Introduction: Vitamin-A deficiency and child health and survival

Reynaldo Martorell

From the early days of its discovery, vitamin A has been popularly known as the anti-infection vitamin [1]. This was due to the evidence that quickly accumulated from animal models which showed that vitamin A deficiency led to impaired immunocompetence and to increased susceptibility to and duration and severity of infections. Documentation of these effects in humans lagged far behind; in fact, it is only recently that well-conducted studies have been initiated.

The Committee on International Nutrition Programs (CINP) was formed in 1970 to provide guidance to the Office of Nutrition of the US Agency for International Development (USAID) on scientific questions relevant to international nutrition programmes. The ClNP's interest in vitamin-A deficiency and child health and survival has had a long history. For example, in 1976 the CINP issued a report entitled Possible Health Benefits of Vitamin A Prophylaxis Programs in Addition to the Prevention of Xerophthalmia and Blindness [1]. The report reviewed the literature and concluded that it was not known whether providing vitamin A to deficient human populations would result in health benefits other than those associated with the eyes. In a report that same year entitled Priorities for Research on Avitaminosis A and Xerophthalmia, the CINP identified, as one of four priority areas, the need for epidemiological studies of avitaminosis A and its relation to infection [2]. The committee went on to offer specific suggestions for studies of vitamin A and immunity and for assessing the impact of vitamin-A control programmes on infection.

A number of important studies carried out subsequent to the publication of these two reports supported a link between vitamin A and child survival.

Studies in Indonesia suggested an association between vitamin-A deficiency and mortality in children with signs no more severe than the World Health Organizations categories of XNC (night blindness) and XIB (Bitot's spots) [3]. In a subsequent community trial, children living in Indonesian villages without a programme of vitamin-A supplementation were found to have a mortality rate 35% greater than those in a control village with a programme 14]. Few studies have created as much interest and debate in the international health and nutrition community as this one. Questions have been raised regarding the validity of the findings because of issues involving design and measurement [5].

The possibility that child mortality in developing countries can be so dramatically reduced by means of a simple and cheap intervention is encouraging if it can be confirmed. There is a marked contrast between the relative ease and low cost of implementing an effective programme of vitamin-A supplementation (i.e. providing two doses of 200,000 IU orally to children a year) and the requirements for other approaches to improving health and survival, such as water and sanitation programmes or the eradication of protein-energy deficiency, which even at their best do not usually show the type of immediate and dramatic effects on mortality that were observed in Indonesia. Given the potential policy implications of the Indonesian findings, the CINP felt it was important to determine whether the mortality effects observed could be replicated in other settings using the best possible epidemiological methods.

The late director of the Office of Nutrition of USAID, Dr. Martin Forman, also saw the need for further research, and at his request the CINP established a Subcommittee on Vitamin A Deficiency Prevention and Control. In August 1986 the subcommittee conducted a workshop that culminated in the report Methodologies for Field Trials of Vitamin A Supplementation [5], which dealt primarily with methods for assessing the impact of vitamin-A supplementation on child mortality in order to provide scientific and technical guidance for the design of the several field trials planned by USAID to verify that impact in developing countries.

Subsequently, the Office of Nutrition requested that the subcommittee explicitly consider the mechanisms that might explain the hypothesized effects of vitamin A on mortality. To this end, a workshop was held on 28 April 1988. The five papers that follow are a result of this workshop.

The first represents the subcommittee's technical recommendations for conducting studies of the mechanisms through which vitamin A might affect mortality, and refers primarily, but not exclusively, to studies of vitamin A and infection. The subcommittee's recommendations took into account the papers presented at the workshop, the general discussion which these generated, and subsequent deliberations by its members. The other four papers were presented at the workshop.

Drs. Chandra and Vyas's review of animal and human research on vitamin A and immunocompetence and De Luca and McDowell's review of the effects of vitamin-A status on hamster tracheal epithelium in viva and in vitro indicate that vitamin A plays a number of physiological functions that contribute to host defence. These two papers offer reasonable mechanisms that would explain a relationship between vitamin-A deficiency and increased incidence and severity of infection.

