Nutritional biomarkers provide objective information on dietary exposure. Objective assessment is highly important to circumvent the fundamental limitation of measurement error in self-reported subjective assessment of dietary exposure [3,14,22].
Among prospective cohort studies assessing incident non-communicable diseases, EPIC-Norfolk is the largest cohort that has measured habitual dietary intake with subjective instruments and also nutritional biomarkers . In one analysis, the investigators compared the association between habitual fruit and vegetable consumption and incident type 2 diabetes with that between plasma vitamin C as a biomarker of fruit and vegetable consumption and type 2 diabetes.
As seen in Figure D.3.2, there was a stronger inverse association when plasma vitamin C biomarker was examined, compared with self-reported fruit and vegetable intake from a food frequency questionnaire, across fifths (quintiles) of their distributions. This indicates a proof of principle that a nutritional biomarker can provide a method with less error than the subjective instrument to examine associations between dietary factors and disease.
Figure D.3.2 Odds ratios for type 2 diabetes according to quintile of self-reported fruit and vegetable intake estimated from food frequency questionnaires [green squares] and plasma vitamin c concentration [orange diamonds]. Analyses are adjusted
for age, sex, family history of diabetes, physical activity, smoking status, social class, education level, vitamin supplement use, body mass index and waist circumference.
Source: EPIC-Norfolk Study – adapted from .
Although many studies demonstrate utility of nutritional biomarkers, they are highly diverse in terms of applications, strengths, and limitations. Dietary dimensions assessed with nutritional biomarkers are shown in Table D.3.9. These vary depending upon the category of biomarker, as described below.
Definition of nutritional biomarkers
In the field, many definitions are proposed. In general terms, a nutritional biomarker can be defined as “any biological specimen that is an indicator of nutritional status with respect to intake or metabolism of dietary constituents. It can be biochemical, functional, or clinical index of status of an essential nutrient or another dietary constituent” .
Categories of nutritional biomarkers
There is more than one scheme used to categorise nutritional biomarkers. Biomarkers can be grouped according to what they assess . A single biomarker may be included in one or more of the following categories:
Another classification scheme distinguishes recovery, concentration, predictive and replacement biomarkers as the followings [6, 10, 13]. The classification is not mutually exclusive.
Table D.3.9 Dietary dimensions assessed by nutritional biomarkers.
|Dietary dimension||Possible to assess?|
|Energy and nutrient intake of total diet||Yes|
|Intake of specific nutrients or food||Yes|
|Infrequently consumed foods||Yes|
|Frequency of eating/meal occasions||No|
|Adult report of diet at a younger age||No|
Nutritional biomarker methods rely upon biological specimen being collected from the participant. Examples of biological specimens include:
Other bio-specimens may include leucocytes, cord blood, breast milk, saliva, sweat, and any biopsied tissue.
Timing of specimen collection
Assessment of nutritional intake and status
Some nutrient intakes are particularly difficult to assess by subjective reporting. An example is dietary sodium, which is present in many manufactured food products and to different degrees even for the same food item. It is also added during cooking or at the table, often inconsistently and without measuring the amount added. Due to the high risk of measurement error associated with subjective reporting, objective measurement of urinary sodium is used to assess sodium intake in epidemiological studies.
The UK implemented a population-based intervention to gradually reduce sodium content in foods over time. Assessment of 24-h urinary sodium before, during, and after the intervention confirmed reduction of salt intake over time .
Nutritional biomarkers are used to identify nutritional adequacy. For example, iron deficiency is determined by using a combination of serum ferritin and transferrin receptors. For clinical diagnosis of deficiency diseases, confirmation is needed with a combination of other clinical examinations (e.g. symptoms).
Biomarkers of nutritional status are not necessarily nutrients. Examples include:
Biomarkers for validation or calibration
Biomarker methods are not without errors and do not necessarily reflect a habitual diet unless repeatedly measured over time. Therefore, strictly speaking, a ‘validation’ study on validity of dietary assessment method is often argued to be called a ‘calibration’ study.
Biomarkers have often been used to examine validity of subjective methods of dietary assessment. Validity assessment of an error-prone instrument is often performed against a reference measure from another instrument (e.g. food-frequency questionnaires compared to weighed food diaries). Comparison between different instruments is informative, but both involve limitations of subjective methods. Use of objective biomarkers is thus important.
Use of recovery biomarkers is ideal, such as doubly labelled water or 24-hour urine collections. However, these are often expensive or inconvenient for participants . Instead, concentration or prediction biomarkers are often used to evaluate the performance of a dietary assessment method. If a valid calibration equation is available, levels of concentration biomarkers and prediction biomarkers can be translated to estimates of absolute intakes. Each type of biomarkers has its own limitation (see below in section 5).
