The "Integrated Phenotype of the Intestinal Lobular Hepatic Axis in the Progression of Chronic Liver Disease" (Igal Axis) (Igal Axis)" project is an observational prospective study aimed at investigating the relationship between intestinal disease based on intestinal liver disease (>18 years of Guided MaSoling) disease, inflammation, inflammation activation, and platelet activation. They were registered and classified according to the level of fibrosis. Specifically, MASLD is defined by the presence of liver steatosis and occurs in subjects with at least one cardiac metabolic risk factor, with no significant alcohol intake (>20 g in women and >30 g daily in men).. This study describes several covariates, including age, gender, body mass index (BMI), diabetes, hypertension, dyslipidemia, and drug use. This study was conducted in full compliance with the principles of the Declaration of Helsinki and was approved by the Local Ethics Committee of Sapienza Umberto I, University of Sapienza, Rome, Italy (Reference 6804, 09/11/2022). Written informed consent was obtained from all patients.

Vibration-controlled transient elastography (Fibroscan, Echesens, Paris, France) was used for liver assessment, and both M and XL probes were used, and an automated probe selection tool was used to determine the most appropriate probe based on real-time measurements of capsule distance from the skin to the liver. This procedure was carried out according to manufacturer guidelines and training. Operators were blinded to all clinical data and patient diagnosis. Staging of fibrosis by transient elastography was classified using the following thresholds to diagnose F2, ≥F3, and F4, respectively: 8.2, 9.7, and 13.6.

The reliability of fibrosis staging was assessed by summarizing LSM values for each fibrosis category (F0-F4). Ten valid measurements were obtained per patient, and results were considered valid if the fibroscan-specific IQR/median value was below 0.30. table 7 Mean ± SD, median report [IQR]minimum and maximum LSM values, and corresponding IQR/median values for each fibrosis category. All categories exhibit IQR/median values well below the 0.30 threshold, exhibiting robust measurement quality, and supporting the accuracy of fibrosis staging in this cohort.

Venous blood samples were collected from study participants at the time of enrollment. Hemogram profiles of the SYSMEX hematology analyzer were obtained using EDTA-stantogogulated blood. Blood samples were collected from all participants after at least 10 hours of overnight speed to minimize variability in metabolite levels caused by dietary intake. This standardized protocol ensured consistent conditions across all samples and reduced the potential confounding effects associated with Fed or fasting conditions. Anticoagulant-free blood was clotted and serum was separated by centrifugation at 3400 g for 20 min. The encoded samples were stored at -80°C until batch analysis.

Inflammatory cytokine concentrations were assessed in serum samples by a multiple bead-based flow cytometry assay (Biolgend, Inflammatory Panel I, Catalog No. 740809) according to the manufacturer's instructions. Briefly, after thawing, serum samples were immediately centrifuged at maximum speed and transferred to a new tube. A small amount (25UL) of each serum (25UL) was diluted 1:1 in the assay buffer provided in the kit. Each serum was incubated with 13 different bead populations, distinguished by size and internal APC fluorescent dye, and bound to 13 different human inflammatory cytokines and chemokines, including IL-1β, IFN-α2, IFN-γ, TNF-α, MCP-1 (CCL2), IL-6, IL-8 (CXCL8), IL-170, and IL-17. IL-18, IL-23, and IL-33. The next day, beads were first incubated with cytokine-specific biotinylated antibodies, then with streptavidin-phycoerythrain, and quickly acquired on a BD precision C6 plus flow cytometer. Cytokine-specific populations were separated based on size and internal APC fluorescence intensity. The concentration of specific cytokines was quantified based on PE fluorescence signals according to the standard curve generated in the same assay. Measurements were confirmed while blinded to the origin of the sample. All samples were assayed in duplicate, and values indicating values above the standard curve were retested with appropriate dilutions.

The FIB-4 index was assessed in 44 patients. It was calculated using the following equation: [age (years) × AST]/[platelet counts (× 10/L) × ALT].

Thawed serum samples were subjected to protein removal using a 3 kDa cutoff Amicon Ultra-0.5 centrifugal filter device. The filter was washed four times with distilled water to remove glycerol (13,800 g, 4 °C, 20 min) and then centrifuged at 13,800 g for 90 min. NMR buffer (250 mm phosphate buffer khrear/kHPOpH 7.4- containing 0.82 mm sodium trimethylsilylpropanoated10% do, and 2%) Added to each filtered serum to reach a final volume of 600 μL. The resulting solution was then transferred to a 5 mM NMR tube.

