Circulating Metabolites Associated with Postprandial Satiety in Overweight/Obese Participants: The SATIN Study

Pilot-Study to Explore Metabolic Signature of Type 2 Diabetes: A Pipeline of Tree-Based Machine Learning and Bioinformatics Techniques for Biomarkers Discovery

BACKGROUND

This study aims to identify unique metabolomics biomarkers associated with Type 2 Diabetes (T2D) and develop an accurate diagnostics model using tree-based machine learning (ML) algorithms integrated with bioinformatics techniques.

METHODS

Univariate and multivariate analyses such as fold change, a receiver operating characteristic curve (ROC), and Partial Least-Squares Discriminant Analysis (PLS-DA) were used to identify biomarker metabolites that showed significant concentration in T2D patients. Three tree-based algorithms [eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Adaptive Boosting (AdaBoost)] that demonstrated robustness in high-dimensional data analysis were used to create a diagnostic model for T2D.

RESULTS

As a result of the biomarker discovery process validated with three different approaches, Pyruvate, D-Rhamnose, AMP, pipecolate, Tetradecenoic acid, Tetradecanoic acid, Dodecanediothioic acid, Prostaglandin E3/D3 (isobars), ADP and Hexadecenoic acid were determined as potential biomarkers for T2D. Our results showed that the XGBoost model [accuracy = 0.831, F1-score = 0.845, sensitivity = 0.882, specificity = 0.774, positive predictive value (PPV) = 0.811, negative-PV (NPV) = 0.857 and Area under the ROC curve (AUC) = 0.887] had the slight highest performance measures.

CONCLUSIONS

ML integrated with bioinformatics techniques offers accurate and positive T2D candidate biomarker discovery. The XGBoost model can successfully distinguish T2D based on metabolites.

NMR metabolite quantification of a synthetic urine sample: an inter-laboratory comparison of processing workflows

INTRODUCTION

Absolute quantification of individual metabolites in complex biological samples is crucial in targeted metabolomic profiling.

OBJECTIVES

An inter-laboratory test was performed to evaluate the impact of the NMR software, peak-area determination method (integration vs. deconvolution) and operator on quantification trueness and precision.

METHODS

A synthetic urine containing 32 compounds was prepared. One site prepared the urine and calibration samples, and performed NMR acquisition. NMR spectra were acquired with two pulse sequences including water suppression used in routine analyses. The pre-processed spectra were sent to the other sites where each operator quantified the metabolites using internal referencing or external calibration, and his/her favourite in-house, open-access or commercial NMR tool.

RESULTS

For 1D NMR measurements with solvent presaturation during the recovery delay (zgpr), 20 metabolites were successfully quantified by all processing strategies. Some metabolites could not be quantified by some methods. For internal referencing with TSP, only one half of the metabolites were quantified with a trueness below 5%. With peak integration and external calibration, about 90% of the metabolites were quantified with a trueness below 5%. The NMRProcFlow integration module allowed the quantification of several additional metabolites. The number of quantified metabolites and quantification trueness improved for some metabolites with deconvolution tools. Trueness and precision were not significantly different between zgpr- and NOESYpr-based spectra for about 70% of the variables.

CONCLUSION

External calibration performed better than TSP internal referencing. Inter-laboratory tests are useful when choosing to better rationalize the choice of quantification tools for NMR-based metabolomic profiling and confirm the value of spectra deconvolution tools.

No Evidence That Circulating GLP-1 or PYY Are Associated with Increased Satiety during Low Energy Diet-Induced Weight Loss: Modelling Biomarkers of Appetite