Examples of validation studies
The Observing Protein and Energy Nutrition (OPEN) Study was conducted in the United States from 1999-2000 and sought to assess the dietary measurement error of two self-reported instruments: the food-frequency questionnaire (FFQ) and two 24-hour recalls [19, 20]. Doubly labelled water, urinary nitrogen, and urinary sugars (fructose and sucrose) from two 24-hour urine samples were used as unbiased biomarkers of energy, protein, and sugar intakes in nearly 500 men and women. Key results were:
The EPIC-Norfolk Study assayed urinary measures of nitrogen, potassium, sodium, and plasma measures of vitamin C and polyunsaturated fatty acids. Then, the Study compared those with corresponding measures of dietary intakes based on a semi-quantitative FFQ and a 7-day diet diary [5, 23]. Key findings indicate that the degree of validity depends on the nutrient:
Accurate assessment of dietary intake using biomarkers depends upon not only sample collection but also the analytical measurement of the biomarker.
Minimising measurement error
Efforts should be made to standardise the collection, storage and analysis of specimens, but this may not be possible in large multi-centre studies which take place over a long time period. Measurement error may result from :
One key step to minimise laboratory measurement errors is blinding of the characteristics of the specimens, such as disease status. Ideally all specimens should be analysed consecutively to reduce between-assay variation. If a matched case-control design is used, the samples from matched individuals (e.g. pairs) should be analysed in the same batch, and samples in a batch should be randomised. Quality control samples are available to ensure compatibility of results between batches or laboratories. It is also possible to test reliability by conducting blinded duplicate analyses of specimens .
A quality control program should :
Assessing measurement error
The coefficient of variation can be used to assess the reliability of laboratory analyses of the same sample, and this can be calculated both within and between runs.
Combination with subjective estimates
Dietary intakes can be estimated by combining biomarkers with subjectively reported data. These estimates should be more valid as one method can account partly for the disadvantages of the other method i.e. biomarkers account for dietary misreporting and self-reporting methods account for errors associated with the metabolism of nutrients . Thus, although concentration or prediction biomarkers themselves cannot indicate absolute levels of dietary intakes, they can be used to predict absolute intakes by combining with subjective estimates.
Key characteristics of nutritional biomarkers are summarised in Table D.3.10. It is important to recognise that all biomarkers have some limitations, but the degree and type of limitation vary by biomarker. Some limitations may or may not be reduced by the study design of sampling, assay, data processing, and statistical adjustment for errors and confounding, and should be appraised appropriately.
Concentration or predictive biomarkers
Here are some examples that a nutrient or food component is specific to a food source:
Concentration or predictive biomarkers
Table D.3.10 Characteristics of nutritional biomarker methods.
|Number of participants||Any|
|Cost of development||High|
|Cost of use||High|
|Participant burden||Varies with sample type|
|Researcher burden of data collection||High|
|Researcher burden of coding and data analysis||High*|
|Risk of reactivity bias||No|
|Risk of recall bias||No|
|Risk of social desirability bias||No|
|Risk of observer bias||Minimised with blinding|
|Participant literacy required||No|
|Suitable for use in free living||Yes|
|Requires individual portion size estimation||No|
* Data coding and analysis here means processing of quantitative data from an assay apparatus.
Considerations relating to the use of nutritional biomarkers for dietary assessment are summarised by population in Table D.3.11.
Table D.3.11 Considerations relating to use of biomarkers for assessing diet in different population groups.
|Pregnancy||Information on dietary supplements should be evaluated in detail. Cord blood samples would be non-invasive biological samples.|
|Infancy and lactation||Breast milk, a unique bio-sample.|
|Toddlers and young||Maturation may influence the utility of biomarkers (e.g. calcium, zinc) due to high turn-over rates of tissues.|
|Adolescents||Adolescent growth and menstruation in girls influence utility of specific biomarkers (e.g. calcium, iron).|
|Adults||In women, menstruation influences utility of specific biomarkers (e.g. iron).|
|Older Adults||Confounding by disease conditions and medications may occur, depending on the aim of the assessment.|
|Ethnic groups||Underlying genetic variability may influence utility of biomarker (e.g. 25-hydroxy vitamin D ) which should be confirmed or used cautiously.|
Sensitivity and specificity of a biomarker
In clinical and epidemiological fields, ‘sensitivity’ and ‘specificity’ have been used in different ways, partly because a biomarker can be used to assess either nutritional intake or nutritional status. Here are some examples of ‘sensitivity’.
Similarly, ‘specificity’ of a biomarker can be documented as the following:
Homocysteine is, for example, sensitive to folate intake but not specific, because it can vary by intakes of other B vitamins and amino acids. Thyroid stimulating hormone is sensitive to iodine deficiency, but not sensitive to a normal range of iodine intake .
Cut-off points of biomarkers
It would be ideal to have cut-off points to diagnose certain states of nutritional status or nutritional intake. However, cut-offs are rarely established. One reason is because any single biomarker has limitations and is often used jointly with other indicators, thus leaving definite cut-off points undetermined. Another reason is diversity of applications of a single biomarker to clinical diagnosis, screening, surveys, monitoring, and other or complex objectives. See perspectives by Raghavan et al.  for more details.
