Next-Generation Nutritional Biomarkers to Guide Better Health Care

E. E. Baetge, A. Dhawan, A. M. Prentice

• Publisher: S. Karger
• Year: 2016
• ISBN:
  9783318055993
• Topic: Medicine
• Subtopic: Endocrinology & Metabolism

There are only a few areas in human nutrition and metabolism where biomarkers are routinely used to predict health and functional outcome. For instance, of the four major nutritional deficiencies, only iron deficiency can be precisely diagnosed by employing biomarkers. They therefore play a limited role in research and decision making, and intervention strategies are still mostly targeted at the population level. What is needed at this stage are biomarkers that are predictive of later functional health and that stay stable from infancy to childhood and adult health. Moreover, individual variability must be considered, taking into account the complexity of foods, lifestyle, and metabolic processes that contribute to health or disease. These factors present significant challenges when it comes to personalizing dietary advice for healthy or diseased individuals. This book focuses on the values and limitations of traditional nutritional biomarkers and on opportunities for new biomarkers. Contributions are divided into three parts: Methodologies with regard to global epidemiology; applications/end users, and future horizons. The main goal is to review recent developments and predict how exciting new technologies could be used to drive advances in nutrition-related health care.

Methodologies: Global Epidemiology

Baetge EE, Dhawan A, Prentice AM (eds): Next-Generation Nutritional Biomarkers to Guide Better
Health Care. Nestlé Nutr Inst Workshop Ser, vol 84, pp 1-13,(DOI: 10.1159/000436947)
Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

______________________

Systems-Level Nutrition Approaches to Define Phenotypes Resulting from Complex Gene-Environment Interactions

Jim Kaput

Systems Nutrition and Health, Nestlé Institute of Health Sciences, Lausanne, Switzerland

______________________

Abstract

High-throughput metabolomic, proteomic, and genomic technologies have delivered 21st-century data showing that humans cannot be randomized into groups: individuals are genetically and biochemically distinct. Gene-environment interactions caused by unique dietary and lifestyle factors contribute to the heterogeneity in physiologies observed in human studies. The risk factors determined for populations (i.e. the population-attributable risk) cannot be applied to the individual. Developing individual risk/benefit factors in light of the genetic diversity of human populations, the complexity of foods, culture and lifestyle, and the variety in metabolic processes that lead to health or disease are significant challenges for personalizing dietary advice for healthy or diseased individuals.

© 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Introduction

Two of the great advances in biomedical research over the past 100 years were the standardization of experimental designs, specifically the case-control design [1], and the one gene-one polypeptide concept that emanated from the groundbreaking work of Beadle and Tatum [2]. These contributions, in addition to a few other seminal discoveries (e.g. elucidation of the DNA structure), laid the conceptual framework for 20th-century biological research. Much of the research of the last 80 years is what Kuhn [3] called normal science. This type of research activity does not produce new concepts, but rather finds facts to match and better explain the existing framework. Normal science standardizes experimental designs that ultimately are considered mandatory for interpreting data and publishing results. Scientific dogma holds, for example, that the only trustworthy results of human studies are from prospective randomized controlled trials (RCTs) [4, 5]. Similarly, many studies in humans or laboratory animals are reductionistic in analyzing how, for example, a single gene or a single nutrient correlates with some physiological effect [2].

Revolutions can occur when normal science produces experimental data that cannot be explained by existing paradigms [3]. The outcomes of this paradigm shift have profound implications for biomarker identification and development that are necessary to assess nutrition and lifestyle choices for maintaining or improving personal and public health.

Analyzed Heterogeneity Supplants Randomization

In addition to the elaborations of human experimental designs, legal developments occurred over a long period that began with the passage of the US Food, Drug, and Cosmetic Act in 1938. The Cosmetic Act required that new drugs undergo premarket safety evaluation, although the legal mandate to prove efficacy was not enacted until 1962. The Drug Amendment Act specifically required adequate and well-controlled clinical investigations with positive findings from at least two clinical studies [6]. The scientific precedents for these legal actions were built on the success of analyzing the efficacy of antibiotics [6]. Infectious bacteria (e.g. Vibrio cholera) are extrinsic agents that affect virtually all humans. Hence, the average response between the treated group and the control group in RCTs can provide proof of the efficacy of the treatment. In contrast, drugs, nutrition, and lifestyle choices become intrinsic factors in that they interact with and affect internal physiological functions.

