Using Explainable Artificial Intelligence Models (ML) to Predict Suspected Diagnoses as Clinical Decision Support.

The complexity of emergency cases and the number of emergency patients have increased dramatically. Due to a reduced or even missing specialist medical staff in the emergency departments (EDs), medical knowledge is often used without professional supervision for the diagnosis. The result is a failure in diagnosis and treatment, even death in the worst case. Secondary: high expenditure of time and high costs. Using accurate patient data from the German national registry of the medical emergency departments (AKTIN-registry, Home - Notaufnahmeregister (aktin.org)), the most 20 frequent diagnoses were selected for creating explainable artificial intelligence (XAI) models as part of the ENSURE project (ENSURE (umg.eu)). 137.152 samples and 51 features (vital signs and symptoms) were analyzed. The XAI models achieved a mean area under the curve (AUC) one-vs-rest of 0.98 for logistic regression (LR) and 0.99 for the random forest (RF), and predictive accuracies of 0.927 (LR) and 0.99 (RF). Based on its grade of explainability and performance, the best model will be incorporated into a portable CDSS to improve diagnoses and outcomes of ED treatment and reduce cost. The CDSS will be tested in a clinical pilot study at EDs of selected hospitals in Germany.

Mycophenolic acid reverses TGF beta-induced cell motility, collagen matrix contraction and cell morphology in vitro.

The aim of this study was to elucidate functional and molecular effects of mycophenolic acid (MPA) on non-lymphatic, kidney epithelial cells treated with transforming growth factor (TGF). MPA effects were studied using HK2 cells incubated with EGF and TGF. The reversibility of these effects was verified using guanosine and 8-aminoguanosine. The following assays were applied: cell proliferation, viability, collagen matrix contraction, scratch wound closure, spindle index, FACS with anti-CD29 and anti-CD326, promoter demethylation of RAS protein activator like 1 (RASAL1), as well as gene expression of RASAL1, integrin 1ß (ITGB1) (CD29) and epithelial cell adhesion molecule (EpCam) (CD326). Cell proliferation was inhibited by increasing concentrations of MPA, whereas neither apoptosis nor cytotoxicity was detected. Stimulation with EGF and/or TGF led to a significant collagen matrix contraction that was successfully inhibited by MPA. In addition, scratch wound closure was inhibited by incubation with TGF alone or with EGF. Under the same conditions, cell morphology (spindle shape) and molecular phenotype (ITGB1(High)EpCam(Low)/ITGB1(Low)EpCam(High)) were both significantly changed, suggesting an epithelial to mesenchymal transformation. Cell morphology and motility, as well as molecular phenotype, were reversible after MPA treatment with TGF transformation in both presence/absence of EGF, thereby suggesting a correlation with the previously described antifibrotic effects of MPA. Dysregulation of TGF signal transduction appears to be related to progression of fibrosis. A TGF-transformed kidney epithelial cell line derived from human proximal tubules was used to study whether the immunosuppressive drug: MPA possesses any functional or molecular antifibrotic effects. Functional and morphological in vitro changes induced by both the TGF and epithelial-growth-factor were reversible by treatment with MPA. An inhibitory effect of MPA on the TGF pathway appears to be responsible for the previously described antifibrotic effects of the MPA in the COL4A3-deficient mouse model of renal fibrosis.

Lack of correlation between Treg quantification assays in inflammatory bowel disease patients.

