Hum Genet (2009) 125:507–525 DOI 10.1007/s00439-009-0662-5

REVIEW ARTICLE

Biomarkers in nutritional : applications, needs and new horizons

Mazda Jenab · Nadia Slimani · Magda Bictash · Pietro Ferrari · Sheila A. Bingham

Received: 16 January 2009 / Accepted: 27 March 2009 / Published online: 9 April 2009 © Springer-Verlag 2009

Abstract Modern epidemiology suggests a potential many functional dietary biomarkers that, if utilized appro- interactive association between diet, lifestyle, and priately, can be very informative, a better understanding of the risk of many chronic . As such, many epidemio- the interactions between diet and genes as potentially deter- logic studies attempt to consider assessment of dietary mining factors in the validity, application and interpretation intake alongside genetic measures and other variables of of dietary biomarkers is necessary. It is the aim of this interest. However, given the multi-factorial complexities of review to highlight how some important biomarkers are dietary exposures, all dietary intake assessment methods being applied in nutrition epidemiology and to address are associated with measurement errors which aVect dietary some associated questions and limitations. This review also estimates and may obscure risk associations. For emphasizes the need to identify new dietary biomarkers and this reason, dietary biomarkers measured in biological highlights the emerging Weld of nutritional metabonomics specimens are being increasingly used as additional or sub- as an analytical method to assess metabolic proWles as mea- stitute estimates of dietary intake and nutrient status. sures of dietary exposures and indicators of dietary pat- Genetic variation may inXuence dietary intake and nutrient terns, dietary changes or eVectiveness of dietary metabolism and may aVect the utility of a dietary interventions. The review will also touch upon new statisti- to properly reXect dietary exposures. Although there are cal methodologies for the combination of dietary question- naire and biomarker data for disease risk assessment. It is clear that dietary biomarkers require much further research M. Jenab (&) Lifestyle, Environment and Cancer Group, in order to be better applied and interpreted. Future priori- International Agency for Research on Cancer (IARC-WHO), ties should be to integrate high quality dietary intake infor- Lyon, France mation, measurements of dietary biomarkers, metabolic e-mail: [email protected] proWles of speciWc dietary patterns, genetics and novel sta- N. Slimani tistical methodology in order to provide important new Nutritional and Database Resource Team, insights into gene-diet-lifestyle-disease risk associations. International Agency for Research on Cancer (IARC-WHO), Lyon, France

M. Bictash Introduction Division of Epidemiology, Public Health and Primary Care, Imperial College London, London, UK Over the past decades, the Weld of nutritional epidemiology has generated a large body of evidence for a potential inter- P. Ferrari Data Collection and Exposure Unit, active association between diet, lifestyle and genetics and European Food Safety Authority (EFSA), Parma, Italy the risk of many chronic diseases. Much of the evidence relating food groups, speciWc S. A. Bingham foods and nutrients to chronic disease risk relies on infor- MRC Centre for Nutritional Epidemiology in Cancer Prevention and Survival, Department of Public Health and Primary Care, mation gathered using various dietary assessment instru- University of Cambridge, Cambridge, UK ments, such as dietary/food frequency questionnaires, food 123 508 Hum Genet (2009) 125:507–525 diaries, food records or 24-h recalls. In most cases, these Applications of dietary biomarkers methodologies require a systematic estimation of the fre- quency of consumption and the portion size of the foods A dietary biomarker can be loosely deWned as a biochemi- consumed as well as more or less detailed information on cal indicator of dietary intake/nutritional status (recent or the recipe ingredients, combinations of foods consumed long term), or it may be an index of nutrient metabolism, or together, and cooking methods, which may aVect the esti- a marker of the biological consequences of dietary intake mation of exposure to a particular dietary component. In (Potischman and Freudenheim 2003). The main advantage addition, the estimation of nutrient intakes relies almost of—or the main assumption behind—dietary biomarkers is entirely on the existence of appropriate and reliable food that they are objective measures and are independent of all composition tables for the target population. When these the biases and errors associated with study subjects and die- issues are coupled to the overwhelming complexities of tary assessment methods (Day et al. 2001; Kaaks et al. diVerent dietary patterns, varying dietary habits, multitudes 2002; et al. 2007). An ‘ideal’ dietary biomarker of lifestyle confounders, numerous reporting biases, daily would accurately reXect its dietary intake level and it would variations in food intake, combinations of foods, timing of be speciWc, sensitive and applicable to many populations. meals, etc., it is of no surprise that all dietary assessment Existing dietary biomarkers are not ‘ideal’, but they are instruments are associated, to one extent or another, with functional and have found wide spread applicability in diVerent, and sometimes considerable, random and systematic modern nutritional epidemiology. In general, dietary bio- measurement errors. markers can be divided into several classes (recovery, pre- In fact, both nutritional epidemiologists and their many dictive, concentration, replacement) which are described in critics are acutely aware of the complexities and limitations more detail below and in Fig. 1. of various dietary assessment methods (Kaaks and Riboli One of the main applications of dietary biomarkers is to 2005; Michels 2005a, b). This reality of nutritional epidemio- use them as reference measurements to assess the validity logy is being met with intense methodological research and and accuracy of dietary assessment methods (Bingham not only are innovative methods (e.g. internet based assess- 2002; Potischman and Freudenheim 2003; Tasevska et al. ment, use of digital cameras, cellular telephones and personal 2005). The most important dietary biomarkers for this digital assistants) being developed and validated (Dowell and application are the ‘recovery’ biomarkers (i.e. doubly Welch 2006; Kikunaga et al. 2007; Subar et al. 2007; Wang labeled water which is utilized to measure the metabolic et al. 2006), but traditional ones are also constantly being rate and total energy expenditure; urinary total nitrogen/ reWned and improved (e.g. computerized 24-h recall: EPIC potassium which are utilized to estimate total daily protein Soft) (Slimani et al. 1999, 2002; Slimani and Valsta 2002). consumption and potassium intake, respectively) (Bingham More recently, various statistical techniques have also 2003; Day et al. 2001; Livingstone and Black 2003). been developed to account for some of the apparent mea- Recovery biomarkers are based on the concept of the meta- surement errors (Fraser and Yan 2007; GorWne et al. 2007; bolic balance between intake and excretion over a Wxed Rosner et al. 2008) and to better estimate usual food intakes period of time and so provide an estimate of absolute intake (Dodd et al. 2006; Tooze et al. 2006). Nevertheless, in the levels (Kaaks et al. 1997). In other words, excretion levels absence of any ‘independent’ observation of food con- are highly correlated with intake (Bingham 2002). How- sumption, true intake cannot really be assessed. In order to ever, before being applied to the task of questionnaire vali- obtain such ‘independent’ observations (i.e. uncorrelated dation, such biomarkers need to be tested in calibration measurement errors), nutrition epidemiologists have uti- studies under controlled conditions (e.g. in a metabolic lized diVerent biomarkers assessed in biological samples suite) in order to establish that their predictability in not only as measures of dietary intake and nutrient status, humans consuming varying diets is comparable to the die- but also as predictors of disease risk. tary intake method under consideration. Unfortunately, the It is the aim of this review to highlight how some impor- cost and complexity of these techniques makes them tant biomarkers are being applied in the Weld of nutrition largely inapplicable for widespread epidemiologic use and epidemiology and also to address some of the questions and they are best applied either in post hoc analyses of on- shortfalls associated with their use. There is a need to going investigations, or built-in to the design of new develop new dietary biomarkers, and in this respect, the studies, for example the use of doubly labeled water in the review will also highlight metabonomics as an analytical OPEN study (Schatzkin et al. 2003) and markers of method that can be utilized to assess metabolic proWles as potassium and nitrogen in 24 h urine collections (Bingham measures of dietary exposures and indicators of dietary pat- 2002). The recently deWned class of ‘predictive’ biomark- terns or dietary changes. Also, new statistical methodolo- ers can also be utilized to assess the degree of measurement gies for the combination of dietary questionnaire and errors in dietary assessment methods. Like recovery biomarker data will be touched upon. biomarkers, predictive biomarkers are sensitive, time 123 Hum Genet (2009) 125:507–525 509

Recovery Dietary Biomarkers Dietary Assessment i.e. doubly labeled water, urinary nitrogen or potassium For estimation of dietary intakes Improvements for Dietary (e.g. questionnaire, diary, 24-hour recall) Uses: Biomarkers in Current Use: •As reference measurements to assess •Enhanced laboratory methods validity / accuracy of diet assessment •Better understanding of: methods Random and systematic dietary -nutrient metabolism -gene-diet, gene-nutrient or measurement errorsaddressed by: Predictive Dietary Biomarkers gene-gene interactions Improvements in diet assessment methods e.g. urinary sucrose or fructose •Refinement of existing methods •Enhancement of food composition databases Uses: •Innovation of new methods •As reference measurements to assess Enhanced Application and Development of novel statistical validity / accuracy of diet assessment methodology methods Interpretation of •Integration of dietary and biomarker data for Dietary Biomarkers identification and correction of measurement Concentration and Replacement errors Dietary Biomarkers e.g. vitamins, carotenoids, individual fatty acids, phytoestrogens, alkylresorcinolsetc. Diet Uses: Observed vs. True Intake •Assess correlation with estimates of dietary intake •Estimation of diet-disease risk Incorporation of individual associations Interactions / SNP, whole genome or -As a substitute or complimentary to Confounders other “omic”data dietary assessments Lifestyle, Environment, Genetic variability Identification of New Dietary Biomarkers: Effect on Disease Risk •Nutritional Metabonomics: Identification • direct or indirect of metabolic profiles as biomarkers specific to different dietary/nutrient patterns and dietary changes •Others?

Fig. 1 DiVerent classes of dietary biomarkers measured in biological samples and their application to the validation of dietary assessment meth- ods, measurement error and estimation of disease risk associations dependent and show a dose-response relationship with 1994), salt (Norat et al. 2008) or metabonomic factors intake levels but the distinction is that their overall recov- (detailed in a later section). Depending on the speciWc die- ery is lower (Tasevska et al. 2005). The only current exam- tary biomarker (e.g. some fatty acids), the distinction ples are 24 h urinary sucrose and fructose which are closely between the concentration and replacement classes may be correlated with intake of , despite the very small vague. fraction of intake which is actually present in urine collec- A common application of concentration or replacement tions (Tasevska et al. 2005). dietary biomarkers is for the estimation of diet-disease risk The class of ‘concentration’ biomarkers (e.g. vita- associations (Potischman 2003). This use is increasingly mins, blood , urinary electrolytes) are also available Wnding application in population studies such as prospec- for comparison with estimates of dietary intake. For exam- tive cohort studies, where biological samples are collected ple, results from a dietary intake method which agreed most before disease onset, or intervention/controlled clinical closely with such biomarkers would be expected to yield studies looking at the eVect of dietary treatments or nutrient more reliable estimates of intake than one which did not supplementation on disease risk or progress. The underly- (Bingham et al. 2008). Concentration biomarkers cannot be ing concept is that the use of such biomarkers may lead to a translated into absolute levels of intake but the biomarker better ranking of subjects for exposure to a particular food concentrations do correlate with intakes of corresponding group or nutrient than would dietary assessment methods. foods or nutrients, although the strength of the correlation In fact, the biomarker level measured in the blood or other is often lower (<0.6) than that expected for recovery bio- biological samples takes into account any eVects of absorp- markers (>0.8). ‘Replacement’ biomarkers are closely tion, inXuences of microbiota (e.g. bioconversion, release related to concentration biomarkers and refer speciWcally to of bioactive dietary compounds, enterohepatic circulation), compounds for which information in food composition data- interactions between nutrients, tissue turnover, metabolism bases is unsatisfactory or unavailable, for example aXatox- and excretion. Additional considerations are issues pertain- ins, some phytoestrogens (Grace et al. 2004; Qian et al. ing to nutrient bioaccessibility and bioavailability (Holst 123 510 Hum Genet (2009) 125:507–525

