Analysis of high-dimensional metabolomics data with complex temporal dynamics using RM-ASCA.

The intricate dependency structure of biological "omics" data, particularly those originating from longitudinal intervention studies with frequently sampled repeated measurements renders the analysis of such data challenging. The high-dimensionality, inter-relatedness of multiple outcomes, and heterogeneity in the studied systems all add to the difficulty in deriving meaningful information. In addition, the subtle differences in dynamics often deemed meaningful in nutritional intervention studies can be particularly challenging to quantify. In this work we demonstrate the use of quantitative longitudinal models within the repeated-measures ANOVA simultaneous component analysis+ (RM-ASCA+) framework to capture the dynamics in frequently sampled longitudinal data with multivariate outcomes. We illustrate the use of linear mixed models with polynomial and spline basis expansion of the time variable within RM-ASCA+ in order to quantify non-linear dynamics in a simulation study as well as in a metabolomics data set. We show that the proposed approach presents a convenient and interpretable way to systematically quantify and summarize multivariate outcomes in longitudinal studies while accounting for proper within subject dependency structures.

Longitudinal associations of plasma metabolites with persistent fatigue among colorectal cancer survivors up to 2 years after treatment.

The underlying biological mechanisms causing persistent fatigue complaints after colorectal cancer treatment need further investigation. We investigated longitudinal associations of circulating concentrations of 138 metabolites with total fatigue and subdomains of fatigue between 6 weeks and 2 years after colorectal cancer treatment. Among stage I-III colorectal cancer survivors (n = 252), blood samples were obtained at 6 weeks, and 6, 12 and 24 months posttreatment. Total fatigue and fatigue subdomains were measured using a validated questionnaire. Tandem mass spectrometry was applied to measure metabolite concentrations (BIOCRATES AbsoluteIDQp180 kit). Confounder-adjusted longitudinal associations were analyzed using linear mixed models, with false discovery rate (FDR) correction. We assessed interindividual (between-participant differences) and intraindividual longitudinal associations (within-participant changes over time). In the overall longitudinal analysis, statistically significant associations were observed for 12, 32, 17 and three metabolites with total fatigue and the subscales "fatigue severity," "reduced motivation" and "reduced activity," respectively. Specifically, higher concentrations of several amino acids, lysophosphatidylcholines, diacylphosphatidylcholines, acyl-alkylphosphatidylcholines and sphingomyelins were associated with less fatigue, while higher concentrations of acylcarnitines were associated with more fatigue. For "fatigue severity," associations appeared mainly driven by intraindividual associations, while for "reduced motivation" stronger interindividual associations were found. We observed longitudinal associations of several metabolites with total fatigue and fatigue subscales, and that intraindividual changes in metabolites over time were associated with fatigue severity. These findings point toward inflammation and an impaired energy metabolism due to mitochondrial dysfunction as underlying mechanisms. Mechanistic studies are necessary to determine whether these metabolites could be targets for intervention.

Comparison of metabolic states using genome-scale metabolic models.

Genome-scale metabolic models (GEMs) are comprehensive knowledge bases of cellular metabolism and serve as mathematical tools for studying biological phenotypes and metabolic states or conditions in various organisms and cell types. Given the sheer size and complexity of human metabolism, selecting parameters for existing analysis methods such as metabolic objective functions and model constraints is not straightforward in human GEMs. In particular, comparing several conditions in large GEMs to identify condition- or disease-specific metabolic features is challenging. In this study, we showcase a scalable, model-driven approach for an in-depth investigation and comparison of metabolic states in large GEMs which enables identifying the underlying functional differences. Using a combination of flux space sampling and network analysis, our approach enables extraction and visualisation of metabolically distinct network modules. Importantly, it does not rely on known or assumed objective functions. We apply this novel approach to extract the biochemical differences in adipocytes arising due to unlimited vs blocked uptake of branched-chain amino acids (BCAAs, considered as biomarkers in obesity) using a human adipocyte GEM (iAdipocytes1809). The biological significance of our approach is corroborated by literature reports confirming our identified metabolic processes (TCA cycle and Fatty acid metabolism) to be functionally related to BCAA metabolism. Additionally, our analysis predicts a specific altered uptake and secretion profile indicating a compensation for the unavailability of BCAAs. Taken together, our approach facilitates determining functional differences between any metabolic conditions of interest by offering a versatile platform for analysing and comparing flux spaces of large metabolic networks.

