of Messina in Italy. We identified all patients with renal disease, characterized by proteinuria and/or CRF, during the period 2007 to 2012 and reviewed their medical records to assess influence on renal disease of genotype, number of pulmonary exacerbation, pancreatic insufficiency, pulmonary function, CF-related diabetes, and antibiotics courses. RESULTS: From a population of 77 adult patients with CF, we identified 9 patients with proteinuria (11.7%), and 11 patients (14.28%) with CRF. Mean age was 35.6 (+5.1 standard deviation) years, 55% were female and 33% had diabetes mellitus. Renal biopsy was performed in 3 patients because of nephrotic syndrome in 1 patient and proteinuria with renal failure in the other 2 patients. Renal amyloidosis was disclosed in 2, whereas IgA nephropathy in 1 patient. The ΔF508 mutation in homozygosis was present in 44% of patients with proteinuria (vs. 27% of our CF population, relative risk 2.07), whereas genotype ΔF508/N1303K in 22%. ΔF508 allele mutation was present in 77.7% of proteinuric patients. CONCLUSIONS: Our study shows a higher prevalence of renal disease in patients with CF, than was previously described. The main reason may be related to increased life expectancy because of better management. Moreover, patients with ΔF508 homozygosis had higher risk of proteinuria.

Celiac disease in the Mediterranean area.

BACKGROUND: The World Gastroenterology Organization recommends developing national guidelines for the diagnosis of Celiac Disease (CD): hence a profile of the diagnosis of CD in each country is required. We aim to describe a cross-sectional picture of the clinical features and diagnostic facilities in 16 countries of the Mediterranean basin. Since a new ESPGHAN diagnostic protocol was recently published, our secondary aim is to estimate how many cases in the same area could be identified without a small intestinal biopsy. METHODS: By a stratified cross-sectional retrospective study design, we examined clinical, histological and laboratory data from 749 consecutive unselected CD children diagnosed by national referral centers. RESULTS: The vast majority of cases were diagnosed before the age of 10 (median: 5 years), affected by diarrhea, weight loss and food refusal, as expected. Only 59 cases (7.8%) did not suffer of major complaints. Tissue transglutaminase (tTG) assay was available, but one-third of centers reported financial constraints in the regular purchase of the assay kits. 252 cases (33.6%) showed tTG values over 10 times the local normal limit. Endomysial antibodies and HLA typing were routinely available in only half of the centers. CD was mainly diagnosed from small intestinal biopsy, available in all centers. Based on these data, only 154/749 cases (20.5%) would have qualified for a diagnosis of CD without a small intestinal biopsy, according to the new ESPGHAN protocol. CONCLUSIONS: This cross-sectional study of CD in the Mediterranean referral centers offers a puzzling picture of the capacities to deal with the emerging epidemic of CD in the area, giving a substantive support to the World Gastroenterology Organization guidelines.

A point-of-care test for facing the burden of undiagnosed celiac disease in the Mediterranean area: a pragmatic design study.

BACKGROUND: We aimed at assessing the factors that can influence results of the dissemination of an already validated, new generation commercial Point-of-Care Test (POCT) for detecting celiac disease (CD), in the Mediterranean area, when used in settings where it was designed to be administered, especially in countries with poor resources. METHODS: Pragmatic study design. Family pediatricians at their offices in Italy, nurses and pediatricians in Slovenia and Turkey at pediatricians', schools and university primary care centers looked for CD in 3,559 (1-14 yrs), 1,480 (14-23 yrs) and 771 (1-18 yrs) asymptomatic subjects, respectively. A new generation POCT detecting IgA-tissue antitransglutaminase antibodies and IgA deficiency in a finger-tip blood drop was used. Subjects who tested positive and those suspected of having CD were referred to a Celiac Centre to undergo further investigations in order to confirm CD diagnosis. POCT Positive Predictive Value (PPV) at tertiary care (with Negative Predictive Value) and in primary care settings, and POCT and CD rates per thousand in primary care were estimated. RESULTS: At tertiary care setting, PPV of the POCT and 95% CI were 89.5 (81.3-94.3) and 90 (56-98.5) with Negative Predictive Value 98.5 (94.2-99.6) and 98.7% (92-99.8) in children and adults, respectively. In primary care settings of different countries where POCT was performed by a different number of personnel, PPV ranged from 16 to 33% and the CD and POCT rates per thousand ranged from 4.77 to 1.3 and from 31.18 to 2.59, respectively. CONCLUSIONS: Interpretation of POCT results by different personnel may influence the performance of POC but dissemination of POCT is an urgent priority to be implemented among people of countries with limited resources, such as rural populations and school children.