The studies of vitamin-A deficiency and risk of respiratory and gastrointestinal infections are critically reviewed by Dr. Forman, who points out that, despite some strong epidemiological support for an association between vitamin-A deficiency and infection, the issue remains clouded by design and measurement problems. In the fifth paper, Dr. Porter reviews ethical issues relating to nutrition field trials. Ethics is always a consideration in research involving humans, and in the case of studies of vitamin A and child health and survival, ethical discussions have loomed large. For example, the Subcommittee on Vitamin A Deficiency Prevention and Control could not come to a consensus regarding the designs to be recommended for trials of vitamin-A supplementation and mortality, and two of its members wrote minority reports questioning the majority's endorsement of controlled trials, a type of study where a control population does not receive vitamin-A supplementation. The majority felt rigorous scientific designs were preferred in order to quickly establish whether or not vitamin-A supplementation to deficient populations has a significant impact on child health and survival as suggested by the Indonesian findings. Some of the recent vitamin-A studies funded by USAID have also generated heated discussions of ethical issues and have had to be modified as a result.

Vitamin-A deficiency is widespread but unevenly distributed among developing countries. For those countries in which vitamin-A deficiency is a public health problem, there are effective programmes that can be implemented, including supplementation, fortification, horticulture, and nutrition education. These must compete for funding with programmes for ameliorating or eradicating other health and nutrition problems. Governments must make difficult choices in selecting the mix of programmes to be implemented; clearly, impact on child health and survival ought to be a salient consideration. Thus it is important for us to know with certainty what effects vitamin-A improvements would have and under what conditions. These answers are best obtained through adequately designed and carefully conducted clinical and epidemiological studies. It is hoped that the work of the Subcommittee on Vitamin A Deficiency and Control and the contributions of the scientists whose papers are included here point in the right directions.

References

1. National Research Council. Possible health benefits of vitamin A prophylaxis programs in addition to the prevention of xerophthalmia and blindness. Report of the Committee on International Programs, Food and Nutrition Board. Washington, D.C.: National Academy of Sciences, 1976.
2. National Research Council. Priorities for research on avitaminosis A and xerophthalmia. Report of the Committee on Intemational Programs, Food and Nutrition Board. Washington, D.C.: National Academy of Sciences, 1976.
3. Sommer A, Tarwotjo I, Hussaini G. Susanto D. Increased mortality in children with mild vitamin A deficiency. Lancet 1983;2:585-88.
4. Sommer A, Tarwatjo I, Djunacdi E, et al. Impact of vitamin A supplementation on childhood mortality: a randomized controlled community trial. Lancet 1986; 1:1169-73.
5. National Research Council, Subcommittee on Vitamin A Deficiency. Vitamin A supplementation: methodologies for field trials. Washington, D.C.: National Academy Press, 1987.

Research priorities for investigation of the influence of vitamin-A supplementation on morbidity

A statement

by the Subcommittee on Vitamin A Deficiency Prevention and Control

The Subcommittee on Vitamin A Deficiency Prevention and Control of the Food and Nutrition Board's Committee on International Programs conducted workshop on Strategies and Priorities for Research on the Influence of Vitamin A Supplementation on Morbidity on 28 April 1988. The purpose of the workshop was to consider the influences of vitamin A on the immune system, the role of vitamin A in the differentiation and maintenance of epithelial tissue, the current research and recent findings on vitamin A and morbidity, and the ethical dimensions of such research.

This report was written by the subcommittee subsequent to the workshop and represents the consensus of its members. It does not give equal weight to all hypotheses considered at the workshop but emphasizes those relating vitamin A to infections within the context of field studies. In addition, the report includes a detailed discussion of methods of assessment of vitamin-A status. It does not discuss the ethical issues in any great detail because these were discussed very thoroughly in its previous report and in Dr. Porter's paper below (pp. 36-40).

The following hypotheses were addressed at the workshop:

• Marginal vitamin-A deficiency adversely affects the integrity of epithelial tissue.
• Marginal vitamin-A deficiency lowers resistance to infection by affecting the immune system.
• Marginal vitamin-A deficiency is a risk factor for the increased incidence and severity of diarrhoeal and respiratory infections.
• Vitamin-A supplementation to populations in which vitamin-A status is marginal increases immunocompetence and reduces the incidence and severity of diarrhoeal and respiratory infections.

