2
Introduction
Human Nutrition Research in the Data Era: Results of 11 Reports on the Effects of a Multiple-
Micronutrient-Intervention Study by2Jim Kaput and Jacqueline Pontes Monteiro
Although molecular and genetic research has been using large datasets for decades, very few
studies focused on nutrition and lifestyle components. An n-of-1 intervention that included 12
vitamins and 5 minerals was performed with Brazilian children aged between nine and thirteen
years (n = 3). This investigation is unique in terms of its experimental design, which included a
replication arm. The Valhalla Project was composed of 26 types of data that included clinical
measures, whole-genome mapping and sequencing, along with proteomic measurements and
multiple other metabolomic assays over two years. This study aimed to leverage these different
data sets in order to identify new biological changes linked with an unhealthy diet that are
potentially traceable from a more comprehensive cohort of micronutrients. We describe key
findings from 11 reports resulting from this study, which evaluated population-level changes in
LDL and total cholesterol, as well as fasting glucose post-intervention accompanied with inter-
individual variability; polygenic risk score associated with baseline vitamin B12 levels;
metabotypes linking diet intake, genetic make-up, and metabolic physiology nutrients & food
groups biomarkers; metabolites and proteins linked to DNA damage. This also shed light on the
challenges and, to some extent, lessons we learned from analyzing diverse omic data.
The type of research design is quantitative methods with an experimental design and the
independent and dependent variables are:
Anthropometry, Body Mass Index (BMI), 24 h recall, Food Frequency Questionnaire, Physical
Activity, Social Economic Status, Clinical Biochemistry, Hematology, Tanner Classification,
3
Proteomics (plasma-Somalogic), Proteomics iTRAQ, Metabolomics (plasma-NMR), Lipidomics,
Fatty acids (Red Blood Cells), Amino acids (Red Blood Cells), Comet Assay (whole blood),
Vitamins (plasma), Methionine Pathway (RBC), Healthy Eating Index (HEI), Whole genome
genotyping, Exome (percent of genome), GIS Food Access & Quality, Food price comparison
In the Harmonized Micronutrient Intervention study, a hypothesis-generating translational
research project, we tested (at nearly 100% of daily intake levels) all twelve vitamins and five
minerals on multiple physiological systems. This study was designed to forecast dietary
responses of pregnant women in low-income settings, partly because actual micronutrient status
might differ from the reported literature or published databases.
Both years decreased from baseline levels of population values in albumen, basophil,
calcium_parse_1 copy 2 parse_5 full b value, and also glucose indeed LDL-cholesterol, either γ-
tocopherol by the intervention and increase level mean corpuscular volume FMN, nudifloramide,
pantothenic acid, pyridoxal, α-tocopherol, folate, and B12 at the year end. A subset of the
population had a significant correlation between improvements in α-tocopherol and several lipid
profile improvements, which again highlights that changes in specific metabolites can vary
depending on metabotype and such must be incorporated into nutritional counseling. The
supplement in isolation increased plasma levels of these vitamins, potentially validating the
nutrient-nutrient interactions involved, favoring an improved metabolic profile.
This was a n-of-1 experimental design with every participant as his or her own control to address
inter-individual variability, and in which an intervention was followed by the same one year later
as a replication arm. A second piece of good news, and arguably the best thing about this study,
is that with 98% compliance in terms of swallowing tablets (via grad students associated with the
project), the kids actually took their vitamins. The exception was the DNA-damage assay and
4
Somalogic proteomic data, for which we did not observe population-level reproducibility of most
clinical and omic measurements.
The research followed guidelines for community-based participatory research, with group
members interviewing principals and staff at 8 schools in Ribeirao Preto. Parent, student, and
community meetings were held in each school to describe the research program this preceding
year, which allows them to input on health concerns. Participants attended, accompanied by at
least one parent/guardian, for both assessments and provided information via dietary
questionnaires completed with their parents on all eating habits, sleep routine, & other behaviors.
A group of four health fairs was organized at the end of a 2-year project for each participant with
their family, consisting of a population-level report on different issues and individual reports
shown by a certified clinical nutritionist. However, the study was limited by a small sample size
and lack of some important metabolite status indicators, pooling samples for one proteomic study
and batch effect in the comet assay, as well as year 2 proteomics analysis.
Guidance on the use of various n-of-1 designs that consider blinding replication carryover effects
randomization/counterbalancing and adaptation has been published recently. Other publications
also consider multiple outcomes for collusion before enrolling subjects in a trial as well as
sample sizes required to achieve power. To display the populations vitamin status the data from
every participant at every time point was stacked. R statistical software version 3. 0 was used to
implement statistical analyses that call for inferential methods. The Statistical Package for the
Social Sciences (SPSS) is another option. Mann-Whitney and Kruskal-Wallis tests which are
nonparametric were used to compare groups and cohorts. Additionally used if required were the
Wilcoxon’s signed rank test and the Spearman’s rank correlation test.
5
RESULTS: To compare the distribution of categorical variables chi-square tests were used.
Analysis of variance (ANOVA) was used for longitudinal analyses involving repeated measures
and linear regression covariance analysis was used to compare groups after confounding
variables were corrected. Five percent was the unadjusted p-value threshold for nominal
associations. When appropriate p-values were adjusted for multiple testing by using the
Benjamini-Hochberg procedure to calculate the false discovery rate (FDR) of adjusted p-values.
