Self-supervised neural network improves tri-exponential intravoxel incoherent motion model fitting compared to least-squares fitting in non-alcoholic fatty liver disease

Recent literature suggests that tri-exponential models may provide additional information and fit liver intravoxel incoherent motion (IVIM) data more accurately than conventional bi-exponential models. However, voxel-wise fitting of IVIM results in noisy and unreliable parameter maps. For bi-exponential IVIM, neural networks (NN) were able to produce superior parameter maps than conventional least-squares (LSQ) generated images. Hence, to improve parameter map quality of tri-exponential IVIM, we developed an unsupervised physics-informed deep neural network (IVIM3-NET). We assessed its performance in simulations and in patients with non-alcoholic fatty liver disease (NAFLD) and compared outcomes with bi-exponential LSQ and NN fits and tri-exponential LSQ fits. Scanning was performed using a 3.0T free-breathing multi-slice diffusion-weighted single-shot echo-planar imaging sequence with 18 b-values. Images were analysed for visual quality, comparing the bi- and tri-exponential IVIM models for LSQ fits and NN fits using parameter-map signal-to-noise ratios (SNR) and adjusted R2. IVIM parameters were compared to histological fibrosis, disease activity and steatosis grades. Parameter map quality improved with bi- and tri-exponential NN approaches, with a significant increase in average parameter-map SNR from 3.38 to 5.59 and 2.45 to 4.01 for bi- and tri-exponential LSQ and NN models respectively. In 33 out of 36 patients, the tri-exponential model exhibited higher adjusted R2 values than the bi-exponential model. Correlating IVIM data to liver histology showed that the bi- and tri-exponential NN outperformed both LSQ models for the majority of IVIM parameters (10 out of 15 significant correlations). Overall, our results support the use of a tri-exponential IVIM model in NAFLD. We show that the IVIM3-NET can be used to improve image quality compared to a tri-exponential LSQ fit and provides promising correlations with histopathology similar to the bi-exponential neural network fit, while generating potentially complementary additional parameters.

Effect of Fecal Microbiota Transplantation Combined With Mediterranean Diet on Insulin Sensitivity in Subjects With Metabolic Syndrome

Background

Recent studies demonstrate that a Mediterranean diet has beneficial metabolic effects in metabolic syndrome subjects. Since we have shown that fecal microbiota transplantation (FMT) from lean donors exerts beneficial effects on insulin sensitivity, in the present trial, we investigated the potential synergistic effects on insulin sensitivity of combining a Mediterranean diet with donor FMT in subjects with metabolic syndrome.

Design

Twenty-four male subjects with metabolic syndrome were put on a Mediterranean diet and after a 2-week run-in phase, the subjects were randomized to either lean donor (n = 12) or autologous (n = 12) FMT. Changes in the gut microbiota composition and bacterial strain engraftment after the 2-week dietary regimens and 6 weeks post-FMT were the primary endpoints. The secondary objectives were changes in glucose fluxes (both hepatic and peripheral insulin sensitivity), postprandial plasma incretin (GLP-1) levels, subcutaneous adipose tissue inflammation, and plasma metabolites.

Results

Consumption of the Mediterranean diet resulted in a reduction in body weight, HOMA-IR, and lipid levels. However, no large synergistic effects of combining the diet with lean donor FMT were seen on the gut microbiota diversity after 6 weeks. Although we did observe changes in specific bacterial species and plasma metabolites, no significant beneficial effects on glucose fluxes, postprandial incretins, or subcutaneous adipose tissue inflammation were detected.

Conclusions

In this small pilot randomized controlled trial, no synergistic beneficial metabolic effects of combining a Mediterranean diet with lean donor FMT on glucose metabolism were achieved. However, we observed engraftment of specific bacterial species. Future trials are warranted to test the combination of other microbial interventions and diets in metabolic syndrome.

Assessment of Imaging Modalities Against Liver Biopsy in Nonalcoholic Fatty Liver Disease: The Amsterdam NAFLD-NASH Cohort

Background

Noninvasive diagnostic methods are urgently required in disease stratification and monitoring in nonalcoholic fatty liver disease (NAFLD). Multiparametric magnetic resonance imaging (MRI) is a promising technique to assess hepatic steatosis, inflammation, and fibrosis, potentially enabling noninvasive identification of individuals with active and advanced stages of NAFLD.