Personalized computational model quantifies heterogeneity in postprandial responses to oral glucose challenge.

Plasma glucose and insulin responses following an oral glucose challenge are representative of glucose tolerance and insulin resistance, key indicators of type 2 diabetes mellitus pathophysiology. A large heterogeneity in individuals' challenge test responses has been shown to underlie the effectiveness of lifestyle intervention. Currently, this heterogeneity is overlooked due to a lack of methods to quantify the interconnected dynamics in the glucose and insulin time-courses. Here, a physiology-based mathematical model of the human glucose-insulin system is personalized to elucidate the heterogeneity in individuals' responses using a large population of overweight/obese individuals (n = 738) from the DIOGenes study. The personalized models are derived from population level models through a systematic parameter selection pipeline that may be generalized to other biological systems. The resulting personalized models showed a 4-5 fold decrease in discrepancy between measurements and model simulation compared to population level. The estimated model parameters capture relevant features of individuals' metabolic health such as gastric emptying, endogenous insulin secretion and insulin dependent glucose disposal into tissues, with the latter also showing a significant association with the Insulinogenic index and the Matsuda insulin sensitivity index, respectively.

Metabolic profiling of tissue-specific insulin resistance in human obesity: results from the Diogenes study and the Maastricht Study.

BACKGROUND: Recent evidence indicates that insulin resistance (IR) in obesity may develop independently in different organs, representing different etiologies toward type 2 diabetes and other cardiometabolic diseases. The aim of this study was to investigate whether IR in the liver and IR in skeletal muscle are associated with distinct metabolic profiles. METHODS: This study includes baseline data from 634 adults with overweight or obesity (BMI ≥ 27 kg/m2) (≤65 years; 63% women) without diabetes of the European Diogenes Study. Hepatic insulin resistance index (HIRI) and muscle insulin sensitivity index (MISI), were derived from a five-point OGTT. At baseline 17 serum metabolites were identified and quantified by nuclear-magnetic-resonance spectroscopy. Linear mixed model analyses (adjusting for center, sex, body mass index (BMI), waist-to-hip ratio) were used to associate HIRI and MISI with these metabolites. In an independent sample of 540 participants without diabetes (BMI ≥ 27 kg/m2; 40-65 years; 46% women) of the Maastricht Study, an observational prospective population-based cohort study, 11 plasma metabolites and a seven-point OGTT were available for validation. RESULTS: Both HIRI and MISI were associated with higher levels of valine, isoleucine, oxo-isovaleric acid, alanine, lactate, and triglycerides, and lower levels of glycine (all p < 0.05). HIRI was also associated with higher levels of leucine, hydroxyisobutyrate, tyrosine, proline, creatine, and n-acetyl and lower levels of acetoacetate and 3-OH-butyrate (all p < 0.05). Except for valine, these results were replicated for all available metabolites in the Maastricht Study. CONCLUSIONS: In persons with obesity without diabetes, both liver and muscle IR show a circulating metabolic profile of elevated (branched-chain) amino acids, lactate, and triglycerides, and lower glycine levels, but only liver IR associates with lower ketone body levels and elevated ketogenic amino acids in circulation, suggestive of decreased ketogenesis. This knowledge might enhance developments of more targeted tissue-specific interventions to prevent progression to more severe disease stages.

A computational model of postprandial adipose tissue lipid metabolism derived using human arteriovenous stable isotope tracer data.

Given the association of disturbances in non-esterified fatty acid (NEFA) metabolism with the development of Type 2 Diabetes and Non-Alcoholic Fatty Liver Disease, computational models of glucose-insulin dynamics have been extended to account for the interplay with NEFA. In this study, we use arteriovenous measurement across the subcutaneous adipose tissue during a mixed meal challenge test to evaluate the performance and underlying assumptions of three existing models of adipose tissue metabolism and construct a new, refined model of adipose tissue metabolism. Our model introduces new terms, explicitly accounting for the conversion of glucose to glyceraldehye-3-phosphate, the postprandial influx of glycerol into the adipose tissue, and several physiologically relevant delays in insulin signalling in order to better describe the measured adipose tissues fluxes. We then applied our refined model to human adipose tissue flux data collected before and after a diet intervention as part of the Yoyo study, to quantify the effects of caloric restriction on postprandial adipose tissue metabolism. Significant increases were observed in the model parameters describing the rate of uptake and release of both glycerol and NEFA. Additionally, decreases in the model's delay in insulin signalling parameters indicates there is an improvement in adipose tissue insulin sensitivity following caloric restriction.

