The Biomarkers of Nutrition for Development (BOND) project is giving impetus to the re - examination of diagnos-tic assessment of nutritional status by pointing out some obvious discrepancies and weaknesses in the current appli-cation and interpretation of nutritional biomarkers (Raiten et al. , 2011 ). After a period of dormancy and reliance on established conventions, renewed emphasis on evidence based decision making and expanding technological and logistic capacity to measure nutrients and metabolites in the clinic and in populations have provided incentives for more objective, transparent, and robust documenta-tion and analysis.
The BOND biomarker paradigm is divided into three levels of interest for any nutrient: (1) magnitude of expo-sure to the nutrient; (2) nutrient (nutritional) status; and (3) functional consequences of exposure. It seeks primarily ard amounts of vitamin A by Irish adults from 20% to 5%
(Kiely et al. , 2001 ).
With respect to juvenile populations, German children and adolescents received 50 – 65% of reference intake needs of vitamin A from non - fortifi ed foods in their diets, and an additional 10 – 20% from fortifi ed foods (Sichert - Hellert et al. , 2001a, b ). This cohort had shown a 5 – 15% increase in the amount of the vitamin obtained from fortifi ed bev-erages during the previous 15 - year period (Sichert - Hellert et al. , 2001b ). On the island of Guam in the South Pacifi c, fruit drinks, milk, and fortifi ed cereals contributed most of the vitamin A in Guamanian children, the median vitamin A intake was 76% of the age - appropriate RDA (Pobocic and Richer, 2002 ). For the children living on the mainland of the United States, however, concern has been expressed that excessive sweetened drink consumption is associated with displacement of milk from children ’ s diets, higher daily energy intake, and greater weight gain (Mrdjenovic and Levitsky, 2003 ).
Human milk can be a relatively important source of vitamin A, even beyond the fi rst year of life. According to a study in Kenya, breast milk supplies more vitamin A than the complementary foods that replace it, making it an “ irreplaceable source of fat and vitamin A ” (Onyango et al. , 2002 ). The role of supplements as the source of vitamin A for juvenile populations is now beginning to emerge, even in the fi rst two years of life. Among a cohort followed from infants to toddlers in the US state of Iowa, 32% were consuming vitamin A - containing supplements 40 – 60% of days by the age of 24 months (Eichenberger Gilmore et al. , 2005 ).
Factors Affecting Vitamin A Consumption The infl uence of cultural, behavioral, and physical factors on vitamin A intake has been studied in various adult populations. In a study among pregnant women in the United States, Mexican - born women consumed more vitamin A than US - born Mexican - American women during pregnancy (Cotton et al. , 2004 ). Edentulous US white elderly consumed less vitamin A and carotene than peers with adequate dentition, although no gradient was seen in African - American contemporaries compared in an analogous manner related to mastication capacity. Smoking behavior – smokers, ex - smokers, or never smokers – had no effect on vitamin A intake in Greater Chicago residents (Dyer et al. , 2003 ). In an analysis of US Department of Agriculture (USDA) survey data, estimated vitamin A
material (chemical) markers, with less tangible (behavio-ral) indicators as a second level of marker. It recognizes for a nutrient such as vitamin A that health and policy issues revolve around how much of the nutrient is actually offered and consumed. Since there is no stoichiometric relationship between consumption with uptake, retention, and utilization, the fi rst level of quantifying exposure is somewhat discrete from nutrient status; the latter refers to the amount of the nutrient within stores and functional pathways. Finally, insofar as nutrients have functions and actions, as discussed earlier, the third element of the para-digm looks for concrete measures of functional conse-quence attributable to a given degree of nutrient exposure.
In general terms, all three of the aforementioned elements can refer to individuals (a specifi c diagnosis), or to groups of individuals (as prevalence and rates).
A virtual plethora of diagnostic options are displayed in Table 11.2 . The key to rational selection depends, in part, on whether the application is for individuals in a clinical context or for a population for epidemiological and public health purposes. In the former, it is our interest to get an accurate measurement for the individual patient to
TABLE 11.2 Clinical assessment of vitamin A status and estimation of population risk of vitamin A defi ciency
diagnose the underlying disease or take a decision for sup-plement therapy. When it comes to a population, the issue is the risk of a substantial number of individuals having suboptimal vitamin A status to merit a collective or tar-geted intervention program. In general, the limits on diagnostic methods for clinical practice derive from the day - to - day variation or distortion by clinical conditions, which combine to create imprecision in refl ecting the underlying status of an individual, or both. The limits on assessment testing at the public health and population level derive from constraints of cost, convenience, accept-ability, and ethical application in otherwise healthy persons.
