New pathways to obesity prevention and metabolic health : the relationship between diet and the gut microbiome : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Nutritional Science at Massey University, Tāmaki Makaurau, Aotearoa New Zealand

Abstract

Background Diet is one of the key drivers of the global obesity epidemic. Based on the results of rodent experiments, the gut microbiota may play an important role in this multifaceted disease. Additionally, the microbiota is known to be influenced by the habitual diets consumed by humans.

Aims and objectives The aim of this PhD research was to characterise the habitual dietary intake of two New Zealand populations (Pacific and New Zealand European (NZE) women) with different metabolic disease risk and body fat profiles (lean and obese). The first objective of the research was to explore the relationship between habitual macronutrient intake in relation to body fat content and metabolic health markers. The second objective was to characterise a posteriori dietary patterns (derived from multiple days of dietary assessment) and to explore the association with body fat content and metabolic health markers. The third objective was to explore the characteristics of microbiota composition in relation to habitual diet (dietary patterns, foods, and nutrients), body fat content and metabolic health markers.

Methods Between July 2016 and September 2017, Pacific (n=126) and NZE (n=161) women, aged 18-45 years, living in Auckland, New Zealand, were recruited to a cross-sectional study, based on their body mass index (lean and obese) and stratified as having low (<35 % body fat) or high (≥35 % body fat) body fat percentage (BF%). Dietary intake was assessed using a 5-day estimated, non-consecutive, food record and a validated semi-quantitative food frequency questionnaire, which were used to calculate habitual dietary intake using the National Cancer Institute (NCI) method. Body composition and BF% were assessed by dual-energy x-ray absorptiometry. Fasting blood samples were analysed for metabolic biomarkers (lipid and glucose profiles). Bulk DNA was extracted from faecal samples and the metagenomic sequences associated with the microbiota were analysed using MetaPhlAN and QIIME2 software. Enterotypes characterising the microbiotas of the participants were predicted in R and the species that defined enterotypes were determined using STAMP software. 𝘈 𝘱𝘰𝘴𝘵𝘦𝘳𝘪𝘰𝘳𝘪 dietary patterns were identified using principal component analysis. Adjusted multivariate regression models were conducted to explore the association between BF% and habitual macronutrient intake and adherence to dietary patterns, as well as the association between microbiota composition and habitual diet.

Results There were no significant differences in BF% between Pacific and NZE women (𝘱=0.498). Higher energy adjusted habitual dietary fibre (DF) intake was associated with lower BF% (β= -0.35, 𝘱≤ 0.001) for both Pacific and NZE women, and this relationship became stronger after further adjustments for protein (g/day), total carbohydrate (g/day), and total fat (g/day) intake (β= -0.47, p≤ 0.001). Women in the highest tertile of DF intake were older, had lower concentrations of fasting plasma insulin, and lower socioeconomic deprivation levels. Four dietary patterns that explained 30.9 % of the observed variance in habitual diet were identified. Higher adherence to dietary patterns characterised by core foods (the “colourful vegetable, plant protein, and dairy” and “fruit, starchy vegetables, and nuts” patterns) were inversely associated with BF%. In contrast, patterns characterised by more ‘discretionary’ foods (“sweet and fat rich carbohydrate”) and less diversity of core foods (“animal meat and fat”) were positively associated with BF% for both Pacific and NZE women. Three enterotypes were identified by higher relative abundance of specific bacterial species: enterotype 1 was characterised by Pacific and NZE women (n=146) and the abundances of 𝘍𝘢𝘦𝘤𝘢𝘭𝘪𝘣𝘢𝘤𝘵𝘦𝘳𝘪𝘶𝘮 𝘱𝘳𝘢𝘶𝘴𝘯𝘪𝘵𝘻𝘪𝘪 and 𝘌𝘶𝘣𝘢𝘤𝘵𝘦𝘳𝘪𝘶𝘮 𝘳𝘦𝘤𝘵𝘢𝘭𝘦. Enterotype 2 (n=70) was characterised by Pacific women, 𝘉𝘪𝘧𝘪𝘥𝘰𝘣𝘢𝘤𝘵𝘦𝘳𝘪𝘶𝘮 𝘢𝘥𝘰𝘭𝘦𝘴𝘤𝘦𝘯𝘵𝘪𝘴, 𝘉𝘪𝘧𝘪𝘥𝘰𝘣𝘢𝘤𝘵𝘦𝘳𝘪𝘶𝘮 𝘣𝘪𝘧𝘪𝘥𝘶𝘮, 𝘢𝘯𝘥 𝘓𝘢𝘤𝘵𝘰𝘣𝘢𝘤𝘪𝘭𝘭𝘶𝘴 𝘳𝘶𝘮𝘪𝘯𝘪𝘴; and by higher BF%, visceral adipose tissue, and concentrations of fasting insulin. Enterotype 3 (n=70) was predominately found in older NZE women with lower deprivation, and characterised by 𝘈𝘬𝘬𝘦𝘳𝘮𝘢𝘯𝘴𝘪𝘢 𝘮𝘶𝘤𝘪𝘯𝘪𝘱𝘩𝘪𝘭𝘢, 𝘙𝘶𝘮𝘪𝘯𝘰𝘤𝘰𝘤𝘤𝘶𝘴 𝘣𝘳𝘰𝘮𝘪𝘪, 𝘚𝘶𝘣𝘥𝘰𝘭𝘪𝘨𝘳𝘢𝘯𝘶𝘭𝘶𝘮 𝘴𝘱𝘦𝘤𝘪𝘦𝘴, 𝘢𝘯𝘥 𝘔𝘦𝘵𝘩𝘢𝘯𝘰𝘣𝘳𝘦𝘷𝘪𝘣𝘢𝘤𝘵𝘦𝘳 𝘴𝘮𝘪𝘵𝘩𝘪𝘪. Adherence to the “colourful vegetables, plant protein, and dairy” dietary pattern was positively associated with enterotypes 1 and 3 and negatively with enterotype 2.

Conclusion Consuming more core foods rich in dietary fibre was associated with enterotypes 1 and 3, including lower adiposity and metabolic disease risks. In contrast, consuming more discretionary foods was associated with enterotype 2, higher adiposity and metabolic disease risks. This PhD research highlights habitual diet-microbiota-host associations, which are similar for a population of women with different metabolic disease risk, body fat profiles, and deprivation levels. Whether the microbiota is a cause or consequence of metabolic health has yet to be elucidated. However, habitually consuming more core foods rich in dietary fibre is associated with microbiota composition, and lower metabolic disease risks.