AIM: To compare the number of regulatory T-cells (Tregs) measured by flow cytometry with those obtained using a real-time quantitative PCR (qPCR) method in patients suffering from inflammatory bowel disease (IBD). METHODS: Tregs percentages obtained by both flow cytometry and qPCR methods in 35 adult IBD patients, 18 out of them with Crohn´s disease (CD) and 17 with ulcerative colitis (UC) were compared to each other as well as to scores on two IBD activity questionnaires using the Harvey Bradshaw Index (HBI) for CD patients and the Simple Colitis Clinical Activity Index (SCCAI) for UC patients. The Treg percentages by flow cytometry were defined as CD4(+)CD25(high)CD127(low)FOXP3(+) cells in peripheral blood mononuclear cells, whereas the Treg percentages by qPCR method were determined as FOXP3 promoter demethylation in genomic DNA. RESULTS: We found an average of 1.56% ± 0.78% Tregs by using flow cytometry, compared to 1.07% ± 0.53% Tregs by using qPCR in adult IBD patients. There were no significant correlations between either the percentages of Tregs measured by flow cytometry or qPCR and the HBI or SCCAI questionnaire scores in CD or UC patients, respectively. In addition, there was no correlation between Treg percentages measured by qPCR and those measured by flow cytometry (r = -0.06, P = 0.73; Spearman Rho). These data suggest that, either Treg-related immune function or the clinical scores in these IBD patients did not accurately reflect actual disease activity. Until the cause(s) for these differences are more clearly defined, the results suggest caution in interpreting studies of Tregs in various inflammatory disorders. CONCLUSION: The two methods did not produce equivalent measures of the percentage of total Tregs in the IBD patients studied which is consistent with the conclusion that Tregs subtypes are not equally detected by these two assays.

Effects of mycophenolate mofetil on kidney function and phosphorylation status of renal proteins in Alport COL4A3-deficient mice.

BACKGROUND: We investigated the effects of mycophenolate mofetil (MMF) on kidney function and on protein phosphorylation in a mouse model for the human Alport syndrome. METHODS: COL4A3-deficient (COL4A3-/-) mice were randomly allocated to receive a placebo (PLC COL4A3-/-) or MMF treatment (MMF COL4A3-/-). Wild type mice (WT) were used as controls. Changes in serum creatinine, total protein and blood urea nitrogen (BUN), concentrations of mycophenolic acid (MPA) and its glucuronide metabolite (MPAG), serum protein electrophoresis, urine dipstick chemistry and sediment were measured. Changes in the phosphorylation status of renal proteins and histology were analyzed. RESULTS: MMF influenced kidney function and protein phosphorylation. Serum creatinine and BUN were lower in MMF treated compared to PLC treated COL4A3-/- mice. Serum albumin and alpha-1 globulins were significantly decreased while serum creatinine, alpha-2 globulins, urine dipstick protein, leukocyte esterase, hemoglobin and red blood cells were all increased in both COL4A3-/- groups compared to WT. Differential 2DE-gel analysis identified six phosphorylated kidney protein spots that were significantly altered by MMF. CONCLUSIONS: These data suggest that the MMF treatment in this murine model moderately improved kidney function and reversed the phosphorylation status of six renal phosphoprotein spots to that seen in WT mice.

Combination of alcohol and fructose exacerbates metabolic imbalance in terms of hepatic damage, dyslipidemia, and insulin resistance in rats.

Although both alcohol and fructose are particularly steatogenic, their long-term effect in the development of a metabolic syndrome has not been studied in vivo. Consumption of fructose generally leads to obesity, whereas ethanol can induce liver damage in the absence of overweight. Here, Sprague-Dawley rats were fed ad libitum for 28 days on five diets: chow (control), liquid Lieber-DeCarli (LDC) diet, LDC +30%J of ethanol (L-Et) or fructose (L-Fr), and LDC combined with 30%J ethanol and 30%J fructose (L-EF). Body weight (BW) and liver weight (LW) were measured. Blood and liver samples were harvested and subjected to biochemical tests, histopathological examinations, and RT-PCR. Alcohol-containing diets substantially reduced the food intake and BW (≤3rd week), whereas fructose-fed animals had higher LW than controls (P<0.05). Additionally, leukocytes, plasma AST and leptin levels were the highest in the fructose-administered rats. Compared to the chow and LDC diets, the L-EF diet significantly elevated blood glucose, insulin, and total-cholesterol levels (also vs. the L-Et group). The albumin and Quick-test levels were the lowest, whereas ALT activity was the highest in the L-EF group. Moreover, the L-EF diet aggravated plasma triglyceride and reduced HDL-cholesterol levels more than 2.7-fold compared to the sum of the effects of the L-Et and L-Fr diets. The decreased hepatic insulin clearance in the L-EF group vs. control and LDC groups was reflected by a significantly decreased C-peptide:insulin ratio. All diets except the control caused hepatosteatosis, as evidenced by Nile red and H&E staining. Hepatic transcription of insulin receptor substrate-1/2 was mainly suppressed by the L-Fr and L-EF diets. The L-EF diet did not enhance the mitochondrial ß-oxidation of fatty acids (Cpt1α and Ppar-α expressions) compared to the L-Et or L-Fr diet. Together, our data provide evidence for the coaction of ethanol and fructose with a high-fat-diet on dyslipidemia and insulin resistance-accompanied liver damage.