Fig. 2 Factors that may aVect • Genetic Variability the measurement and utility of a • genes that may affect dietary intake patterns, taste, attraction to specific foods or food types etc. dietary biomarker to properly reXect dietary exposures in indi- • biological variation in nutrient absorption, metabolism, tissue turnover, excretion viduals or target populations • epigenetic variation, gene-gene interactions • Lifestyle or Physiologic Factors • smoking, alcohol consumption, , gender, age, body weight / size, socioeconomic status • influence of colonic microbiota (bioconversion, release of bioactive dietary compounds) • enterohepatic circulation of nutrients (e.g. phytoestrogens, lignans) • metabolic and inflammation related disorders, stress, occult / underlying disease • Dietary Factors • range or frequency of intake for a particular nutrient • nutrient-nutrient interactions • nutrient bioavailability, influence of food matrix • Biological Sample • type of sample collected for analysis of biomarkers (e.g. whole blood, plasma, serum, urine) • conditions of sample collection, transport, treatment, storage conditions, length of storage • diurnal variation, day of the week or season of sample collection • Analytical Methodology • precision, accuracy, detection limits of the analytical technique • variations from method to method or laboratory to laboratory and Williamson 2008). All of these points are very impor- aVect the measurement and utility of a dietary biomarker to tant because they highlight that food components and nutri- properly reXect dietary exposures and suggests that the validity ents are inXuenced by a large number of host factors, both of some dietary biomarkers may well be population (or metabolic and genetic, that may aVect the correlation of a even individual) speciWc with respect to genetic background biomarker with the relevant dietary exposure. In addition, or other characteristics (Fig. 3) (Kaput 2008). Nevertheless, other factors such as the type of biological sample obtained, very few studies to date have considered possible gene-diet/ how the sample was collected, treated, and stored, the labo- nutrient or gene-gene interactions as potentially determining ratory methodology used to measure the biomarker (preci- factors in the validity and application of dietary biomarkers. sion, accuracy and detection limits of the analytical The interaction of genes and diet is engendered in the con- technique; variations from method to method or laboratory cepts of nutrigenomics (study of how diet inXuences gene to laboratory) can also aVect the measurement and utility of transcription, protein expression and metabolism) and nutri- dietary biomarkers (Fig. 2). The combination of all of the genetics (study of how genetic disposition aVects response to above factors makes it very diYcult to compare absolute diet and its components), which are extensively reviewed concentrations of certain dietary biomarkers across various elsewhere (Mutch et al. 2005; Ordovas and Mooser 2004). A studies which are based on diVerent populations and utilize key message is that the consideration of genetics is important diVerent biological samples and analytical techniques. for nutritional scientists and the consideration of dietary Currently, very little is known about how genetic variation assessment methodology and dietary biomarker measurement may inXuence dietary intake, food choices, nutrient metabo- is of relevance to geneticists. lism, or aVect the bio-availability, absorption, transport, bio- The issues raised above are all pertinent to the validity, transformation, and excretion of nutrients or bio-active dietary application and interpretation of dietary biomarkers. In the sec- components. Extensive information already exists on genetic tions below, these concepts are dealt with in the context of var- variation in taste and how that may aVect food preferences ious foods and their corresponding biomarkers, but they are and dietary habits (Garcia-Bailo et al. 2009). It is probably relevant and applicable to almost all other dietary biomarkers. safe to assume that genetic variability, gene-diet/nutrient interactions and gene-gene (epistatic) interactions may result in diVerential response to dietary factors along with changes Biomarkers of fruits and vegetables in nutrient metabolism and dietary biomarker levels. Some examples that have yet to be conWrmed in diVerent popula- In their 1997 comprehensive review of the literature the tions are folate and the MTHFR gene, vitamin D and the World Cancer Research Fund (WCRF) listed the strength VDR gene and iron and the HFE gene. This may signiWcantly of the evidence for a cancer preventive role of the fruits 123 Hum Genet (2009) 125:507–525 511

Diet Effect on Disease Risk • direct or indirect

Use of Biomarkers to Assess Dietary Exposures Gene-Gene Interactions Genetic Influence on Risk May Be Dietary Choices and Modulated By Food Intake Variability in Genes Differences in the metabolic Related to Impact on Dietary effects of nutrients Nutrient Metabolism Biomarker Other factors that may affect Measures biomarker measurement: •Lifestyle or physiologic factors •Dietary factors Differences in digestion, •Type of biological sample absorption, transport, Effect on Body / •Analytical methodology metabolism, bio- Tissue Exposure (see Figure 2 for details) transformation, excretion etc of Levels nutrients or bio-active food components

Gene-Diet/Nutrient Interactions

Fig. 3 Possible interactions of dietary intake and dietary biomarker measures with genetic variability to aVect disease risk and/or vegetables food group as ‘convincing’ (for cancers (vitamin C content very low) and urine (appreciable levels of the mouth/pharynx, esophagus, lung, stomach, colon/ only in individuals in excess of adequate intake) do not rectum) or ‘probable’ (for cancers of the larynx, pancreas, appear to be appropriate for the measurement of vitamin C breast bladder) (World Cancer Research Fund and Ameri- status (Benzie 1999). There is little information on the use can Institute for Cancer Research 1997). However, these of tissues or individual cells for the measure of vitamin C as judgments were downgraded in the 2003 IARC Handbooks a dietary biomarker. Thus, most biomarker studies analyze of Cancer Prevention on Fruits and Vegetables (IARC vitamin C concentrations in plasma. However, the results of Working Group on the Evaluation of Cancer Preventive a recent meta-analysis show that plasma vitamin C concen- Strategies 2003) and in the 2007 update of the WCRF com- tration and estimates of dietary vitamin C intake from vari- prehensive report (World Cancer Research Fund 2007). ous dietary assessment instruments are only modestly The revised conclusions were largely based on accumulat- correlated (r = »0.4) (Dehghan et al. 2007). The magnitude ing evidence tending towards null, particularly from pro- of the correlation varies between populations [e.g. range of spective studies (World Cancer Research Fund 2007). r values: Germany = 0.24 (Boeing et al. 1997); India = 0.12 Although such null Wndings based on assessment of dietary (Chiplonkar et al. 2002); Iran = 0.35 (Malekshah et al. fruit and vegetable intake may indeed be real, it is equally 2006); Spain = 0.53 (Schroder et al. 2001); USA = 0.25 possible that the disease risk association may be attenuated (Cooney et al. 1995), 0.39 (Jacques et al. 1993)] based on as a result of measurement errors associated with dietary the range of vitamin C intake, and tends to be stronger in assessment instruments, the concomitant loss of statistical men than women. They may also be aVected to some power and probable genetic variations in response to diet. It degree by other determinants such as: errors in food com- is for these reasons that more and more studies are begin- position tables from which dietary vitamin C values were ning to measure blood biomarkers of fruit and vegetable derived, un-/mis-reported intake of vitamin C from dietary intake in their assessment of the disease risk associations supplements or food additives, diVerent food processing for this important food group. In this section the biomarkers techniques which may aVect vitamin C content (e.g. expo- vitamin C and carotenoids are discussed. sure to high temperatures may lead to break down of vita- min C), various lifestyle factors (e.g. smoking, exercise, Vitamin C chronic low-grade inXammation) which may reduce vita- min C level, biases in the assessment of dietary vitamin C Blood measures of vitamin C (ascorbate and dihydroascor- intake, or potential inXuence of genetic variability on vita- bate) may be considered as a surrogate for dietary vitamin min C metabolism. C intake and of the main fruit and vegetable sources of The measurement of vitamin C in biological specimens vitamin C. Of the easily sampled biological Xuids, saliva can also be aVected by sample handling and treatment. 123 512 Hum Genet (2009) 125:507–525

