Matrix-Variate Regression for Sparse, Low-Rank Estimation of Brain Connectivities Associated With a Clinical Outcome.

OBJECTIVE: We address the problem of finding brain connectivities that are associated with a clinical outcome or phenotype. METHODS: The proposed framework regresses a (scalar) clinical outcome on matrix-variate predictors which arise in the form of brain connectivity matrices. For example, in a large cohort of subjects we estimate those regions of functional connectivities that are associated with neurocognitive scores. We approach this high-dimensional yet highly structured estimation problem by formulating a regularized estimation process that results in a low-rank coefficient matrix having a sparse set of nonzero entries which represent regions of biologically relevant connectivities. In contrast to the recent literature on estimating a sparse, low-rank matrix from a single noisy observation, our scalar-on-matrix regression framework produces a data-driven extraction of structures that are associated with a clinical response. The method, called Sparsity Inducing Nuclear-Norm Estimator (SpINNEr), simultaneously constrains the regression coefficient matrix in two ways: a nuclear norm penalty encourages low-rank structure while an l1 norm encourages entry-wise sparsity. RESULTS: Our simulations show that SpINNEr outperforms other methods in estimation accuracy when the response-related entries (representing the brain's functional connectivity) are arranged in well-connected communities. SpINNEr is applied to investigate associations between HIV-related outcomes and functional connectivity in the human brain. CONCLUSION AND SIGNIFICANCE: Overall, this work demonstrates the potential of SpINNEr to recover sparse and low-rank estimates under scalar-on-matrix regression framework.

GENERALIZED MATRIX DECOMPOSITION REGRESSION: ESTIMATION AND INFERENCE FOR TWO-WAY STRUCTURED DATA.

Motivated by emerging applications in ecology, microbiology, and neuroscience, this paper studies high-dimensional regression with two-way structured data. To estimate the high-dimensional coefficient vector, we propose the generalized matrix decomposition regression (GMDR) to efficiently leverage auxiliary information on row and column structures. GMDR extends the principal component regression (PCR) to two-way structured data, but unlike PCR, GMDR selects the components that are most predictive of the outcome, leading to more accurate prediction. For inference on regression coefficients of individual variables, we propose the generalized matrix decomposition inference (GMDI), a general high-dimensional inferential framework for a large family of estimators that include the proposed GMDR estimator. GMDI provides more flexibility for incorporating relevant auxiliary row and column structures. As a result, GMDI does not require the true regression coefficients to be sparse, but constrains the coordinate system representing the regression coefficients according to the column structure. GMDI also allows dependent and heteroscedastic observations. We study the theoretical properties of GMDI in terms of both the type-I error rate and power and demonstrate the effectiveness of GMDR and GMDI in simulation studies and an application to human microbiome data.

Intestinal microbiota controls graft-versus-host disease independent of donor-host genetic disparity.

Acute graft-versus-host disease (aGVHD) remains a major limitation of allogeneic stem cell transplantation (SCT), and severe intestinal manifestation is the major cause of early mortality. Intestinal microbiota control MHC class II (MHC-II) expression by ileal intestinal epithelial cells (IECs) that promote GVHD. Here, we demonstrated that genetically identical mice of differing vendor origins had markedly different intestinal microbiota and ileal MHC-II expression, resulting in discordant GVHD severity. We utilized cohousing and antibiotic treatment to characterize the bacterial taxa positively and negatively associated with MHC-II expression. A large proportion of bacterial MHC-II inducers were vancomycin sensitive, and peri-transplant oral vancomycin administration attenuated CD4+ T cell-mediated GVHD. We identified a similar relationship between pre-transplant microbes, HLA class II expression, and both GVHD and mortality in a large clinical SCT cohort. These data highlight therapeutically tractable mechanisms by which pre-transplant microbial taxa contribute to GVHD independently of genetic disparity.

Multi-Omic Biomarkers Improve Indeterminate Pulmonary Nodule Malignancy Risk Assessment.