Diet high in fructose leads to an overexpression of lipocalin-2 in rat fatty liver.

AIM: To explore lipocalin-2 (LCN-2) expression and its possible role and mechanism(s) of production in rat models of diet-inducible fatty liver. METHODS: Fatty liver was triggered in male Sprague-Dawley rats fed either with liquid Lieber-DeCarli (LDC) or LDC + 70% cal fructose (L-HFr) diet for 4 or 8 wk. Chow-nourished animals served as controls. Hepatic expression of LCN-2 and other metabolic and inflammatory mediators was assessed by quantitative reverse transcription polymerase chain reaction and Western blotting. Serum LCN-2, fasting leptin, and lipid profile were evaluated via Enzyme-Linked Immunosorbent Assay, Radioimmunoassay, and colorimetric assays, respectively. The localization of LCN-2 in the liver was detected by using immunofluorescence staining. Furthermore, HE stain was used to evaluate hepatic fat degeneration and inflammation. RESULTS: Both LDC-fed and L-HFr-fed rat histologically featured fatty liver. In the liver, mRNA transcriptions of Mcp-1, a2-m, Il-8 and Glut5 were increased in the L-HFr group at both time points (P < 0.001), while the transcription of Tlr4, Inos, and Tnf-α was significantly up-regulated at week 4. Interestingly, hepatic Lcn-2 expression was 90-fold at week 4 and 507-fold at week 8 higher in L-HFr-subjected rats vs control (P < 0.001). In contrast to HDL-cholesterol, systemic levels of LCN-2, fasting leptin and triglycerides were elevated in the L-HFr regimen (P < 0.001). Moreover, protein expression of hepatic LCN-2, CD14, phospho-MAPK, caspase-9, cytochrome c and 4-hydroxynonenal was increased in the L-HFr group. Conversely, the hepatic expression of PGC-1α (a mitochondrial-biogenic protein) was reduced in the L-HFr category at week 8. The localization of LCN-2 in the liver was predominantly restricted to MPO⁺ granulocytes. CONCLUSION: Fructose diet up-regulates hepatic LCN-2 expression, which correlates with the increased indicators of oxidative stress and mitochondrial dysfunction. The LCN-2 may be involved in liver protection.

FoxP3 demethylation is increased in human colorectal cancer and rat cholangiocarcinoma tissue.

OBJECTIVES: FoxP3 expression is a marker for Tregs which are known to be involved in tumor immunity. We aimed to evaluate FoxP3 promoter demethylation in human colorectal cancer (CRC) and rat intrahepatic cholangiocarcinoma (ICC). DESIGN AND METHODS: Bisulfite-treated genomic DNA templates of shock frozen paired samples were studied from 13 anonymous CRC patients and from 10 male rats (n=6 ICC induced by thioacetamide and n=4 age-matched controls). Real-time PCR was carried out using a LightCycler 480 system. Human FoxP3 and CD3 promoter demethylations were estimated using previously described assays; and rat FoxP3 promoter demethylation using a newly developed assay. RESULTS: A significant 3.5-fold increase of the demethylation in FoxP3 promoter region was found in human CRC and rat ICC (P<0.05). The average frequency of cells with FoxP3 demethylation in patients suffering from CRC was 0.26% in normal tissue and 0.92% in tumor tissue (n=11 paired samples). Although, no significant difference was found between the mean frequency of CD3 demethylation in normal tissue (4.80%, n=6) and in tumor tissue (4.14%, n=6) from CRC patients, the ratio of demethylated CD3/FoxP3 promoter areas was significantly lower in tumor specimens (P<0.05). Using our novel assay, we found a significant increase in mean frequencies of cells with FoxP3 demethylation in rats with ICC (7.42%, n=6) in comparison to controls (2.14%, n=4). CONCLUSION: FoxP3 seems to be an interesting biomarker for immune response to epithelial tumors. Functional consequences from the increase of Tregs remain to be demonstrated. Further studies with outcome data are necessary.

