15.2: Indices of vitamin A status (18a.2)

Most of the vitamin A in the body is stored in the form of retinyl ester in the liver. Therefore, a measure of liver vitamin A stores is the best index of vitamin A status. The definition of vitamin A defi­ciency is when liver stores of retinol are below 0.1µmol/g (Tanumihardjo, 2021 ). Vitamin A is not uniformly distributed in the liver (Olson et al., 1979) and can vary by 8-13% among the liver lobes in humans (Olsen et al., 2018). Furthermore, liver biopsies are impractical in popu­la­tion studies. Instead, serum retinol concen­tra­tions are often deter­mined. However, the serum or plasma contains only about 1% of the total body reserve of vitamin A, and concen­tra­tions do not reflect body stores until they are severely depleted. Consequently, it is best to use other bio­chem­ical and physio­logical func­tional tests of vitamin A status in combination with serum retinol concen­tra­tions. The most used biomarkers of vitamin A status are shown in Figure 18a.2. Details of tests currently used are summar­ized in a review article (Tanumihardjo et al., 2016) and are discussed briefly in the following sections.

Figure 18a.2: Vitamin A status indicators as they relate to liver vitamin A concen­tra­tions, which is considered the reference standard. The length of the black bar indicates the working optimal range (Tanumihardjo, 2020).

18a.2.1 Serum retinol concen­tra­tions

Vitamin A in the plasma circulates largely in a 1:1 complex of retinol and retinol-binding protein (RBP). The remainder is in the form of retinyl ester and very small amounts of retinoic acid and other metabolites (Olson, 1984). Retinyl esters are elevated post-prandially after a meal containing pre­formed or pro­vita­min A. If retinyl esters are elevated in the fasted state, this is an indication of vitamin A toxicity (Tanumihardjo et al., 2016). In humans, this occurs at a liver vitamin A concen­tra­tion of 3µmol/g liver (Olsen et al., 2018).

Serum retinol concen­tra­tions are a common indicator for popu­la­tion studies; however, they only reflect vitamin A status when liver vitamin A stores are severely depleted or excessively high, in which case retinyl esters are also in circulation. When liver vitamin A concen­tra­tions are low or adequate, serum retinol concen­tra­tions are homeo­statically controlled, and levels remain relatively constant and do not reflect total body reserves of vitamin A (Olson, 1984). Hence, it is not surprising that in popu­la­tions from developed countries, such as the United States, where the liver vitamin A concen­tra­tions are generally adequate, positive rela­tion­ships between serum retinol concen­tra­tions and usual intakes of vitamin A are rare (Hallfrisch et al., 1994; Nierenberg et al., 1997).

Func­tional impairment has been seen in undernourished children with extremely low serum vitamin A concen­tra­tions. For example, results in India (Pirie and Anbunathan, 1981 ) and Indonesia (Sommer, 1982). showed that at least 75% of the children with xerophthalmia had serum vitamin A concen­tra­tions <0.35mol/L. In contrast, in a sample of 252 clinically normal Indonesian children, only 8% had serum vitamin A at this low concen­tra­tion. Table 18a.1.

Table 18a.1 Serum vitamin A levels (µmol/L) in a sample of Indonesian children with and without ocular lesions. Compiled from Sommer et al., (1980).

		Percent with serum Vit A
	n	<0.35 μmol/L	0.35–0.69 μmol/L	≥0.7 μmol/L
Normal children	252	8	37	55
Night-blindness or Bitot's spots	325	30	55	15
Corneal xerophthalmia	98	75	24	1

A cutoff for serum vitamin A concen­tra­tions of <0.70mol/L is a less specific measure of vitamin A defi­ciency; its ability to predict a defi­ciency of vitamin A varies widely by region, probably depending on the presence and severity of other risk factors such as infection. None­the­less, the World Health Organization has guidance on the use of this cutoff (WHO, 2011). For example, in Sri Lanka, serum vitamin A concen­tra­tions <0.70mol/L were found in only 28% of children (n=29) with clinical signs of vitamin A defi­ciency (Bitot's spots, corneal scars, or blindness) and in 5% of the children without positive eye findings (Brink et al., 1979). In Indonesia, 85% of children <6y (n=325) with either night blindness or Bitot's spots had serum vitamin A concen­tra­tions <0.70mol/L (Sommer et al., 1980). In contrast, 28% of the children with eye findings and 48% of the children without eye findings had serum vitamin A concen­tra­tions >1.05µmol/L. Additional tests may therefore be required to confirm vitamin A depletion at the individual level when a cutoff of <0.70mol/L is used. At a popu­la­tion level, however, a cutoff of <0.70mol/L can be used to indicate whether vitamin A defi­ciency is likely to be a public health problem (WHO, 2011).

Factors affecting serum retinol concen­tra­tions

Age, sex, and race influence concen­tra­tions, as indicated by the NHANES II and NHANES III results (Pilch 1987; Stephensen and Gildengorin, 2000; Ballew et al., 2001). African American children and adolescents had lower serum retinol concen­tra­tions than Caucasians (Ballew et al., 2001; Neuhouser et al. 2001). In the U.K. National Diet and Nutrition Survey of young people, mean values increased with increasing age in both boys and girls Gregory et al., 2000). In the U.K. Survey of British Adults, mean values were highest for males and females aged 50-64y (Gregory et al., 1990). Age-specific criteria should always be used when interpreting serum retinol concen­tra­tions.

Low fat diets (i.e., <5 to 10g daily) impair the absorption of pro­vita­min A carot­enoids (Jayarajan et al., 1980), and over the longer-term, lower plasma retinol concen­tra­tions. In contrast, the ingestion of meals containing relatively large amounts of either dietary or supple­mental vitamin A during the previous 4h does not alter serum retinol concen­tra­tions in children or adults (Mejia and Arroyave, 1983; Mejia et al., 1984). Consequently, the collection of fasting blood samples is not necessary for serum retinol determinations, unless retinyl esters and carot­enoids are also of interest, in which case fasting is a requirement for accurate interpretation (Mondloch et al., 2o15; Williams et al., 2021).

Other nutrient deficiencies can affect serum retinol concen­tra­tions. Protein- energy malnu­tri­tion decreases liver apo-retinol-binding protein (RBP) production because of a limited supply of protein substrate (Russell et al., 1983). Consequently, hepatic release of vitamin A is impaired, resulting in decreased serum retinol levels, despite the presence of adequate vitamin A stores in the liver. Zinc defi­ciency decreases serum retinol concen­tra­tions because of its role in the hepatic synthesis or secretion of RBP even in the presence of adequate liver vitamin A stores (Christian and West, 1998) Deficient niacin status sequesters vitamin A in the liver and lowers serum retinol concen­tra­tions (Titcomb et al., 2023).

Disease may significantly alter serum retinol concen­tra­tions. Chronic renal disease increases serum retinol concen­tra­tions by reducing catabolism of vitamin A and its carriers. In contrast, liver disease decreases serum retinol concen­tra­tions, probably as a result of a combination of decreased synthesis and secretion of RBP. Cystic fibrosis is also asso­ci­ated with decreased levels of circulating retinol and RBP, probably because of a defect in the transport of vitamin A from the hepatic stores to the periphery.

