The influx of intestinal bacteria-derived lipopolysaccharide (LPS) into the blood has attracted attention as a cause of diseases. The aim of this study is investigating the associations between the influx of LPS, dietary factors, gut microbiota, and health status in the general adult population. Food/nutrient intake, gut microbiota, health status and plasma LPS-binding protein (LBP; LPS exposure indicator) were measured in 896 residents (58.1% female, mean age 54.7 years) of the rural Iwaki district of Japan, and each correlation was analyzed. As the results, plasma LBP concentration correlated with physical (right/left arms’ muscle mass [β = −0.02, −0.03]), renal (plasma renin activity [β = 0.27], urine albumin creatinine ratio [β = 0.50]), adrenal cortical (cortisol [β = 0.14]), and thyroid function (free thyroxine [β = 0.05]), iron metabolism (serum iron [β = −0.14]), and markers of lifestyle-related diseases (all Qs < 0.20). Plasma LBP concentration were mainly negatively correlated with vegetables/their nutrients intake (all βs ≤ −0.004, Qs < 0.20). Plasma LBP concentration was positively correlated with the proportion of Prevotella (β = 0.32), Megamonas (β = 0.56), and Streptococcus (β = 0.65); and negatively correlated with Roseburia (β = −0.57) (all Qs < 0.20). Dietary factors correlated with plasma LBP concentration correlated with positively (all βs ≥ 0.07) or negatively (all βs ≤ −0.07) the proportion of these bacteria (all Qs < 0.20). Our results suggested that plasma LBP concentration in the Japanese general adult population was associated with various health issues, and that dietary habit was associated with plasma LBP concentration in relation to the intestinal bacteria.

Recently, web-based dietary assessment tools for the targeted population have been developed and used to estimate the dietary intake level in several epidemiological studies. This study aimed to examine the validity of estimating energy and nutrient intake by the web-based 24 h dietary recall (Web24HR), which we developed for the Japanese population. Overall, 228 adults aged ≥20 years who agreed to participate were included. Web24HR was administered three times per person: twice within 3 weeks and once 3 months later. The data on 3-day weighed food records (WFR) at 3-month intervals in the four seasons were collected using the reference method. The intake of energy and nutrients between Web24HR and WFR were compared using Pearson’s correlation coefficients and the Bland–Altman analysis. As results, the correlations were moderate for both men (median r = 0.51) and women (median r = 0.38) except for iodine, retinol, retinol equivalents, and β-tocopherol. The Bland–Altman method revealed that the bias in intake was within ±10% for most nutrients, except for cholesterol, iodine, vitamin C, and the water content, in both sexes. Additionally, monounsaturated fatty acids in men and β-cryptoxanthin in women exhibited an underestimation of more than 10%. In conclusion, the Web24HR intake assessment showed moderate correlations for most nutrients in both sexes. The bias in intake was within ±10% for most nutrients, but there were discrepancies for some nutrients. This tool’s performance is comparable to Japan’s standard dietary exposure assessment methods and will be helpful for future applications in epidemiological studies, though caution is needed for certain nutrient assessments.

Background: The average birth weight in Japan has decreased by 200 g in the last 40 years. Only three studies were reported for the association between maternal dietary patterns and birth weight in East Asia, whose results were inconsistent. We examined what maternal dietary patterns were associated with the birth weight in Japan. Methods: Totally 22,493 pregnant women were recruited between July 2013 and September 2016 into the Tohoku Medical Megabank Project Birth and Three-Generation Cohort Study. We included 17,287 women who had a full-term single healthy baby into the analysis. Consumption of food and beverage items was evaluated based on food frequency questionnaire at the first-trimester. Dietary patterns were analyzed using a machine learning method of k-means clustering algorithm. Birth weight was obtained from the medical record. The association between dietary patterns and birth weight was analyzed using multiple liner regression model adjusted for potential confounders with multiple imputation method for missing values. Results: Dietary patterns were classified into seven groups by cluster analysis: “high in rice (reference) (n = 8046)”, “middle in vegetables, beans, mushrooms, seaweeds and miso-soup (n = 4842)”, “high in fruits (n = 1176)”, “high in bread, dairy and alcohol (n = 1091)”, “high in meat and fish (n = 1049)”, “high in coffee, black tea, soft drinks and confections (high in coffee) (n = 585)”, and “high in vegetables, beans, mushrooms, seaweeds and miso-soup (n = 498)” groups. In multiple liner regression models, birth weight was 22.6 g (95%CI: 0.1 to 45.2 g) heavier in “high in fruits” group than in reference group and 39.4 g (95%CI: 8.6 to 70.3 g) lighter in “high in coffee” group than in reference group. The other groups were not statistically significant. Conclusions: High consumption of fruits and high consumption of coffee, black tea, soft drinks and confections during early pregnancy were associated with increased and decreased birth weight, respectively. Key messages: We found maternal dietary patterns during early pregnancy using a machine learning method of k-means clustering algorithm. We found maternal dietary patterns which associated with the birth weight in Japan.