The FDR-adjusted p-value threshold was set at 0. 1. When extreme values regress back to
population parameters we also conducted a second analysis to account for effect size above
regression towards the mean. This technique is evident in situations involving repeated
measurements. Before and after the intervention almost every metabolite measured differed
between individuals however there were insufficient amounts in each category of groups
(combining those that responded similarly) for us to understand how they were reacting. In order
to do so they used n-of-1 analysis and population-level data aggregated at each time point within
years and in both years. The scientists put these findings to the test in a replication arm to see if
they could be repeated. They used the glmnet 2. 0-7 R package to predict response to
intervention using training data from the first year which included repeat dosage information and
test data from the second year. Elastic net regression was used to fit the data.
Critique
All 12 vitamins and five minerals were assessed at 100% of the daily consumption across
physiological systems in the Harmonized Micronutrient Intervention a hypothesis-generating
translational trial. The dietary responses of low-income pregnant women were predicted by this
study despite the possibility that their micronutrient status differed from earlier research or
databases. FMN has increased over the last two years while albumen basophils glucose LDL
6
cholesterol and γ-tocopherol have all decreased. Together with a previously underutilized
contrast element a one-year course of nudifloramide pantothenic acid pyridoxal α-tocopherol
folate and B12 was provided. In certain people higher levels of α-tocopherol are occasionally
associated with better lipid profiles. Nutrition should be considered because metabolite changes
are influenced by metabolotype. By raising plasma vitamin levels, the supplement may have
increased metabolism and interactions between nutrients. To lessen inter-individual variability
this study used n-of-1 experiments with individual controls. The intervention was repeated a year
later. 98 percent of the vitamins were consumed by the children, which is the second and most
important positive development. The Somalogic proteome data and the DNA-damage assay
demonstrated poor population-level reproducibility for the majority of omic and clinical findings.
The researchers conducted staff and administrator interviews at eight Ribeirao Preto schools
using community-based participatory research techniques. Parents students and members of the
community were given presentations by each school about the health issues and research
program from the previous year. For each assessment participants filled out nutritional
questionnaires with at least one parent or guardian regarding their sleeping eating and other
routines. After a two-year project a certified clinical nutritionist produced individual reports for
each participant and their family at four health fairs as well as population-level research. The
lack of essential metabolite status markers pooling samples for a single proteome analysis comet
assay batch effects year 2 proteomics analysis and a small sample size were the study’s
limitations.
Was the article clear, relevant, and accurate? (Include examples from the article to support your
critique
Conclusion
7
This research project was initially designed in 2011 to perform a nutritional-challenge study, as
recommended by others, guided by physiological, biochemical, and behavioral data (e.g., based
on dietary habits), environmental factors (potential poisons ingested from the polluted
environment), together with distinctive socio-demographic indices, followed a discovery-
informed statistical and algorithmic strategy to explore crucial variables hiding behind the
responsiveness of individuals towards adaptation to multiple-micronutrient supplementation
treatment. Taken in the context of training for a doctoral student and postdoctoral fellow — each
needing their own, directed research project — as well as current publication practices (notably
topic-focused articles), this investigation helped identify what impact supplementation with
vitamins and minerals had on physiology among children and adolescents that demonstrated such
poor diet quality extremes based upon Brazilian Healthy Eating Index; low intake of fruit,
vegetables, legumes or whole grains/milk/dairy was considered lower central nutritional issues
faced by these youths while high consumption levels appearing instead to be limited almost
strictly sugar-rich foods/foods/meats fans nutrients.
Given that current political and public health policy, as well as the habits of these individuals in
terms of food consumption, have shown themselves to be insufficient to provide or support
access to healthy foods, we hope supplementation with a multi-micronutrient formulation at
doses below upper tolerable levels can supplement Brazilians 9–13 years old's demands for
optimum metabolism. Lessons from this research project point to the imperative of data source
diversification (e.g., omics plus features derived from unstructured sources), excluding sources
with important unmeasured confounding, and enabling covariate adjustment in statistical
analyses. Future research with similar design conducted in other populations or re-analyses of the
data were also required to test and confirm our findings. We recommend that the context‐
8
dependent utilization of big data in nutrition science could provide additional resources to grasp
dietary complexity and its impact on health maintenance, which has significance for delaying or
preventing chronic diseases.
The research project, initiated in 2011, aimed to study the nutritional challenges faced by
children and adolescents in Brazil. The study used physiological, biochemical, and behavioral
data, environmental factors, and socio-demographic indices to explore the variables influencing
individuals' adaptation to multiple-micronutrient supplementation treatment. The study identified
the impact of vitamin and mineral supplementation on physiology, specifically lipid profile
changes in those who received them. The research found that low intake of fruits, vegetables,
legumes, whole grains, milk, and dairy was considered lower nutritional issues, while high
consumption levels were limited to sugar-rich foods. The study suggests that current political and
public health policy and individual habits are insufficient to provide access to healthy foods. The
researchers suggest that supplementation with a multi-micronutrient formulation at doses below
upper tolerable levels could supplement Brazilians 9-13 years old's needs for optimum
metabolism.
The research emphasizes the importance of data source diversification, excluding unmeasured
confounding, and enabling covariate adjustment in statistical analyses. Future research with
similar designs or re-analyses of data is needed to test and confirm the findings.
Reference for article