Vitamin C is one of the most labile vitamins. It can easily or correlations between the level of dietary vitamin C degrade in biological samples and should ideally be stabi- intake and blood vitamin C concentrations. Some recent lized by addition of protein precipitating agents such as data suggest that African Americans may be more aVected metaphosphoric acid (Bates 1994; Ching et al. 2002). But, by the functional consequences of a variation in SVCT1 in the case of large prospective studies biologic samples are (Eck et al. 2004), but this remains to be conWrmed. It is pos- collected for a multitude of purposes, are not usually sible that the consideration of genetic variability in this treated speciWcally for the preservation of vitamin C and transporter protein may allow an improvement in the corre- are often stored for many years prior to analysis. Vitamin C lation between dietary and blood vitamin C measures in may degrade during sample handling and processing diVerent population groups and permit a better utilization of because of various factors such as temperature, exposure to plasma vitamin C measures as a biomarker of fruit and vege- light, choice of anticoagulant, time sitting on the shelf, or table consumption. Perhaps an equally important aspect is freezing procedures (Ching et al. 2002; Chung et al. 2001; whether such a Wnding may alter previously identiWed asso- Key et al. 1996; Lykkesfeldt et al. 1995; Terzuoli et al. ciations between dietary and plasma vitamin C levels and 2004). Also, loss of vitamin C during long periods of stor- risk of various diseases. age has been noted (loss per year of frozen storage in those Another issue pertaining to vitamin C is whether its use with low/high baseline vitamin C levels: men = 0.26/1.96, as a dietary biomarker is valid in all populations, or only in and women = 0.69/2.35 mol/l plasma) (Jenab et al. 2005). some populations depending on their range of intakes. Sim- Although some studies suggest that immediate sample sta- ilar to many other nutrients, the absorption eYciency and bilization, freezing and analysis within a short period of resulting plasma concentration of vitamin C (dose-response time are required for best results (Karlsen et al. 2007), oth- curve) is not linear at all intake levels. At lower intake lev- ers show that vitamin C can be still be measured with rea- els (approx. <100 mg/day), a linear relationship may exist, sonable reliability in un-stabilized plasma samples frozen whereas it may plateau at higher intake levels (approx. for long periods of time at ultra low temperatures (Jenab >120 mg/day) such that over exposure may not at all be et al. 2005). The choice of analytical technique (e.g. chro- well represented by plasma vitamin C measurements. The matographic versus Xuorometric versus spectrophotometric general understanding is that at lower dietary vitamin C methods) is also an important consideration in the analysis intake levels, vitamin C is eYciently absorbed from the of vitamin C. Initially, vitamin C was often measured using intestines and renal excretion is minimized (Nelson et al. enzymatic or spectrophotometric methods, but recent Wnd- 1978), whereas at higher concentrations the SVCT1 is satu- ings suggest that these methods are subject to much inter- rated and down-regulated (MacDonald et al. 2002), leading ference, that chromatographic methods are far superior and to reduced intestinal absorption and limiting blood vitamin that high sample throughput is necessary to avoid vitamin C C concentrations (Li and Schellhorn 2007). Thus, measure- degradation (Karlsen et al. 2005; Levine et al. 1999). ment of blood vitamin C as a marker of dietary intake may Genetic variability could conceivably aVect the kinetics be reasonable in populations with low to moderate or heter- of vitamin C absorption (transporter saturation, down-regu- ogeneous vitamin C intake levels, but it may be less infor- lation) and metabolism, yet very few studies have consi- mative in a well fed population with a homogeneous high dered this aspect (Wilson 2005). Indeed, genetic level of dietary vitamin C intake. A related issue is whether polymorphisms may represent a large source of variation the measure of plasma vitamin C is a short- or long-term for many dietary biomarkers, such as antioxidant nutrients marker of intake or nutritional status. Plasma vitamin C (Nowell et al. 2004; Reszka et al. 2006). This concept has concentration has been shown to peak within hours of been referred to as metabolic confounding (Saracci 1997). intake in relation to the size of the dose (Benzie and Strain A well described example is the gene coding for human 1997), and tends to plateau at around 100 mol/l because of apolipoprotein E whose three common alleles are associ- renal excretion and feedback regulation of intestinal ated with diVerent levels of serum cholesterol (Burnett and absorption (Graumlich et al. 1997). Thus, it is likely that Hooper 2008; Ordovas 2002, 2007; Ribalta et al. 2003). In plasma vitamin C concentration indicates functional the case of vitamin C, the sodium dependent vitamin C reserves of vitamin C (Benzie 1999) and is a biomarker of transporter protein 1 (SVCT1) coded for by the SLC23A1 short- to medium-term intake (Mayne 2003), raising con- gene (Eck et al. 2004; Li and Schellhorn 2007) controls the cerns of whether a single blood sampling is suYcient for intestinal absorption of dietary vitamin C and may also the objectives of the study at hand or whether multiple mea- aVect blood and/or tissue-speciWc vitamin C concentrations. sures are necessary. Despite, intense research on the role of SVCT1 in control- A number of studies suggest that the absorption ling vitamin C concentration in various cell types (Wilson eYciency of vitamin C from the diet is also aVected by life- 2005), very little information exists on the inXuence of its style factors such as age and smoking status (Brubacher genetic variability in terms of vitamin C pharmacokinetics et al. 2000). Some in vitro studies further suggest that 123 Hum Genet (2009) 125:507–525 513 vitamin C absorption or transport into certain cell types Working Group on the Evaluation of Cancer Preventive may also be inhibited by dietary factors such as glucose or Strategies 1998; Reboul et al. 2006) and their blood con- Xavonoids (Wilson 2005) or by use of drugs such as aspirin centrations appear to be moderately correlated with diet (r (Ioannides et al. 1982). In addition, blood vitamin C con- ranges from »0.2 to over 0.5) (Kaaks et al. 1997), perhaps centration also depends in large part on the level or better in some populations than others (e.g. stronger corre- eYciency of renal re-absorption/excretion (Li and Schell- lation in normal weight than obese subjects (Vioque et al. horn 2007), and presumably variability in their genetic 2007)), and do not appear to be under homeostatic control determinants. In the setting of an observational study, many (IARC Working Group on the Evaluation of Cancer Pre- of these factors are diYcult or impossible to control for, and ventive Strategies 1998). These points suggest that the it is quite probable that they may interact to modulate blood measurement of their circulating blood levels probably pro- vitamin C concentration, and aVect its correlation with die- vides a good estimation of their bioavailability, or overall tary vitamin C intake (Kaaks et al. 1997). level of body exposure, in diVerent populations. The above are all sources of variation that are at issue in The consideration of carotenoids as dietary biomarkers the reproducibility and validity of many biomarkers and go raises an issue that was lightly touched upon in the descrip- to the of the question of whether a dietary assessment tion of vitamin C: does the metabolism of a nutrient aVect method can provide suYcient information or whether a bio- its validity as a dietary biomarker? The metabolism of marker assessment is also necessary. A case in point carotenoids can aVect their blood concentrations and regarding vitamin C is a recent study based on the Euro- decrease their correlation with estimates of their dietary pean Prospective Investigation into Cancer and Nutrition intake or result in a mis-interpretation of their overall expo- (EPIC) which showed that higher concentration of plasma sure levels. The carotenoids -carotene, -carotene, -cryp- vitamin C concentration, but not the level of dietary vita- toxanthin can be partially metabolized to retinol (Fraser and min C intake assessed by country speciWc questionnaires Bramley 2004). This metabolism may be unimportant in and other dietary methods, is associated with a decreased subjects who are vitamin A replete but may be vital in sub- risk for gastric cancer (Jenab et al. 2006b). jects who have a lower vitamin A status (Nagao 2004), and may lead to diVerential ranking of subjects based on bio- Carotenoids marker measures, even if their carotenoid intake levels are similar. As another example, one of the most important Another class of compounds that have been widely mea- roles ascribed to carotenoids is their potential as antioxi- sured as dietary biomarkers of fruit and vegetable intake are dants (Krinsky and Johnson 2005; Rao and Rao 2007). carotenoids. Similar to vitamin C, and many other nutri- Subjects with increased oxidative stress, such as smokers ents, the estimation of carotenoid intakes from dietary (Saintot et al. 1995) or those consuming higher levels of assessment instruments is prone to many measurement alcohol (Albanes et al. 1997), have been shown to have errors and may not reXect their actual bioavailability. Most decreased blood concentrations of some carotenoids. How- food composition tables contain values for only a few ever, it is relatively unknown how diVerent carotenoids carotenoids, primarily -carotene and lycopene. However, may interact with each other and other antioxidant nutrients the measurement of their blood concentrations as biomark- in vivo, or whether they have the same antioxidant poten- ers of their intake is subject to many of the same issues tial. Some biomarker studies suggest that not all carote- detailed above for vitamin C. Of the hundreds of carote- noids may have a negative cancer risk association (Jenab noids existing in plants, only a few (mostly: -carotene, - et al. 2006a), and it has been suggested that the total sum of carotene, -cryptoxanthin, canthanxanthin, lycopene, all blood carotenoids measured at any one time may be a lutein, zeaxanthin) can be found in signiWcant amounts in better estimate of their exposure than comparisons of indi- the blood (Barua et al. 1993; Crews et al. 2001). They are vidual carotenoids (Liu 2004). But there is currently little amenable to use in prospective cohort studies because their evidence to support this notion because few studies to date long term frozen storage has only a minor eVect on their have actually measured a battery of carotenoids and most reliability (Al-Delaimy et al. 2008). Carotenoids are epidemiologic data focus on a few select compounds from soluble and their intestinal absorption, and hence bioavail- this group. ability, may be modulated by the lipid content of the diet, To date, very little is known about genetic variability food matrix, competition with other carotenoids (Reboul pertaining to carotenoid bioavailability and metabolism. It et al. 2007), degree of colonic fermentation (Goni et al. may be safe to assume that such variability exists, and may 2006), menstrual cycle and hormonal factors (Erdman contribute in part to the applicability and validity of carote- 2005), other host factors and, presumably, genetic variability. noids as dietary biomarkers. For example, interactions have Their intestinal absorption can range from less than 10% to been noted between carotenoid intake and polymorphisms over 50% depending on the source and type (IARC in the oxidative stress genes MPO and COMT on the risk of 123 514 Hum Genet (2009) 125:507–525 (He et al. 2009) and between lycopene intake should be treated with much caution, and may even be an and XRCC1 gene polymorphisms on the risk of prostate underlying reason for conXicting literature reports relating cancer (Goodman et al. 2006). But, it is unknown whether sugars intake to cancer risk (Key et al. 2002). these gene-nutrient interactions may aVect carotenoid Without an adequate biomarker of sugar intake, it is metabolism, their antioxidant functionality or modify the diYcult to judge the extent of unreliability or bias of sugar validity of plasma carotenoids as dietary biomarkers of fruit intake from food records or questionnaires. Recently, a pre- and vegetable intake in diVerent populations. Similar to dictive biomarker for sugar intake has been identiWed many other nutrients, more research is required to better (Tasevska et al. 2008). In volunteers consuming their nor- understand the underlying genetic variability in carotenoid mal diet, about 100 mg of sucrose and fructose in 24 h metabolism. It should be noted that genetic variants associ- urine samples, predicts an intake of about 200 g of total ated with biomarker levels are applicable to the principle of sugars intake (r =0.77, P < 0.001) (Tasevska et al. 2008). Mendelian randomization, where an observed disease risk The sucrose measured in urine is derived from dietary association of the genetic variant may strongly suggest that sucrose that has escaped enzymatic hydrolysis in the small any observed cancer risk association with the relevant bio- intestine and is excreted from the general circulation, while marker is not confounded by other lifestyle or dietary vari- the fructose in the urine represents a small fraction of die- ables (Hunter 2006). However, so little is known about tary fructose and fructose derived endogenously from genetic variants associated with dietary biomarker levels hepatic metabolism of sucrose (Tasevska et al. 2008). that the principle of Mendelian randomization is not often Results from subjects housed in metabolic suites and fed exploited in diet-biomarker-cancer studies (Schatzkin et al. controlled diets suggest that BMI does not aVect the valid- 2009). ity of urinary sugars as a biomarker of sugar intake (Joosen et al. 2008), which is not well estimated from dietary assessments in obese individuals (Bingham et al. 2007). Biomarkers of energy, energy yielding nutrients Thus, these predictive biomarkers may be utilized for the and dietary Wber purposes of dietary questionnaire validation or as biomark- ers of exposure in the overall population. Nevertheless, One of the most important aspects of diet is the measure- these biomarkers may be prone to some of the issues dis- ment of total energy and estimation of energy yielding cussed earlier in terms of biomarker validity and interpreta- nutrients. As mentioned above, satisfactory recovery bio- tion, i.e. potential inXuences of sample collection, treatment markers exist for total energy (doubly labeled water) and and possible deterioration with long term storage. In addi- total protein (urinary nitrogen), but currently no recovery or tion, it has recently been suggested that variability in genes predictive biomarkers have been identiWed for total fat, or controlling bitter/sweet taste reception may alter glucose carbohydrate. Similarly, the accurate assessment of dietary homeostasis and insulin metabolism (Dotson et al. 2008) Wber would be enhanced if a recovery or predictive bio- and that sugar consumption may be modulated to some marker existed for its intake. degree by genetic variation in the glucose transporter type 2 (GLUT2) gene (Eny et al. 2008). It remains to be deter- Predictive biomarker of sugar intake mined whether diVerences in sugar intake, absorption or metabolism due to potential genetic control or as of yet Sugars, in the form of monosaccharides (e.g. glucose, fruc- unknown gene-diet/nutrient or gene-gene interactions may tose) and disaccharides (e.g. sucrose, lactose) are important aVect the validity of urinary sugars as predictive biomark- contributors to total energy intake. Estimates suggest that ers across diVerent populations with varying genetic back- they may supply as much as 22% of total adult energy grounds, dietary habits and lifestyles. intake in US adults about 8–20% in the European Commu- nity (Gibney et al. 1995). The food sources of sugars are Blood lipids and fatty acid proWles diverse, and much sugar in human diets is often derived from hidden sources and from processed foods. For these Like urinary sugars, some blood lipids may also be deemed reasons, the intake of total and individual sugars is very as predictive biomarkers for the intake of dietary fats and diYcult to assess. This is complicated by the fact that obese possibly dietary Wber (Bingham et al. 2008). In a recent individuals may underestimate their usual intakes of total publication, plasma levels of low density cholesterol (LDL) energy and sugar and fat containing foods (Bingham et al. were positively associated with the intake of saturated fats, 1995). Indeed, the reliability of dietary reports of obese and inverse associations were shown between plasma high individuals has been questioned via comparisons with sev- density cholesterol (HDL) and intake of dietary carbohy- eral dietary biomarkers, including sugars (Bingham et al. drate, and between plasma triglycerides (TG) and consump- 2007). Thus, intake information from food intake surveys tion levels of dietary Wber (Bingham et al. 2008). Plasma 123 Hum Genet (2009) 125:507–525 515 cholesterol and TG levels may to some extent be indicative of HDL, LDL and TG applied as dietary biomarkers in of level of dietary Wber intake, but the Wndings to date are diVerent populations, but this has not been fully addressed conXicting (Hunninghake et al. 1994; Panlasigui et al. in the literature to date. 2003; Sonnenberg et al. 1996; Truswell 1995; Zunft et al. Some studies have considered the fatty acid composition 2003) and some Wndings suggest that some of the inter- of adipose tissues or erythrocyte/plasma phospholipid fatty individual variation in the plasma lipoprotein response to acid proWles as biomarkers of dietary fats from various dietary Wber may be attributable to variation in the fatty foods. This concept is reviewed in great detail elsewhere acid binding protein 2 gene (Hegele 1998). More recent (Hodson et al. 2008; Poppitt et al. 2005). These biomarkers observations suggest that the correlation may also be can be aVected by a number of diVerent dietary and lifestyle dependent on the type and quality of the dietary assessment factors as well as endogenous fatty acid synthesis and com- instrument used to determine dietary Wber intake (Bingham plex fatty acid biochemistry. But, any eVects of genetic var- et al. 2008). In other words, consideration of genetic vari- iability in these pathways, direct/indirect determinants of ability is as important as the choice of the type of dietary fat absorption, or gene-diet/nutrient or gene–gene inter- assessment instrument to use. For now, these measures may actions on fatty acid proWles as dietary biomarkers is largely Wnd applicability at ecological levels for comparisons of unexplored (Fig. 3). For example, fatty acid proWles may be diVerences in dietary patterns, but their use as dietary bio- altered by variations in genes encoding for enzymes in the markers in other than very controlled settings still requires elongase/desaturase pathway of n-3 and n-6 fatty acid much validation. metabolism (Baylin et al. 2007) or by interactions between Results from carefully controlled metabolic settings or high intake of dietary fat, obesity and variations in fatty cross sectional studies suggest that saturated fat intake may acid binding protein 2 gene which may result in a modula- be a main determinant of blood cholesterol levels (Hegsted tion of insulin metabolism (Weiss et al. 2002). As another et al. 1993). It has been recently observed that plasma LDL example, a recent animal study has shown that disruption of level is strongly associated with total saturated fat intake as hepatic P450 reductase activity can cause diVerential gene well as percentage energy from total saturated fat (Wu et al. expression resulting in changes in fatty acid metabolism 2007). It is well known that LDL levels can be aVected by with system wide eVects (Mutch et al. 2007). From a meta- many diVerent factors including diet, lifestyle and variation bolomic proWling point of view, it has recently been shown in a number of genes (Burnett and Hooper 2008). Some that genetic variants of FADS1 and LIPC induce diVer- examples from diVerent populations include a modiWed ences in metabolites related to the long chain fatty acid association of: (a) polyunsaturated fatty acid (PUFA) intake metabolism pathways in which the coded enzymes are and HDL levels by polymorphisms in the APOA1 (Ordovas active (Gieger et al. 2008). Although far from being conclu- et al. 2002a), hepatic lipase (LIPC) (Ordovas et al. 2002b), sive, these examples indicate that genetic factors that aVect TNF (Fontaine-Bisson et al. 2007) and NFKB1 (Fontaine- fatty acid metabolism may also aVect the utility and appli- Bisson et al. 2009) genes, (b) PUFA intake and TG levels cation of these compounds as dietary biomarkers— by polymorphisms in the PPAR gene (Tai et al. 2005), and although this remains to be validated. Nevertheless, there (c) alcohol intake and LDL levels by polymorphisms in the are strong suggestions that levels of certain fatty acids in APOE locus (Corella et al. 2001). In addition, genetic vari- the serum or plasma phospholipids may reXect medium- ation in the NPC1L1 gene has been shown to aVect sterol term intake (weeks to months) of various foods (Riboli absorption from the gastro-intestinal tract (Fahmi et al. et al. 1987) and can correlate with habitual intake of spe- 2008) and may also have consequences in the induction of ciWc dietary fats or fatty acids in diVerent populations sterol responsive genes in the liver, particularly the gene (Kobayashi et al. 2001; Saadatian-Elahi et al. 2009; Sasaki that codes for LDL receptor thus potentially modulating et al. 2000; Wolk et al. 2001). In general, correlations with LDL levels, mainly in terms of dietary sterol intake (Zhao diet tend to be better for fatty acids that are either Wsh oil et al. 2008). Furthermore, the T-del variant of the FADS2 derived or not produced endogenously. gene (codes for the delta-6 desaturase enzyme in the elong- ase/desaturase pathway of n-3 and n-6 fatty acid metabo- lism) has been associated with higher plasma TG and lower Biomarkers of meat and meat products eicosapentanoic acid concentrations (Baylin et al. 2007). The genetic modulation of such relationships is likely inter- Biomarkers for meat are important in cancer research active between several genes, as has been observed for because meat has been linked with cancer risk at a variety plasma TG and polymorphisms of the LIPC and APOE of sites (World Cancer Research Fund 2007). Meats may genes (Wood et al. 2008). It is clear that inconsideration of contain various compounds such as 3-methyl histidine gene-nutrient and gene-gene interactions as well as their (formed in muscle breakdown and released and rapidly metabolic consequences may aVect the validity of measures excreted in urine) or creatinine, that may serve as possible 123 516 Hum Genet (2009) 125:507–525 predictive biomarkers of total meat consumption, as was components. Diet, which is a very complex mixture of suggested in some early research (Bingham 1987). More compounds, and single nutrients may both aVect or modu- recent metabonomic studies have identiWed a variety of late biological systems diVerently and their associations metabolites, including creatine, carnitine, and trimethyla- with disease risk are multi-faceted. The current understand- mine-N-oxide as biomarkers for meat based diets (Stella ing of dietary biomarkers is limited by a very incomplete et al. 2006), and have shown diVerences in metabonomic comprehension of how genetics, diet and nutrients interact proWles of meat protein or vegetable protein based diets in to aVect metabolism. In this regard, there is not only a need various populations (Holmes et al. 2008a). In some recent to better genetically characterize populations in epidemio- epidemiologic studies, the meat-disease risk association has logic studies but also to integrate this knowledge with high been more focused on the intake of red and processed quality dietary assessment methodology, established/tradi- meats. This is true for diabetes and several cancers, particu- tional dietary biomarkers (as detailed above) as well as larly those of the gastro-intestinal tract (Cross et al. 2007; newer dietary biomarkers which can provide a more com- Gonzalez et al. 2006; Norat et al. 2005; Vang et al. 2008) prehensive or holistic assessment of dietary exposures, die- where gene-diet interactions have been shown between red tary/nutrient patterns and dietary changes. Thus, integrative meat intake and polymorphisms of cytochrome P450 approaches are required that encompass genetic and non- (CYP) genes (Kury et al. 2007) and other genes involved in genetic factors and that can capture dietary and environ- nucleotide excision and mismatch repair (Joshi et al. 2009). mental exposures. In this respect, the emerging science of Randomized controlled diet intervention studies with metabonomics is reviewed below with an emphasis on its human volunteers show that consumption of red and pro- application to the study of the metabolism of dietary com- cessed meats (but not white meat) can increase endogenous pounds and observation of population and individual diVer- gastro-intestinal formation of N-nitroso compounds ences in dietary exposures. (NOCs) (Bingham et al. 2002), many of which are known carcinogens (SaVhill et al. 1985). In fact, red meats are rich in heme which is the likely responsible agent for the endog- Metabonomics: the identiWcation of metabolic proWles enous NOCs formation (Cross et al. 2003; Kuhnle et al. as dietary biomarkers 2007). DNA adducts speciWc for endogenous NOCs have been found in cells exfoliated from volunteers with a sig- A broad biological representation of gene function, gene niWcant increase following a high red meat diet (Lewin regulation and gene–gene and gene-diet–lifestyle–environ- et al. 2006). In epidemiologic settings, the direct measure- mental interactions can be observed using some newly ment of NOCs is logistically diYcult because it requires developed ‘omic’ technologies such as transciptomics fecal samples, which are not routinely collected. Currently, (study of gene expression at the RNA level), it is only possible to estimate the endogenous formation of (study of protein structure and function), and metabonom- NOCs using data from diet intervention studies with volun- ics (study of metabolic responses). Metabonomics, which is teers. Thus, data on endogenous NOC formation, as a bio- the focus of this section, closely reXects the visible pheno- marker of red and processed meat intake, gathered from type of interest and is deWned as ‘the quantitative measure- carefully controlled intervention trials, can be used to for- ment of the dynamic multi-parametric metabolic response mulate an estimate of endogenous NOC exposure which of living systems to patho-physiological stimuli or genetic can then be extrapolated to epidemiologic studies. One such modiWcation (Nicholson et al. 1999). study observed an association between estimated endoge- Metabonomics is often confused with the Weld of meta- nous NOC production and risk of non-cardia gastric cancer bolomics. Although they both utilize similar analytical which was signiWcant, whereas exposure to exogenous tools with pattern recognition techniques, metabolomics NOC was not (Jakszyn et al. 2006). involves the comprehensive analysis of all measurable It is clear from the above sections that the multi-factorial metabolite concentrations under a given set of conditions, complexities of dietary exposures should not be—but often whereas metabonomics refers more to a systems biology are—simplistically assessed. The science of dietary bio- approach and aims to measure the metabolic response to markers requires much further research in order to better biological, genetic, environmental or dietary stimuli (Fiehn understand, use and interpret existing biomarkers. How- 2002; Nicholson and Lindon 2008). Of particular relevance ever, the development of new dietary biomarkers should to both Welds are technologies such as high resolution also be a subject of intense research. This involves develop- nuclear magnetic resonance (NMR) spectroscopy, mass ing a much clearer understanding of food chemistry, inter- spectrometry and a range of other hyphenated technologies, actions between nutrients, the role of genetic variation in all of which are platforms that collectively support the com- modulation of nutrient absorption and metabolism, as well prehensive analysis of metabolite proWles in biological as mechanisms of action of nutrients or other bioactive food Xuids or tissue samples. Coupled with multivariate statistical 123 Hum Genet (2009) 125:507–525 517 analysis, these platforms enable the generation of character- shown that increased intake of Wsh is associated with a istic patterns of metabolites that reXect the metabolic phe- reduction in risk (Kimura et al. 2007; notype or status of the organism. The metabolic phenotype Norat et al. 2005), while higher consumption of some can be used either to explore molecular mechanisms of meats is associated with an increased risk (Chao et al. 2005; chronic disease etiology or to determine the metabolic con- Cross et al. 2007; Norat et al. 2005). It would be very inter- sequences of dietary changes or diVerent dietary/nutrient esting to validate these observed cancer risk associations patterns. using metabonomic proWling in the same populations. This Even a simple diet contains a complex matrix of foods, is potentially possible, because metabonomic studies have which are a composite of multiple nutrients and which shown distinct proWles for higher Wsh intake in a free living undergo digestion through diverse pathways involving co- population (Zuppi et al. 1997), and for a diet rich in animal metabolism by the host and its symbiotic gut microbiota. proteins in a controlled cross-over feeding study of three Macro- and micro-nutrients are intimately involved in and diVerent protein source diets (Stella et al. 2006). can potentially modulate almost all metabolic pathways. In fact, a key application of metabonomic methodology Indeed, dietary imbalances and diVerences in nutrient is to repositories of biological samples provided by popula- intake and dietary patterns may result in many metabolic tion based studies. The data can then be combined with changes or disturbances. Nutrients also interact with many genetic information as well as dietary and other lifestyle diVerent genes in association with environmental factors. In exposure data. However, the typically large sample sizes of this light, metabonomics can be a valuable tool to consider modern epidemiologic studies and requirements for rapid individual or population responses to dietary exposures or sample characterization mean that the metabonomic analyt- diVerences in dietary patterns in order to better understand ical platforms need to be adapted for high throughput. diet-disease associations. In other words, certain metabo- Indeed, there are constant improvements in the automation nomic proWles may potentially be considered as dietary bio- and miniaturization of these analytical platforms resulting markers. In fact, this is a cornerstone of the emerging Weld in increased capacity and the use of such platforms in pre- of nutritional metabonomics (or nutri-metabonomics) liminary metabonomic characterization of population stud- which aims to combine the insights gained from metabo- ies has shown great promise, despite the fact that these nomic studies pertaining to health and disease with the cur- studies were not originally designed to include metabolic rent understanding of how diet may aVect those states. This proWling, and that they are confounded by the presence of Weld is still in its infancy and to date studies have largely extensive inter-subject metabolic variation due to greater focused on the experimental design and the various techni- diversity in genetic factors, environmental factors and cal challenges in the measurement of the subtle diVerences health status compared to controlled clinical or animal stud- of dietary factors and the exerted multiple, pleiotropic ies. Metabolic proWling strategies have recently been eVects on metabolism (Gibney et al. 2005; Rezzi et al. adopted in a retrospective manner to improve information 2007; Walsh et al. 2006). Nevertheless, examples of tar- recovery from established large-scale populations studies geted approaches at nutri-metabonomic proWling exist in such as INTERSALT (INTERSALT Co-operative the current literature and include the evaluation of metabo- Research Group 1988) and the INTERMAP study of nutri- nomic diVerences according to the type of fats consumed ents and blood pressure (Stamler et al. 2003) where a clear (Watson 2006), consideration of the adequacy of amino diVerentiation of urinary metabolic signatures was achieved acid intakes (Noguchi et al. 2003), green versus black tea within diVerent western (USA and UK) and East Asian consumption (Van Dorsten et al. 2006), and animal studies (China and Japan) populations (Holmes et al. 2008a, b). In looking at increased longevity associated with caloric the INTERMAP study, NMR spectroscopy was applied to restriction (Wang et al. 2007) or the reversal of the bio-banked urine samples and a map of metabolic pheno- metabolic consequences of a high fat diet by dietary supple- types was established that reXected observed diVerences in mentation of catenin (Fardet et al. 2008). In other nutri- diet and cardiovascular disease risk. A list was produced of metabonomic studies, the consumption of diets with the top twenty discriminatory metabolites for all combina- varying phytochemical contents has been shown to alter tions of country pair-wise comparisons included hippurate, metabolic proWles (Walsh et al. 2007) and the ingestion of trimethlamine-N-oxide, amino acids and ethanol, which soy has also been shown to result in an identiWable derived from a combination of endogenous, gut microbial, metabolic proWle (Solanky et al. 2003) and to alter several dietary and xenobiotic sources. The relationships between metabolic pathways related to energy metabolism (Solanky selected quantiWed metabolites with the recorded dietary et al. 2005). intakes of macro and micronutrients and also blood pres- This technology may also be used to validate Wndings sure values were investigated and identiWed novel candi- from observational studies. For example, several studies date urinary biomarkers that were positively (e.g. alanine) using data derived from dietary intake assessments have or inversely associated (e.g. formate, hippurate) with blood 123 518 Hum Genet (2009) 125:507–525 pressure. The INTERMAP metabonomic investigations bolite classes with both high accuracy and sensitivity. It is also showed metabolic distinctions within the Asian popu- an integrative approach, encompassing genetic, dietary, lation samples from China and Japan and even within lifestyle and environmental factors. In the foreseeable China between the northern (Beijing, Shanxi) and southern future, an integration of metabonomic and whole genomic (Guangxi) centers. In addition, the study metabolically platforms can be envisioned for a detailed exploration of diVerentiated between Japanese and American populations genetic variability, dietary patterns/changes and metabo- living in Honolulu, with similar dietary, lifestyle and envi- lism. However, it must be cautioned that in order to be ronmental exposures to American populations, from Japa- meaningfully applied, the identiWcation of distinct meta- nese living in Japan, whereas the Japanese–Americans bolic proWles as dietary biomarkers must be validated in (Honolulu) were not signiWcantly metabolically diVerenti- diVerent populations under various experimental protocols ated from other American populations (Holmes et al. and using standardized methodology. Much progress is 2008a, b). Thus, it is possible that at an ecological level, being made in this respect (Castle et al. 2006). If all this can utilization of metabonomic proWling as biomarkers of diet be accomplished, in the foreseeable future Metabolome can serve to diVerentiate populations with similar genetic Wide Association studies may highlight new dietary bio- backgrounds but varying dietary and lifestyle habits and markers and provide novel insights into chronic disease eti- environmental exposures. ology as well as the gene-diet-lifestyle-disease connection. Another application of nutritional metabonomics could be to address the important role of the colonic microbiota in nutrient metabolism and possible implications for disease Statistical considerations to integrate dietary risk associations. The diverse and active microbial popula- assessments with biomarker data tions within the colonic environment are able to metabolize a wide range of dietary components and contribute to The accuracy of dietary questionnaire measurements is energy production, enterohepatic circulation of nutrients often evaluated in validation studies, where dietary ques- and bioactive compounds (as well as toxins) and metabolic tionnaire measurements are compared to more detailed and activation of some compounds. Historically the gut reliable reference (R) measurements (e.g. 24-h dietary microbes and the host organisms were largely considered recalls, food diaries, etc.) (Riboli and Kaaks 2000; Stram individually but metabonomics can enable the study of their et al. 2000; Thompson et al. 1997). These approaches are complex interactions and reveal changes that will aid the based on two important statistical assumptions: (a) errors in understanding of dietary modulation of the gut microbes questionnaire and reference measurements are independent, and the consequences for disease risk (Nicholson et al. (b) reference measurements follow the classical measure- 2005). For example, in the INTERMAP population, diVer- ment error structure, i.e. errors are strictly random   ences were observed in several metabolites known to be (R = T + R, where T indicates unknown true intake and R produced by gut microbes or by mammalian–microbial co- models random error in R measurements). In practice, these metabolism, including phenylacetylglutamine and hippu- assumptions are likely to be violated for several important rate, for every population investigated (Holmes et al. reasons: the tendency of study subjects to consistently 2008a). In animal models, metabolic proWling approaches under- or over-report speciWc food intakes, the limited per- have shown that gut microbiota play an active role in insu- formance of self-reported dietary assessment instruments to lin resistance (Dumas et al. 2006), which can be modulated accurately capture a large spectrum of dietary diversity, and by diet (Minich and Bland 2008). Thus, it is feasible that the presence of sizeable systematic measurement errors in the integration of information on dietary intake, lifestyle the reference assessment measurements. Thus, when possi- and metabonomic proWling may help to identify metabolic ble, biomarkers can be incorporated into the validation and/ signatures that are key to diet related disease mechanisms. or calibration of dietary assessment methods as objective Future studies should further attempt to utilize metabo- measures of intake, or in other words measurements whose nomic proWles related to diVerent gut bacteria or their meta- errors are assumed to be independent from self-reported bolic activities as biomarkers to observe eVects of dietary dietary measurements. modulation or diVerences in dietary regimes, intakes and In a validation study questionnaire, reference and bio- patterns. For example, the metabonomic identiWcation of marker measurements are linearly related to unknown true diVerent short chain fatty acid proWles arising from diVerent intake levels by means of latent factor modeling (Bentler diets or dietary patterns may provide an estimate of the and Weeks 1980; Bollen 1989). The error correlation degree of colonic fermentation and butyrate exposure, or between self-reported dietary assessments can be estimated conversely production of toxic metabolites (O’Keefe 2008). when replicate measurements of a biomarker are available, Metabonomics enables simultaneous measurement of or when more than one dietary factor is being evaluated, for large numbers of metabolites across a wide range of meta- example in a multivariate study (Kaaks et al. 2002). Moreover, 123 Hum Genet (2009) 125:507–525 519 recovery biomarkers make it possible to Wt models where respect, genetic information accounting for the eVect of the classical measurement error structure for self-reported various gene-diet/nutrient or gene-gene interactions on bio- reference measurements can be relaxed (Day et al. 2001; marker levels or disease risk may also be incorporated, Kipnis et al. 2003; Spiegelman et al. 2005). Unfortunately, although this has not yet been attempted in practice. as described in detail in the previous sections, only a few Complex statistical modeling to integrate dietary assess- recovery biomarkers are available (Kaaks et al. 2002). ments with biomarker data for the purposes of shedding To overcome the cost and availability limitations of light onto the measurement error structure of self-reported recovery biomarkers, latent factor models using two con- dietary estimates is being used more readily in modern centration or replacement dietary biomarkers have been nutrition epidemiology. Calibration and validation studies developed, where the Wrst biomarker is assumed to provide are being built a priori into the design of large cohorts estimates of unknown true intake, while the other bio- allowing for statistical error correction later down the line. marker is a biologic correlate only and not directly measur- Nevertheless, although these techniques are themselves at ing the variable of interest (Fraser et al. 2005). For best only approximations, with improving dietary assess- example, vitamin E measured in the blood can be comple- ment methodology along with greater availability and mented by a measure of blood folic acid in order to validate selection of dietary biomarkers and more knowledge about questionnaire measurements of folic acid. However, these gene-diet/nutrient interactions, statistical measurement models have raised concerns about the speciWcity of addi- error correction will allow for the production of more accu- tional dietary biomarkers to measure a given food or nutri- rate evidence on the relationship between dietary factors ent (Kaaks and Ferrari 2006). In addition, it must be and the risk of major chronic conditions. assumed that error correlations between diVerent dietary biomarkers are close to zero. But, this assumption may not generally hold in the case of between-person variation in Overall summary biomarker levels among people with the same dietary intake (Rosner et al. 2008). To overcome this latter limita- This review has outlined the major advantages and short- tion, two very diVerent biomarkers could be favored, for falls of dietary biomarkers, with a particular emphasis on example PUFA together with -carotene measured in blood the established/traditional biomarkers of fruit and vegetable to validate PUFA questionnaire measurements. This solu- intake, i.e. vitamin C and carotenoids, and a consideration tion diminishes the probability of non-zero error correla- of potential biomarkers for other food groups such as fats, tions between biomarkers, though it introduces issues of carbohydrates and meats. Aside from the many analytical, large conWdence intervals in the parameter estimates, unless environmental and lifestyle factors that can modulate die- studies of very large in size are implemented. tary biomarkers (Fig. 2), there is a strong potential for Novel methods have been proposed to complement self- gene-diet/nutrient or gene–gene interactions to aVect the reported measurements with objective biomarkers of validity of dietary biomarker measures (Fig. 3). However, dietary intake in the evaluation of the association between the contribution of these interactions to the application, dietary exposure and risk of chronic disease (Prentice et al. measurement and interpretation of dietary biomarkers 2002). This statistical model allows the potential depen- remains to be elucidated. A better understanding of these dence of the systematic component of measurement errors interactions is an urgent need that can be addressed eVec- upon individual characteristics (e.g. body mass, age, ethni- tively by multidisciplinary collaboration and the combined city) to be estimated (Sugar et al. 2007). To date, these eVorts of nutritional scientists and geneticists. This can lead methods have been proposed for recovery biomarkers only to better application of dietary biomarkers for exposure but in theory additional types of biomarkers, for example assessment in diVerent populations (e.g. based on genetic metabonomic proWling, or even possibly genetic informa- background) either for disease risk estimation in observa- tion, may also be incorporated. In the same framework, tional studies or for assessing the eYcacy of dietary inter- innovative research could focus on the use of recovery and ventions. Much of the existing information on dietary concentration biomarkers, possibly addressing departures biomarker validity and potential gene-diet/nutrient inter- from the classical measurement error model for the refer- actions has been derived from North American or European ence measurement, and relaxing assumptions of zero corre- populations and needs to be substantiated across many lations between self-reported dietary assessments. For this diVerent populations and groups. The overall limitations of purpose, in a Bayesian framework (Richardson and Gilks the speciWc dietary biomarkers discussed here apply to 1993), a measurement error model to relate observed quan- some extent to the many other dietary biomarkers in current tities to unknown true intakes could integrate a risk model use. Careful attention to factors that can aVect dietary bio- where the association between true intake and the risk of marker measures can help to enhance their application and disease could be quantiWed (Ferrari et al. 2008). In this interpretation. With the appropriate study design and 123 520 Hum Genet (2009) 125:507–525 methodology, the assessment of dietary intake and quantiW- Barua AB, Kostic D, Olson JA (1993) New simpliWed procedures for cation of exposure via the use of dietary biomarkers can the extraction and simultaneous high-performance liquid chro- matographic analysis of retinol, tocopherols and carotenoids in provide very meaningful results. human serum. J Chromatogr 617:257–264 Out of the many thousands of chemical compounds pres- Bates CJ (1994) Plasma vitamin C assays: a European experience. EC ent in foods, not all bio-active dietary components have FLAIR concerted action no. 10: Micronutrient measurement, been identiWed or characterized and many could potentially absorption and status. Int J Vitam Nutr Res 64:283–287 V Baylin A, Ruiz-Narvaez E, Kraft P, Campos H (2007) Alpha-linolenic a ect gene expression, modulate cellular/physiologic pro- acid, delta 6-desaturase gene polymorphism, and the risk of non- cesses and be associated with disease risk. This review has fatal . Am J Clin Nutr 85:554–560 highlighted the need for the development of new dietary bio- Bentler PM, Weeks DG (1980) Linear structural equations with latent markers. In this respect, the emerging Weld of nutritional variables. Psychometrika 45:289–308 W W W Benzie IF (1999) Vitamin C: prospective functional markers for de n- metabonomics was outlined. Speci c metabolic pro les, ing optimal nutritional status. Proc Nutr Soc 58:469–476 whether in distinct populations or even individuals, may be Benzie IF, Strain JJ (1997) Acute post-ingestion changes in plasma utilized as dietary biomarkers which are speciWc to various ascorbic acid concentration: relationship to the dose and to exist- dietary/nutrient intake patterns or dietary changes. With the ing body stores. Nutr Res 17:187–190 Bingham S (1987) The dietary assessment of individuals: methods, consideration of the potential importance of gene-diet/nutri- accuracy, new techniques and recommendations. Nutr Abstr Rev ent interactions in the application and validity of dietary bio- 57:705–742 markers, the inclusion of a more holistic approach to dietary Bingham SA (2002) Biomarkers in nutritional epidemiology. Public biomarker identiWcation may be very informative for assess- Health Nutr 5:821–827 Bingham SA (2003) Urine nitrogen as a biomarker for the validation ing gene-diet-lifestyle-disease associations. This technology of dietary protein intake. J Nutr 133(Suppl 3):921S–924S can only get better and cheaper, and combined with mea- Bingham SA, Cassidy A, Cole TJ, Welch A, Runswick SA, Black AE, sures of established/traditional dietary biomarkers and stan- Thurnham D, Bates C, Khaw KT, Key TJ (1995) Validation of dardized validated dietary intake assessments it can evaluate weighed records and other methods of dietary assessment using V the 24 h urine nitrogen technique and other biological markers. Br dietary exposures and metabolic di erences arising from J Nutr 73:531–550 various dietary regimes, patterns or interventions. There is a Bingham SA, Hughes R, Cross AJ (2002) EVect of white versus red need to bring together high quality dietary intake informa- meat on endogenous N-nitrosation in the human colon and further tion (i.e. maximal information with minimal measurement evidence of a dose response. J Nutr 132:3522S–3525S W Bingham S, Luben R, Welch A, Tasevska N, Wareham N, Khaw KT errors), traditional dietary biomarkers and metabolic pro les (2007) Epidemiologic assessment of sugars consumption using as dietary biomarkers with information about gene-diet/ biomarkers: comparisons of obese and nonobese individuals in nutrient or gene-gene interactions. This information can also the European prospective investigation of cancer Norfolk. Cancer be integrated with data from other relevant technologies, Epidemiol Biomarkers Prev 16:1651–1654 Bingham S, Luben R, Welch A, Low YL, Khaw KT, Wareham N, Day N namely other ‘omic’ platforms and whole genome scans, in (2008) Associations between dietary methods and biomarkers, and the same population groups or individuals to provide a glo- between fruits and vegetables and risk of ischaemic heart disease, in bal overview of diet-lifestyle-disease associations and the the EPIC Norfolk Cohort Study. Int J Epidemiol 37:978–987 many intervening genetic and metabolic interactions. None Boeing H, Bohlscheid-Thomas S, Voss S, Schneeweiss S, Wahrendorf J (1997) The relative validity of vitamin intakes derived from a of these ambitious possibilities can be achieved without the food frequency questionnaire compared to 24-hour recalls and development of novel statistical methodologies and new biological measurements: results from the EPIC pilot study in database management techniques to assess, combine, cali- Germany. European Prospective Investigation into Cancer and brate, analyze and make sense of the generated data. With Nutrition. Int J Epidemiol 26(Suppl 1):S82–S90 Bollen KA (1989) A new incremental Wt index for general structural all of these tools at its disposal in the near future, nutritional equation models. Sociol Methods Res 17:303–316 epidemiology should provide even further insight into the Brubacher D, Moser U, Jordan P (2000) Vitamin C concentrations in association of dietary, lifestyle and environmental exposures plasma as a function of intake: a meta-analysis. Int J Vitam Nutr and the risk of cancer and other chronic diseases. Res 70:226–237 Burnett JR, Hooper AJ (2008) Common and rare gene variants aVect- ing plasma LDL cholesterol. Clin Biochem Rev 29:11–26 Castle AL, Fiehn O, Kaddurah-Daouk R, Lindon JC (2006) Metabolo- mics Standards Workshop and the development of international References standards for reporting metabolomics experimental results. Brief Bioinform 7:159–165 Albanes D, Virtamo J, Taylor PR, Rautalahti M, Pietinen P, Heinonen Chao A, Thun MJ, Connell CJ, McCullough ML, Jacobs EJ, Flanders OP (1997) EVects of supplemental beta-carotene, cigarette smok- WD, Rodriguez C, Sinha R, Calle EE (2005) Meat consumption ing, and alcohol consumption on serum carotenoids in the alpha- and risk of colorectal cancer. JAMA 293:172–182 tocopherol, beta-carotene cancer prevention study. Am J Clin Ching SY, Prins AW, Beilby JP (2002) Stability of ascorbic acid in se- Nutr 66:366–372 rum and plasma prior to analysis. Ann Clin Biochem 39:518–520 Al-Delaimy WK, Natarajan L, Sun X, Rock CL, Pierce JJ (2008) Reli- Chiplonkar SA, Agte VV, Mengale SS, Tarwadi KV (2002) Are life- ability of plasma carotenoid biomarkers and its relation to study style factors good predictors of retinol and vitamin C deWciency power. Epidemiology 19:338–344 in apparently healthy adults? Eur J Clin Nutr 56:96–104 123 Hum Genet (2009) 125:507–525 521