Plasma lipid profiling of tissue-specific insulin resistance in human obesity.

BACKGROUND/OBJECTIVES: Obesity-associated insulin resistance (IR) may develop in multiple organs, representing different aetiologies towards cardiometabolic diseases. This study aimed to identify distinct plasma lipid profiles in overweight/obese individuals who show muscle-IR and/or liver-IR. SUBJECTS/METHODS: Baseline data of the European multicenter DiOGenes project were used (n = 640; 401 women, nondiabetic BMI: 27-45 kg/m2). Muscle insulin sensitivity index (MISI) and hepatic insulin resistance index (HIRI) were derived from a 5-point oral glucose tolerance test. The 140 plasma lipids were quantified by liquid chromatography-mass spectrometry. Linear mixed models were used to evaluate associations between MISI, HIRI and plasma lipids. RESULTS: MISI was comparable between sexes while HIRI and triacylglycerol (TAG) levels were lower in women than in men. MISI was associated with higher lysophosphatidylcholine (LPC) levels (standardized (std)ß = 0.126; FDR-p = 0.032). Sex interactions were observed for associations between HIRI, TAG and diacylglycerol (DAG) lipid classes. In women, but not in men, HIRI was associated with higher levels of TAG (44 out of 55 species) and both DAG species (stdß: 0.139-0.313; FDR-p < 0.05), a lower odd-chain/even-chain TAG ratio (stdß = -0.182; FDR-p = 0.005) and a lower very-long-chain/long-chain TAG ratio (stdß = -0.156; FDR-p = 0.037). CONCLUSIONS: In overweight/obese individuals, muscle insulin sensitivity is associated with higher plasma LPC concentrations. Women have less hepatic IR and lower TAG than men. Nevertheless, hepatic IR is associated with higher plasma TAG and DAG concentrations and a lower abundance of odd-chain and very-long-chain TAG in women, but not in men. This suggests a more pronounced worsening of plasma lipid profile in women with the progression of hepatic IR.

Classical Pathway of Complement Activation: Longitudinal Associations of C1q and C1-INH With Cardiovascular Outcomes: The CODAM Study (Cohort on Diabetes and Atherosclerosis Maastricht)-Brief Report.

OBJECTIVE: The classical complement pathway has been assigned both protective and pathological effects in cardiovascular disease (CVD), but human data are lacking. We determined the associations of the pattern recognition factor C1q and the regulator C1-INH (C1-inhibitor) with incident CVD, carotid intima-media thickness, endothelial dysfunction, and low-grade inflammation. APPROACH AND RESULTS: Baseline concentrations of C1q and C1-INH were measured in the CODAM study (Cohort on Diabetes and Atherosclerosis Maastricht; n=574; 61% men; age, 60±7 years). The 7-year incidence of CVD in participants free of CVD at baseline was evaluated using logistic regression analyses (n=342; 73 cases). The lowest incidence of CVD was observed in the middle tertile of C1q (Tlow compared with Tmiddle: odds ratio, 2.38 [95% confidence interval, 1.14-4.95]; Thigh compared with Tmiddle: odds ratio, 1.96 [95% confidence interval, 0.94-4.07]). C1-INH was not associated with CVD. During the 7-year follow-up period, C1q and C1-INH were not, or inconsistently, associated with carotid intima-media thickness or with biomarker scores reflecting endothelial dysfunction and low-grade inflammation. CONCLUSIONS: Our results suggest a nonlinear association between C1q and incident CVD. This supports the concept that early steps in classical pathway activation may have both protective and pathological effects on human CVD.

Prospective associations of dietary carbohydrate, fat, and protein intake with ß-cell function in the CODAM study.