Infections including HIV ( Kafwembe et al., 2001), measles (Thurnham et al., 1997), and parasitic infections (e.g., malaria and Ascaris lumbricoides) are asso­ci­ated with low serum retinol concen­tra­tions, in some cases because of malab­sorption of vitamin A (Sivakumar and Reddy, 1972; 1975). Infections may also cause low serum retinol as a result of transient decreases in the concen­tra­tions of acute phase proteins, RBP, and transthyretin, even in the presence of adequate liver vitamin A stores. In NHANES III, inflammation, indicated by a serum C‑reactive protein concen­tra­tion above 10mg/L, was asso­ci­ated with lower serum retinol concen­tra­tions. This, in turn, led to misclassification of vitamin A status (Stephensen and Gildengorin, 2000. The confounding effect of the acute-phase response on serum retinol was not evaluated in earlier NHANES investigations. Clearly, it is impor­tant to include a measure of this response to interpret serum retinol concen­tra­tions correctly. A variety of methods have been developed to adjust serum retinol concen­tra­tions for inflammation using acute phase protein concen­tra­tions (Larson et al., 2018). For a practical guide on applying the "Biomarkers Reflecting Inflammation and Nutritional Determinants of Anemia" (BRINDA), see Luo et al., (2023)

Estrogens, either endogenous or those used in contraceptive agents or hormone replacement therapy, increase serum retinol and RBP apparently as a result of increased mobilization of vitamin A from the liver.

Interpretive criteria

Data for mean and selected percentiles for serum retinol concen­tra­tions classified by age, sex, and race or ethnicity for NHANES III are presented in Ballew et al. (2001) and IOM (2001). Serum retinol concen­tra­tions were also deter­mined in the U.K. national surveys (Gregory et al., 1990; 1995; 2000; Finch et al., 1998). Mean, median, and lower and upper 2.5th or 5th percentiles by age and sex are given.

Age-specific cutoff points for serum retinol concen­tra­tions developed for NHANESII are shown in Table 18a.2. These same cutoffs were used for the interpretation of the serum vitamin A values in the Hispanic Health and Nutrition Exam­ination Survey (HHANES) and in NHANES III (Ballew et al., 2001}. In the latter survey, the preva­lence of serum retinol <0.70µmol/L was very low for all age and sex groups. However, the preva­lence of potentially suboptimal concen­tra­tions of serum retinol, defined as <1.05mol/L, was high among certain minority groups of children. Racial and ethnic differences were significant, even after controlling for confounding factors; non-Hispanic African American and Mexican American children were more at risk than were non-Hispanic Caucasian children.

Table 18a.2 Cutoff points recommended by the NHANES II committee for interpreting low serum total vitaminA concen­tra­tions in three age categories. Vitamin A status refers to serum and tissue levels of the nutrient. Impairment of function may include impaired dark adaptation, night blindness, and ocular lesions. From Pilch (1987).

Vit. A μmol/L	Age 3–11 y	Age 12–17 y	Age 18–74 y
<0.35	Vitamin A status is very likely to improve with increased consumption of vitamin A. Impairment of function is likely.
<0.70	Vitamin A status is likely to improve with increased con- sumption of vitamin A	Vitamin A status is likely to improve with increased con- sumption of vitamin A; some individuals might exhibit impairment of function.	Vitamin A status is likely to improve with increased con- sumption of vitamin A; impairment of function is likely.
0.70 - 1.05	Vitamin A status of some sub- jects may improve with in- creased consumption of vita- min A. Improvement is most likely in those with values 0.70 - 0.95µmol/L.	Vitamin A status may improve with increased consumption of vitamin A. Improvement is more likely in those with values 0.70 - 0.95µmol/L.	Vitamin A status may improve with increased consumption of vitamin A. Some individuals may exhibit impairment of function.

The U.S. serum retinol cutoff points used for NHANES III are based on their rela­tion­ship with func­tional indices of vitamin A defi­ciency, such as impaired dark adaptation, night blindness, ocular lesions, and possibly impaired immune function and on the serum vitamin A distri­butions from the earlier NHANES investigations (Pilch 1987). It is not known if these cutoffs are appropriate for other countries.

The World Health Organization uses a single cutoff for serum retinol concen­tra­tions of <0.70µmol/L as indicative of a low vitamin A status for children 6‑71mo(WHO, 2011). The cutoff for infants <6mo is not defined. Currently, this cutoff is applied irrespective of the preva­lence of infection in a partic­ular popu­la­tion.

WHO (2011) has also proposed that preva­lence levels of serum retinol concen­tra­tions <0.70µmol/L be used to indicate the possible presence of a public health problem. These define whether vitamin A is either a mild (2% to 9% preva­lence), moderate (10% to 19% preva­lence), or severe (>20% preva­lence) public health problem.

Mea­sure­ment of serum retinol

Special precautions must be taken in the preparation and storage of serum samples for vitamin A. Serum should be centrifuged soon after the blood is drawn, where possible, and hemolysis and exposure to bright light should be avoided. If separation is not possible within a short time, then the blood sample should be stored in an icebox with ice (Mejia and Arroyave, 1983). If serum is to be stored for prolonged periods of time prior to analysis, it should be quickly frozen in temper­ature resistant tubes, and then flushed with argon or oxygen-free nitrogen. The tubes should be closed tightly with a screw top, and only a small gas space should be left within the tube. Tubes should be stored, together with a reference serum sample, in the dark at temper­atures below −70°C (Olson, 1984). Thawing and refreezing should be kept to a minimum because this may disrupt RBP.

Several methods are available for the analysis of total serum vitamin A or retinol. Only high-perfor­mance liquid chromatography can distinguish retinol from the retinyl esters. All the other methods measure total serum vitamin A and generally give comparable results, provided that conditions of collection, storage, and analysis are clearly controlled.

A standard reference material (SRM 968 Fat-Soluble Vitamins in Human Serum) is available from the National Institute of Standards and Technology (Gaithersburg, MD) which is certified for both retinol and retinyl palmitate The within-subject daily CV for serum retinol for subjects following their normal dietary pattern was 11.3% (Gallagher et al., 1992).

High-perfor­mance liquid chromatography (HPLC) is the method of choice for the separation and analysis of serum retinol because it has the required precision at low concen­tra­tions (de Pee and Dary, 2002). It is also specific and easy to use (WHO, 1996). Usually 100‑200µL of serum or plasma are required for analysis. Either reverse- (Bieri et al., 1979; Furr et al., 1984), or normal-phase HPLC is used (Bankson et al., 1986). The former technique is preferred for serum and was used in NHANES III (Ballew et al., 2001). All interfering compounds, such as retinyl esters and phyto­fluene, can be separated from retinol using HPLC. Moreover, mild hemolysis does not affect serum retinol concen­tra­tions measured by HPLC (Marinovic et al., 1997).

Fluorometric methods based on the direct mea­sure­ment of fluorescence of the retinol-binding complex are used for measuring total serum vitamin A. They are highly sensitive and critical timing is not necessary. Nevertheless, other highly fluorescent substances, partic­ularly phyto­fluene and phytoene, interfere but can be removed by column chromatography. Alternatively, a correction factor can be used (Thompson et al., 1971). Severe but not moderate hemolysis may be a confounding factor which could lead to serum vitamin A concen­tra­tions being underestimated (Marinovic et al., 1997).