Chung WY, Chung JK, Szeto YT, Tomlinson B, Benzie IF (2001) multicenter study with exposures measured with error: the EPIC Plasma ascorbic acid: measurement, stability and clinical utility study. Stat Med 27:6037–6054 revisited. Clin Biochem 34:623–627 Fiehn O (2002) Metabolomics—the link between genotypes and phe- Cooney RV, Franke AA, Hankin JH, Custer LJ, Wilkens LR, Harwood notypes. Plant Mol Biol 48:155–171 PJ, Le ML (1995) Seasonal variations in plasma micronutrients Fontaine-Bisson B, Wolever TM, Chiasson JL, Rabasa-Lhoret R, and antioxidants. Cancer Epidemiol Biomarkers Prev 4:207–215 Maheux P, Josse RG, Leiter LA, Rodger NW, Ryan EA, Connelly Corella D, Tucker K, Lahoz C, Coltell O, Cupples LA, Wilson PW, PW, Corey PN, El-Sohemy A (2007) Genetic polymorphisms of Schaefer EJ, Ordovas JM (2001) Alcohol drinking determines the tumor necrosis factor-alpha modify the association between die- eVect of the APOE locus on LDL-cholesterol concentrations in tary polyunsaturated fatty acids and fasting HDL-cholesterol and men: the Framingham OVspring Study. Am J Clin Nutr 73:736– apo A-I concentrations. Am J Clin Nutr 86:768–774 745 Fontaine-Bisson B, Wolever TM, Connelly PW, Corey PN, El-Sohe- Crews H, Alink G, Andersen R, Braesco V, Holst B, Maiani G, Ovesen my A (2009) NF-kappaB ¡49Ins/Del ATTG polymorphism mod- L, Scotter M, Solfrizzo M, van den BR, Verhagen H, Williamson iWes the association between dietary polyunsaturated fatty acids G (2001) A critical assessment of some biomarker approaches and HDL-cholesterol in two distinct populations. Atherosclerosis linked with dietary intake. Br J Nutr 86(Suppl 1):S5–35 (e-pub ahead of print, PMID: 19070859) Cross AJ, Pollock JR, Bingham SA (2003) Haem, not protein or inor- Fraser PD, Bramley PM (2004) The biosynthesis and nutritional uses ganic iron, is responsible for endogenous intestinal N-nitrosation of carotenoids. Prog Lipid Res 43:228–265 arising from red meat. Cancer Res 63:2358–2360 Fraser GE, Yan R (2007) A multivariate method for measurement error Cross AJ, Leitzmann MF, Gail MH, Hollenbeck AR, Schatzkin A, Sin- correction using pairs of concentration biomarkers. Ann Epidem- ha R (2007) A prospective study of red and processed meat intake iol 17:64–73 in relation to cancer risk. PLoS Med 4:e325 Fraser GE, Butler TL, Shavlik D (2005) Correlations between esti- Day N, McKeown N, Wong M, Welch A, Bingham S (2001) Epidemi- mated and true dietary intakes: using two instrumental variables. ological assessment of diet: a comparison of a 7-day diary with a Ann Epidemiol 15:509–518 food frequency questionnaire using urinary markers of nitrogen, Garcia-Bailo B, Toguri C, Eny KM, El-Sohemy A (2009) Genetic var- potassium and sodium. Int J Epidemiol 30:309–317 iation in taste and its inXuence on food selection. OMICS 13:69– Dehghan M, khtar-Danesh N, McMillan CR, Thabane L (2007) Is plas- 80 (PMID: 18687042) ma vitamin C an appropriate biomarker of vitamin C intake? A Gibney M, Sigman-Grant M, Stanton JL Jr, Keast DR (1995) Con- systematic review and meta-analysis. Nutr J 6:41 sumption of sugars. Am J Clin Nutr 62:178S–193S Dodd KW, Guenther PM, Freedman LS, Subar AF, Kipnis V, Midth- Gibney MJ, Walsh M, Brennan L, Roche HM, German B, van OB une D, Tooze JA, Krebs-Smith SM (2006) Statistical methods for (2005) Metabolomics in human nutrition: opportunities and chal- estimating usual intake of nutrients and foods: a review of the the- lenges. Am J Clin Nutr 82:497–503 ory. J Am Diet Assoc 106:1640–1650 Gieger C, Geistlinger L, Altmaier E, Hrabe de AM, Kronenberg F, Dotson CD, Zhang L, Xu H, Shin YK, Vigues S, Ott SH, Elson AE, Meitinger T, Mewes HW, Wichmann HE, Weinberger KM, Ad- Choi HJ, Shaw H, Egan JM, Mitchell BD, Li X, Steinle NI, Mun- amski J, Illig T, Suhre K (2008) Genetics meets metabolomics: a ger SD (2008) Bitter taste receptors inXuence glucose homeosta- genome-wide association study of metabolite proWles in human sis. PLoS ONE 3:e3974 serum. PLoS Genet 4:1000282 Dowell SA, Welch JL (2006) Use of electronic self-monitoring for Goni I, Serrano J, Saura-Calixto F (2006) Bioaccessibility of beta-car- food and Xuid intake: a pilot study. Nephrol Nurs J 33:271–277 otene, lutein, and lycopene from fruits and vegetables. J Agric Dumas ME, Barton RH, Toye A, Cloarec O, Blancher C, Rothwell A, Food Chem 54:5382–5387 Fearnside J, Tatoud R, Blanc V, Lindon JC, Mitchell SC, Holmes Gonzalez CA, Jakszyn P, Pera G, Agudo A, Bingham S, Palli D, E, McCarthy MI, Scott J, Gauguier D, Nicholson JK (2006) Met- Ferrari P, Boeing H, Del GG, Plebani M, Carneiro F, Nesi G, abolic proWling reveals a contribution of gut microbiota to fatty Berrino F, Sacerdote C, Tumino R, Panico S, Berglund G, Siman liver phenotype in insulin-resistant mice. Proc Natl Acad Sci USA H, Nyren O, Hallmans G, Martinez C, Dorronsoro M, Barricarte 103:12511–12516 A, Navarro C, Quiros JR, Allen N, Key TJ, Day NE, Linseisen J, Eck P, Erichsen HC, Taylor JG, Yeager M, Hughes AL, Levine M, Nagel G, Bergmann MM, Overvad K, Jensen MK, Tjonneland A, Chanock S (2004) Comparison of the genomic structure and var- Olsen A, Bueno-de-Mesquita HB, Ocke M, Peeters PH, Numans iation in the two human sodium-dependent vitamin C transport- ME, Clavel-Chapelon F, Boutron-Ruault MC, Trichopoulou A, ers, SLC23A1 and SLC23A2. Hum Genet 115:285–294 Psaltopoulou T, Roukos D, Lund E, Hemon B, Kaaks R, Norat T, Eny KM, Wolever TM, Fontaine-Bisson B, El-Sohemy A (2008) Ge- Riboli E (2006) Meat intake and risk of stomach and esophageal netic variant in the glucose transporter type 2 is associated with adenocarcinoma within the European Prospective Investigation higher intakes of sugars in two distinct populations. Physiol Into Cancer and Nutrition (EPIC). J Natl Cancer Inst 98:345–354 Genomics 33:355–360 Goodman M, Bostick RM, Ward KC, Terry PD, van Gils CH, Taylor Erdman JW Jr (2005) How do nutritional and hormonal status modify JA, Mandel JS (2006) Lycopene intake and prostate cancer risk: the bioavailability, uptake, and distribution of diVerent isomers of eVect modiWcation by plasma antioxidants and the XRCC1 geno- lycopene? J Nutr 135:2046S–2047S type. Nutr Cancer 55:13–20 Fahmi S, Yang C, Esmail S, Hobbs HH, Cohen JC (2008) Functional GorWne M, Lipshtat N, Freedman LS, Prentice RL (2007) Linear mea- characterization of genetic variants in NPC1L1 supports the surement error models with restricted sampling. Biometrics sequencing extremes strategy to identify complex trait genes. 63:137–142 Hum Mol Genet 17:2101–2107 Grace PB, Taylor JI, Low YL, Luben RN, Mulligan AA, Botting NP, Fardet A, Llorach R, Martin JF, Besson C, Lyan B, Pujos-Guillot E, Dowsett M, Welch AA, Khaw KT, Wareham NJ, Day NE, Bingham Scalbert A (2008) A liquid chromatography-quadrupole time-of- SA (2004) Phytoestrogen concentrations in serum and spot urine as Xight (LC-QTOF)-based metabolomic approach reveals new met- biomarkers for dietary phytoestrogen intake and their relation to abolic eVects of catechin in rats fed high-fat diets. J Res breast cancer risk in European prospective investigation of cancer and 7:2388–2398 nutrition-norfolk. Cancer Epidemiol Biomarkers Prev 13:698–708 Ferrari P, Carroll RJ, Gustafson P, Riboli E (2008) A Bayesian Graumlich JF, Ludden TM, Conry-Cantilena C, Cantilena LR Jr, multilevel model for estimating the diet/disease relationship in a Wang Y, Levine M (1997) Pharmacokinetic model of ascorbic 123 522 Hum Genet (2009) 125:507–525