The clinical management of patients with indeterminate pulmonary nodules is associated with unintended harm to patients and better methods are required to more precisely quantify lung cancer risk in this group. Here, we combine multiple noninvasive approaches to more accurately identify lung cancer in indeterminate pulmonary nodules. We analyzed 94 quantitative radiomic imaging features and 41 qualitative semantic imaging variables with molecular biomarkers from blood derived from an antibody-based microarray platform that determines protein, cancer-specific glycan, and autoantibody-antigen complex content with high sensitivity. From these datasets, we created a PSR (plasma, semantic, radiomic) risk prediction model comprising nine blood-based and imaging biomarkers with an area under the receiver operating curve (AUROC) of 0.964 that when tested in a second, independent cohort yielded an AUROC of 0.846. Incorporating known clinical risk factors (age, gender, and smoking pack years) for lung cancer into the PSR model improved the AUROC to 0.897 in the second cohort and was more accurate than a well-characterized clinical risk prediction model (AUROC = 0.802). Our findings support the use of a multi-omics approach to guide the clinical management of indeterminate pulmonary nodules.

Posttranslational modifications induce autoantibodies with risk prediction capability in patients with small cell lung cancer.

Small cell lung cancer (SCLC) elicits the generation of autoantibodies that result in unique paraneoplastic neurological syndromes. The mechanistic basis for the formation of such autoantibodies is largely unknown but is key to understanding their etiology. We developed a high-dimensional technique that enables detection of autoantibodies in complex with native antigens directly from patient plasma. Here, we used our platform to screen 1009 human plasma samples for 3600 autoantibody-antigen complexes, finding that plasma from patients with SCLC harbors, on average, fourfold higher disease-specific autoantibody signals compared with plasma from patients with other cancers. Across three independent SCLC cohorts, we identified a set of common but previously unknown autoantibodies that are produced in response to both intracellular and extracellular tumor antigens. We further characterized several disease-specific posttranslational modifications within extracellular proteins targeted by these autoantibodies, including citrullination, isoaspartylation, and cancer-specific glycosylation. Because most patients with SCLC have metastatic disease at diagnosis, we queried whether these autoantibodies could be used for SCLC early detection. We created a risk prediction model using five autoantibodies with an average area under the curve of 0.84 for the three cohorts that improved to 0.96 by incorporating cigarette smoke consumption in pack years. Together, our findings provide an innovative approach to identify circulating autoantibodies in SCLC with mechanistic insight into disease-specific immunogenicity and clinical utility.

Prediction of Busulfan Clearance by Predose Plasma Metabolomic Profiling.

Intravenous busulfan doses are often personalized to a target plasma exposure (targeted busulfan) using an individual's busulfan clearance (BuCL). We evaluated whether BuCL could be predicted by a predose plasma panel of 841 endogenous metabolomic compounds (EMCs). In this prospective cohort of 132 hematopoietic cell transplantation (HCT) patients, all had samples collected immediately before busulfan administration (preBU) and 96 had samples collected 2 weeks before busulfan (2-week-preBU). BuCL was significantly associated with 37 EMCs after univariate linear regression analysis and controlling for false discovery (< 0.05) in the 132 preBU samples. In parallel, with preBU samples, we included all 841 EMCs in a least absolute shrinkage and selection operator-penalized regression which selected 13 EMCs as predominantly associated with BuCL. Then, we constructed a prediction model by estimating coefficients for these 13 EMCs, along with sex, using ordinary least-squares. When the resulting linear prediction model was applied to the 2-week-preBU samples, it explained 40% of the variation in BuCL (adjusted R2  = 0.40). Pathway enrichment analysis revealed 18 pathways associated with BuCL. Lysine degradation followed by steroid biosynthesis, which aligned with the univariate analysis, were the top two pathways. BuCL can be predicted before busulfan administration with a linear regression model of 13 EMCs. This pharmacometabolomics method should be prioritized over use of a busulfan test dose or pharmacogenomics to guide busulfan dosing. These results highlight the potential of pharmacometabolomics as a precision medicine tool to improve or replace pharmacokinetics to personalize busulfan doses.