A portable fluorometer optimized for vitamin A fluorescence that operates from a direct current power supply (battery, automobile cigarette lighter, or AC/DC power supply) has been developed for field use. It weighs less than 6kg, has no moving parts and uses a cuvet-shaped fluorescent calibration block. Use of this instrument eliminates the necessity for the transport and storage of samples and can generate immediate results provided a portable centrifuge is also available. It reduces the cost of serum retinol analyses (Craft, 2001).

Dried blood spots (DBS) can be used for the mea­sure­ment of retinol (Erhardt et al., 2002). Positive correlations have been recorded between retinol concen­tra­tions in dried blood spots and conventional plasma samples (Figure 18a.3, Craft et al., 2000).

Figure 18a.3 HPLC chromatogram of the separation of retinol extracted from a DBS. (Craft et al., 2000). Retinol and the internal standard, retinyl acetate, elute at 3.6 and 4.0min, respectively. HPLC conditions: Betasil C8, 3µm, 4.6 × 150mm column, methanol / water (95/5 v/v) 1.0mL/min flow rate, 325nm detection.

Blood should be collected from a fingerprick onto filter paper (Schleicher and Schuell 903 specimen collection paper), dried for 3h in the dark, and stored with a desiccant in the dark in bags and at room temper­ature (i.e., up to 25°C) before analysis. Storage under humid conditions causes loss of retinol in DBS and should be avoided.

Extraction of the DBS retinol can be followed by analysis using normal-phase HPLC with a highly sensitive detector. To reduce the variability of the method, it is recommended that the individual volume of the spots be considered by measuring their sodium content by flame photometry or by weighing the blood spots before extraction. The method has been used on blood spot samples from India, Nepal, Nicaragua, Indonesia, and Guatemala.

18a.2.2 Serum retinol-binding protein

Retinol-binding protein (RBP) is a specific vitamin A transport protein. It is called holo‑RBP when it is bound to retinol; the portion without retinol is called apo‑RBP. If the liver becomes depleted in retinol, as occurs in the late stages of vitamin A defi­ciency, RBP accumulates in the liver as apo‑RBP, and levels of both serum retinol and RBP decline.

Retinol-binding protein has a single high affinity binding site for a molecule of retinol. The resultant complex of RBP ⨦ retinol, together with one molecule of plasma transthyretin, form a trimolecular complex. Thus, the molar ratio of retinol to RBP in circulation is about one to one. Concen­tra­tions of serum RBP may thus be used as a surrogate measure for serum retinol. The assay of RBP is partic­ularly useful in popu­la­tions where resources and technical support are limited: sample collection and the analytical procedures are easier and cheaper than for serum retinol, and the analysis can be performed on serum from a fingerprick blood sample.

Several studies have confirmed the positive corre­lation of serum retinol and serum RBP(Solomons et al., 1990; Almekinder et al., 2000; Hix et al., 2004). In a study in the Republic of the Marshall Islands of children who were deficient in vitamin A, Gamble et al. (2001) showed significant correlations between serum retinol and serum RBP (r=0.94) across all retinol concen­tra­tions. Severe vitamin A defi­ciency (serum retinol <0.35µmol/L) was predicted with 96% sensitivity and 91% specificity using serum RBP at 0.48µmol/L as the cutoff, and more moderate vitamin A defi­ciency (serum retinol <0.70µmol/L) with 87% sensitivity and 98% specificity using a serum RBP cutoff of 0.70µmol/L. Hence, the mea­sure­ment of serum RBP concen­tra­tions appeared to be a practical alternative to using serum retinol in this popu­la­tion. Similar findings were noted for pregnant women in Malawi ((Almekinder et al., 2000).

Nevertheless, a variety of factors may influence the binding of RBP to retinol. These include the presence and magnitude of the acute-phase response, protein-energy malnu­tri­tion, liver disease, and chronic renal failure. To overcome these confounding effects, use of the ratio of serum RBP: transthyretin (TTR) has been proposed (Rosales and Ross, 1998; Rosales et al., 2002). Transthyretin is unaffected by vitamin A status but, like RBP and serum retinol, decreases in infection and trauma. Hence, theoretically, the ratio of RBP:TTR should decline during vitamin A defi­ciency but not during an acute-phase response, thus enabling these two conditions to be distinguished. Indeed, Rosales et al. (2002) suggest that a cutoff value of 0.36 for RBP:TTR is indicative of marginal vitamin A defi­ciency. However, use of this ratio has been questioned (Filteau et al., 2000). and more research on the use of the RBP:TTR ratio is warranted.

Interpretive criteria for RBP

There is no consensus on a cutoff value for RBP equivalent to a serum retinol concen­tra­tion of <0.70µmol/L (de Pee and Dary, 2002). More research on the rela­tion­ship of serum RBP and serum retinol concen­tra­tions in popu­la­tions with a range of serum retinol concen­tra­tions is required before such a cutoff can be defined and is likely dependent on the kit or method used to evaluate RBP. In the interim, investigators are advised to first establish the rela­tion­ship between serum retinol and serum RBP in a subsample of the popu­la­tion under study by HPLC before using serum RBP as a surrogate for serum retinol concen­tra­tions. This is especially impor­tant during pregnancy, when both the transport and the metabolism of RBP are altered (Sapin et al., 2000).

Mea­sure­ment of RBP

Retinol-binding protein, unlike serum retinol, is not photosensitive and is less temper­ature sensitive and more stable during refrigeration — all attributes that facilitate its use in field surveys.

The RBP assay can be performed using a specific and sensitive radioimmunoassay procedure in which the RBP is bound to radioactively labeled antibodies. This method has a lower limit of detection of 0.5nmol/L (Blaner, 1990; Rosales, 1998). Alternatively, a rapid quantitative enzyme immunoassay (EIA) can be used that has an average intra‑ and inter-assay variability of 6.7% and 8.0%, respectively (Hix et al., 2004). Tests indicate that the RBP EIA correlates well with serum retinol measured by HPLC. Positive correlations between serum RBP measured by radial immunodiffusion and EIA have been reported (Hix et al., 2004).

An EIA method can also measure RBP in dried blood spots. The kit contains three calibrants over the deficient to normal range for RBP. In the future, it may become feasible to measure RBP directly in a drop of whole blood using a portable fluorometer in the field (Craft, 2001).

A sandwich ELISA method has been developed where RBP, C‑reactive protein and α‑1‑acid glyco­protein can be analyzed in a single run (Erhardt et al., 2004). When properly standardized, this assay could replace serum retinol analysis by HPLC at a fraction of the cost.

18a.2.3 Serum retinyl ester

In normal healthy people, retinyl esters constitute less than 5% of the total vitamin A content in fasting serum samples. However, when the capacity of the liver to store vitamin A is exceeded — for example, after the chronic ingestion of excessive amounts of vitamin A (hyper­vitamin­osis A), or in liver disease — vitamin A is released into the circulation as retinyl esters, and then elevated concen­tra­tions of serum retinyl esters are observed. For example, in three patients with hyper­vitamin­osis A, retinyl esters made up 67%, 65%, and 33% of the total vitamin A present in the plasma (Smith and Goodman, 1976).