Serum Lipocalin2 is a potential biomarker of liver irradiation damage.

BACKGROUND/AIM: IL-6 - IL-1- lipocalin2 (LCN2) - liver irradiation - oxidative stress - TNF-a Lipocalin2 (LCN2) is an acute phase protein. The source of its increased serum level in oxidative stress conditions (ROS) remains still unknown. We prospectively evaluate the serum LCN2 increase after single dose liver irradiation along with hepatic LCN2 gene and protein expression. METHODS: A single dose of 25 Gray was administered percutaneously to the liver of randomly paired rats after a planning CT scan. Male Wistar rats were sacrificed 1, 3, 6, 12, 24 and 48 h after irradiation along with sham-irradiated controls. ELISA, RT-PCR, Western blot and immunofluorescence staining was performed. Furthermore, hepatocytes, myofibroblasts and Kupffer cells were isolated from the liver of healthy rats and irradiated ex-vivo. RESULTS: After liver irradiation, LCN2 serum levels increased significantly up to 2.7 µg/ml within 6 h and stayed elevated over 24 h. LCN2 specific transcripts increased significantly up to 552 ± 109-fold at 24 h after liver irradiation, which was further confirmed at protein level. α2-macroglobulin and hemoxygenase-1 also showed an increase, but the magnitude was less as compared to LCN2. LCN2+ granulocytes were detected within 1 h after irradiation around central and portal fields and remained high during the course of study. Ex-vivo irradiated hepatocytes (2.4 ± 0.6-fold) showed a higher LCN2 gene expression as compared to myofibroblasts and Kupffer cells. IL-1ß treatment further increased LCN2 gene expression in cultured hepatocytes. CONCLUSIONS: Single dose liver irradiation induces a significant increase in LCN2 serum levels, comparable to the induction of acute phase proteins. We suggest LCN2 as marker for the early phase of radiation-induced tissue damage.

Characterization of the erythropoietin/erythropoietin receptor axis in a rat model of liver damage and cholangiocarcinoma development.

It has been recently shown that the biological effects of erythropoietin (EPO) are not limited to the hematopoietic compartment but, as pleiotropic glycoprotein, this hormone can exert pro-angiogenic and tissue-protective functions also in a wide range of non-hematopoietic organs. The role of EPO and the effective functionality of its receptor in solid tumors are still a controversial point of debate. In the present work we analyzed the gene expression of EPO and its cognate receptor (EpoR) in a rat model of thioacetamide-induced damage and tumor. An analysis of the EPO/EpoR axis was also performed on human cholangiocarcinoma (CC) cell lines. A progressive increase of EPO and EpoR mRNA can already be observed during the fibrotic-cirrhotic development with a peak of expression rising at tumor formation (24.7 ± 9.9-fold increase and 15.5 ± 1.1-fold increase, respectively, for the two genes). Co-localization studies by immunofluorescence revealed hepatocytes in the regenerative cirrhotic nodules (Hep Par-1(+)) and in the dysplastic bile duct cells (CK19(+)) as the major EPO producers in this specific condition. The same cell populations, together with endothelial cells, exhibited an increased expression of EpoR, although all the non-parenchymal cell populations in the liver exhibited modest basal mRNA levels. Challenging human CC cells, Mz-Cha-2, with a combination of EPO and SCF resulted in a synergistic effect on the gene expression of EPO, CyclinD1 and PCNA. This study suggests that the autocrine and paracrine release of endogenous EPO in the microenvironment may contribute to the development and maintenance of the CC possibly in cooperation with other signaling pathways.