Purpose

To examine the diagnostic performance of multiparametric MRI for the assessment of disease severity along the NAFLD disease spectrum with comparison to histological scores.

Study Type

Prospective, cohort.

Population

Thirty‐seven patients with NAFLD.

Field Strength/Sequence

Multiparametric MRI at 3.0 T consisted of magnetic resonance (MR) spectroscopy (MRS) with multi‐echo stimulated‐echo acquisition mode, magnitude‐based and three‐point Dixon using a two‐dimensional multi‐echo gradient echo, MR elastography (MRE) using a generalized multishot gradient‐recalled echo sequence and intravoxel incoherent motion (IVIM) using a multislice diffusion weighted single‐shot echo‐planar sequence.

Assessment

Histological steatosis grades were compared to proton density fat fraction measured by MRS (PDFF_MRS), magnitude‐based MRI (PDFF_MRI‐M), and three‐point Dixon (PDFF_Dixon), as well as FibroScan® controlled attenuation parameter (CAP). Fibrosis and disease activity were compared to IVIM and MRE. FibroScan® liver stiffness measurements were compared to fibrosis levels. Diagnostic performance of all imaging parameters was determined for distinction between simple steatosis and nonalcoholic steatohepatitis (NASH).

Statistical Tests

Spearman's rank test, Kruskal–Wallis test, Dunn's post‐hoc test with Holm‐Bonferroni P‐value adjustment, receiver operating characteristic curve analysis. A P‐value <0.05 was considered statistically significant.

Results

Histological steatosis grade correlated significantly with PDFF_MRS (r_s = 0.66, P < 0.001), PDFF_MRI‐M (r_s = 0.68, P < 0.001), and PDFF_Dixon (r_s = 0.67, P < 0.001), whereas no correlation was found with CAP. MRE and IVIM diffusion and perfusion significantly correlated with disease activity (r_s = 0.55, P < 0.001, r_s = −0.40, P = 0.016, r_s = −0.37, P = 0.027, respectively) and fibrosis (r_s = 0.55, P < 0.001, r_s = −0.46, P = 0.0051; r_s = −0.53, P < 0.001, respectively). MRE and IVIM diffusion had the highest area‐under‐the‐curve for distinction between simple steatosis and NASH (0.79 and 0.73, respectively).

Data Conclusion

Multiparametric MRI is a promising method for noninvasive, accurate, and sensitive distinction between simple hepatic steatosis and NASH, as well as for the assessment of steatosis and fibrosis severity.

Level of Evidence

2

Technical Efficacy

2

Donor Fecal Microbiota Transplantation Alters Gut Microbiota and Metabolites in Obese Individuals With Steatohepatitis

The intestinal microbiota has been linked to the development and prevalence of steatohepatitis in humans. Interestingly, steatohepatitis is significantly lower in individuals taking a plant-based, low-animal-protein diet, which is thought to be mediated by gut microbiota. However, data on causality between these observations in humans is scarce. In this regard, fecal microbiota transplantation (FMT) using healthy donors is safe and is capable of changing microbial composition in human disease. We therefore performed a double-blind randomized controlled proof-of-principle study in which individuals with hepatic steatosis on ultrasound were randomized to two study arms: lean vegan donor (allogenic n = 10) or own (autologous n = 11) FMT. Both were performed three times at 8-week intervals. A liver biopsy was performed at baseline and after 24 weeks in every subject to determine histopathology (Nonalcoholic Steatohepatitis Clinical Research Network) classification and changes in hepatic gene expression based on RNA sequencing. Secondary outcome parameters were changes in intestinal microbiota composition and fasting plasma metabolomics. We observed a trend toward improved necro-inflammatory histology, and found significant changes in expression of hepatic genes involved in inflammation and lipid metabolism following allogenic FMT. Intestinal microbial community structure changed following allogenic FMT, which was associated with changes in plasma metabolites as well as markers of . Conclusion: Allogenic FMT using lean vegan donors in individuals with hepatic steatosis shows an effect on intestinal microbiota composition, which is associated with beneficial changes in plasma metabolites and markers of steatohepatitis.