A cutoff point of retinyl esters >10% of total vitamin A was used to reflect potential hyper­vitamin­osis. Ballew et al. (2001). studied the validity of this cutoff by examining the distri­bution of concen­tra­tions of serum retinyl esters in samples from NHANES III. They noted that 37% of the sample of adults aged 18y had serum retinyl ester concen­tra­tions above the cutoff point, but they were unable to find any associations between serum retinyl ester concen­tra­tions and five bio­chem­ical indices of liver dysfunction. None­the­less, elevated retinyl esters >7.5% of total was asso­ci­ated with a liver reserve of 3µmol/g (Olsen et al., 2018). and 5% of total was suggested for use in children where more sensitive biomarkers of liver reserves of vitamin A noted hyper­vitamionsis A (Mondloch et al., 2015; Williams et al., 2021).

Fasting blood samples are essential for serum retinyl ester mea­sure­ments because concen­tra­tions rise transiently after the ingestion of a vitamin A-rich meal or of vitamin A supple­ments. Mea­sure­ment of retinyl ester concen­tra­tions in serum can be performed using HPLC. The low levels in fasting serum can be measured concurrently with serum retinol concen­tra­tions.

18a.2.4 Serum carot­enoids

Approximately 50 carot­enoids show pro­vita­min A activity and provide about 50% of the total vitamin A intake in the United States, as well as larger percentages in Asia, Africa, and parts of South America. In those countries where dietary carot­enoids from plants are the major source of vitamin A and where dietary patterns are relatively constant, serum carot­enoids may serve as a useful secondary index of vitamin A intake. For popu­la­tions receiving most of their vitamin A from animal sources, serum carotenoid concen­tra­tions provide no inform­ation on vitamin A status.

Major components of serum carot­enoids are lutein, zeaxanthin, β‑crypto­xanthin, lycopene, α‑carotene and β‑carotene. Serum carot­enoids are of increasing interest because of their possible antioxidant properties and their rela­tion­ship to the risk of certain cancers, cardioЩvascular disease, macular degeneration, and the onset of cataracts (Omenn et al., 1996).

Unfortunately, several non-nutritional factors are known to influence serum carotenoid concen­tra­tions, including age, sex, alcohol intake, physio­logical state, body mass index, and season (Brady et al., 1996; Neuhouser et al. 2001). Smoking may also modify the rela­tion­ship of dietary-carotene and serum-carotene levels (Järvinen et al., 1993).

Serum carot­enoids were measured following NHANES III. Ford (2000) and Ford et al. (2002) measured serum concen­tra­tions of α‑carotene, β‑crypto­xanthin, lutein, zeaxanthin, and lycopene in samples from NHANES III and presented data for the mean and selected percentiles by age, sex, and race or ethnicity. Serum carotenoid concen­tra­tions were not uniformly distributed among the popu­la­tion groups. Children, adolescents, and adults of African descent had the highest total carotenoid concen­tra­tions whereas overweight children, Caucasian adolescents, and adults had the lowest.

Plasma β‑carotene, α‑crypto­xanthin, lycopene, and lutein results are available for the U.K. national surveys (Gregory et al., 1990; 1995; 2000; Finch et al., 1998). Mean, median, and lower and upper 2.5th or 5th percentiles by age and sex are presented.

carot­enoids can be both separated and measured by HPLC combined with ultraviolet / visual detection (Sowell et al., 1994) or photodiode array(Mondloch et al., 2o15; Williams et al., 2021). Care must be taken to store frozen serum samples for carotenoid analysis at −70°C or −80°C to avoid decay. A standard reference material for β‑carotene can be obtained from NIST (Gaithersburg, MD). The within subject daily CV for serum β‑carotene for subjects following their normal dietary pattern is 7.2% (Gallagher et al., 1992).

18a.2.5 Breast milk retinol

Breast milk retinol concen­tra­tions can indicate when the maternal vitamin A intake is sub­optimal, because lactating women will secrete breast milk with a reduced content of retinol. This usually reflects inade­quacies in both maternal dietary intake and body stores of vitamin A. Breast milk retinol concen­tra­tions can also be used as an indirect indicator of the vitamin A status of breastfed infants (Stoltzfus and Underwood, 1995).

Most of the vitamin A in breast milk is in the form of retinyl palmitate in the milk fat. Concen­tra­tions are very high in colostrum and transitional breast milk (days 7‑21 postpartum). After about day 21, the concen­tra­tion stabilizes, so that breast milk samples taken after the first month postpartum are the most useful. The concen­tra­tion of retinol in breast milk varies over the course of a feed: the lowest concen­tra­tion is in the first milk expressed from a full breast because the fat content is highest at the end of a feed. Hence, preferably all the milk from a full breast that has not been used to feed an infant for at least 2h should be collected.

Stoltzfus et al. (1993). used full breast milk samples to investigate the effect of high‑dose vitamin A sup­plement­ation on the vitamin A status of mothers and their infants in Indonesia. At 1‑8mo postpartum, the breast milk retinol concen­tra­tions of the supple­mented mothers were significantly higher than in the placebo group (Figure 18a.4).

Figure 18a.4 Retinol concen­tra­tions (µmol/L) in breast milk at baseline and during the subsequent 8mo in supple­ment and placebo groups. Data from (Stoltzfus et al., 1993).

In practice, collecting the entire contents of one breast is difficult to achieve. Instead, a standardized collection procedure can be used to assess vitamin A status at the individual level. This involves collecting a small sample of breast milk (8‑10mL) from a full breast before the infant starts suckling, either by manual self-expression or by using a breast pump (Stoltzfus and Underwood, 1995).

For popu­la­tion assess­ment, casual breast milk samples can be taken randomly through­out the day and at varying times after the infants were last fed. This will ensure that the variation in milk fat is randomly sampled. However, if random sampling is not possible, then the breast milk retinol concen­tra­tions can be expressed in terms of the fat content (Table 18a.3). (WHO, 1996). In a study of rural women from Bangladesh, Rice et al. (2000) noted that breast milk vitamin A concen­tra­tions expressed per gram of milk fat were a partic­ularly responsive indicator of vitamin A status.

Table 18a.3 Response to postpartum vitamin A supplementation measured by maternal and infant indicators. The values shown are means ±SD. A natural log transformation was used in all cases to improve normality except for the serum retinol data. The means and SDs of the transformed values are presented. [n], number of samples. Data from Rice et al., 2000.

Indicator (month post-partum)	Vitamin A group [n]	Placebo group [n]	Standardized difference
Breast milk vit.A (μg/g fat) in casual samples (3 mo)	2.05 ± 0.44 [36]	1.70 ± 0.47 [37]	0.76
Breast milk vit.A (μmol/L) in casual samples (3 mo)	0.12 ± 0.70 [36]	–0.18 ± 0.48 [37]	0.50
Maternal serum retinol (μmol/L) (3 mo)	1.45 ± 0.47 [34]	1.33 ± 0.42 [35]	0.27
Breast milk vit.A (μmol/L) in full samples (3 mo)	–0.33 ± 0.74 [33]	–0.45 ± 0.53 [35]	0.19
Breast milk vit.A (μg/g fat) in full samples (3 mo)	1.87 ± 0.51 [33]	1.82 ± 0.45 [35]	0.10

There are two circumstances when it may be preferable to express breast milk retinol concen­tra­tions per unit volume rather than per gram of fat. First, when a field survey includes mothers with a wide range in the stage of lactation, there will be a large standard deviation around the mean retinol concen­tra­tions because of the marked variation in milk retinol concen­tra­tion from early to late lactation. This large variation may be masked by expressing the retinol concen­tra­tion in terms of fat content. Secondly, in studies in which both the β‑carotene and the fat content of breast milk are increased because of a vitamin A intervention, retinol concen­tra­tions expressed per unit volume will provide a better measure of the response than when expressed in terms of the fat content (de Pee et al., 1997).