The word marginal is used to describe the range of vitamin-A deficiency of interest to the subcommittee; it refers to the subclinical stage preceding the appearance of night blindness and the conjunctival and corneal changes characteristic of xerophthalmia. Although it is difficult to define marginal vitamin-A deficiency adequately, it is nonetheless clear that in countries where xerophthalmia is a significant public health problem a large proportion of children are afflicted with marginal deficiency. The World Health Organization considers prevalences of corneal xerosis and keratomalacia (X2 + X3A + X3B)* in excess of 0.01% to indicate a significant public health problem affecting large numbers of children [1]. In such areas, many more children will show less severe clinical signs (X1A, X1B) and night blindness (my), and even more will have low plasma vitamin A, low vitamin-A stores in the liver, and vitamin-A-deficient diets. Xerophthalmia is known to be associated with diminished host defences and increased risk of infection. Similarly, marginal vitamin-A deficiency may lower resistance to infection, a possibility that needs to be confirmed and quantified through adequately designed studies. Because so many children are likely to be affected by marginal vitamin-A deficiency, the study of this condition and its implications constitutes a significant research priority.

The subcommittee's first report considered effects of vitamin A on mortality and provided advice on the conduct of field studies designed to test the hypothesis that vitamin-A supplementation lowers child mortality rates [2]. Mortality studies require very large samples in order to achieve satisfactory statistical power.

The study of morbidity related to vitamin A is more feasible since infections occur more frequently and fewer subjects need be studied. Mortality and morbidity studies are complementary. The demonstration of an effect of vitamin-A supplementation on the incidence and severity of gastrointestinal and respiratory infections would strengthen the persuasiveness of findings on mortality because such a demonstration would provide a plausible pathway leading to the mortality effects. Similarly, the demonstration of effects of marginal vitamin-A deficiency on epithelialtissue differentiation and maintenance and on immunocompetence would further validate both morbidity and mortality findings.

Research gaps

The subcommittee has assessed progress to date in regard to morbidity effects and has determined that further research is required. Salient research gaps are as follows.

Immuncompetence

Vitamin-A deficiency both in man and in laboratory animals impairs immunological responses [3]. Although not all of the immune responses have been tested in man, the consensus of results in published studies indicates that the following parameters are altered. Reduced vitamin-A intake and low serum-rehnol concentrations are associated with decreased delayed hypersensitivity response, lower serum-antibody response to some antigens, decreased lymphocyte response to mitogens, decreased natural-killer-cell activity, delayed rejection of grafts, changes in complement level, and phagocyte dysfunction. Several published reviews are available [4-6]. In animals deficient in vitamin A, morbidity and mortality increased after infectious challenge 13]. The significance of these findings in humans remains to be established.

In general, changes in immunocompetence often precede obvious infection and can be viewed as a functional index of nutritional deficiency. Changes in immune responses may mediate heightened susceptibility to infection in populations deficient in vitamin A.

More information should be obtained through research in the following areas:

• epithelial-tissue integrity and barrier function in vitamin-A-deficient individuals, particularly in marginal stages of deficiency, including examination of the effect of vitamin-A supplementation;
• the pattern of recovery of immunocompetence following either small, frequent physiological amounts of vitamin A or massive pharmacological doses of vitamin A;
• secretory immunoglobulin and cell-mediated immune responses to immunization and the extent of protective immunity achieved in vitamin-A-deficient groups;
• investigation of the mechanism(s) by which vitamin-A deficiency alters immunocompetence;
• examination of the quantitative and temporal relationship between vitamin-A status and different indices of immune function.

Morbidity

The literature suggests that marginal vitamin-A deficiency is associated with increased incidence or severity of infections (or both). The evidence seems to be much stronger for respiratory than for gastrointestinal infections. The interpretation of the results to date is difficult because studies have failed to fully document vitamin-A status or to control for factors associated with both vitamin-A deficiency and the risk of infection.

An example is the need to control for environmental sanitation. Vitamin-A-deficient children may be from lower-income families with poorer housing and sanitary facilities. Children with adequate vitamin-A status, even when living in the same communities, may be from better-off homes. Thus, failure to control for environmental sanitation would overestimate the importance of vitamin A for morbidity. Similarly, vitamin-A deficiency generally coexists with other nutritional problems that are thought to affect the risk of morbidity. Studies of marginal vitamin-A deficiency and morbidity must at least control for protein-energy malnutrition and other deficiencies such as of zinc and iron.