Evolution of Vitamin A Assessment
It was the emergence of interest in vitamin A and child-hood mortality (Sommer et al. , 1983 ), in which pre clinical ( “ marginal ” ) vitamin A defi ciency was a risk factor, that stimulated creation of assessment tools. In 1993, in a guideline book from the International Vitamin A Consultative Group, Underwood and Olson (1993) set forth its mission: to provide tools for “ assessing the regional distribution and magnitude of vitamin A defi ciency. ” It covered 13 assessment areas, 10 of which would be fi rmly considered biomarkers. This was followed, in 1996, by a largely derivative offi cial WHO publication (WHO, 1996 ) which claimed to provide “ principles governing the use of biological indicators for vitamin A defi ciency (VAD) sur-veillance [providing] the rationale behind each indicator and its limitations and cut - off points for interpretation in terms of public health signifi cance. ” Almost a decade later, Tanumihardjo (2004) comments, in published proceed-ings: “ having many choices of vitamin A assessment methods, laboratory sophistication and resources available will usually dictate which methods are chosen. ” As part of the BOND, she has updated her consideration: “ biomark-ers of vitamin A status are still needed for the near future in order to more specifi cally identify populations at risk for vitamin A defi ciency and to evaluate the effectiveness of different interventions or programs ” (Tanumihardjo, 2011 ).
Gold Standards of Vitamin A Status for Reference to Biomarkers
A chemical rendering of a cadaver postmortem is the most defi nitive gold standard, but of little clinical relevance.
Since 80% of vitamin stores are in the liver, chemical analysis of a percutaneous liver biopsy is the second best approximation (Figures 11.4 and 11.5 ), and refl ection of
FIG. 11.4 The horizontal axis represents a continuum of hepatic vitamin A concentrations in μ mol/g, with corresponding division into range bands categorizing specifi c states of vitamin A status. For each of the six approaches (one clinical method and fi ve biomarkers), the range of liver concentration in which an abnormal result can provide diagnostic validity is represented by the horizontal solid bars in each row. Reproduced with permission from Tanumihardjo (2011) .
VITAMIN A STATUS Indicator
VA STATUS LIVER VA
Deficient
< 0.07
Marginal 0.07 − 0.1
Adequate
0.1 − 1.0 Sub-toxic
>1.0
Toxic 10 mmol/g Clinical signs and tests
Serum retinol Breast milk retinol Dose response tests Isotope dilution Liver sample
FIG. 11.5 The relationship of the modifi ed relative dose – response value to liver retinol concentration in piglets. Below 17 μ g/g liver, the MRDR value is invari-ably positive, i.e. > 0.060; between 17 and 29 μ g/g, the response is split; and above 29 μ g/g liver, the MRDR value is usually < 0.060. Reproduced with permission from Tanumihardjo (2011) .
0 0.25
0.20
0.15
0.10
0.05
0.00 10
MRDR value
20
Liver reserves μg retinol/g liver30 40 50
hepatic stores is the metric of other biomarkers (WHO, 1996 ; Tanumihardjo, 2004, 2011 ). Underwood and Olson (1993) consider 1.05 μ mol of vitamin A per gram of liver tissue (equivalent to 30 μ g/g in gravimetric units) the lower threshold for normal vitamin A status. The safest
surrogate “ gold standard ” method for assessment of total body vitamin A is deuterated - retinol, stable - isotopic dilution technique (Tang et al. , 2002 ; Ribaya - Mercado et al. , 2003, 2004a,b ); it refl ects hepatic vitamin A levels across the entire spectrum from depletion to intoxication (Figure 11.4 ).