Association of Habitual Intake of Probiotic Supplements and Yogurt with Characteristics of the Gut Microbiome in the Multiethnic Cohort Adiposity Phenotype Study.

Consumption of probiotics and/or yogurt could be a solution for restoring the balance of the gut microbiota. This study examined associations of regular intake of probiotic supplements or yogurt with the gut microbiota among a diverse population of older adults (N=1,861; 60-72 years). Fecal microbial composition was obtained from 16S rRNA gene sequencing (V1-V3 region). General Linear Models were used to estimate the associations of probiotic supplement or yogurt intake with microbiome measures adjusting for covariates. Compared to non-yogurt consumers (N=1,023), regular yogurt consumers (≥once/week, N=818) had greater Streptococcus (ß=0.29, P=0.0003) and lower Odoribacter (ß=-0.33, P<0.0001) abundance. The directions of the above associations were consistent across the five ethnic groups but stronger among Japanese Americans (Streptococcus: ß=0.56, P=0.0009; Odoribacter: ß=-0.62, P=0.0005). Regular intake of probiotic supplements (N=175) was not associated with microbial characteristics (i.e., alpha diversity and the abundance of 152 bacteria genera). Streptococcus is one of the predominant bacteria genera in yogurt products, which may explain the positive association between yogurt consumption and Streptococcus abundance. Our analyses suggest that changes in Odoribacter were independent of changes in Streptococcus abundance. Future studies may investigate whether these microbial genera and their sub-level species mediate potential pathways between yogurt consumption and health.

Annexin A2/TLR2/MYD88 pathway induces arginase 1 expression in tumor-associated neutrophils.

Myeloid lineage cells suppress T cell viability through arginine depletion via arginase 1 (ARG1). Despite numerous studies exploring the mechanisms by which ARG1 perturbs lymphocyte function, the cellular populations responsible for its generation and release remain poorly understood. Here, we showed that neutrophil lineage cells and not monocytes or macrophages expressed ARG1 in human non-small cell lung cancer (NSCLC). Importantly, we showed that approximately 40% of tumor-associated neutrophils (TANs) actively transcribed ARG1 mRNA. To determine the mechanism by which ARG1 mRNA is induced in TANs, we utilized FPLC followed by MS/MS to screen tumor-derived factors capable of inducing ARG1 mRNA expression in neutrophils. These studies identified ANXA2 as the major driver of ARG1 mRNA expression in TANs. Mechanistically, ANXA2 signaled through the TLR2/MYD88 axis in neutrophils to induce ARG1 mRNA expression. The current study describes what we believe to be a novel mechanism by which ARG1 mRNA expression is regulated in neutrophils in cancer and highlights the central role that neutrophil lineage cells play in the suppression of tumor-infiltrating lymphocytes.

Incorporation of spatial- and connectivity-based cortical brain region information in regularized regression: Application to Human Connectome Project data.

Studying the association of the brain's structure and function with neurocognitive outcomes requires a comprehensive analysis that combines different sources of information from a number of brain-imaging modalities. Recently developed regularization methods provide a novel approach using information about brain structure to improve the estimation of coefficients in the linear regression models. Our proposed method, which is a special case of the Tikhonov regularization, incorporates structural connectivity derived with Diffusion Weighted Imaging and cortical distance information in the penalty term. Corresponding to previously developed methods that inform the estimation of the regression coefficients, we incorporate additional information via a Laplacian matrix based on the proximity measure on the cortical surface. Our contribution consists of constructing a principled formulation of the penalty term and testing the performance of the proposed approach via extensive simulation studies and a brain-imaging application. The penalty term is constructed as a weighted combination of structural connectivity and proximity between cortical areas. Simulation studies mimic the real brain-imaging settings. We apply our approach to the study of data collected in the Human Connectome Project, where the cortical properties of the left hemisphere are found to be associated with vocabulary comprehension.

Batch effects removal for microbiome data via conditional quantile regression.