acid in healthy male volunteers during depletion and repletion. Chapelon F, Boeing H, Schulz M, Linseisen J, Nagel G, Pharm Res 14:1133–1139 Trichopoulou A, Naska A, Oikonomou E, Berrino F, Panico S, He C, Tamimi RM, Hankinson SE, Hunter DJ, Han J (2009) A pro- Palli D, Sacerdote C, Tumino R, Peeters PH, Numans ME, Bue- spective study of genetic polymorphism in MPO, antioxidant sta- no-de-Mesquita HB, Buchner FL, Lund E, Pera G, Chirlaque MD, tus, and breast cancer risk. Breast Cancer Res Treat 113:585–594 Sanchez MJ, Arriola L, Barricarte A, Quiros JR, Johansson I, Hegele RA (1998) A review of intestinal fatty acid binding protein Johansson A, Berglund G, Bingham S, Khaw KT, Allen N, Key gene variation and the plasma lipoprotein response to dietary T, Carneiro F, Save V, Del GG, Plebani M, Kaaks R, Gonzalez components. Clin Biochem 31:609–612 CA (2006a) Plasma and dietary carotenoid, retinol and tocopherol Hegsted DM, Ausman LM, Johnson JA, Dallal GE (1993) Dietary fat levels and the risk of gastric adenocarcinomas in the European and serum lipids: an evaluation of the experimental data. Am J prospective investigation into cancer and nutrition. Br J Cancer Clin Nutr 57:875–883 95:406–415 Hodson L, SkeaV CM, Fielding BA (2008) Fatty acid composition of Jenab M, Riboli E, Ferrari P, Sabate J, Slimani N, Norat T, Friesen M, adipose tissue and blood in humans and its use as a biomarker of Tjonneland A, Olsen A, Overvad K, Boutron-Ruault MC, Clavel- dietary intake. Prog Lipid Res 47:348–380 Chapelon F, Touvier M, Boeing H, Schulz M, Linseisen J, Nagel Holmes E, Loo RL, Stamler J, Bictash M, Yap IK, Chan Q, Ebbels T, G, Trichopoulou A, Naska A, Oikonomou E, Krogh V, Panico S, De IM, Brown IJ, Veselkov KA, Daviglus ML, Kesteloot H, Ue- Masala G, Sacerdote C, Tumino R, Peeters PH, Numans ME, shima H, Zhao L, Nicholson JK, Elliott P (2008a) Human meta- Bueno-de-Mesquita HB, Buchner FL, Lund E, Pera G, Sanchez bolic phenotype diversity and its association with diet and blood CN, Sanchez MJ, Arriola L, Barricarte A, Quiros JR, Hallmans G, pressure. Nature 453:396–400 Stenling R, Berglund G, Bingham S, Khaw KT, Key T, Allen N, Holmes E, Wilson ID, Nicholson JK (2008b) Metabolic phenotyping Carneiro F, Mahlke U, Del GG, Palli D, Kaaks R, Gonzalez CA in health and disease. Cell 134:714–717 (2006b) Plasma and dietary vitamin C levels and risk of gastric Holst B, Williamson G (2008) Nutrients and phytochemicals: from cancer in the European Prospective Investigation into Cancer and bioavailability to bioeYcacy beyond antioxidants. Curr Opin Bio- Nutrition (EPIC-EURGAST). Carcinogenesis 27:2250–2257 technol 19:73–82 Joosen AM, Kuhnle GG, Runswick SA, Bingham SA (2008) Urinary Hunninghake DB, Miller VT, LaRosa JC, Kinosian B, Jacobson T, sucrose and fructose as biomarkers of sugar consumption: com- Brown V, Howard WJ, Edelman DA, O’Connor RR (1994) Long- parison of normal weight and obese volunteers. Int J Obes (Lond) term treatment of hypercholesterolemia with dietary Wber. Am J 32:1736–1740 Med 97:504–508 Joshi AD, Corral R, Siegmund KD, Haile RW, Le ML, Martinez ME, Hunter DJ (2006) The inXuence of genetic polymorphism. J Nutr Ahnen DJ, Sandler RS, Lance P, Stern MC (2009) Red meat and 136:2711S–2713S poultry intake, polymorphisms in the nucleotide excision repair IARC Working Group on the Evaluation of Cancer Preventive Strate- and mismatch repair pathways, and colorectal cancer risk. Carci- gies (1998) Carotenoids. In: Vaino H, Heseltine E (eds) IARC nogenesis 30(3):472–479 (PMID: 19029193) Handbooks of cancer prevention, vol 2. IARC Press, Lyon Kaaks R, Ferrari P (2006) Dietary intake assessments in epidemiology: IARC Working Group on the Evaluation of Cancer Preventive Strate- can we know what we are measuring? Ann Epidemiol 16:377– gies (2003) Fruit and vegetables. In: Vaino H, Bianchini F (eds) 380 IARC handbooks of cancer prevention, vol 8. IARC Press, Lyon Kaaks R, Riboli E (2005) Epidemiologic studies of nutrition and can- INTERSALT Co-operative Research Group (1988) Sodium, potas- cer: let us not throw out the baby with the bath water. Int J Cancer sium, body mass, alcohol and blood pressure: the INTERSALT 116:662–664 Study. The INTERSALT Co-operative Research Group. J Hyper- Kaaks R, Riboli E, Sinha R (1997) Biochemical markers of dietary in- tens Suppl 6:S584–S586 take. IARC Sci Publ 10:3–126 Ioannides C, Stone AN, Breacker PJ, Basu TK (1982) Impairment of Kaaks R, Ferrari P, Ciampi A, Plummer M, Riboli E (2002) Uses and absorption of ascorbic acid following ingestion of aspirin in guin- limitations of statistical accounting for random error correlations, ea pigs. Biochem Pharmacol 31:4035–4038 in the validation of dietary questionnaire assessments. Public Jacques PF, Sulsky SI, Sadowski JA, Phillips JC, Rush D, Willett WC Health Nutr 5:969–976 (1993) Comparison of micronutrient intake measured by a dietary Kaput J (2008) Nutrigenomics research for personalized nutrition and questionnaire and biochemical indicators of micronutrient status. medicine. Curr Opin Biotechnol 19:110–120 Am J Clin Nutr 57:182–189 Karlsen A, BlomhoV R, Gundersen TE (2005) High-throughput analy- Jakszyn P, Bingham S, Pera G, Agudo A, Luben R, Welch A, Boeing sis of vitamin C in human plasma with the use of HPLC with H, Del GG, Palli D, Saieva C, Krogh V, Sacerdote C, Tumino R, monolithic column and UV-detection. J Chromatogr B Analyt Panico S, Berglund G, Siman H, Hallmans G, Sanchez MJ, Technol Biomed Life Sci 824:132–138 Larranaga N, Barricarte A, Chirlaque MD, Quiros JR, Key TJ, Karlsen A, BlomhoV R, Gundersen TE (2007) Stability of whole blood Allen N, Lund E, Carneiro F, Linseisen J, Nagel G, Overvad K, and plasma ascorbic acid. Eur J Clin Nutr 61:1233–1236 Tjonneland A, Olsen A, Bueno-de-Mesquita HB, Ocke MO, Key T, Oakes S, Davey G, Moore J, Edmond LM, McLoone UJ, Peeters PH, Numans ME, Clavel-Chapelon F, Trichopoulou A, Thurnham DI (1996) Stability of vitamins A, C, and E, carote- Fenger C, Stenling R, Ferrari P, Jenab M, Norat T, Riboli E, noids, lipids, and in whole blood stored at 4 degrees Gonzalez CA (2006) Endogenous versus exogenous exposure to C for 6 and 24 hours before separation of serum and plasma. Can- N-nitroso compounds and gastric cancer risk in the European cer Epidemiol Biomarkers Prev 5:811–814 Prospective Investigation into Cancer and Nutrition (EPIC-EUR- Key TJ, Allen NE, Spencer EA, Travis RC (2002) The eVect of diet on GAST) study. Carcinogenesis 27:1497–1501 risk of cancer. Lancet 360:861–868 Jenab M, Bingham S, Ferrari P, Friesen MD, Al-Delaimy WK, Luben Kikunaga S, Tin T, Ishibashi G, Wang DH, Kira S (2007) The applica- R, Wareham N, Khaw KT, Riboli E (2005) Long-term cryocon- tion of a handheld personal digital assistant with camera and mo- servation and stability of vitamin C in serum samples of the Euro- bile phone card (Wellnavi) to the general population in a dietary pean prospective investigation into cancer and nutrition. Cancer survey. J Nutr Sci Vitaminol (Tokyo) 53:109–116 Epidemiol Biomarkers Prev 14:1837–1840 Kimura Y, Kono S, Toyomura K, Nagano J, Mizoue T, Moore MA, Jenab M, Riboli E, Ferrari P, Friesen M, Sabate J, Norat T, Slimani N, Mibu R, Tanaka M, Kakeji Y, Maehara Y, Okamura T, Ikejiri K, Tjonneland A, Olsen A, Overvad K, Boutron-Ruault MC, Clavel- Futami K, Yasunami Y, Maekawa T, Takenaka K, Ichimiya H, 123 Hum Genet (2009) 125:507–525 523