PURPOSE: Type 2 diabetes mellitus (T2DM) is characterized by both impaired pancreatic ß-cell function (BCF) and insulin resistance. In the etiology of T2DM, BCF basically determines whether a person with a certain degree of insulin resistance develops T2DM, as ß-cells are able to compensatorily increase insulin secretion. The effects of dietary intake on BCF are largely unknown. Our study aim was to investigate whether dietary macronutrient intake predicts BCF. METHODS: Prospective data (median follow-up 7 years) of 303 individuals recruited from the CODAM study population (aged 40-70 years, 39% women) were analyzed. BCF was measured by C-peptide deconvolution and physiological modeling of data from a 5-point, 75-g, 2-h oral glucose tolerance test. Macronutrient intake was estimated by a 178-item Food Frequency Questionnaire. RESULTS: Associations adjusted for relevant covariates of baseline macronutrient intake with model-derived parameters describing BCF (glucose sensitivity, rate sensitivity or potentiation) or C-peptidogenic index were detected for trans fat [standardized regression coefficient (95%-CI) glucose sensitivity - 0.14 (- 0.26, - 0.01)] per g, cholesterol [potentiation 0.20 (0.02, 0.37)] per 100 mg, dietary fiber [glucose sensitivity 0.21 (0.08, 0.33)] per 10 g, MUFA glucose sensitivity 0.16 (0.02, 0.31) per 10 g, and polysaccharide [potentiation - 0.24 (- 0.43, - 0.05), C-peptidogenic index - 0.16 (- 0.29 - 0.03); odds ratio lowest versus highest tertile (95%-CI) rate sensitivity 1.51 (1.06, 2.15)) per 50 g. CONCLUSIONS: In this population at high risk for developing T2DM, polysaccharide and trans fat intake were associated with worse BCF, whereas increased intake of MUFA, dietary cholesterol, and fiber were associated with better BCF.

Medication adherence among gout patients initiated allopurinol: a retrospective cohort study in the Clinical Practice Research Datalink (CPRD).

Objectives: When urate lowering therapy is indicated in patients with gout, medication adherence is essential. This study assesses non-persistence and non-adherence in patients with newly diagnosed gout, and identifies factors associated with poor medication adherence. Methods: A retrospective data analysis was performed within the UK Clinical Practice Research Datalink (1987-2014) among incident gout patients, aged ⩾40 years and starting allopurinol (n = 48 280). The proportion of patients non-persistent (a first medication gap of ⩾90 days) after 1 and 5 years, and median time until a first 90-day gap was estimated using Kaplan-Meier statistics in those starting allopurinol and restarting after a first interruption. Non-adherence (proportion of days covered <80%) over the full observation period was calculated. Multivariable Cox- or logistic regressions assessed factors associated with non-persistence or non-adherence, respectively. Results: Non-persistence increased from 38.5% (95% CI: 38.1, 38.9) to 56.9% (95% CI: 56.4, 57.4) after 1 and 5 years of initiation. Median time until a first 90-day gap was 1029 days (95% CI: 988, 1078) and 61% were non-adherent. After a first gap, 43.3% (95% CI: 42.7, 43.9) restarted therapy within 1 year, yet only 52.3% (95% CI: 51.4, 53.1) persisted for 1 year. Being female and a current smoker increased the risk for non-persistence and non-adherence, while older age, overweight, receiving anti-hypertensive medication or colchicine and suffering from dementia, diabetes or dyslipidaemia decreased the risk. Conclusion: Medication adherence among gout patients starting allopurinol is poor, particularly among females and younger patients and patients with fewer comorbidities. Medication adherence remains low in those reinitiating after a first gap.

Estimating real cell size distribution from cross-section microscopy imaging.

MOTIVATION: Microscopy imaging is an essential tool for medical diagnosis and molecular biology. It is particularly useful for extracting information about disease states, tissue heterogeneity and cell specific parameters such as cell type or cell size from biological specimens. However, the information obtained from the images is likely to be subjected to sampling and observational bias with respect to the underlying cell size/type distributions. RESULTS: We present an algorithm, Estimate Tissue Cell Size/Type Distribution (EstiTiCS), for the adjustment of the underestimation of the number of small cells and the size of measured cells while accounting for the section thickness independent of the tissue type. We introduce the sources of bias under different tissue distributions and their effect on the measured values with simulation experiments. Furthermore, we demonstrate our method on histological sections of paraffin-embedded adipose tissue sample images from 57 people from a dietary intervention study. This data consists of measured cell size and its distribution over the dietary intervention period at four time points. Adjusting for the bias with EstiTiCS results in a closer fit to the true/expected adipocyte size distribution with earlier studies. Therefore, we conclude that our method is suitable as the final step in estimating the tissue wide cell type/size distribution from microscopy imaging pipeline. AVAILABILITY AND IMPLEMENTATION: Source code and its documentation are available at https://github.com/michaelLenz/EstiTiCS The whole pipeline of our method is implemented in R and makes use of the 'nloptr' package. Adipose tissue data used for this study are available on request. CONTACT: Michael.Lenz@Maastrichtuniversity.nl, Gokhan.Ertaylan@Maastrichtuniversity.nl.