Batch effects in microbiome data arise from differential processing of specimens and can lead to spurious findings and obscure true signals. Strategies designed for genomic data to mitigate batch effects usually fail to address the zero-inflated and over-dispersed microbiome data. Most strategies tailored for microbiome data are restricted to association testing or specialized study designs, failing to allow other analytic goals or general designs. Here, we develop the Conditional Quantile Regression (ConQuR) approach to remove microbiome batch effects using a two-part quantile regression model. ConQuR is a comprehensive method that accommodates the complex distributions of microbial read counts by non-parametric modeling, and it generates batch-removed zero-inflated read counts that can be used in and benefit usual subsequent analyses. We apply ConQuR to simulated and real microbiome datasets and demonstrate its advantages in removing batch effects while preserving the signals of interest.

Lipidomics of cyclophosphamide 4-hydroxylation in patients receiving post-transplant cyclophosphamide.

Biomarker-guided dosing may improve the efficacy and toxicity of cyclophosphamide (CY); however, clinical studies evaluating their association with the area under the plasma concentration-time curve (AUC) of CY and its metabolites are time- and resource-intensive. Therefore, we sought to identify lipidomic biomarkers associated with the time-varying differences in CY formation clearance to 4-hydroxycyclophosphamide (4HCY), the principal precursor to CY's cytotoxic metabolite. Hematopoietic cell transplant (HCT) patients receiving post-transplant CY (PT-CY) were enrolled, cohort 1 (n = 25) and cohort 2 (n = 26) donating longitudinal blood samples before they started HCT (pre-HCT), before infusion of the donor allograft (pre-graft), before the first dose of PT-CY (pre-CY) and 24 h after the first dose of PT-CY (24-h post-CY) which is also immediately before the second dose of CY. A total of 409 and 387 lipids were quantitated in the two cohorts, respectively. Associations between lipids, individually and at a class level, and the ratio of 4HCY/CY AUC (i.e., 4HCY formation clearance) were evaluated using linear regression with a false discovery rate <0.05. There were no individual lipids that passed control for false discovery at any time point. These results demonstrate the feasibility of lipidomics, but future studies in larger samples with multiple omic tools are warranted to optimize CY dosing in HCT.

Personalized Nutrition Using Microbial Metabolite Phenotype to Stratify Participants and Non-Invasive Host Exfoliomics Reveal the Effects of Flaxseed Lignan Supplementation in a Placebo-Controlled Crossover Trial.

High-fiber plant foods contain lignans that are converted to bioactive enterolignans, enterolactone (ENL) and enterodiol (END) by gut bacteria. Previously, we conducted an intervention study to gain mechanistic insight into the potential chemoprotective effects of flaxseed lignan supplementation (secoisolariciresinol diglucoside; SDG) compared to a placebo in 42 men and women. Here, we expand on these analyses to further probe the impact of the microbial metabolite phenotype on host gene expression in response to lignan exposure. We defined metabolic phenotypes as high- or low-ENL excretion based on the microbial metabolism of SDG. RNA-seq was used to assess host gene expression in fecal exfoliated cells. Stratified by microbial ENL excretion, differentially expressed (DE) genes in high- and low-ENL excreter groups were compared. Linear discriminant analysis using the ENL phenotypes identified putative biomarker combinations of genes capable of discriminating the lignan treatment from the placebo. Following lignan intervention, a total of 165 DE genes in high-ENL excreters and 1450 DE genes in low-ENL excreters were detected. Functional analysis identified four common upstream regulators (master genes): CD3, IFNG, IGF1 and TNFRSF1A. Our findings suggest that the enhanced conversion of flaxseed lignan to ENL is associated with a suppressed inflammatory status.

Dietary Intake Mediates Ethnic Differences in Gut Microbial Composition.