Bariatric surgery and pharmacology treatments increase circulating glucagon-like peptide-1 (GLP-1) and peptide YY (PYY), in turn promoting satiety and body weight (BW) loss. However, the utility of GLP-1 and PYY in predicting appetite response during dietary interventions remains unsubstantiated. This study investigated whether the decrease in hunger observed following low energy diet (LED)-induced weight loss was associated with increased circulating 'satiety peptides', and/or associated changes in glucose, glucoregulatory peptides or amino acids (AAs). In total, 121 women with obesity underwent an 8-week LED intervention, of which 32 completed an appetite assessment via a preload challenge at both Week 0 and Week 8, and are reported here. Visual analogue scales (VAS) were administered to assess appetite-related responses, and blood samples were collected over 210 min post-preload. The area under the curve (AUC_0-210), incremental AUC (iAUC_0-210), and change from Week 0 to Week 8 (∆) were calculated. Multiple linear regression was used to test the association between VAS-appetite responses and blood biomarkers. Mean (±SEM) BW loss was 8.4 ± 0.5 kg (-8%). Unexpectedly, the decrease in ∆AUC_0-210 hunger was best associated with decreased ∆AUC_0-210 GLP-1, GIP, and valine (p < 0.05, all), and increased ∆AUC_0-210 glycine and proline (p < 0.05, both). The majority of associations remained significant after adjusting for BW and fat-free mass loss. There was no evidence that changes in circulating GLP-1 or PYY were predictive of changes in appetite-related responses. The modelling suggested that other putative blood biomarkers of appetite, such as AAs, should be further investigated in future larger longitudinal dietary studies.

Does a Higher Protein Diet Promote Satiety and Weight Loss Independent of Carbohydrate Content? An 8-Week Low-Energy Diet (LED) Intervention

Both higher protein (HP) and lower carbohydrate (LC) diets may promote satiety and enhance body weight (BW) loss. This study investigated whether HP can promote these outcomes independent of carbohydrate (CHO) content. 121 women with obesity (BW: 95.1 ± 13.0 kg, BMI: 35.4 ± 3.9 kg/m2) were randomised to either HP (1.2 g/kg BW) or normal protein (NP, 0.8 g/kg BW) diets, in combination with either LC (28 en%) or normal CHO (NC, 40 en%) diets. A low-energy diet partial diet replacement (LEDpdr) regime was used for 8 weeks, where participants consumed fixed-energy meal replacements plus one ad libitum meal daily. Four-day dietary records showed that daily energy intake (EI) was similar between groups (p = 0.744), but the difference in protein and CHO between groups was lower than expected. Following multiple imputation (completion rate 77%), decrease in mean BW, fat mass (FM) and fat-free mass (FFM) at Week 8 in all was 7.5 ± 0.7 kg (p < 0.001), 5.7 ± 0.5 kg (p < 0.001), and 1.4 ± 0.7 kg (p = 0.054) respectively, but with no significant difference between diet groups. LC (CHO×Week, p < 0.05), but not HP, significantly promoted postprandial satiety during a preload challenge. Improvements in blood biomarkers were unrelated to LEDpdr macronutrient composition. In conclusion, HP did not promote satiety and BW loss compared to NP LEDpdr, irrespective of CHO content.

Postprandial glycine as a biomarker of satiety: A dose-rising randomised control trial of whey protein in overweight women

This study aimed to identify biomarkers of appetite response, modelled using a dose-rising whey protein preload intervention. Female participants (n = 24) with body mass index (BMI) between 23 and 40 kg/m² consumed preload beverages (0 g protein water control, WC; 12.5 g low-dose protein, LP; or 50.0 g high-dose protein, HP) after an overnight fast, in a randomised cross over design. Repeated venous blood samples were collected to measure plasma biomarkers of appetite response, including glucose, glucoregulatory peptides, gut peptides, and amino acids (AAs). Appetite was assessed using Visual Analogue Scales (VAS) and ad libitum energy intake (EI). Dose-rising protein beverage significantly changed the postprandial trajectory of almost all biomarkers (treatment*time, p < 0.05), but did not suppress postprandial appetite (treatment*time, p > 0.05) or EI (ANOVA, p = 0.799). Circulating glycine had the strongest association with appetite response. Higher area under the curve (AUC_0-240) glycine was associated with lower EI (p = 0.026, trend). Furthermore, circulating glycine was associated with decreased Hunger in all treatment groups, whereas the associations of glucose, alanine and amylin with appetite were dependent on treatment groups. Multivariate models, incorporating multiple biomarkers, improved the estimation of appetite response (marginal R², range: 0.13-0.43). In conclusion, whilst glycine, both alone and within a multivariate model, can estimate appetite response to both water and whey protein beverage consumption, a large proportion of variance in appetite response remains unexplained. Most biomarkers, when assessed in isolation, are poor predictors of appetite response, and likely of utility only in combination with VAS and EI.