Cross-attention enables deep learning on limited omics-imaging-clinical data of 130 lung cancer patients.

Deep-learning tools that extract prognostic factors derived from multi-omics data have recently contributed to individualized predictions of survival outcomes. However, the limited size of integrated omics-imaging-clinical datasets poses challenges. Here, we propose two biologically interpretable and robust deep-learning architectures for survival prediction of non-small cell lung cancer (NSCLC) patients, learning simultaneously from computed tomography (CT) scan images, gene expression data, and clinical information. The proposed models integrate patient-specific clinical, transcriptomic, and imaging data and incorporate Kyoto Encyclopedia of Genes and Genomes (KEGG) and Reactome pathway information, adding biological knowledge within the learning process to extract prognostic gene biomarkers and molecular pathways. While both models accurately stratify patients in high- and low-risk groups when trained on a dataset of only 130 patients, introducing a cross-attention mechanism in a sparse autoencoder significantly improves the performance, highlighting tumor regions and NSCLC-related genes as potential biomarkers and thus offering a significant methodological advancement when learning from small imaging-omics-clinical samples.

The importance of disease prevalence in assessing the diagnostic value of a test: endoscopic markers in celiac disease.

BACKGROUND/AIMS: We evaluated the diagnostic variability and reproducibility of endoscopic signs in two populations with a different pretest likelihood of celiac disease (CD). METHODS: We recruited 289 CD patients (both adults and children) in a multicenter prospective study. Group 1 (high risk) included 111 patients referred for positive serology. Group 2 (low risk) included 178 unselected patients. Mosaic pattern, reduction/loss of Kerckring's folds, scalloping of the valvulae conniventes and a nodular pattern were the endoscopic findings looked for in the duodenum. RESULTS: In group 1, the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of endoscopic findings were 100, 84.6, 94.2 and 100% in adults, and 86.8, 9.1, 82.1 and 12.5% in children. In group 2, the sensitivity, specificity, PPV and NPV of endoscopic findings were 33.3, 91.4, 7.7 and 98.5% in adults, and noncalculable, 78.3, 0.0 and 100% in children. Comparing group 1 and group 2, there was a statistically significant difference in sensitivity and PPV in adults, and in specificity, PPV and NPV in children. Concerning the reproducibility of endoscopic findings, a wide variability of κ values was found. CONCLUSION: Endoscopic signs have low reproducibility for CD, and their diagnostic value in selecting patients for multiple intestinal biopsies is unacceptable, especially in populations with low disease prevalence.

Effect of thalidomide on clinical remission in children and adolescents with refractory Crohn disease: a randomized clinical trial.