Biomarkers for Patient Management in Clinical Practice
If vitamin A status is severely defi cient, clinical acumen in the physical examination may provide suspicion of hypo-vitaminosis A. In general, the majority of the screening tests listed in Table 11.2 are not relevant to hospital or clinic practice. Only the circulating retinol for routine application and the tissue biopsy for extraordinary use are of major interest. With respect to the former, however, for clinical assessment retinol has a series of recognized limita-tions. It is often low in situations of defi ciency. It can be falsely normal in the situation of recent ingestion of preformed vitamin, dehydration, or hyperproteinemia, or falsely low with hypoproteinemia, infection, infl ammatory states, obesity, hormonal replacement, or oral contracep-tion. It has been suggested that only circulating concentra-tions < 10 μ g/dL (0.35 μ mol) have any robust diagnostic signifi cance for vitamin A inadequacy (Underwood and Olson, 1993 ), but it is probably more prudent to be guided by the established limits of normality for circulating retinol in the clinical laboratory of record. A lower -than (or higher - than) limit result should guide the practitioner to pursue the implications for differential diagnosis in an individual case.
liver vitamin A in the adequate range, for a virtually 100%
diagnostic specifi city to detect vitamin A - adequate indi-viduals. Abnormal ratios begin to appear below the critical liver value, but the test is only 100% sensitive for a defi -cient classifi cation at liver vitamin A concentrations less than 18 μ g/g (0.63 μ mol/g). For applications, such as to exclude populations from interventions and to monitor improvement (correction) of vitamin A status to adequate, the MRDR is a minimally invasive biomarker of excep-tional promise.
We should not neglect or discard, however, the other creative – and even less invasive and certainly less expen-sive – assessment tools that came out of the test - generating brainstorming of the 1980s and 1990s. Among these are ocular - response tests (Wondmikun, 2002 ; Taren et al. , 2004 ), conjunctival histology (Courtright et al. , 2002 ), buccal sampling (Sobeck et al. , 2002 ), and others, which are yet to be calibrated with hepatic stores. However, by cross - calibrating the rates of test - result abnormalities in a population known to meet the objective threshold for intervention, one might develop population - based criteria for the diagnosis of hypovitaminosis A endemicity. The caveats are two - fold: (1) selecting appropriate criteria for normal and abnormal rating of the tests, and (2) making any feasible adjustments for ethnic, environmental or other confounders of test response. As it is the population itself that is the target of policy action and programmatic intervention, the status of the collective – rather than the individuals within it – would seem to be a viable view-point for the interpretation of screening biomarkers for vitamin A. The BOND process (Raiten et al. , 2011 ) must grapple with a subtle conceptual issue of whether popula-tion assessment inherently needs to be based on measure-ment of the prevalence of individuals with truly marginal vitamin A stores in a population of interest or merely on an index of the rate of abnormal screening test occurrences.
Assessment for Vitamin A Excess
The concern for exposure and status assessment does not only reside at the lower end and the center of the vitamin A continuum. The upper end of individual status or col-lective exposure can produce adverse consequences on unborn fetuses and on overly exposed individuals. Early in the course of toxic exposure, circulating levels of retinol may rise above 300 μ g/dL (10.5 μ mol), in association with elevations of fasting retinyl esters (Underwood and Olson, Biomarkers for Population Assessment
in Surveys and Epidemiology
In the era of concern surrounding “ marginal ” vitamin A defi ciency and the populations in which it is endemic, the relevance of biomarkers relates to action for public health (Underwood and Olson, 1993 ; WHO, 1996 ; Tanumihardjo, 2004, 2011 ). Among the selected biomarkers listed in Table 11.2 , those with the lowest degree of invasiveness, complexity, and cost are the initial candidates for applica-tion in the fi eld survey setting. According to the recent review by Tanumihardjo (2011) , only four biomarkers (serum retinol, breast - milk retinol, dose – response tests, isotope dilution) have been cross - calibrated with grades of liver vitamin A concentration, and only the fi rst three are appropriate for the fi eld assessment for marginal vitamin A status of a population. Serum retinol and breast - milk retinol would both be susceptible to distortion by infl am-mation and infection, which are prevalent in low - income populations; they would be limited as monitoring indices for improvement in response to interventions by not bridging from the marginal to the adequate band of hepatic vitamin stores.