Background: The human gut microbiome (GM) has been observed to vary by race/ethnicity. Objective: Assess whether racial/ethnic GM variation is mediated by differences in diet. Design: Stool samples collected from 2013 to 2016 from 5267 healthy Multiethnic Cohort participants (age 59−98) were analyzed using 16S rRNA gene sequencing to estimate the relative abundance of 152 bacterial genera. For 63 prevalent genera (>50% in each ethnic group), we analyzed the mediation of GM differences among African Americans, Japanese Americans, Latinos, Native Hawaiians, and Whites by overall diet quality (Healthy Eating Index score (HEI-2015)) and intake amounts of 14 component foods/nutrients assessed from 2003 to 2008. For each significant mediation (p < 1.3 × 10−5), we determined the percent of the total ethnicity effect on genus abundance mediated by the dietary factor. Results: Ethnic differences in the abundance of 12 genera were significantly mediated by one or more of eight dietary factors, most frequently by overall diet quality and intakes of vegetables and red meat. Lower vegetable intake mediated differences in Lachnospira (36% in African Americans, 39% in Latinos) and Ruminococcus-1 (−35% in African Americans, −43% in Latinos) compared to Native Hawaiians who consumed the highest amount. Higher red meat intake mediated differences in Lachnospira (−41%) and Ruminococcus-1 (36%) in Native Hawaiians over African Americans, who consumed the least. Dairy and alcohol intakes appeared to mediate and counterbalance the difference in Bifidobacterium between Whites and Japanese Americans. Conclusions: Overall diet quality and component food intakes may contribute to ethnic differences in GM composition and to GM-related racial/ethnic health disparities.

Pharmacometabonomic association of cyclophosphamide 4-hydroxylation in hematopoietic cell transplant recipients.

The widely used alkylating agent cyclophosphamide (CY) has substantive interpatient variability in the area under the curve (AUC) of it and its metabolites. Numerous factors may influence the drug-metabolizing enzymes that metabolize CY to 4-hydroxycyclophosphamide (4HCY), the principal precursor to CY's cytotoxic metabolite. We sought to identify endogenous metabolomics compounds (EMCs) associated with 4HCY formation clearance (ratio of 4HCY/CY AUC) using global metabolomics. Patients who undergo hematopoietic cell transplantation receiving post-transplant CY (PT-CY) were enrolled, cohort 1 (n = 26) and cohort 2 (n = 25) donating longitudinal blood samples before they started HCT (pre-HCT), before infusion of the donor allograft (pre-graft), before the first dose of PT-CY (pre-CY), and 24 h after the first dose of PT-CY (24-h post-CY), which is also immediately before the second dose of CY. A total of 512 and 498 EMCs were quantitated in two cohorts, respectively. Both univariate linear regression with false discovery rate (FDR), and pathway enrichment analyses using a global association test were performed. At the pre-CY time point, no EMCs were associated at FDR less than 0.1. At pre-HCT, cohort 1 had one EMC (levoglucosan) survive the FDR threshold. At pre-graft, cohort 1 and cohort 2 had 20 and 13 EMCs, respectively, exhibiting unadjusted p values less than 0.05, with the only EMCs having an FDR less than 0.1 being two unknown EMCs. At 24-h post-CY, there were three EMCs, two ketones, and threitol, at FDR less than 0.1 in cohort 2. These results demonstrate the potential of pharmacometabonomics, but future studies in larger samples are needed to optimize CY.

The Gut Microbiome Is Associated with Circulating Dietary Biomarkers of Fruit and Vegetable Intake in a Multiethnic Cohort.