Interpretive criteria

The average retinol content of breast milk from vitamin A-sufficient lactating women ranges from 1.7 to 2.5µmol/L; values are often <1.4µmol/L for vitamin A-deficient mothers. The cutoff value defined by WHO (1996) at which both mother and breastfed infant are likely to be at risk for vitamin A defi­ciency is <1.05µmol retinol/L or <28nol/g milk fat. The sensitivity of breast milk retinol using <1.0µmol/L milk was 42% using total liver reserves of 0.10µmol retinol/g as the reference standard for definition of vitamin A defi­ciency in Zambian women (Kaliwile et al., 2021).

WHO (1996) defined two popu­la­tion preva­lence levels that are indicative of a public health problem for breast milk concen­tra­tions. When 10% to <25% of the sample popu­la­tion of mothers have breast milk retinol concen­tra­tions <1.05 µmol/L, then the public health problem for vitamin A defi­ciency is classified as moderate; when the preva­lence is >25%, the problem is severe.

Mea­sure­ment of breast milk retinol

After collecting breast milk samples, they must be placed imme­di­ately in an insu­lated ice box for transport from the field to the base laboratory. If the analysis cannot be completed imme­di­ately, samples may be frozen below −20°C. Care must be taken to ensure that the thawed samples are thoroughly mixed before aliquots are removed for analysis into amber or yellow poly­propy­lene tubes with airtight caps. Analysis of retinol in breast milk can be performed by HPLC, after saponifying the breast milk samples. The fat content of breast milk can be deter­mined in the field by using the creamatocrit method. This involves collecting a sample (about 75µL) of well‑mixed breast milk into a hematocrit capillary tube, which is then sealed at one end. The tubes are spun in a hematocrit centrifuge, after which the length of the cream layer is measured. The amount of fat is deter­mined by comparison with a standard curve of fat concen­tra­tions compiled using a standard lipid assay (Ferris and Jensen, 1984).

18a.2.6 Relative dose response

The relative dose response (RDR) test can be used to estimate the liver stores of vitamin A and thereby identify individuals with marginal vitamin A defi­ciency. The test is based on the observation that during vitamin A defi­ciency, when liver stores are diminished, RBP accumulates in the liver as apo‑RBP. Following the admin­istra­tion of a test dose of vitamin A, some of the vitamin A binds to the excess apo‑RBP in the liver. It is then released as holo‑RBP (i.e., RBP bound to retinol) into the circulation (Loerch et al., 1979). Consequently, in vitamin A- depleted individuals, there is a rapid and sustained increase in serum retinol after a small oral dose of vitamin A, whereas in individuals with normal liver vitamin A stores, this rise in serum retinol is either very small or does not occur.

The validity of the RDR test as an index of body stores of vitamin A has been studied by comparing vitamin A concen­tra­tions in liver biopsy samples with corresponding RDR test results for otherwise healthy surgical patients (Amédée-Manesme et al., 1984; 1987). Of the twelve surgical patients, the two with the lowest liver vitamin A concen­tra­tions had the highest RDR values. Following sup­plement­ation with vitamin A, RDR values fell to <5%. Patients with liver vitamin A concen­tra­tions ranging from 0.2 to 1.5µmol/g had RDR values from 0% to 12% (Amédée-Manesme et al., 1984).

In a study of Brazilian children from low-income families, all the children with serum retinol concen­tra­tions <0.70µmol/L had elevated RDR values. Moreover, 86% of the children with serum retinol concen­tra­tions of 0.74‑1.02µmol/L, and 26% with serum retinol concen­tra­tions of 1.05‑1.40µmol/L also had elevated RDR values Flores et al., 1984) (Table 18a.4).

Table 18a.4 Percentage of children with a positive RDR test classified by serum retinol levels. From Flores et al. (1984)

Serum retinol (µmol/L)	%	No. tested
≤0.70	100	12
0.71 - 1.04	86	21
1.05 - 1.39	26	19
>1.40	3	39
Total tested		91

After sup­plement­ation with vitamin A, all the elevated RDR values reverted to normal. These results indicate that the RDR is a more sensitive index of marginal vitamin A status than using serum vitamin A concen­tra­tions <0.70µmol/L, but it does not distinguish among different levels of adequate vitamin A reserves (Solomons et al., 1990).

Other factors asso­ci­ated with low RDR values include malab­sorp­tion, infection, liver disease, severe protein-energy malnu­tri­tion, and zinc defi­ciency. Such factors reduce the sensitivity and the specificity of the RDR test (Mobarhan et al., 1981; Russell et al., 1983. Campos et al., 1987). For example, when an oral dose of vitamin A was given to patients with varying degrees of liver dysfunction and protein-energy malnu­tri­tion, no corre­lation was observed between the vitamin A content of liver biopsies and the RDR test result (Mobarhan et al., 1981; Russell et al., 1983). These results were attributed to malab­sorption of the oral dose because when an intravenous injection of retinyl palmitate was given to children with liver disease, the RDR test proved to be a reliable and sensitive index of vitamin A status (Amédée-Manesme et al., 1987).

Protein defi­ciency interferes with the RDR test by decreasing liver synthesis of the rapid turnover apo-RBP, while zinc defi­ciency interferes with the liver synthesis or secretion of RBP (Christian and West, 1998). A further limitation of the RDR test is low precision, which may lead to serious errors in classi­fying subjects with vitamin A defi­ciency, especially in popu­la­tions where the preva­lence is moderate (Solomons et al., 1990).

Interpretive criteria for the RDR test

Vitamin A-replete subjects have RDR values ranging from zero to 14%. Relative dose response values >20% are generally considered indicative of inadequate hepatic stores of vitamin A and marginal vitamin A status (WHO, 1996). When >20% to <30% of the sample popu­la­tion show abnormal RDR values (i.e., >20%), then a public health problem of moderate importance may be assumed. When the preva­lence is >30%, the public health problem is severe (WHO, 1996). A systematic review and meta-analysis revealed 68% sensitivity and 85% specificity for the RDR test using total liver reserves of 0.10µmol retinol/g as the reference standard for definition of vitamin A defi­ciency (Figure 18a.5) (Sheftel and Tanumihardjo, 2021).