Intestinal Ralstonia pickettii augments glucose intolerance in obesity

An altered intestinal microbiota composition has been implicated in the pathogenesis of metabolic disease including obesity and type 2 diabetes mellitus (T2DM). Low grade inflammation, potentially initiated by the intestinal microbiota, has been suggested to be a driving force in the development of insulin resistance in obesity. Here, we report that bacterial DNA is present in mesenteric adipose tissue of obese but otherwise healthy human subjects. Pyrosequencing of bacterial 16S rRNA genes revealed that DNA from the Gram-negative species Ralstonia was most prevalent. Interestingly, fecal abundance of Ralstonia pickettii was increased in obese subjects with pre-diabetes and T2DM. To assess if R. pickettii was causally involved in development of obesity and T2DM, we performed a proof-of-concept study in diet-induced obese (DIO) mice. Compared to vehicle-treated control mice, R. pickettii-treated DIO mice had reduced glucose tolerance. In addition, circulating levels of endotoxin were increased in R. pickettii-treated mice. In conclusion, this study suggests that intestinal Ralstonia is increased in obese human subjects with T2DM and reciprocally worsens glucose tolerance in DIO mice.

Carriers of loss-of-function mutations in EXT display impaired pancreatic beta-cell reserve due to smaller pancreas volume

Exotosin (EXT) proteins are involved in the chain elongation step of heparan sulfate (HS) biosynthesis, which is intricately involved in organ development. Loss of function mutations (LOF) in EXT1 and EXT2 result in hereditary exostoses (HME). Interestingly, HS plays a role in pancreas development and beta-cell function, and genetic variations in EXT2 are associated with an increased risk for type 2 diabetes mellitus. We hypothesized that loss of function of EXT1 or EXT2 in subjects with hereditary multiple exostoses (HME) affects pancreatic insulin secretion capacity and development. We performed an oral glucose tolerance test (OGTT) followed by hyperglycemic clamps to investigate first-phase glucose-stimulated insulin secretion (GSIS) in HME patients and age and gender matched non-affected relatives. Pancreas volume was assessed with magnetic resonance imaging (MRI). OGTT did not reveal significant differences in glucose disposal, but there was a markedly lower GSIS in HME subjects during hyperglycemic clamp (iAUC HME: 0.72 [0.46–1.16] vs. controls 1.53 [0.69–3.36] nmol·l⁻¹·min⁻¹, p<0.05). Maximal insulin response following arginine challenge was also significantly attenuated (iAUC HME: 7.14 [4.22–10.5] vs. controls 10.2 [7.91–12.70] nmol·l⁻¹·min⁻¹ p<0.05), indicative of an impaired beta-cell reserve. MRI revealed a significantly smaller pancreatic volume in HME subjects (HME: 72.0±15.8 vs. controls 96.5±26.0 cm³ p = 0.04). In conclusion, loss of function of EXT proteins may affect beta-cell mass and insulin secretion capacity in humans, and render subjects at a higher risk of developing type 2 diabetes when exposed to environmental risk factors.

NADP+ -dependent IDH1 R132 mutation and its relevance for glioma patient survival

The isocitrate dehydrogenase 1 (IDH1) mutation occurs in high frequency in glioma and secondary glioblastoma (GBM). Mutated IDH1 produces the oncometabolite 2-hydroxyglutarate rather than α-ketoglutarate or isocitrate. The oncometabolite is considered to be the major cause of the association between the IDH1 mutation and gliomagenesis. On the other hand, the IDH1 mutation in GBM is associated with prolonged patient survival. This association is not well understood yet but IDH1 involvement in epigenetic silencing of O-6-methylguanine-DNA methyltransferase (MGMT), a DNA repair enzyme is considered to be an important mechanism. However, it was shown recently that the IDH1 mutation and MGMT silencing are independent prognostic factors. Here, we hypothesize that the IDH1 mutation reduces the capacity to produce NADPH and thus reduces the capacity to scavenge reactive oxygen species that are generated during irradiation and chemotherapy. IDH1 activity is responsible for two-thirds of the NADPH production capacity in normal brain, whereas the IDH1 mutation reduces this capacity by almost 40%. Therefore, we hypothesize that the reduced NADPH production capacity due to the IDH1 mutation renders GBM cells more vulnerable to irradiation and chemotherapy thus prolonging survival of the patients.