BACKGROUND: Results from observational studies suggest high diet quality favorably influences the human gut microbiome. Fruit and vegetable consumption is often a key contributor to high diet quality. OBJECTIVE: To evaluate measures of gut bacterial diversity and abundance in relation to serum biomarkers of fruit and vegetable intake. DESIGN: Secondary analysis of cross-sectional data. PARTICIPANTS AND SETTING: Men and women from Los Angeles, CA, and Hawai'i who participated in the Multiethnic Cohort-Adiposity Phenotype Study from 2013 to 2016 (N = 1,709). MAIN OUTCOME MEASURES: Gut microbiome diversity and composition in relation to dietary biomarkers. STATISTICAL ANALYSIS: Carotenoid (beta carotene, alpha carotene, cryptoxanthins, lutein, lycopene, and zeaxanthin), tocopherol (α, ß + γ, and δ), and retinol concentrations were assessed in serum. The α and ß diversity and composition of the gut microbiome were classified based on 16S rRNA gene sequencing of bacterial DNA from self-collected fecal samples. Global differences in microbial community profiles in relation dietary biomarkers were evaluated using multivariable permutational analysis of variance. Associations of α diversity (Shannon index), ß diversity (weighted and unweighted UniFrac) with center log-ratio-transformed phyla and genera abundances were evaluated using linear regression, adjusted for covariates. RESULTS: Increasing total carotenoid, beta carotene, alpha carotene, cryptoxanthin, and lycopene concentrations were associated with higher gut bacterial diversity (Shannon Index) (P < 0.001). Total tocopherol, α-tocopherol, and δ-tocopherol concentrations contributed significantly to more than 1% of the microbiome variation in gut bacterial community: total tocopherol: 1.74%; α-tocopherol: 1.70%; and δ-tocopherol: 1.16% (P < 0.001). Higher total carotenoid was associated with greater abundance of some genera relevant for microbial macronutrient metabolism (P < 0.001). CONCLUSIONS: Objective biomarkers of fruit and vegetable intake, particularly carotenoids, were favorably associated with gut bacterial composition and diversity in this multiethnic population. These observations provide supportive evidence that fruit and vegetable intake is related to gut bacterial composition; more work is needed to elucidate how this influences host health.

Associations of the Dietary Inflammatory Index with total adiposity and ectopic fat through the gut microbiota, LPS, and C-reactive protein in the Multiethnic Cohort-Adiposity Phenotype Study.

BACKGROUND: Mechanisms linking a proinflammatory diet to obesity remain under investigation. The ability of diet to influence the gut microbiome (GM) in creating chronic low-grade systemic inflammation provides a plausible connection to adiposity. OBJECTIVES: Assess whether any associations seen between the Energy-Adjusted Dietary Inflammatory Index (E-DII score), total fat mass, visceral adipose tissue (VAT), or liver fat (percentage volume) operated through the GM or microbial related inflammatory factors, in a multiethnic cross-sectional study. METHODS: In the Multiethnic Cohort-Adiposity Phenotype Study (812 men, 843 women, aged 60-77 y) we tested whether associations between the E-DII and total adiposity, VAT, and liver fat function through the GM, LPS, and high-sensitivity C-reactive protein (hs-CRP). DXA-derived total fat mass, MRI-measured VAT, and MRI-based liver fat were measured. Participants provided stool and fasting blood samples and completed an FFQ. Stool bacterial DNA was amplified and the 16S rRNA gene was sequenced at the V1-V3 region. E-DII score was computed from FFQ data, with a higher E-DII representing a more proinflammatory diet. The associations between E-DII score, GM (10 phyla, 28 genera, α diversity), and adiposity phenotypes were examined using linear regression and mediation analyses, adjusting for confounders. RESULTS: There were positive total effects (c) between E-DII and total fat mass (c = 0.68; 95% CI: 0.47, 0.90), VAT (c = 4.61; 95% CI: 2.95, 6.27), and liver fat (c = 0.40; 95% CI: 0.27, 0.53). The association between E-DII score and total fat mass was mediated by LPS, Flavonifractor, [Ruminococcus] gnavus group, and Tyzzerella. The association between E-DII score and ectopic fat occurred indirectly through Fusobacteria, Christensenellaceae R-7 group, Coprococcus 2, Escherichia-Shigella, [Eubacterium] xylanophilum group, Flavonifractor, Lachnoclostridium, [Ruminococcus] gnavus group, Tyzzerella, [Ruminococcus] gnavus group (VAT only), and α diversity (liver fat only). There was no significant association between E-DII score and adiposity phenotype through hs-CRP. CONCLUSIONS: Associations found between E-DII and adiposity phenotypes occurred through the GM and LPS.

Associations of the gut microbiome with hepatic adiposity in the Multiethnic Cohort Adiposity Phenotype Study.