Figure 18a.5 The dose response tests compared with liver retinol concen­tra­tions from multiple studies. The symbols represent individual or group values, triangles: rats, squares: calves, diamonds: swine, and circles: humans. A relative dose response (RDR) value of >20% and an A2/A1 for the modified relative dose response (MRDR) test of >0.060 are considered vitamin A defi­ciency

Mea­sure­ment of relative dose response

Two blood samples are needed for the RDR test. A baseline fasting blood sample is taken imme­di­ately before the admin­istra­tion of a small oral dose (450 to 1000µg) of vitamin A (as retinyl acetate or retinol palmitate). Next, a high‑fat snack that contains minimal vitamin A is consumed to ensure absorption of vitamin A. The second blood sample is taken 5h later. The RDR (%) is calculated as:

\[\frac{\text { plas. retinol }(5 \mathrm{~h})-\text { plas. retinol }(0 \mathrm{~h}) \times 100 \%}{\text { plas. retinol }(5 \mathrm{~h})}\nonumber\]

Concen­tra­tions of serum retinol for the RDR test should be measured by reverse-phase HPLC because this method has the required precision at low concen­tra­tions. Even a small analytical error can alter the RDR value significantly, especially when serum retinol concen­tra­tions are low.

18a.2.7 Modified relative dose response

The modified relative dose response (MRDR) test requires only one blood sample, avoiding the necessity of taking a baseline sample. This modified test has been used to assess the vitamin A status in children and in pregnant and lactating women (Tanumihardjo et al., 1990; 1994; 1995; 1996).

For the test, a small oral dose of vitamin A is administered, usually 3,4‑di­dehydro­retinyl acetate (DRA). This is followed by a high-fat, low-vitamin A snack to facilitate the absorption of the DRA. The latter combines with the RBP in the same way as retinol but, unlike retinol, is not normally found in human plasma except when high levels of freshwater fish are consumed. The DRA is hydrolyzed in the gastrointestinal tract to 3,4‑di­dehydro­retinol (DR), absorbed, and reesterified, mainly with palmitic acid in the intestinal mucosal cells (Tanumihardjo et al., 1995). The serum concen­tra­tions of DR and retinol (R) in the single blood sample are measured by HPLC. The molar ratio of [DR]:[R] in the blood sample is a measure of the response.

Interpretive criteria for the MRDR

Ratios >0.060 are considered indicative of marginal vitamin A status (subclinical defi­ciency), whereas those <0.030 are satisfactory (Tanumihardjo et al., 1996). The WHO (1996). criteria for a moderate public health problem are a preva­lence of MRDR ratios >0.060 of between 20% and 30%. If the preva­lence is >30%, the public health problem is severe. Because of the preferred use of the MRDR in subsets of popu­la­tion studies, these criteria are being reexamined.

A limitation is that DRA is not yet commercially available and presently must be synthesized in the laboratory. None­the­less, the test has been used in several popu­la­tion health surveys (Faye et al., 2021; Williams et al., 2021; Suri et al., 2021). A systematic review and meta-analysis revealed 80% sensitivity and 69% specificity for the MRDR test using total liver reserves of 0.10µmol retinol/g as the reference standard for definition of vitamin A defi­ciency (Figure 18a.5) (Sheftel and Tanumihardjo, 2021).

Mea­sure­ment of the MRDR

The recommended doses of DRA for use in the field are 3.0µmol for infants, 5.3µmol for preschool children (2 to 6y), 7.0µmol for preadolescent children (6‑12y), and 8.8µmol for teenagers and adults (>12y) (Tanumihardjo et al., 1996). The acceptable time between administering the oral DRA dose and obtaining a single blood sample (about 0.5‑2ml) is 4‑7h. To enhance the stability of the DRA dose, it should be dissolved in corn oil and stored in amber vials inside a cooler on ice in the field, or at −20°C to −70°C for long‑term storage. The blood sample should also be stored on ice in a light-protected cooler after collection, prior to separation of the serum. Analyses of DR and retinol (R) concen­tra­tions in the serum should be performed by HPLC. The molar ratio of [DR]:[R] (sometimes reported as A2/A1) is then calculated.

18a.2.8 Subjective assess­ment of night blindness

Night blindness, or the inability to see after dusk or at night, has been reported among young children and women of reproductive age in developing countries with moderate to severe vitamin A defi­ciency (Escoute et al., 1991; Christian et al., 1998). It is the most common ocular manifestation of vitamin A defi­ciency and is often described by specific terms in countries or cultures where the preva­lence is high. For example, in some cultures it is known as "chicken eyes" or "chicken blindness" (Christian P. 2002.). Poor dark adaptation resulting in night blindness arises when there is reduced production in the rods of the visual pigment rhodopsin, or opsin protein bound to the retinal form of vitamin A.

To assess the preva­lence of night blindness, the local term for night blindness must first be identified through focus group discussion (Dawson et al., 1993), and its reliability field tested. Next, a night blindness history is elicited via interviews. Care should always be taken during this stage to exclude those individuals whose night blindness results from other causes, such as the rare hereditary eye disease retinitis pigmentosa.

Table 18a.5 Mean serum vitamin A level in Indonesian preschool children by clinical indicators of vitamin A status: night blindness and conjuctival xerosis with Bitot's spots. Data from Sommer et al., (1980).

Clinical status	n	Mean (µmol/L)
Night blindness reported; no conjunctival xerosis or Bitot’s Spots	174	0.49
Controls	161	0.62
No night blindness; conjunctival xerosis and Bitot’s Spots present	51	0.47
Controls	45	0.60
Night blindness reported; with conjunctival xerosis and Bitot’s Spots	79	0.42
Controls	76	0.64

Some studies of children >2y have been based on interviews to elicit any history of night blindness. In general, reliable data on night blindness cannot be deter­mined for children <2y in this way because they are not able to move around freely after dusk and bump into objects. In studies of children >2y, WHO (1996) recommends the use of specific questions to increase the specificity and reduce misclassification of self-reported night blindness. Sommer et al. (1980) reported that 85% of Indonesian preschool children with reported night blindness had low serum vitamin A concen­tra­tions (i.e., <0.70µmol/L). This strong association between night blindness and serum retinol levels is shown in Table 18a.5. This also shows similar positive findings for a second clinical sign of vitamin A defi­ciency — Bitot's spots with conjunctival xerosis.

Determination of the preva­lence of night blindness is recommended as a tool to assess vitamin A defi­ciency in women of reproductive age. When using maternal night blindness in this way, only women with a previous pregnancy that ended in a live birth in the past 3y should be included (Christian P. 2002). Significant associations between maternal night blindness and low serum and breast milk retinol concen­tra­tions, as well as func­tional indices such as abnormal CIC and impaired dark adaptation (Section 18.1.9), have been reported among pregnant Nepalese women, as shown in Table 18a.6.

Table 18a.6 Association of night blindness with other indicators of vitaminA defi­ciency in pregnant women in Nepal. N, sample size; n, number of subjects below cutoff; OR, odds ratio; CI, confidence interval; CIC, conjunctival impression cytology. From Christian, (2002).

Vitamin A status indicators	Night blind N n(%)	Not night blind N n(%)	OR	95% CI
Serum retinol <0.7 μmol/L	85 44 (51.0)	90 19 (21.1)	4.0	2.2–7.4
Serum retinol <1.05 μmol/L	85 65 (76.5)	90 50 (55.5)	2.5	1.4–4.6
CIC abnormal	85 24 (28.2)	90 11 (12.2)	2.8	1.3–6.1
Dark adaptation abnormal	94 67 (71.2)	98 42 (43.8)	3.3	1.8–6.0
Breast milk vit. A <1.05 μmol/L	94 56 (59.6)	97 41 (42.3)	2.0	1.1–3.6

In a large popu­la­tion-based control study in Nepal, women with night blindness were also less likely to consume dietary sources of vitamin A (Christian et al., 1998). In a randomized placebo-controlled sup­plement­ation trial in Nepal, women in the placebo group with night blindness during pregnancy had a mortality rate four times higher than did women without night blindness. This mortality level was reduced by 68% after weekly vitamin A sup­plement­ation (Christian et al., 1998).