Unlike extrinsic agents, intrinsic factors may be metabolized by or interact with physiological systems. Humans (and all species) not only show biochemical individuality [7] at homeostasis, but may also respond differently to drug or food chemicals [8-10]. Differential responses occur because each individual is unique [11] and genetic variations may be differently expressed in response to nutrients and other factors. Heterogeneity within a species is, of course, the fundamental basis by which natural selection acts to produce evolutionary changes. The statistical result of RCTs is the population-attributable risk (PAR), defined as the number (or proportion) of cases that would not occur in a population if the factor were eliminated [12] - they are not individual risk factors [13]. While PARs are applicable for large effect sizes and extrinsic agents, the utility of PARs is diminished by the heterogeneity of many individual genetic makeups added to the calculations, the latent effects of many interacting environmental factors (dietary chemicals and activity levels), and the resulting individuality of metabolic responses [14-16]. The human system, which is comprised of a person's genetic makeup, microbiome, diet, lifestyle, and resulting physiology, acts on (metabolizes) treatments, and the treatments differentially alter the physiology depending upon the individual. The result of this heterogeneity can be (and often is) that the distribution of metabolic measures (or responses) in the case group can overlap with the distribution of metabolic measures (or responses) in the control group [17].

Since the pregenomic era had limited methodological tools to analyze human variation at the genetic or physiological level, randomization was essential for 20th-century biomedical research. However, genome sequencing has not only demonstrated genetic individuality [reviewed in ref. 11, 18, 19] but can now be used to completely characterize genetic makeups of study participants. Recent data [20, 21] suggest that each person differs from others and the reference genomes by about 3.5 million single nucleotide polymorphisms (SNPs), almost 1,000 large copy number variants (CNVs), and large numbers of insertions and deletions. Different levels of DNA methylation and therefore epigenetic regulation have also been demonstrated [22-24]. Variation in the (epi)genetic makeup may express itself in variation in RNA abundance [25, 26] and therefore levels of proteins and enzymes. To add to this complexity, physiological variability is influenced by the human microbiome [27, 28], the combination of all microorganisms that reside on the skin, in saliva, in the oral mucosa, in the conjunctiva, in the gastrointestinal tract, and in the vagina. Each of these factors alone or in combination could alter the level of a biomarker.

Nutrition and lifestyle not only alter the expression of information encoded in the genome [29], they can also modify the epigenome [30] and the microbial composition [31]. No experiments are needed to demonstrate that individuals have heterogeneous dietary intakes and physical activity levels. The result of interactions among (epi)genetic, microbiome, nutritional, and environmental factors is that humans have heterogeneous metabolomic profiles [32-34] in health and disease states. Physiological variability has been recognized for centuries and summarized in the modern pregenomic era by Williams [7] in 1956 in a book entitled Biochemical Individuality, and by nutritionists of the 1960s and 1970s [35, 36]. Many (but not all) of the previously unmeasurable, molecular characteristics of individuals can now be analyzed.

Medical practitioners and health professionals treat individuals and not population groups [35]. Individual risk factors are unknown, even though the concepts of personal determinants of health were taught by Hippocrates and Galen centuries ago. In addition to Williams’ [7] treatise in 1956, others also expressed the need and approach to analyze individuals rather than groups [35-40]. More recently, we [41-44] and others [45-47] have been promoting or using n-of-1 aggregation and analysis methods [45, 48]. The concept of n-of-1 studies is that each person serves as his or her own control. Physiological assessments are usually done before and after a treatment [39] or intervention. The results of each trial (that is, from one individual) can then be aggregated for statistical analysis [45]. For example, we aggregated results from data obtained at homeostasis to analyze group average differences between males and females [48], but also found that each individual varies in micronutrient levels. This variation could then be used to associate patterns of plasma protein levels and variations in the genetic makeup [41].

Most -omes will remain incompletely characterized because they are so complex in composition and because they vary in time and space. Nevertheless, the incomplete molecular data sets can be analyzed to generate a model that predicts new markers to test in remaining samples of the experiment, or as new markers in the follow-on experiment in an iterative process of refining the model. Regardless of the new methodological and analytical approaches, the quantitative postgenomic data demonstrating human metabolomic and physiological heterogeneity should have profound impacts on the design of human research studies and, specifically, the validity of RCTs for determining optimal nutrition or drug treatments. Hence, even though not all physiological variables can be analyzed, randomization b...