Nonalcoholic fatty liver disease (NAFLD) is a risk factor for liver cancer and prevalence varies by ethnicity. Along with genetic and lifestyle factors, the gut microbiome (GM) may contribute to NAFLD and its progression to advanced liver disease. Our cross-sectional analysis assessed the association of the GM with hepatic adiposity among African American, Japanese American, White, Latino, and Native Hawaiian participants in the Multiethnic Cohort. We used MRI to measure liver fat and determine nonalcoholic fatty liver disease (NAFLD) status (n = 511 cases) in 1,544 participants, aged 60-77 years, with 12-53% overall adiposity (BMI of 17.8-46.2 kg/m2). The GM was measured by 16S rRNA gene sequencing and, on a subset, by metagenomic sequencing. Alpha diversity was lower overall with NAFLD and in certain ethnicities (African Americans, Whites, and Latinos). In models regressing genus on NAFLD status, 62 of 149 genera (40%) exhibited a significant interaction between NAFLD and ethnicity stratified analysis found 69 genera significantly associated with NAFLD in at least one ethnic group. No single genus was significantly associated with NAFLD across all ethnicities. In contrast, the same bacterial metabolic pathways were over-represented in participants with NAFLD regardless of ethnicity. Imputed secondary bile acid and carbohydrate pathways were associated with NAFLD, the latter of which was corroborated by metagenomics, although different genera in different ethnicities were associated with these pathways. Overall, we found that NAFLD was associated with altered bacterial composition and metabolism, and that bacterial endotoxin, assessed by plasma lipopolysaccharide binding protein (LBP), may mediate liver fat-associated systemic inflammation in a manner that seems to vary by ethnicity.

Long-term association between diet quality and characteristics of the gut microbiome in the multiethnic cohort study.

As past usual diet quality may affect gut microbiome (GM) composition, we examined the association of the Healthy Eating Index (HEI)-2015 assessed 21 and 9 years before stool collection with measures of fecal microbial composition in a subset of the Multiethnic Cohort. A total of 5936 participants completed a validated quantitative FFQ (QFFQ) at cohort entry (Q1, 1993-1996), 5280 at follow-up (Q3, 2003-2008) and 1685 also at a second follow-up (Adiposity Phenotype Study (APS), 2013-2016). All participants provided a stool sample in 2013-2016. Fecal microbial composition was obtained from 16S rRNA gene sequencing (V1-V3 regions). HEI-2015 scores were computed based on each QFFQ. Using linear regression adjusted for relevant covariates, we calculated associations of HEI-2015 scores with gut microbial diversity and 152 individual genera. The mean HEI-2015 scores increased from Q1 (67 (sd 10)) to Q3 (71 (sd 11)) and APS (72 (sd 10)). Alpha diversity assessed by the Shannon Index was significantly higher with increasing tertiles of HEI-2015. Of the 152 bacterial genera tested, seven (Anaerostipes, Coprococcus_2, Eubacterium eligens, Lachnospira, Lachnospiraceae_ND3007, Ruminococcaceae_UCG-013 and Ruminococcus_1) were positively and five (Collinsella, Parabacteroides, Ruminiclostridium_5, Ruminococcus gnavus and Tyzzerella) were inversely associated with HEI-2015 assessed in Q1, Q3 and APS. The estimates of change per unit of the HEI-2015 score associated with the abundance of these twelve genera were consistent across the three questionnaires. The quality of past diet, assessed as far as ∼20 years before stool collection, is equally predictive of GM composition as concurrently assessed diet, indicative of the long-term consistency of this relation.

The gut microbiome and type 2 diabetes status in the Multiethnic Cohort.