Interpretive criteria

A recommended cutoff at which vitamin A defi­ciency is considered a significant public health problem within a community is when the preva­lence of maternal night blindness is 5% or greater (Christian, 2002). This cutoff value takes into account the misclassification that may occur (i.e., false positives) among women who report problems with daytime vision (i.e., daytime blindness). Efforts should be made during surveys to exclude those women whose night blindness is likely due to daytime blindness. WHO (1996) have proposed cutoffs for the preva­lence of night blindness for children 24‑71mo. Three recommended cutoffs for the preva­lence of night blindness to define a public health problem and its level of importance (i.e., mild, moderate, or severe) are given; these are shown in Table 18a.7 together with minimum sample sizes required. However, the large minimum sample sizes required together with the increasing use of vitamin A supple­ments among young children in most developing countries preclude the use of these WHO cutoffs for children. Instead, in most low‑income countries, the assess­ment of nightblindness in women of reproductive age is probably more useful.

Table 18a.7: preva­lence of night blindness in children 24‑71mo, and the minimum sample sizes for identifying a mild, moderate, or severe vitamin A defi­ciency public health problem with a relative precision of 20% and 50% at the 95% confidence level. From WHO (1996).

	preva­lence	Minimum sample for 20% relative precision	Minimum sample for 50% relative precision
Mild	<1%	—	—
Moderate	≥1% to <5%	4706	753
Severe	≥5%	1825	292

18a.2.9 Rapid dark adaptation test

Before night blindness develops, disturbances in dark adaptation occur. These can be detected by specially designed noninvasive tests (Congdon et al., 1995). The conventional laboratory-based, formal dark adaptometry test is a tedious and time- consuming procedure. Instead, a rapid dark adaptation test (RDAT), suitable for field conditions, has been developed. This is based on the mea­sure­ments of the timing of the Purkinje shift (Thornton, 1977), in which the peak wavelength sensitivity of the retina shifts from the red toward the blue end of the visual spectrum during the transition from photopic or cone‑mediated day vision to scotopic or rod‑mediated night vision. This shift causes the intensity of blue light to appear brighter than that of red light under scotopic lighting conditions.

The RDAT requires a light‑proof room, a light source, a dark, nonreflective work surface, a standard X‑ray view box, and sets of red, blue, and white discs; details are given in Vinton and Russell (1981).

Mea­sure­ments for the RDAT are undertaken during the first few minutes of dark adaptation. This is a disadvantage because the mea­sure­ments rely mainly on the cones or light vision cells in the retina instead of the rods or dark vision cones. As a result, the test is not very sensitive to the early signs of vitamin A defi­ciency (Favaro et al., 1986; Kemp et al., 1988). An additional disadvantage is that the RDAT is not appropriate for preschool children, who are too young to perform the test accurately. This is unfortunate because preschool children are the group most at risk for vitamin A defi­ciency. False positives may also occur and between examiner variability may produce inconsistencies in the results.

Age influences dark adaptation and hence must be taken into account when determining the normal range of the rapid test adaptation times for healthy reference popu­la­tions (Vinton and Russell, 1981).

18a.2.10 Pupillary and visual threshold test

A scotopic (dim light vision) device has been developed to assess the responsiveness and sensitivity of the pupil to light as an indication of an individual's dark adaptation threshold. The test measures the threshold of light at which pupillary contraction occurs under dark-adapted conditions (Congdon et al., 1995). A portable field dark adaptometer was developed that includes a digital camera, retinal bleaching flash, and Ganzfeld light source housed within a pair of light-obscuring goggles (Labrique et al., 2015). Pupillary response mea­sure­ments may be more affordable for countries wishing to use this device.

Unlike the RDAT, the pupillary and visual threshold test can be conducted in the field on individuals of all ages, including preschool children, who cannot be tested by the RDAT. The test requires minimal cooperation from the subjects and takes about 20 min per subject. A darkened facility is required for this test. A portable tent has been developed and tests devised to ensure that the darkness in the testing area is adequate (Sanchez et al., 1997). The apparatus consists of two handheld illuminators, each having a yellow-green, light-emitting diode light source (dominant wavelength = 572nm) with 12 intensity settings. Each illuminator is designed to fit entirely over one eye and to illuminate the entire retina. One of the illuminators is designed to measure light of "low intensity" (illumination range: 8.75 to ‑3.00 log cd/m²) and the other "high intensity" (illumination range ‑4.16 to 0.44 log cd/m²).

Before testing, participants are subjected to a camera-flash "partial bleach" of the full retina, which involves placing the subjects in a dark room and exposing them to a camera flash reflected through a foil-lined cone, after which they are allowed 10 min of dark adaptation. The visual threshold is measured first by placing the low‑intensity illuminator over the subject's left eye. The light intensity is then incremented over 11 intensity settings (roughly a 4‑log unit range) at 10s intervals until a pupillary response (i.e., quick contraction of the pupil on presenting the stimulus) is seen in the uncovered right eye on two successive trials. The uncovered right eye is observed with an obliquely mounted red LED light source (dominant wavelength = 626nm), which preserves dark adaptation in both the subjects and the observer. Next, the pupillary threshold is measured, as described above, using the high intensity illuminator; further details are given in Congdon et al. (1995; 2000). and in Christian et al. (2001). All tests should be performed using standardized procedures and well-trained examiners, as discussed in Christian et al. (2001).

The stimulus for the visual or the pupillary threshold is defined as the lowest level at which the subject can correctly distinguish stimulus from non-stimulus on three successive trials. High pupillary and visual scores reflect a pupillary response achieved at a greater light intensity, and indicate poorer dark adaptability (Congdon et al., 1995).

Additional studies are needed to establish whether testing for both pupillary and visual thresholds is necessary, because the latter requires less training and standardization of personnel. In some studies on young children aged 1 to 2y, only pupillary testing has been performed because complete visual testing on such young children is not always possible (Congdon et al., 1995).

The pupillary and visual threshold tests have been validated as an index of vitamin A status using controlled vitamin A sup­plement­ation trials on children and pregnant women (Congdon et al., 1995; 2000; Congdon and West, 2002). Significant improvements in dark adaptation as assessed by visual and/or pupillary testing were reported in those subjects supple­mented with vitamin A but not with a placebo. Moreover, dark adaptation scores, measured by visual and/or pupillary testing, correlate well with serum retinol, as shown in Figure 18a.6, and with RDR (Congdon et al., 1995; 2000; Sanchez et al., 1997).

Figure 18a.6 Dark-adaptation threshold and serum retinol concen­tra­tion of pregnant Nepali women receiving β-carotene, vitamin A, and placebo. Dark-adaptation threshold and serum retinol concen­tra­tion were inversely correlated in the placebo (regression coefficient = −0.014, P = 0.0005) and β‑carotene (regression coefficient = −0.009, P = 0.005) groups. Modified from Congdon et al. (2000).