Ferroportin-1 is a 'nuclear'-negative acute-phase protein in rat liver: a comparison with other iron-transport proteins.

Liver is the central organ of iron metabolism. During acute-phase-response (APR), serum iron concentration rapidly decreases. The current study aimed to compare expression and localization of iron transport protein ferroportin-1 (Fpn-1) and of other iron import proteins after experimental tissue damage induced by injecting turpentine oil in the hind limbs of rats and mice. Serum and spleen iron concentration decreased with an increase in total liver, cytoplasmic and nuclear iron concentration. In liver, mRNA amount of Fpn-1, Fpn-1a, Fpn-1b, HFE, hemojuvelin (HJV) and hephaestin (heph) genes showed a rapid decrease. Hepcidin, divalent metal transporter-1 (DMT-1), transferrin (Tf) and Tf-receptor-1 (TfR1), TfR-2 (TfR2) gene expression was increased. Western blot analysis of liver tissue lysate confirmed the changes observed at mRNA level. In spleen, a rapid decrease in gene expression of Fpn-1, Fpn-1a, Fpn-1b, DMT-1, Tf, TfR1 and TfR2, and an increase in hepcidin was observed. Immunohistochemistry of DMT-1 and TfR2 were mainly detected in the nucleus of rat liver and spleen, whereas TfR1 was clearly localized in the plasma membrane. Fpn-1 was mostly found in the nuclei of liver cells, whereas in spleen, the protein was mainly detected in the cell membrane. Western blot analysis of liver fractions confirmed immunohistochemical results. In livers of wild-type mice, gene expression of Fpn-1, Fpn-1a and Fpn-1b was downregulated, whereas hepcidin gene expression was increased. In contrast, these changes were less pronounced in IL-6ko-mice. Cytokine (IL-6, IL-1b and TNF-a) treatment of rat hepatocytes showed a downregulation of Fpn-1, Fpn-1a and Fpn-1b, and upregulation of hepcidin gene expression. Moreover, western blot analysis of cell lysate of IL-6-treated hepatocytes detected, as expected, an increase of a2-macroglobulin (positive acute-phase protein), whereas albumin (negative acute-phase protein) and Fpn-1 were downregulated. Our results demonstrate that liver behaves as a 'sponge' for iron under acute-phase conditions, and Fpn-1 behaves as a negative acute-phase protein in rat hepatocytes mainly, but not exclusively, because of the effect of IL-6. These changes could explain iron retention in the cytoplasm and in the nucleus of hepatocytes during APR.

Immunodetection of cyclooxygenase-2 (COX-2) is restricted to tissue macrophages in normal rat liver and to recruited mononuclear phagocytes in liver injury and cholangiocarcinoma.

It has been suggested that cyclooxygenase-2 (COX-2)-mediated prostaglandin synthesis is associated with liver inflammation and carcinogenesis. The aim of this study is to identify the cellular source of COX-2 expression in different stages, from acute liver injury through liver fibrosis to cholangiocarcinoma (CC). We induced in rats acute and "chronic" liver injury (thioacetamide (TAA) or carbon tetrachloride (CCl(4))) and CC development (TAA) and assessed COX-2 gene expression in normal and damaged liver tissue by RT-PCR of total RNA. The cellular localization of COX-2 protein in liver tissue was analyzed by immunohistochemistry as well as in isolated rat liver cells by Western blotting. The findings were compared with those obtained in human cirrhotic liver tissue. The specificity of the antibodies was tested by 2-DE Western blot and mass spectrometric identification of the positive protein spots. RT-PCR analysis of total RNA revealed an increase of hepatic COX-2 gene expression in acutely as well as "chronically" damaged liver. COX-2-protein was detected in those ED1(+)/ED2(+) cells located in the non-damaged tissue (resident tissue macrophages). In addition COX-2 positivity in inflammatory mononuclear phagocytes (ED1(+)/ED2(-)), which were also present within the tumoral tissue was detected. COX-2 protein was clearly detectable in isolated Kupffer cells as well as (at lower level) in isolated "inflammatory" macrophages. Similar results were obtained in human cirrhotic liver. COX-2 protein is constitutively detectable in liver tissue macrophages. Inflammatory mononuclear phagocytes contribute to the increase of COX-2 gene expression in acute and chronic liver damage induced by different toxins and in the CC microenvironment.