Distinct Longitudinal Associations of MBL, MASP-1, MASP-2, MASP-3, and MAp44 With Endothelial Dysfunction and Intima-Media Thickness: The Cohort on Diabetes and Atherosclerosis Maastricht (CODAM) Study.

OBJECTIVE: Previous studies suggested that the lectin-complement pathway plays a complex role in cardiovascular disease (CVD). To date, no prospective human studies have investigated the relationship between the initiating factor of the lectin pathway, that is, mannose-binding lectin (MBL), and low-grade inflammation, endothelial dysfunction, or carotid intima-media thickness (cIMT). Moreover, MBL-associated proteases (MASPs) and MBL-associated proteins (MAps), which mediate downstream complement activation, have not been studied in the development of CVD. APPROACH AND RESULTS: In a prospective cohort (n=574; age 60±7 years; 7-year follow-up), we investigated longitudinal associations of plasma MBL, MASP-1, MASP-2, MASP-3, and MAp44 with biomarker scores that reflect low-grade inflammation and endothelial dysfunction, respectively, and with cIMT. We also investigated their associations with incident CVD (n=73). In adjusted analyses, low-grade inflammation was lowest in the middle tertile (TMiddle) of MBL, that is, TMiddle was 0.19 SD (0.03 to 0.34) lower than TLow, and 0.15 SD (-0.02 to 0.31) lower than THigh. cIMT was 28 µm (-50 to -5) lower in the highest MBL tertile (THigh) than in TMiddle and did not differ between TLow and TMiddle. MBL was not associated with endothelial dysfunction or CVD. MASP-1 and MASP-2 were not associated with any cardiovascular outcomes. MASP-3 and MAp44 were, independently of MBL levels, associated with endothelial dysfunction (per 1 SD higher MASP-3: ß=0.10 SD [0.02 to 0.18]; per 1 SD higher MAp44 ß=0.12 SD [0.04 to 0.20]) but not with low-grade inflammation, cIMT, or CVD. CONCLUSIONS: High MBL may contribute to low cIMT, whereas the association of MBL with low-grade inflammation was nonlinear. MASP-1 and MASP-2 were not associated with adverse cardiovascular outcomes. MASP-3 and MAp44 may play a role in endothelial dysfunction, potentially independent of lectin-pathway activation.

Early Life Antibiotic Exposure and Weight Development in Children.

OBJECTIVE: To examine the timing, frequency, and type of antibiotic exposure during the first 10 years of life in association with (over)weight across this period in a cohort of 979 children. STUDY DESIGN: Within the Child, Parents and Health: Lifestyle and Genetic Constitution Birth Cohort Study, antibiotic exposure record was obtained from general practitioners. Anthropometric outcomes (age- and sex-standardized body mass index, weight and height z-scores, and overweight) were measured repeatedly at 7 time points during the first 10 years of life. Generalized estimating equations method was used for statistical analysis. RESULTS: After adjusting for confounding factors, children exposed to one course of antibiotics compared with none in the first 6 months of life had increased weight- (adjusted generalized estimating equations estimates [adjß] 0.24; 95% CI 0.03-0.44) and height (adjß 0.23; 95% CI 0.0002-0.46) z-scores; exposure to ≥2 courses during the second year of life was associated with both increased weight (adjß 0.34; 95% CI 0.07-0.60), and height z-scores (adjß 0.29; 95% CI -0.003 to 0.59). Exposure later in life was not associated with anthropometric outcomes. Associations with weight z-scores were mainly driven by exposure to broad- (≥2 courses: adjß 0.11; 95% CI 0.003-0.22) and narrow-spectrum ß-lactams (1 course: adjß 0.18; 95% CI 0.005-0.35) during the follow-up period. Specific antibiotic used was not associated with body mass index z-scores and overweight. CONCLUSIONS: Repeated exposure to antibiotics early in life, especially ß-lactam agents, is associated with increased weight and height. If causality of obesity can be established in future studies, this further highlights the need for restrictive antibiotic use and avoidance of prescriptions when there is minimal clinical benefit.