Factors influencing sensitivity and specificity of nutritional biomarkers
Other factors might influence nutrient metabolism and may be potential confounders for an association between nutrient intake and nutrient biomarker. Many of the following considerations are based on clinical and biological knowledge. Impact of such influences on biomarkers in population-based research remains unestablished. Examples include :
Lifestyle or Physiologic Factors
Table D.3.12 Examples of nutritional biomarkers* and associated analytical methods.
|Total energy||Doubly-labelled water||Mass spectrometry to assay isotope ratios (2H/1H and 18O/16O)|
|Protein||Urinary nitrogen||Kjeltec method|
|Omega-3 and -6 fatty acids||Fatty acid concentration in blood or tissue lipid compartment||Gas chromatography after transmethylation (Folch method)|
|Potassium||Urinary potassium||Flame photometer|
|Sodium||Sodium potassium||Flame photometer|
|Calcium||Serum ionized calcium||Ion-specific electrodes|
|Phosphorus||Serum phosphorus||Colorimetry using molybdenum blue|
|Magnesium||Serum magnesium||Atomic absorption spectrometry|
|Serum ionized magnesium||Ion-specific electrodes|
|Copper||Erythrocyte superoxide dismutase||Spectrophotometric assay or ELISA|
|Iodine||Urinary iodine||Acid digestion, followed by spectrophotometric assay using the Sandell-Kolthoff reaction|
|Iodine||Urinary iodine||Acid digestion, followed by spectrophotometric assay using the Sandell-Kolthoff reaction|
|Thyroid stimulating hormone||ELISA with dried blood spots or serum|
|Iron||Serum ferritin||ELISA (in absence of infection)|
|Haemoglobin||Cyanmethemoglobin method. 'HemoCue' in field studies|
|Serum transferrin receptor||ELISA|
|Selenium||Plasma selenium||Atomic absorption spectrometry with Zeeman background correction or hydride generation atomic absorption spectrometry|
|Plasma glutathione peroxidase||ELISA: only useful if Se intakes are habitually low|
|Zinc||Serum/plasma zinc||Flame atomic absorption spectrometry (in absence of infection)|
|Hair Zinc||Serum/plasma zinc||Measure by neutron activation analysis or atomic absorption spectrometry in children with low height percentiles and/or hypogeusia|
|Vitamin A||Liver retinol stores||HPLC|
|Modified relative dose response||Serum retinol and dehydroretinol via HPLC 4-6 h after oral dose of 3,4-didehydroretinol acetate (100 μg/kg)|
|Vitamin D||Serum 25-hydroxyvitamin D||Separation of serum 25(OH)-D, followed by a competitive binding assay or radioimmunoassay, or tandem mass spectrometry|
|Vitamin E||Ratio of serum tocopherol to serum cholesterol||Reverse phase HPLC with a high sensitivity fluorescence detector|
|Vitamin K||Plasma vitamin K||HPLC, mass spectrometry|
|Thiamin||Erythrocyte activity of transketolase with and without added thiamin pyrophosphate||Semi-automated spectrophotometry using glyceraldehyde as an internal standard|
|Riboflavin||Erythrocyte activity of glutathione reductase with and without added prosthetic group flavin adenine dinucleotide||Enzyme-coupled kinetic assay where glutathione reductase activity is measured spectrophotometrically via oxidation of NADP to NADP+|
|Niacin||NAD:NADP ratio in erythrocytes||HPLC|
|Pyridoxine||Plasma pyridoxal-5’-phosphate||Cation-exchange HPLC with fluorescence detection|
|Vitamin C||Serum or leucocyte ascorbic acid||HPLC with electrochemical detection. Should use preservative (metaphosphoric acid)|
|Folate||Erythrocyte and serum folate||Microbiological assay using L. casei|
|Serum homocysteine||Reversed-phase HPLC with fluorescence detection|
|Vitamin B12||Serum vitamin B12||radioimmunoassay|
|Serum methylmalonic acid||Mass spectrometry|
*Adapted from: [7, 8, 22].
Other molecules have been studied as markers of specific dietary intakes, such as:
- carotenoids, caffeine metabolites, alkylresorcinols, flavones, isoflavones, phytosterols, and phytochemicals
- mercury, cadmium, arsenic, and other ultra-trace minerals
- trans fatty acids, conjugated linolenic acid, pentadecanoic acid, heptadecanoic acid (fatty acids produced in ruminants)
- dioxins, heterocyclic amines, bisphenol A, and other contaminants
Statistical approaches to combine multiple methods including biomarker information and self-reported dietary data have been under rigorous methodological investigations over more than a decade . While the methodology has been well conceptualised, there remain limitations in application, empirical evidence, and availability of software for wide use.
Some high-dimensional assays, including ‘omics’ technologies, have been used to identify new biomarkers, including:
- Metabolomics, measuring a large number of low-weight metabolites including essential amino acids and molecules not synthesized in the human body (e.g. trimethylamine N-oxide, TMAO, as a marker of fish/animal-based foods ).
- Epigenetics, including DNA methylation reflecting intakes of methyl donors (e.g. methionine) from diets and B vitamins (folate) playing key roles in the methylation cycle.
- Proteomics, potentially reflecting functional markers of nutritional intake and status.