BACKGROUND: The gut microbiome may play a role in inflammation associated with type 2 diabetes (T2D) development. This cross-sectional study examined its relation with glycemic status within a subset of the Multiethnic Cohort (MEC) and estimated the association of circulating bacterial endotoxin (measured as plasma lipopolysaccharide-binding protein (LBP)) with T2D, which may be mediated by C-reactive protein (CRP). METHODS: In 2013-16, cohort members from five ethnic groups completed clinic visits, questionnaires, and stool and blood collections. Participants with self-reported T2D and/or taking medication were considered T2D cases. Those with fasting glucose >125 and 100-125 mg/dL were classified as undiagnosed (UT2D) and pre-diabetes (PT2D) cases, respectively. We characterized the gut microbiome through 16S rRNA gene sequencing and measured plasma LBP and CRP by standard assays. Linear regression was applied to estimate associations of the gut microbiome community structure and LBP with T2D status adjusting for relevant confounders. RESULTS: Among 1,702 participants (59.9-77.4 years), 735 (43%) were normoglycemic (NG), 506 (30%) PT2D, 154 (9%) UT2D, and 307 (18%) T2D. The Shannon diversity index decreased (ptrend = 0.05), while endotoxin, measured as LBP, increased (ptrend = 0.0003) from NG to T2D. Of 10 phyla, Actinobacteria (ptrend = 0.007), Firmicutes (ptrend = 0.003), and Synergistetes (ptrend = 0.02) were inversely associated and Lentisphaerae (ptrend = 0.01) was positively associated with T2D status. Clostridium sensu stricto 1, Lachnospira, and Peptostreptococcaceae were less, while Escherichia-Shigella and Lachnospiraceae were more abundant among T2D patients, but the associations with Actinobacteria, Clostridium sensu stricto 1, and Escherichia-Shigella may be due metformin use. PT2D/UT2D values were closer to NG than T2D. No indication was detected that CRP mediated the association of LBP with T2D. CONCLUSIONS: T2D but not PT2D/UT2D status was associated with lower abundance of SCFA-producing genera and a higher abundance of gram-negative endotoxin-producing bacteria suggesting that the gut microbiome may contribute to chronic systemic inflammation and T2D through bacterial translocation.

Connectivity-informed adaptive regularization for generalized outcomes.

One of the challenging problems in neuroimaging is the principled incorporation of information from different imaging modalities. Data from each modality are frequently analyzed separately using, for instance, dimensionality reduction techniques, which result in a loss of mutual information. We propose a novel regularization method, generalized ridgified Partially Empirical Eigenvectors for Regression (griPEER), to estimate associations between the brain structure features and a scalar outcome within the generalized linear regression framework. griPEER improves the regression coefficient estimation by providing a principled approach to use external information from the structural brain connectivity. Specifically, we incorporate a penalty term, derived from the structural connectivity Laplacian matrix, in the penalized generalized linear regression. In this work, we address both theoretical and computational issues and demonstrate the robustness of our method despite incomplete information about the structural brain connectivity. In addition, we also provide a significance testing procedure for performing inference on the estimated coefficients. Finally, griPEER is evaluated both in extensive simulation studies and using clinical data to classify HIV+ and HIV- individuals.

L'un des défis en imagerie cérébrale consiste à établir les principes pour incorporer de l'information provenant de différentes modalités d'imagerie. Les données de chaque modalité sont fréquemment analysées séparément, exploitant par exemple des techniques de réduction de la dimension, ce qui conduit à une perte d'information mutuelle. Les auteurs proposent une nouvelle méthode de régularisation, griPEER (ou par vecteurs propres ridgifiés partiellement empiriques généralisés pour la régression) afin d'estimer l'association entre des caratéristiques de structures du cerveau et une variable réponse scalaire dans le cadre d'une régression linéaire généralisée. Les griPEER améliorent l'estimation des coefficients de régression en établissant les principes d'une approche permettant d'utiliser des informations externes de connectivité des structures du cerveau. À cet effet, les auteurs ajoutent au modèle de régression pénalisée généralisé un terme de pénalité dérivé de la matrice laplacienne de connectivité structurelle. Les auteurs résolvent des problèmes théoriques et calculatoires, puis démontrent la robustesse de leur méthode lorsque l'information à propos de la connectivité du cerveau est incomplète. De plus, ils présentent une procédure de test d'hypothèse permettant de l'inférence au sujet des paramètres estimés. Finalement, les auteurs évaluent les griPEER dans de vastes études de simulation et en utilisant des données cliniques afin de classifier les individus en VIH+ et VIH−.