The H-NMR spectra of each sample were obtained using a Bruker Avance 700 MHz spectrometer equipped with a triple-resonant TXI probe and a SampleXPress Lite AutoSampler. Spectra were collected at 25°C at 25°C using 1024 scans, four dummy scans, a spectral width of 16 ppm, an acquisition time of 2 s, a relaxation delay of 3 s, and a mixing time of 100 ms. After the acquisition, the spectrum was processed with a line broadening of 0.5 Hz, followed by manual phase and baseline correction. Metabolites were quantified with TSP- using Chenomx Nmrsuite 8.5 (Chenomx Inc.). d It acts as an internal standard. 52 metabolites were quantified using this method in almost all samples.

All statistical analyses were performed using r v4.3.1 (https://www.r-project.org), and the plot was created using Excel. All inference tests are 2 tailed with a small alpha level of 0.05. Due to the exploratory nature of the analysis, no adjustments for multiplicity were made. The p-values of the most important association between metabolites and fibrosis were further validated using Bonferroni correction. Metabolite concentrations were normalized prior to statistical analysis using normalization of stochastic quotientsand data from the inflammation panel were log-transformed. The combined dataset, including the hemogram, inflammatory panel and metabolic panel, consisted of 84 variables.

Additionally, variables describing the demographic and health status of subjects, including age, gender (for women), BMI, fibrosis, diabetes, hypertension, and dyslipidemia, were included in the analysis. Fibrosis variables were created by assigning 0th values to the control group and assigning values from 1 (f = 0,1) to 4 (f = 2,3,4) based on the fibrosis scale.. table S1 View all data used in this analysis.

Collinearity of the variables was tested before calculating interaction terms. As age and fibrosis were found to be collinear, all conditions were extended using unit variation (UV).. The interaction terms between age and fibrosis were then calculated and expanded. Finally, all measured variables were uVed. This allows the coefficients to be comparable and allows you to determine the relative importance of the effect.

To calculate a profile linked to an individual condition, use each of the 84 measurement variables (m) was linearly correlated with seven parameters (c) Describe the demographics and health conditions mentioned above, and the interaction terms that follow the formula. (1)As mentioned above:

For each of the 84 equations, the values of nine coefficients, standard error, t-value, p-value, and 95% CI are calculated, and the complete results are shown in the table. S2. A set of metabolites that exhibit statistically significant associations with a particular condition represent the corresponding profile. This approach was used to investigate the correlation between different levels and fibrosis stages across the F0-F4 spectrum. This model is described in equations. (1), highlighted the need to convert fibrosis status into binary variables as required by more frequently used logistic regression techniques.

An unparalleled logistic regression model was employed in assessing the predictive ability of different indices to explain the severity of liver fibrosis (F = 0-1 vs F = 2-4) without adjusting for age and gender. Potential problems of logistic regression bias and separation, particularly when small sample sizes or events were rare, were utilized using penalty maximum likelihood regression implemented using penalty maximum likelihood logistic regression. writerf function. Using this same approach, we constructed a mixed model combining metabolite concentrations.

To construct the GP index, 84 inflammation, hematology, and metabolic variables were first analyzed using multivariate linear regression tailored to age and stage of fibrosis. Variables that are significantly related to fibrosis and not age-related (Bonferroni-Adjusted) p<0.001) was retained as a candidate marker. This procedure reduced the dimensions of the dataset and eliminated age-related confounding factors. To avoid overparameterization and to assess the model with independent data, a cohort of 63 MASLD patients was randomly divided into a training set (n = 48) and a validation set (n = 15). All combinations of selected candidate metabolites were tested on the training set using Pseudo-Ralias and Bayesian Information Criteria (AIC, BIC)Cross-validation Accuracy (CVA)and the area under the ROC curve (AUC). AIC) and BIC used the (along with CVA) to provide insights into model fit and predictive performance. Pseudo R Statistics measure the explanatory power of the model and show how well the model explains the variance of the dependent variables. Finally, Delong's nonparametric test was used to compare AUCs of two ROC curves to assess the discriminant ability of the model. To determine the optimal model with all metrics in mind, we created a combined index that takes into account the relative importance of each metric. pseudo-rCVA, and AUC values are scaled between 0 and 1, with one representing the best performance. We used inverted scales for AIC and BIC to ensure that low scores for AIC and BIC correspond to higher normalized values, indicating the model that best describes the data with the fewest parameters. The best overall performance and saving combination was chosen to define the GP index. The final model coefficients were calculated only on the training set and were then applied without modifying the validation set to assess the performance of the invisible data. We used the Likelihood Ratio Test (LRT) and its associated P values to test whether inclusion of age or age and gender as covariants provided a better index.