Determinants of the prevalence of gout in the general population: a systematic review and meta-regression.

Studies on the occurrence of gout show a large range in estimates. However, a clear insight into the factors responsible for this variation in estimates is lacking. Therefore, our aim was to review the literature on the prevalence and incidence of gout systematically and to obtain insight into the degree of and factors contributing to the heterogeneity. We searched MEDLINE, EMBASE, and Web of Science (January 1962 to July 2012) to identify primary studies on the prevalence and incidence of gout in the general population. Data were extracted by two persons on sources of clinical heterogeneity, methodological heterogeneity, and variation in outcome reporting. Meta-analysis and meta-regression analysis were performed for the prevalence of gout. Of 1,466 articles screened, 77 articles were included, of which 71 reported the prevalence and 12 the incidence of gout. The pooled prevalence (67 studies; N = 12,226,425) based on a random effects model was 0.6% (95% CI 0.4; 0.7), however there was a high level of heterogeneity (I(2) = 99.9%). Results from a mixed-effects meta-regression model indicated that age (p = 0.019), sex (p < 0.001), continent (p < 0.001), response rate (p = 0.016), consistency in data collection (p = 0.002), and case definition (p < 0.001) were significantly associated with gout prevalence and jointly accounted for 88.7% of the heterogeneity. The incidence in the total population ranged from 0.06 to 2.68 per 1,000 person-years. In conclusion, gout is a common disease and the large variation in the prevalence data on gout is explained by sex, continent on which the study was performed, and the case definition of gout.

Dietary nucleotide and nucleoside exposure in infancy and atopic dermatitis, recurrent wheeze, and allergic sensitization.

We hypothesized that early life exposure to nucleotides and nucleosides lowers the risk of recurrent wheeze, atopic dermatitis, and allergic sensitization among n = 429 children. Concentrations in breast milk were established by high-performance liquid chromatography; concentrations in formula milks were obtained from manufacturers. Questionnaires and home visits were used to assess outcomes. Adjusted odds ratios in the highest tertile compared with those in the lowest tertile of exposure ranged from 1.11 to 1.99 in predominantly formula-fed children, and from 0.40 to 0.53 in predominantly breast-fed children, but were not significant. Thus, we found no evidence for association between nucleotide and nucleoside exposure and the development of atopic outcomes in children up to 2 years.

The cross-sectional association between uric acid and atherosclerosis and the role of low-grade inflammation: the CODAM study.

OBJECTIVES: The aims of this study were to investigate (i) associations between uric acid and prevalent cardiovascular disease (CVD), ankle-arm blood pressure index (AAIx) and carotid intima-media thickness (CIMT) in the total population and in predefined subgroups according to glucose metabolism status and (ii) the extent to which these associations are explained by low-grade inflammation. METHODS: Cross-sectional analyses were conducted among 530 individuals [60.6% men, mean age 58.9 years (s.d. 6.9), 52.6% normal glucose metabolism (NGM)] at increased risk of CVD from the Cohort of Diabetes and Atherosclerosis Maastricht study. A low-grade inflammation score was computed by averaging the z-scores of eight inflammation markers [CRP, TNF-α, IL-6, IL-8, serum amyloid A, intercellular adhesion molecule 1 (ICAM-1), ceruloplasmin and haptoglobin]. RESULTS: After adjustment for traditional CVD risk factors, plasma uric acid (per s.d. of 81 µmol/l) was associated with CVD in individuals with NGM [odds ratio (OR) = 1.66, 95% CI 1.06, 2.58] but not with disturbed glucose metabolism (DGM) (OR = 0.81, 95% CI 0.55, 1.19, P interaction = 0.165). Uric acid was associated with CIMT in the total population (ß = 0.024, 95% CI 0.007, 0.042) and slightly more strongly in individuals with NGM (ß = 0.030, 95% CI 0.006, 0.054) than DGM (ß = 0.018, 95% CI -0.009, 0.044, P interaction = 0.443). There was no association between uric acid and AAIx in any group (P interaction = 0.058). Uric acid was associated with low-grade inflammation in the total population (ß = 0.074, 95% CI 0.013, 0.134, P interaction = 0.737). Adding low-grade inflammation to the models did not attenuate any of the associations. CONCLUSION: The associations for uric acid with CIMT, and with CVD in NGM only, were not explained by low-grade inflammation. A difference in the strength of the associations between individuals with NGM and DGM was suggested.