Imaizumi N (2007) Meat, Wsh and fat intake in relation to subsite- Nagao A (2004) Oxidative conversion of carotenoids to retinoids and speciWc risk of colorectal cancer: the Fukuoka Colorectal Cancer other products. J Nutr 134:237S–240S Study. Cancer Sci 98:590–597 Nelson EW, Lane H, Fabri PJ, Scott B (1978) Demonstration of satu- Kipnis V, Subar AF, Midthune D, Freedman LS, Ballard-Barbash R, ration kinetics in the intestinal absorption of vitamin C in man and Troiano RP, Bingham S, Schoeller DA, Schatzkin A, Carroll RJ the guinea pig. J Clin Pharmacol 18:325–335 (2003) Structure of dietary measurement error: results of the Nicholson JK, Lindon JC (2008) Systems biology: metabonomics. OPEN biomarker study. Am J Epidemiol 158:14–21 Nature 455:1054–1056 Kobayashi M, Sasaki S, Kawabata T, Hasegawa K, Akabane M, Tsug- Nicholson JK, Lindon JC, Holmes E (1999) ‘Metabonomics’: under- ane S (2001) Single measurement of serum phospholipid fatty standing the metabolic responses of living systems to pathophys- acid as a biomarker of speciWc fatty acid intake in middle-aged iological stimuli via multivariate statistical analysis of biological Japanese men. Eur J Clin Nutr 55:643–650 NMR spectroscopic data. Xenobiotica 29:1181–1189 Krinsky NI, Johnson EJ (2005) Carotenoid actions and their relation to Nicholson JK, Holmes E, Wilson ID (2005) Gut microorganisms, health and disease. Mol Aspects Med 26:459–516 mammalian metabolism and personalized health care. Nat Rev Kuhnle GG, Story GW, Reda T, Mani AR, Moore KP, Lunn JC, Bing- Microbiol 3:431–438 ham SA (2007) Diet-induced endogenous formation of nitroso Noguchi Y, Sakai R, Kimura T (2003) Metabolomics and its potential compounds in the GI tract. Free Radic Biol Med 43:1040–1047 for assessment of adequacy and safety of amino acid intake. Kury S, Buecher B, Robiou-du-Pont S, Scoul C, Sebille V, Colman H, J Nutr 133:2097S–2100S Le HC, Le NT, Bourdon J, Faroux R, Ollivry J, Lafraise B, Chu- Norat T, Bingham S, Ferrari P, Slimani N, Jenab M, Mazuir M, Over- pin LD, Bezieau S (2007) Combinations of cytochrome P450 vad K, Olsen A, Tjonneland A, Clavel F, Boutron-Ruault MC, gene polymorphisms enhancing the risk for sporadic colorectal Kesse E, Boeing H, Bergmann MM, Nieters A, Linseisen J, Trich- cancer related to red meat consumption. Cancer Epidemiol Bio- opoulou A, Trichopoulos D, Tountas Y, Berrino F, Palli D, Panico markers Prev 16:1460–1467 S, Tumino R, Vineis P, Bueno-de-Mesquita HB, Peeters PH, Levine M, Wang Y, Rumsey SC (1999) Analysis of ascorbic acid and Engeset D, Lund E, Skeie G, Ardanaz E, Gonzalez C, Navarro C, dehydroascorbic acid in biological samples. Methods Enzymol Quiros JR, Sanchez MJ, Berglund G, Mattisson I, Hallmans G, 299:65–76 Palmqvist R, Day NE, Khaw KT, Key TJ, San JM, Hemon B, Sar- Lewin MH, Bailey N, Bandaletova T, Bowman R, Cross AJ, Pollock acci R, Kaaks R, Riboli E (2005) Meat, Wsh, and colorectal cancer J, Shuker DE, Bingham SA (2006) Red meat enhances the colonic risk: the European Prospective Investigation into cancer and formation of the DNA adduct O6-carboxymethyl guanine: impli- nutrition. J Natl Cancer Inst 97:906–916 cations for colorectal cancer risk. Cancer Res 66:1859–1865 Norat T, Bowman R, Luben R, Welch A, Khaw KT, Wareham N, Li Y, Schellhorn HE (2007) New developments and novel therapeutic Bingham S (2008) Blood pressure and interactions between the perspectives for vitamin C. J Nutr 137:2171–2184 angiotensin polymorphism AGT M235T and sodium intake: a Liu RH (2004) Potential synergy of phytochemicals in cancer preven- cross-sectional population study. Am J Clin Nutr 88:392–397 tion: mechanism of action. J Nutr 134:3479S–3485S Nowell SA, Ahn J, Ambrosone CB (2004) Gene-nutrient interactions Livingstone MB, Black AE (2003) Markers of the validity of reported in cancer etiology. Nutr Rev 62:427–438 energy intake. J Nutr 133(Suppl 3):895S–920S O’Keefe SJ (2008) Nutrition and colonic health: the critical role of the Lykkesfeldt J, Loft S, Poulsen HE (1995) Determination of ascorbic microbiota. Curr Opin Gastroenterol 24:51–58 acid and dehydroascorbic acid in plasma by high-performance Ordovas JM (2002) Gene-diet interaction and plasma lipid responses liquid chromatography with coulometric detection—are they reli- to dietary intervention. Biochem Soc Trans 30:68–73 able biomarkers of oxidative stress? Anal Biochem 229:329–335 Ordovas JM (2007) Gender, a signiWcant factor in the cross talk be- MacDonald L, Thumser AE, Sharp P (2002) Decreased expression of tween genes, environment, and health. Gend Med 4(Suppl the vitamin C transporter SVCT1 by ascorbic acid in a human B):S111–S122 intestinal epithelial cell line. Br J Nutr 87:97–100 Ordovas JM, Mooser V (2004) Nutrigenomics and nutrigenetics. Curr Malekshah AF, Kimiagar M, Saadatian-Elahi M, Pourshams A, Nou- Opin Lipidol 15:101–108 raie M, Goglani G, Hoshiarrad A, Sadatsafavi M, Golestan B, Ordovas JM, Corella D, Cupples LA, Demissie S, Kelleher A, Coltell Yoonesi A, Rakhshani N, Fahimi S, Nasrollahzadeh D, Salahi R, O, Wilson PW, Schaefer EJ, Tucker K (2002a) Polyunsaturated Ghafarpour A, Semnani S, Steghens JP, Abnet CC, Kamangar F, fatty acids modulate the eVects of the APOA1 G-A polymorphism Dawsey SM, Brennan P, BoVetta P, Malekzadeh R (2006) Valid- on HDL-cholesterol concentrations in a sex-speciWc manner: the ity and reliability of a new food frequency questionnaire com- Framingham Study. Am J Clin Nutr 75:38–46 pared to 24 h recalls and biochemical measurements: pilot phase Ordovas JM, Corella D, Demissie S, Cupples LA, Couture P, Coltell of Golestan cohort study of esophageal cancer. Eur J Clin Nutr O, Wilson PW, Schaefer EJ, Tucker KL (2002b) Dietary fat in- 60:971–977 take determines the eVect of a common polymorphism in the he- Mayne ST (2003) Antioxidant nutrients and chronic disease: use of patic lipase gene promoter on high-density lipoprotein biomarkers of exposure and oxidative stress status in epidemiol- metabolism: evidence of a strong dose eVect in this gene-nutrient ogic research. J Nutr 133(Suppl 3):933S–940S interaction in the Framingham Study. Circulation 106:2315–2321 Michels KB (2005a) Diet and cancer: current knowledge, methodolog- Panlasigui LN, Baello OQ, Dimatangal JM, Dumelod BD (2003) ic pitfalls and future directions. Int J Cancer 116:665–666 Blood cholesterol and lipid-lowering eVects of carrageenan on Michels KB (2005b) The role of nutrition in cancer development and human volunteers. Asia Pac J Clin Nutr 12:209–214 prevention. Int J Cancer 114:163–165 Poppitt SD, Kilmartin P, Butler P, Keogh GF (2005) Assessment of Minich DM, Bland JS (2008) Dietary management of the metabolic erythrocyte phospholipid fatty acid composition as a biomarker syndrome beyond macronutrients. Nutr Rev 66:429–444 for dietary MUFA, PUFA or saturated fatty acid intake in a con- Mutch DM, Wahli W, Williamson G (2005) Nutrigenomics and nutri- trolled cross-over intervention trial. Lipids Health Dis 4:30 genetics: the emerging faces of nutrition. FASEB J 19:1602–1616 Potischman N (2003) Biologic and methodologic issues for nutritional Mutch DM, Klocke B, Morrison P, Murray CA, Henderson CJ, Seifert biomarkers. J Nutr 133(Suppl 3):875S–880S M, Williamson G (2007) The disruption of hepatic cytochrome Potischman N, Freudenheim JL (2003) Biomarkers of nutritional p450 reductase alters mouse lipid metabolism. J Proteome Res exposure and nutritional status: an overview. J Nutr 133(Suppl 6:3976–3984 3):873S–874S 123 524 Hum Genet (2009) 125:507–525