Nevertheless, there are limitations associated with the pupillary and visual threshold tests. For example, some investigators have shown overlapping values for deficient and nondeficient popu­la­tion groups, so that scotopic threshold testing may be more suitable for popu­la­tion assess­ment than for individuals. Further, the time required to administer the test is about 20 min per subject, so it is probably not feasible to test large numbers of individuals. Sanchez et al. (1997) suggest that for popu­la­tions at high risk for vitamin A defi­ciency, as few as six subjects may be sufficient to show that the group mean thresholds differ significantly from normal. In popu­la­tions with mild vitamin A defi­ciency, testing about 100 subjects is probably adequate (Sanchez et al., 1997).

Another limitation is that recovery of the normal pupillary response after dosing with vitamin A supple­ments takes about 4‑6wk. One must wait at least this long before retesting treated individuals. Finally, the period required to train examiners to recognize the pupillary response is 1‑3d.

Two cutoff values were proposed to indicate whether vitamin A defi­ciency is a problem or not in an area, based on a mean pupillary dark adaptation score for a popu­la­tion. Scores worse than the cutoff of −1.11 log cd/m² are said to indicate vitamin A defi­ciency, whereas scores better than the normal cutoff of −1.24 log cd/m² should indicate that a popu­la­tion is normal, or that an intervention has successfully improved vitamin A status (Congdon and West, 2002). The popu­la­tion mean should be calculated for this test, rather than the proportion falling below a specific value, as has been stipulated for the other tests of vitamin A nutriture. This approach is recommended because in popu­la­tions with marginal vitamin A status, the small proportion of subjects with abnormal values for the test could potentially markedly increase the sample size required for testing.

18a.2.12 Stable isotope methods and total body stores of vitamin A

Retinol isotope dilution (RID) tests are the only method that indirectly provide a quantitative measure of the hepatic stores of vitamin A. It involves the admin­istra­tion of an oral dose of isotopically labeled vitamin A. The dose is allowed to equilibrate with the vitamin A pool in the body. A post-dose blood sample is taken, and the ratio of labeled to non-labeled vitamin A in serum is measured by mass spectrometry. The amount of total body stores of vitamin A is related to the extent of dilution of the labeled tracer and is calculated using prediction equations (Gannon and Tanumihardjo, 2015).

The isotope dilution method was validated in adult surgical patients in the United States and Bangladesh by measuring liver vitamin A biopsy samples (Furr et al., 1989; Haskell et al., 1997). For example, in ten U.S. surgical patients, the corre­lation coefficient between calculated and measured liver vitamin A concen­tra­tions was 0.88.

The length of time required for the isotopic dose to equilibrate with the vitamin A pool in the body varies according to the age and total body vitamin A stores of the study group. Equilibration periods after the test dose ranging from 11 to 26d have generally been used (Furr et al., 1989; Haskell et al., 1998; Ribaya-Mercado et al., 1999). although Tang et al. (2002) suggest that 3d may suffice. A shorter equilibration time facilitates the use of this isotope-dilution method in field settings, but may not offer an accurate quantitative estimate. The MRDR test should be considered if enough time is not available in the field to obtain the post-dose sample.

The RID test has been used mostly in research settings, but the application for use in subsets of popu­la­tion surveys is on the horizon (Tanumihardjo, 2020). The International Atomic Energy Authority has assisted many low-income countries on its use (Sheftel et al., 2018). A combined analysis from five African countries revealed stark differences in vitamin A status based on total liver vitamin A reserves estimated by RID. Ethiopia was experiencing vitamin A defi­ciency, while South Africa had a high rate of hyper­vitamin­osis A (Suri et al., 2023).

Interpretive criteria

The RID test using appropriate assumptions, gives a prediction of the total liver vitamin A reserves (Gannon and Tanumihardjo, 2015). Therefore, a value that represents vitamin A status is deter­mined, which reflect that actual hepatic reserve. While best used at the group level, vitamin A status can be viewed across the continuum, such as that represented in Figure X.

Hepatic liver reserves to define vitamin A defi­ciency and toxicity

The criteria to define vitamin A defi­ciency were evaluated in light of the Dietary Reference Intakes (DRIs) for North Americans (Tanumihardjo, 2021 ). The criteria used to formulate DRIs were clinical eyes signs, circulating plasma retinol concen­tra­tions, bile excretion containing vitamin A metabolites, and long-term vitamin A storage. In consideration of DRI criteria, induced biliary excretion and long-term vitamin A storage do not occur until liver vitamin A concen­tra­tions are >0.10µmol/g (Tanumihardjo, 2021 ). Therefore, in line with recom­men­dations of an expert group working on Biomarkers of Nutrition for Development (Tanumihardjo et al., 2016), vitamin A defi­ciency should be defined as <0.10 µmol/g liver.

Direct measure of hepatic vitamin A concen­tra­tions are only feasible in special circum­stances, either during surgery (Furr et al., 1989; Haskell et al., 1997). or on cadavers (Olsen et al., 2018). In children (n=366) dying from multiple causes, vitamin A defi­ciency was found in 34.2% and hyper­vitamin­osis A was deter­mined in 8.7% (Gupta et al., 2024). This underlines the fact that vitamin A defi­ciency is still a public health concern in young children.

18a.2.13 Multiple indices

Vitamin A defi­ciency disorders are not defined with any certainty using a single measure of vitamin A status. WHO (2011). has recommended a combination of bio­chem­ical, func­tional, and clinical indicators for children aged 6-71mo; the indicators are given in Table 18a.8. together with the corresponding cutoff values used in popu­la­tion assess­ment. Some caution is needed when applying these cutoffs to individuals because of the influence of the many confounding factors. Table 18a.8 also provides inform­ation on the popu­la­tion preva­lence levels of each indicator for defining a mild, moderate, or severe public health problem in relation to vitamin A.

Table 18a.8: Indicators of subclinical vitamin A defi­ciency in children 6–71mo of age and their use in detecting a public health problem. The preva­lence in a popu­la­tion of at least two of the above indicators of vitamin A status, at levels below the cutoff, indicates a public health problem. RDR, relative dose response; MRDR, modified relative dose response. From (WHO, 1996).

Indicator (cutoff)	Mild preva­lence below cutoff	Moderate preva­lence below cutoff	Severe preva­lence below cutoff
Night blindness present	>0 to <1%	≥1% to 5%	≥5%
Serum retinol ≤ 0.70 μmol/L	≥2% to <10%	≥10% to <20%	≥20%
Breast milk retinol ≤ 1.05 μmol/L or ≤ 0.028 μmol/g (≤ 8μg/g) milk fat	<10%	≥10% to <25%	≥25%
RDR ≥20%	<20%	≥20% to <30%	≥30%
MRDR ratio ≥0.06	<20%	≥20% to <30%	≥30%

According to WHO (2011). a public health problem exists when a popu­la­tion has at least two biological indicators with a preva­lence above the level corresponding to a defi­ciency, or one biological indicator of defi­ciency supported by a composite of at least four demographic and ecological risk factors, two of which must be nutrition‑ and diet-related. Details of these demographic and ecological risk factors are given in WHO (2011).

The selection of the most appropriate combination of criteria depends on the purpose of the study, the expected range of vitamin A status for the study group, and the resources available. WHO often uses mortality outcomes to define its guidance, especially for young children (WHO, 2011).