Further research is recommended in the following areas:

• Identification of the extent to which the incidence and severity of infections are affected by marginal vitamin-A deficiency.
• The relation of marginal vitamin-A deficiency to specific types of morbidity: What types of morbidity are most affected by vitamin-A deficiency? What is the natural outcome of commonly occurring infections in the community and the role of immunocompetence in mediating the outcome? Are there different effects of vitamin-A supplementation or depletion on different infections - for example, on measles, rotavirus, and bacterial or viral pneumonias?
• Interaction of vitamin-A status with other deficiencies (for example, protein-energy, zinc, and iron) in regard to effects on morbidity.
• Interaction of vitamin-A status with factors related to socio-economic status, such as level of exposure to infection and access to health services.
• Effectiveness of massive doses of vitamin A versus lower doses given daily or weekly, adverse effects of massive doses, and mechanisms responsible for these different effects.

Indicators of vitamin-A status

In its first report [2], the subcommittee noted the importance of adequately characterizing the vitamin-A status of study populations. Such data will allow for stratification by vitamin-A status if required in the data analysis and will increase flexibility in extrapolating the results to other populations. The report also noted the problems in the measurement of vitamin-A status that require further methodological work:

• Biochemical indicators: How reliable and accurate are serum concentrations of vitamin A as indicators of overall vitamin-A status? How useful is the relative dose-response test?
• Cytological indicators: How useful is impression cytology as an index of vitamin-A status? Can buccal cytology be used instead of conjunctival cytology?
• Clinical history: How accurate are historical reports of night blindness?
• Dietary histories: How well do dietary histories for consumption of vitamin-A-rich foods actually correlate with vitamin-A intake and vitamin-A status?

Vitamin-A status is determined by total body stores and can be thought of in relative terms as deficient, depleted, adequate, excessive, or toxic. Deficient, excessive, and toxic concentrations are manifested by clinical signs and symptoms or by biochemical measures interpretable both for individuals and populations. Quantitative measurement of intermediate levels of vitamin A (depleted and adequate) is problematic and in need of research.

Vitamin-A status is also affected by absorption and other factors that increase metabolism, including infection and protein-energy malnutrition. Such factors should be considered in the design of studies to investigate the impact of vitamin A on morbidity.

Field-applicable indicators of marginal vitamin-A status currently recognized include serum concentrations, the relative dose-response test, and conjunctival-impression cytology. Each of these has limitations in feasibility and practicality in the field as well as in quantitative interpretation. Detailed discussions of methodologies for clinical histories and dietary histories are not specific to studies of vitamin A and morbidity and thus will not be discussed in detail here.

Serum concentrations

The interpretation of serum concentrations of vitamin A as reflectors of status on an individual and population basis have been discussed in a previous publication [2]. In the past, the interpretation of population distribution curves was largely based on comparisons with non-representative population surveys from developing countries from which cutoff points for deficient, low, and acceptable levels were derived [7]. Recently available are age- and sex-specific distribution curves for representative, relatively well nourished populations living in the United States [8; 9]. Although these reference curves appear to be independent of ethnic differences, they are influenced by environment [10; 11], and their usefulness as a universal reference needs to be validated. The most appropriate reference curve for serum levels for field studies of vitamin A and morbidity would be from a representative population known to have adequate body stores while living in an environment characteristic of an at-risk population. Studies are needed to establish appropriate reference curves for populations residing outside the United States and in high-risk areas. To obtain such references, distribution curves established before and repeated after large-dose supplements - for example, 30 days after a 200,000IU supplement - would be appropriate. These curves could then be compared with the US reference curves to determine whether there are differences.

Relative dose-response test

Whereas to demonstrate a rise in homeostatically controlled serum concentrations in response to increased intake or supplementation requires serially obtained samples with an intervening period for stabilization, the relative dose-response (RDR) test can detect depleted stores with only a five-hour interval between dosing and sampling [12; 13]. This test, modified to suit various conditions, has been successfully applied in both clinical and field-survey conditions. It has been validated by an intravenous procedure for dosing coupled with direct liver biopsy in a limited number of children with liver disease and adults [14;15]. The number of direct quantitative validation studies needs to be expanded to obtain greater assurance as to the range and confidence intervals of the association of a positive RDR test with vitamin-A concentrations in the liver. However, this can be accomplished only under specialized clinical circumstances, not in field studies in developing countries. Indirect validation with a before-and-after supplementation procedure also needs to be extended to population groups in which heavy intestinal parasitism, malabsorption, and protein-energy malnutrition are common. In addition, a proposed adaptation of the RDR method using didehydroretinol (DR, or vitamin A₂), a natural derivative of retinal (vitamin Al, the common form of vitamin A in foods and formed in the gut from carotenoids) [16], needs to be tested in human populations. If the DR adaptation proves reliable in human populations, only a single blood sample after five hours might be required, making the test much more feasible for field morbidity studies [17]. The RDR approach to assessment of depletion requires additional validity tests under field conditions to verify that mild to moderate protein-energy malnutrition, infections, or other conditions that may lower absolute serum concentration do not impede this short-term response and hence limit the applicability of the test for reliably indicating depletion of body stores.