Prentice RL, Sugar E, Wang CY, Neuhouser M, Patterson R (2002) of a food frequency questionnaire with a 24-hour recall for use in Research strategies and the use of nutrient biomarkers in studies an epidemiological cohort study: results from the biomarker- of diet and chronic disease. Public Health Nutr 5:977–984 based Observing Protein and Energy Nutrition (OPEN) study. Int Qian GS, Ross RK, Yu MC, Yuan JM, Gao YT, Henderson BE, Wo- J Epidemiol 32:1054–1062 gan GN, Groopman JD (1994) A follow-up study of urinary mark- Schatzkin A, Abnet CC, Cross AJ, Gunter M, PfeiVer R, Gail M, Lim ers of aXatoxin exposure and liver cancer risk in Shanghai, U, vey-Smith G (2009) Mendelian randomization: how it can— People’s Republic of China. Cancer Epidemiol Biomarkers Prev and cannot—help conWrm causal relations between nutrition and 3:3–10 cancer. Cancer Prev Res (Phila Pa) 2:104–113 Rao AV, Rao LG (2007) Carotenoids and human health. Pharmacol Schroder H, Covas MI, Marrugat J, Vila J, Pena A, Alcantara M, Masia Res 55:207–216 R (2001) Use of a three-day estimated food record, a 72-hour re- Reboul E, Richelle M, Perrot E, smoulins-Malezet C, Pirisi V, Borel P call and a food-frequency questionnaire for dietary assessment in (2006) Bioaccessibility of carotenoids and vitamin E from their a Mediterranean Spanish population. Clin Nutr 20:429–437 main dietary sources. J Agric Food Chem 54:8749–8755 Slimani N, Valsta L (2002) Perspectives of using the EPIC-SOFT pro- Reboul E, Thap S, Tourniaire F, Andre M, Juhel C, Morange S, Amiot gramme in the context of pan-European nutritional monitoring MJ, Lairon D, Borel P (2007) DiVerential eVect of dietary antiox- surveys: methodological and practical implications. Eur J Clin idant classes (carotenoids, , vitamins C and E) on Nutr 56(Suppl 2):S63–S74 lutein absorption. Br J Nutr 97:440–446 Slimani N, Deharveng G, Charrondiere RU, van Kappel AL, Ocke Reszka E, Wasowicz W, Gromadzinska J (2006) Genetic polymor- MC, Welch A, Lagiou A, van LM, Agudo A, Pala V, Brandstetter phism of xenobiotic metabolising enzymes, diet and cancer sus- B, Andren C, Stripp C, Van Staveren WA, Riboli E (1999) Struc- ceptibility. Br J Nutr 96:609–619 ture of the standardized computerized 24-h diet recall interview Rezzi S, Ramadan Z, Martin FP, Fay LB, van BP, Lindon JC, Nichol- used as reference method in the 22 centers participating in the son JK, Kochhar S (2007) Human metabolic phenotypes link di- EPIC project. European Prospective Investigation into Cancer rectly to speciWc dietary preferences in healthy individuals. and Nutrition. Comput Methods Programs Biomed 58:251–266 J Proteome Res 6:4469–4477 Slimani N, Kaaks R, Ferrari P, Casagrande C, Clavel-Chapelon F, Lot- Ribalta J, Vallve JC, Girona J, Masana L (2003) Apolipoprotein and ze G, Kroke A, Trichopoulos D, Trichopoulou A, Lauria C, Belle- apolipoprotein receptor genes, blood lipids and disease. Curr gotti M, Ocke MC, Peeters PH, Engeset D, Lund E, Agudo A, Opin Clin Nutr Metab Care 6:177–187 Larranaga N, Mattisson I, Andren C, Johansson I, Davey G, Riboli E, Kaaks R (2000) Invited commentary: the challenge of multi- Welch AA, Overvad K, Tjonneland A, Van Staveren WA, Saracci center cohort studies in the search for diet and cancer links. Am J R, Riboli E (2002) European Prospective Investigation into Can- Epidemiol 151:371–374 cer and Nutrition (EPIC) calibration study: rationale, design and Riboli E, Ronnholm H, Saracci R (1987) Biological markers of diet. population characteristics. Public Health Nutr 5:1125–1145 Cancer Surv 6:685–718 Solanky KS, Bailey NJ, Beckwith-Hall BM, Davis A, Bingham S, Richardson S, Gilks WR (1993) Conditional independence models for Holmes E, Nicholson JK, Cassidy A (2003) Application of bio- epidemiological studies with covariate measurement error. Stat Xuid 1H nuclear magnetic resonance-based metabonomic tech- Med 12:1703–1722 niques for the analysis of the biochemical eVects of dietary Rosner B, Michels KB, Chen YH, Day NE (2008) Measurement error isoXavones on human plasma proWle. Anal Biochem 323:197– correction for nutritional exposures with correlated measurement 204 error: use of the method of triads in a longitudinal setting. Stat Solanky KS, Bailey NJ, Beckwith-Hall BM, Bingham S, Davis A, Med 27:3466–3489 Holmes E, Nicholson JK, Cassidy A (2005) BioXuid 1H NMR- Saadatian-Elahi M, Slimani N, Chajes V, Jenab M, Goudable J, Biessy based metabonomic techniques in nutrition research - metabolic C, Ferrari P, Byrnes G, Autier P, Peeters PH, Ocke M, Bueno de eVects of dietary isoXavones in humans. J Nutr Biochem 16:236– MB, Johansson I, Hallmans G, Manjer J, Wirfalt E, Gonzalez CA, 244 Navarro C, Martinez C, Amiano P, Suarez LR, Ardanaz E, Sonnenberg LM, Quatromoni PA, Gagnon DR, Cupples LA, Franz Tjonneland A, Halkjaer J, Overvad K, Jakobsen MU, Berrino F, MM, Ordovas JM, Wilson PW, Schaefer EJ, Millen BE (1996) Pala V, Palli D, Tumino R, Vineis P, de Magistris MS, Spencer Diet and plasma lipids in women. II. Macronutrients and plasma EA, Crowe FL, Bingham S, Khaw KT, Linseisen J, Rohrmann S, triglycerides, high-density lipoprotein, and the ratio of total to Boeing H, Noethlings U, Olsen KS, Skeie G, Lund E, Trichopou- high-density lipoprotein cholesterol in women: the Framingham lou A, Oustoglou E, Clavel-Chapelon F, Riboli E (2009) Plasma nutrition studies. J Clin Epidemiol 49:665–672 phospholipid fatty acid proWles and their association with food in- Spiegelman D, Zhao B, Kim J (2005) Correlated errors in biased sur- takes: results from a cross-sectional study within the European rogates: study designs and methods for measurement error correc- Prospective Investigation into Cancer and Nutrition. Am J Clin tion. Stat Med 24:1657–1682 Nutr 89(1):331–346 (PMID: 19056549) Stamler J, Elliott P, Appel L, Chan Q, Buzzard M, Dennis B, Dyer AR, SaVhill R, Margison GP, O’Connor PJ (1985) Mechanisms of carcino- Elmer P, Greenland P, Jones D, Kesteloot H, Kuller L, Labarthe genesis induced by alkylating agents. Biochim Biophys Acta D, Liu K, Moag-Stahlberg A, Nichaman M, Okayama A, Okuda 823:111–145 N, Robertson C, Rodriguez B, Stevens M, Ueshima H, Horn LV, Saintot M, Astre C, Scali J, Gerber M (1995) Within-subjects seasonal Zhou B (2003) Higher blood pressure in middle-aged American variation and determinants of inter-individual variations of plas- adults with less education-role of multiple dietary factors: the IN- ma beta-carotene. Int J Vitam Nutr Res 65:169–174 TERMAP study. J Hum Hypertens 17:655–775 Saracci R (1997) Comparing measurements of biomarkers with other Stella C, Beckwith-Hall B, Cloarec O, Holmes E, Lindon JC, Powell J, measurements of exposure. IARC Sci Publ 30:3–312 van der OF, Bingham S, Cross AJ, Nicholson JK (2006) Suscep- Sasaki S, Ushio F, Amano K, Morihara M, Todoriki O, Uehara Y, tibility of human metabolic phenotypes to dietary modulation. Toyooka E (2000) Serum biomarker-based validation of a self- J Proteome Res 5:2780–2788 administered diet history questionnaire for Japanese subjects. Stram DO, Hankin JH, Wilkens LR, Pike MC, Monroe KR, Park S, Hen- J Nutr Sci Vitaminol (Tokyo) 46:285–296 derson BE, Nomura AM, Earle ME, Nagamine FS, Kolonel LN Schatzkin A, Kipnis V, Carroll RJ, Midthune D, Subar AF, Bingham (2000) Calibration of the dietary questionnaire for a multiethnic co- S, Schoeller DA, Troiano RP, Freedman LS (2003) A comparison hort in Hawaii and Los Angeles. Am J Epidemiol 151:358–370 123 Hum Genet (2009) 125:507–525 525