Complement c3 is inversely associated with habitual intake of provitamin A but not with dietary fat, fatty acids, or vitamin E in middle-aged to older white adults and positively associated with intake of retinol in middle-aged to older white women.

Complement factor 3 (C3) has been identified as a novel risk factor for obesity-associated cardiometabolic diseases. Data in the literature suggest that C3 concentrations may be influenced by diet. Therefore, we investigated the associations of intake of total fat, specific fatty acids, and fat-soluble vitamin E (and individual tocopherols) and vitamin A (and its dietary precursors) with circulating C3. In a white cohort [Cohort on Diabetes and Atherosclerosis Maastricht (CODAM); n = 501; 59.4 ± 7.1 y; 61% men], associations of habitual nutrient intake (assessed by a food-frequency questionnaire) with circulating C3 were evaluated by using cross-sectional multiple linear regression analyses. Adjustments were first performed for age, sex, glucose metabolism status (i.e., impaired glucose metabolism or type 2 diabetes), and energy intake and subsequently for BMI, waist circumference, alcohol intake, smoking behavior, and season of blood collection. No associations with C3 were observed for total dietary fat intake or intake of specific fatty acids [saturated, monounsaturated, polyunsaturated, n-6 (ω-6), and n-3 (ω- 3) fatty acids], vitamin E, or individual tocopherols. We observed an inverse association with intake of provitamin A carotenoids α-carotene (in µg/d; regression coefficient ß = -0.075; 95% CI: -0.140, -0.010; P = 0.025) and ß-carotene (in µg/d; ß = -0.021; 95% CI: -0.044, 0.002; P = 0.068) with C3 (in mg/L). In contrast, and only in women, dietary retinol intake (in µg/d) was positively associated with C3 (ß = 0.116; 95% CI: 0.014, 0.218; P = 0.026; n = 196). In conclusion, these data suggest that fasting concentrations of C3 may, in a complex manner, be modifiable by variation in dietary provitamin A carotenoids and/or retinol content of the usual diet but most likely not by variations in fat composition and vitamin E content.

Leveraging continuous glucose monitoring for personalized modeling of insulin-regulated glucose metabolism.

Continuous glucose monitoring (CGM) is a promising, minimally invasive alternative to plasma glucose measurements for calibrating physiology-based mathematical models of insulin-regulated glucose metabolism, reducing the reliance on in-clinic measurements. However, the use of CGM glucose, particularly in combination with insulin measurements, to develop personalized models of glucose regulation remains unexplored. Here, we simultaneously measured interstitial glucose concentrations using CGM as well as plasma glucose and insulin concentrations during an oral glucose tolerance test (OGTT) in individuals with overweight or obesity to calibrate personalized models of glucose-insulin dynamics. We compared the use of interstitial glucose with plasma glucose in model calibration, and evaluated the effects on model fit, identifiability, and model parameters' association with clinically relevant metabolic indicators. Models calibrated on both plasma and interstitial glucose resulted in good model fit, and the parameter estimates associated with metabolic indicators such as insulin sensitivity measures in both cases. Moreover, practical identifiability of model parameters was improved in models estimated on CGM glucose compared to plasma glucose. Together these results suggest that CGM glucose may be considered as a minimally invasive alternative to plasma glucose measurements in model calibration to quantify the dynamics of glucose regulation.

Quantifying the effect of nutritional interventions on metabolic resilience using personalized computational models.

The manifestation of metabolic deteriorations that accompany overweight and obesity can differ greatly between individuals, giving rise to a highly heterogeneous population. This inter-individual variation can impede both the provision and assessment of nutritional interventions as multiple aspects of metabolic health should be considered at once. Here, we apply the Mixed Meal Model, a physiology-based computational model, to characterize an individual's metabolic health in silico. A population of 342 personalized models were generated using data for individuals with overweight and obesity from three independent intervention studies, demonstrating a strong relationship between the model-derived metric of insulin resistance (ρ = 0.67, p < 0.05) and the gold-standard hyperinsulinemic-euglycemic clamp. The model is also shown to quantify liver fat accumulation and ß-cell functionality. Moreover, we show that personalized Mixed Meal Models can be used to evaluate the impact of a dietary intervention on multiple aspects of metabolic health at the individual level.