IMPORTANCE: Pediatric-onset Crohn disease is more aggressive than adult-onset disease, has high rates of resistance to existing drugs, and can lead to permanent impairments. Few trials have evaluated new drugs for refractory Crohn disease in children. OBJECTIVE: To determine whether thalidomide is effective in inducing remission in refractory pediatric Crohn disease. DESIGN, SETTING, AND PATIENTS: Multicenter, double-blind, placebo-controlled, randomized clinical trial of 56 children with active Crohn disease despite immunosuppressive treatment, conducted August 2008-September 2012 in 6 pediatric tertiary care centers in Italy. INTERVENTIONS: Thalidomide, 1.5 to 2.5 mg/kg per day, or placebo once daily for 8 weeks. In an open-label extension, nonresponders to placebo received thalidomide for an additional 8 weeks. All responders continued to receive thalidomide for an additional minimum 52 weeks. MAIN OUTCOMES AND MEASURES: Primary outcomes were clinical remission at week 8, measured by Pediatric Crohn Disease Activity Index (PCDAI) score and reduction in PCDAI by ≥25% or ≥75% at weeks 4 and 8. Primary outcomes during the open-label follow-up were clinical remission and 75% response. RESULTS: Twenty-eight children were randomized to thalidomide and 26 to placebo. Clinical remission was achieved by significantly more children treated with thalidomide (13/28 [46.4%] vs 3/26 [11.5%]; risk ratio [RR], 4.0 [95% CI, 1.2-12.5]; P = .01; number needed to treat [NNT], 2.86). Responses were not different at 4 weeks, but greater improvement was observed at 8 weeks in the thalidomide group (75% response, 13/28 [46.4%] vs 3/26 [11.5%]; RR, 4.0 [95% CI, 1.2-12.5]; NNT = 2.86; P = .01; and 25% response, 18/28 [64.2%] vs 8/26 [30.8%]; RR, 2.1 [95% CI, 1.1-3.9]; NNT = 2.99; P = .01). Of the nonresponders to placebo who began receiving thalidomide, 11 of 21 (52.4%) subsequently reached remission at week 8 (RR, 4.5 [95% CI, 1.4-14.1]; NNT = 2.45; P = .01). Overall, 31 of 49 children treated with thalidomide (63.3%) achieved clinical remission, and 32 of 49 (65.3%) achieved 75% response. Mean duration of clinical remission in the thalidomide group was 181.1 weeks (95% CI, 144.53-217.76) vs 6.3 weeks (95% CI, 3.51-9.15) in the placebo group (P < .001). Cumulative incidence of severe adverse events was 2.1 per 1000 patient-weeks, with peripheral neuropathy the most frequent severe adverse event. CONCLUSIONS AND RELEVANCE: In children and adolescents with refractory Crohn disease, thalidomide compared with placebo resulted in improved clinical remission at 8 weeks of treatment and longer-term maintenance of remission in an open-label follow-up. These findings require replication to definitively determine clinical utility of this treatment. TRIAL REGISTRATION: clinicaltrials.gov Identifier: NCT00720538.

Endoscopic and histological pitfalls in the diagnosis of celiac disease: A multicentre study assessing the current practice.

BACKGROUND AND AIMS: the diagnosis of celiac disease requires small bowel biopsies to identify the characteristic mucosal changes. The current biopsy practice among endoscopists for celiac disease is in most part unknown. The aim of this study was to compare the different diagnostic policies in various centers in their current practice. METHOD: information from a total of 931 confirmed celiac disease patients was retrospectively obtained retrospectively from nine centers in European and Middle Eastern countries. The number of small-bowel biopsies obtained from the duodenal bulb and the second part of the duodenum was compared among different centers. RESULTS: the most frequent stage of mucosal changes amongst Iranian subjects was Marsh IIIa whereas in the rest of the study population was Marsh IIIc. Marsh I and Marsh II were more prevalent in adults (P < 0.05) and Marsh IIIc was significantly higher in pediatric ages between 1 and 15 (P < 0.05). The most common number of biopsy specimens obtained from Romanian subjects was 1 (52% of cases), followed by 2 for Iranian (56%), 3 for Lithuanian (66.7%) and British patients (65%) and 4 for Italian patients (48.3%). For majority of cases, anemia was the most prevalent symptom (18.7%) followed by malabsorption (10.5%), diarrhea (9.3%) and dyspepsia (8.2%), respectively. CONCLUSIONS: despite the evidence-based recommendations, this study revealed a poor compliance with major guidelines on diagnosis of celiac disease. We emphasize that taking adequate number of duodenal biopsies should be implemented for an accurate diagnosis and also for the exclusion of celiac disease.