Subar AF, Thompson FE, Potischman N, Forsyth BH, Buday R, Rich- ical intake on human urinary metabolomic proWles. Am J Clin ards D, McNutt S, Hull SG, Guenther PM, Schatzkin A, Baranow- Nutr 86:1687–1693 ski T (2007) Formative research of a quick list for an automated Wang DH, Kogashiwa M, Kira S (2006) Development of a new instru- self-administered 24-hour dietary recall. J Am Diet Assoc ment for evaluating individuals’ dietary intakes. J Am Diet Assoc 107:1002–1007 106:1588–1593 Sugar EA, Wang CY, Prentice RL (2007) Logistic regression with Wang Y, Lawler D, Larson B, Ramadan Z, Kochhar S, Holmes E, exposure biomarkers and Xexible measurement error. Biometrics Nicholson JK (2007) Metabonomic investigations of aging and 63:143–151 caloric restriction in a life-long dog study. J Proteome Res Tai ES, Corella D, Demissie S, Cupples LA, Coltell O, Schaefer EJ, 6:1846–1854 Tucker KL, Ordovas JM (2005) Polyunsaturated fatty acids inter- Watson AD (2006) Thematic review series: systems biology approach- act with the PPARA-L162 V polymorphism to aVect plasma tri- es to metabolic and cardiovascular disorders. Lipidomics: a glo- glyceride and apolipoprotein C-III concentrations in the bal approach to lipid analysis in biological systems. J Lipid Res Framingham Heart Study. J Nutr 135:397–403 47:2101–2111 Tasevska N, Runswick SA, McTaggart A, Bingham SA (2005) Weiss EP, Brown MD, Shuldiner AR, Hagberg JM (2002) Fatty acid Urinary sucrose and fructose as biomarkers for sugar consump- binding protein-2 gene variants and insulin resistance: gene and tion. Cancer Epidemiol Biomarkers Prev 14:1287–1294 gene-environment interaction eVects. Physiol Genomics 10:145– Tasevska N, Runswick SA, Welch AA, McTaggart A, Bingham SA 157 (2008) Urinary sugars biomarker relates better to extrinsic than to Wilson JX (2005) Regulation of vitamin C transport. Annu Rev Nutr intrinsic sugars intake in a metabolic study with volunteers con- 25:105–125 suming their normal diet. Eur J Clin Nutr Wolk A, Furuheim M, Vessby B (2001) Fatty acid composition of adi- Terzuoli L, Pagani R, Frosi B, Galli A, Felici C, Barabesi L, Porcelli B pose tissue and serum lipids are valid biological markers of dairy (2004) Stability of serum and plasma ascorbic acid. J Lab Clin fat intake in men. J Nutr 131:828–833 Med 143:67 Wood KC, Fullerton MD, El-Sohemy A, Bakovic M (2008) Interac- Thompson FE, Moler JE, Freedman LS, CliVord CK, Stables GJ, Wil- tions between hepatic lipase and apolipoprotein E gene polymor- lett WC (1997) Register of dietary assessment calibration-valida- phisms aVect serum lipid proWles of healthy Canadian adults. tion studies: a status report. Am J Clin Nutr 65:1142S–1147S Appl Physiol Nutr Metab 33:761–768 Tooze JA, Midthune D, Dodd KW, Freedman LS, Krebs-Smith SM, World Cancer Research Fund and American Institute for Cancer Re- Subar AF, Guenther PM, Carroll RJ, Kipnis V (2006) A new sta- search (1997) Food, nutrition and the prevention of cancer: a glo- tistical method for estimating the usual intake of episodically con- bal perspective. World Cancer Research Fund; American Institute sumed foods with application to their distribution. J Am Diet for Cancer Research, Washington, DC Assoc 106:1575–1587 World Cancer Research Fund (2007) Food, nutrition, physical activity Truswell AS (1995) Dietary Wbre and plasma lipids. Eur J Clin Nutr and the prevention of cancer: a global perspective. World Cancer 49(Suppl 3):S105–S109 Research Fund; American Institute for Cancer Research, Wash- Van Dorsten FA, Daykin CA, Mulder TP, Van Duynhoven JP (2006) ington, DC Metabonomics approach to determine metabolic diVerences be- Wu K, Bowman R, Welch AA, Luben RN, Wareham N, Khaw KT, tween green tea and black tea consumption. J Agric Food Chem Bingham SA (2007) Apolipoprotein E polymorphisms, dietary fat 54:6929–6938 and Wbre, and serum lipids: the EPIC Norfolk study. Eur Heart J Vang A, Singh PN, Lee JW, Haddad EH, Brinegar CH (2008) Meats, 28:2930–2936 processed meats, obesity, weight gain and occurrence of diabetes Zhao HL, Houweling AH, Vanstone CA, Jew S, Trautwein EA, among adults: Wndings from Adventist Health Studies. Ann Nutr Duchateau GS, Jones PJ (2008) Genetic variation in ABC G5/G8 Metab 52:96–104 and NPC1L1 impact cholesterol response to plant sterols in Vioque J, Weinbrenner T, Asensio L, Castello A, Young IS, Fletcher hypercholesterolemic men. Lipids 43:1155–1164 A (2007) Plasma concentrations of carotenoids and vitamin C are Zunft HJ, Luder W, Harde A, Haber B, Graubaum HJ, Koebnick C, better correlated with dietary intake in normal weight than over- Grunwald J (2003) Carob pulp preparation rich in insoluble Wbre weight and obese elderly subjects. Br J Nutr 97:977–986 lowers total and LDL cholesterol in hypercholesterolemic Walsh MC, Brennan L, Malthouse JP, Roche HM, Gibney MJ (2006) patients. Eur J Nutr 42:235–242 EVect of acute dietary standardization on the urinary, plasma, and Zuppi C, Messana I, Forni F, Rossi C, Pennacchietti L, Ferrari F, salivary metabolomic proWles of healthy humans. Am J Clin Nutr Giardina B (1997) 1H NMR spectra of normal urines: reference 84:531–539 ranges of the major metabolites. Clin Chim Acta 265:85–97 Walsh MC, Brennan L, Pujos-Guillot E, Sebedio JL, Scalbert A, Fagan A, Higgins DG, Gibney MJ (2007) InXuence of acute phytochem-

123