The family of dose – response tests, the relative dose response (RDR) (Stephensen et al. , 2002 ; Verhoef and West, 2005 ) and the modifi ed RDR (Tanumihardjo and Olson, 1988 ; Tanumihardjo et al. , 1990a,b ) merit special attention because of their putative ability to respond across a wider range of vitamin A status and through the range of the spectrum most relevant to public health (Figure 11.4 ). Both are based on the same physiological principle, namely that apo - RBP builds up in the liver during vitamin A deprivation; hence, an acute dosing with oral retinol leads to an exaggerated export of vitamin A.
This response can be tracked as a quantitative post - dose elevation in retinol concentrations in the standard RDR, which requires two serial blood samples at a 5 - hour inter-val (Verhoef and West, 2005 ), or an increment in a variant vitamin A isomer (3,4 - didehydroretinol) in the modifi ed RDR, which requires only a single (4 - hour post - dose) blood sampling (Tanumihardjo et al. , 1990b ; Tanumihardjo, 2011 ). The sensitivity and specifi city of the modifi ed RDR (MRDR) in relation to the 30 μ g/g (1.05 μ mol/g) hepatic vitamin A concentration, posited as the cutoff for marginal versus adequate vitamin A status (Underwood and Olson, 1993 ) are illustrated by original work by Tanumihardjo (2011) in Figure 11.5 . A normal MRDR ratio value of < 0.06 was seen in all but one instance of
economics, however, often interfere with achieving ade-quate intakes in low - income societies. To succeed with dietary approaches, strategies appropriate to the endog-enous possibilities and limitations of the settings must be devised.
Preformed vitamin A is the more utilizable form. In Bangladesh, the consumption of small indigenous fi sh is widespread and promotes adequacy of intake (Roos et al. , 2003a,b ). In Kenya, the addition of milk or meat to the school snack (a vegetable stew) improved overall vitamin A intake (Murphy et al. , 2003 ). When animal source foods can be made affordable and directed to vulnerable seg-ments of the population, they constitute the most potent underpinning for a food - based strategy.
Plant sources of vitamin A are considerably more pro blematic as strategies for ensuring adequacy given the bioconversion issues of plant matrices. Nevertheless, home - gardening interventions in South Africa (Faber et al. , 2002a ) and Thailand (Schipani et al. , 2002 ) were shown to support better vitamin A status. Varieties of bananas, extraordinarily rich in provitamin A carotenes, are cultivated throughout the Micronesian islands of the Pacifi c and have been recommended as a complement for that region (Englberger et al. , 2003 ). To enhance the bioef-fi cacy of mango ’ s provitamin A, its consumption with fat has been advanced in Gambia (Drammeh et al. , 2002 ).
A number of oil - bearing fruits with high contents of vitamin A are available but underutilized. Gac fruit (Momordica cochinchinensis ) from Vietnam (Vuong and King, 2003 ) and aguage or buriti fruit ( Mauritia vinifera Mart) from the Amazon valley (Mariath et al. , 1989 ) are the two tropical fatty fruits with the highest specifi c con-centrations of provitamin A. Derivatives of the palm fruit (Elaeis spp.), such as red palm oil, are in third place, con-taining in excess of 50 mg of mixed provitamin A per 100 g of oil (Nagendran et al. , 2000 ). Intervention trials in India (Sivan et al. , 2002 ; Radhika et al. , 2003 ) and Burkina Faso (Zagre et al. , 2003 ) demonstrate the effi cacy of cooking with red palm oil to improve vitamin A status. Tailored shortenings (Benade, 2003 ) or dishes of local cuisine (Solomons and Orozco, 2003 ) prepared with red palm oil represent the most highly developed approaches to dietary fortifi cation with derivatives of oil - bearing fruits. With the caveat of potential health risks from certain saturated fatty acid patterns, the provitamin A in oil matrices optimizes the nutritional potential for the carotenes. (Institute of Medicine, 2001 ).
1993 ). When severe hypervitaminosis A is suspected on clinical grounds, a liver biopsy is medically justifi ed.
The risk of damage to the developing fetus is based on dietary intake criteria (Institute of Medicine, 2001 ). In the absence of any imminent prospects for a salivary, urinary or fecal metabolite as a biomarker for individuals ’ chronic exposure, mathematical modeling, using data on food selection, food preformed vitamin A content, and extent of fortifi cation (Dary, 2006 ), is likely to offer the best promise for identifying community risk and taking reme-dial action at the public education or food regulation levels or both.