A score that verifies adherence to a gluten-free diet: a cross-sectional, multicentre validation in real clinical life.

A dietary interview performed by expert personnel is the best method to check whether patients with coeliac disease follow a strict gluten-free diet (GFD). We previously developed a score based on four fast and simple questions that can be administered even by non-expert personnel. The aim of the present study is to verify the reliability of our questionnaire in a new cohort of patients. The questionnaire has a five-level score. From March 2008 to January 2011, the questionnaire was administered to 141 coeliac patients on a GFD, who were undergoing re-evaluation. The score obtained was compared with persistence of both villous atrophy and endomysial antibodies (EMA). The rate of lower scores was higher among the patients with persistence of either villous atrophy (Fisher's exact, P < 0·001; test for trend, P < 0·001) or positive EMA (Fisher's exact, P = 0·001; test for trend, P = 0·018). Given that the coeliac patients have been well instructed on what a GFD means and on how to follow it, our questionnaire is a reliable and simple method to verify compliance to a GFD.

Clinical stratification improves the diagnostic accuracy of small omics datasets within machine learning and genome-scale metabolic modelling methods.

BACKGROUND: Recently, multi-omic machine learning architectures have been proposed for the early detection of cancer. However, for rare cancers and their associated small datasets, it is still unclear how to use the available multi-omics data to achieve a mechanistic prediction of cancer onset and progression, due to the limited data available. Hepatoblastoma is the most frequent liver cancer in infancy and childhood, and whose incidence has been lately increasing in several developed countries. Even though some studies have been conducted to understand the causes of its onset and discover potential biomarkers, the role of metabolic rewiring has not been investigated in depth so far. METHODS: Here, we propose and implement an interpretable multi-omics pipeline that combines mechanistic knowledge from genome-scale metabolic models with machine learning algorithms, and we use it to characterise the underlying mechanisms controlling hepatoblastoma. RESULTS AND CONCLUSIONS: While the obtained machine learning models generally present a high diagnostic classification accuracy, our results show that the type of omics combinations used as input to the machine learning models strongly affects the detection of important genes, reactions and metabolic pathways linked to hepatoblastoma. Our method also suggests that, in the context of computer-aided diagnosis of cancer, optimal diagnostic accuracy can be achieved by adopting a combination of omics that depends on the patient's clinical characteristics.

A Practical Guide to Integrating Multimodal Machine Learning and Metabolic Modeling.

Complex, distributed, and dynamic sets of clinical biomedical data are collectively referred to as multimodal clinical data. In order to accommodate the volume and heterogeneity of such diverse data types and aid in their interpretation when they are combined with a multi-scale predictive model, machine learning is a useful tool that can be wielded to deconstruct biological complexity and extract relevant outputs. Additionally, genome-scale metabolic models (GSMMs) are one of the main frameworks striving to bridge the gap between genotype and phenotype by incorporating prior biological knowledge into mechanistic models. Consequently, the utilization of GSMMs as a foundation for the integration of multi-omic data originating from different domains is a valuable pursuit towards refining predictions. In this chapter, we show how cancer multi-omic data can be analyzed via multimodal machine learning and metabolic modeling. Firstly, we focus on the merits of adopting an integrative systems biology led approach to biomedical data mining. Following this, we propose how constraint-based metabolic models can provide a stable yet adaptable foundation for the integration of multimodal data with machine learning. Finally, we provide a step-by-step tutorial for the combination of machine learning and GSMMs, which includes: (i) tissue-specific constraint-based modeling; (ii) survival analysis using time-to-event prediction for cancer; and (iii) classification and regression approaches for multimodal machine learning. The code associated with the tutorial can be found at https://github.com/Angione-Lab/Tutorials_Combining_ML_and_GSMM .