Fetal calf serum heat inactivation and lipopolysaccharide contamination influence the human T lymphoblast proteome and phosphoproteome.

BACKGROUND: The effects of fetal calf serum (FCS) heat inactivation and bacterial lipopolysaccharide (LPS) contamination on cell physiology have been studied, but their effect on the proteome of cultured cells has yet to be described. This study was undertaken to investigate the effects of heat inactivation of FCS and LPS contamination on the human T lymphoblast proteome. Human T lymphoblastic leukaemia (CCRF-CEM) cells were grown in FCS, either non-heated, or heat inactivated, having low (< 1 EU/mL) or regular (< 30 EU/mL) LPS concentrations. Protein lysates were resolved by 2-DE followed by phospho-specific and silver nitrate staining. Differentially regulated spots were identified by nano LC ESI Q-TOF MS/MS analysis. RESULTS: A total of four proteins (EIF3M, PRS7, PSB4, and SNAPA) were up-regulated when CCRF-CEM cells were grown in media supplemented with heat inactivated FCS (HE) as compared to cells grown in media with non-heated FCS (NHE). Six proteins (TCPD, ACTA, NACA, TCTP, ACTB, and ICLN) displayed a differential phosphorylation pattern between the NHE and HE groups. Compared to the low concentration LPS group, regular levels of LPS resulted in the up-regulation of three proteins (SYBF, QCR1, and SUCB1). CONCLUSION: The present study provides new information regarding the effect of FCS heat inactivation and change in FCS-LPS concentration on cellular protein expression, and post-translational modification in human T lymphoblasts. Both heat inactivation and LPS contamination of FCS were shown to modulate the expression and phosphorylation of proteins involved in basic cellular functions, such as protein synthesis, cytoskeleton stability, oxidative stress regulation and apoptosis. Hence, the study emphasizes the need to consider both heat inactivation and LPS contamination of FCS as factors that can influence the T lymphoblast proteome.

Differential kidney proteome profiling in a murine model of renal fibrosis under treatment with mycophenolate mofetil.

AIM: The aim of this study was to investigate the effect of mycophenolate mofetil (MMF) using differential kidney proteome profiling of COL4A3-deficient mice as a model of progressive renal disease. METHODS: Histological evaluation of kidney sections was performed. Total protein lysate from kidneys of placebo- and MMF-treated COL4A3-deficient mice was studied for significant differences in protein abundance using 2-dimensional electrophoresis and mass spectrometry. RESULTS: While tubulointerstitial fibrosis in COL4A3-deficient mice was inhibited by MMF, 19 proteins in the kidneys were regulated: 12 with lower (ATPO, TAGL2, CAH1, TPD52, VA0D1, SERPH, GNAL, PSB6, EF1D, OTUB1, NDUS8, and NAPSA) and 7 with higher (ACADM, ACY3, CK054, ACTB/G, ACTB, UBP5, and ACY1) spot intensity. Nine differentially expressed proteins showed interaction potential (ATPO, TPD52, PSB6, EF1D, OTUB1, NAPSA, ACTB, ACTG, and UBP5). CONCLUSIONS: The identified proteins take part in different signaling pathways. With the highest probability, the VEGF signaling pathway (TAGL2, EF1D, and ACTB) and hypoxia (CAH1, PSB6, and ACTG) were influenced by MMF in fibrotic conditions. Moreover, MMF may modulate antifibrotic and apoptotic pathways as well as epithelial-mesenchymal transition (EMT). Different signaling pathways may be influenced by MMF therapy. These results suggest an inhibitory effect of MMF on renal EMT in COL4A3-deficient mice. Further studies are necessary to validate these findings.