Conjunctival-impression cytology

The histologically based conjunctival-impression cytology (CIC) test requires fewer logistical and technological resources than a direct biochemical determination of vitamin A. In theory, this increases its practicality for field studies in developing countries [18|. Validation against liver biopsies and the RDR test has been reported in a clinical setting among a very limited number of children with liver disease [19]. These studies suggest that the absence of goblet cells in the impressions corresponded to a liver concentration of <20 microg per gram, or a positive RDR, or both. Additional quantitative validation studies are needed but, as with the RDR, can be accomplished only under special clinical conditions. In a larger field study conducted in Guatemala, the CIC evaluated against the RDR was reported to lack sensitivity although it was specific [20; 21]. This field validation trial needs to be repeated. The effect of potential confounders such as concurrent eye infections, e.g. conjunctivitis and trachoma, should also be studied. The report of a strong linear correlation between abnormal CIC and relative plasma concentrations is not consistently reported by others under clinical or field conditions and needs confirmation [18]. Interpretation of the histological picture relative to degrees of vitamin-A depletion needs to be standardized.

Research needs in relation to vitamin-A assessment are as follows:

• Population reference curves of serum vitamin-A concentrations should be validated for developing countries.
• The relative dose-response test should be further validated under appropriate field conditions.
• Field trials to validate the conjunctival-impression cytology technique should be repeated.
• Standardization of the interpretation of histological findings for conjunctival-impression cytology relative to vitamin-A status is needed.

Indicators of nutritional status

The following issues related to indicators of nutritional status should be considered in designing studies of vitamin A and morbidity:

• What is the nature of the relation among nutritional status indicators and morbidity and immunocompetence? Is the relation linear, or are thresholds evident?
• While wasting is known to be related to increased risk of morbidity, the relation with stunting is less clear and should be explored further. In particular, studies should assess whether the process of stunting (linear growth retardation) is associated with diminished immunocompetence and increased morbidity.
• How should feeding histories be recorded, and which variables are most essential (i.e., duration of exclusive breast-feeding, type and timing of supplementary feeding, and duration of breastfeeding)?

Indicators of morbidity

• What are the validity and reliability of different methods of morbidity measurement?
• What is the definition of a "case" or episode?
• What are the diagnostic criteria for causal agents?

Measures of immunocompetence

The following measures of immunocompetence may be considered in the evaluation of vitamin-A-deficient individuals and populations before and after any proposed intervention. The rationale for choosing these four tests is based on the consistency of abnormalities observed in published studies and will be discussed further in the subcommittee's next report. The ultimate choice will be dictated by the nature of the study and practical considerations such as the availability of laboratory facilities and costs.

• delayed hypersensitivity,
• lymphocyte response to mitogens,
• salivary-immunoglobulin-A concentration,
• natural-killer-cell activity.

It would be important to role out the confounding effect of concurrent or recent infection on the basis of clinical findings and estimation of complement-reactive protein and endotoxins in the blood.

Strategies for research

The subcommittee recognizes that no single research design and set of procedures can or should be followed in a variety of settings in different countries with differing social contexts and resource limitations that affect feasibility. Nonetheless, the subcommittee wishes to present some guidelines that may prove useful in designing future studies.

Choice of population

It is desirable to test the major hypotheses in populations with various levels of vitamin-A nutriture.

Choice of deign

The subcommittee strongly recommends that double-blind randomized studies with placebo controls be used where feasible (as always, with due consideration for scientific and ethical issues). A controlled study is likely to yield information of greater scientific validity than a non-experimental study (e.g. a case-control study). Thus, experimental studies are preferable for unequivocally testing the hypotheses of interest.

In its previous report [2], the subcommittee reviewed theoretical and practical aspects of controlled trials. It noted, for example, that baseline comparability is ensured on average by random assignment of treatment and control interventions. The choice of the level of randomization - by individual, household, or community - is a matter of balancing considerations of study size with feasibility. If cluster sampling is used, appropriate statistical methods should be followed.

The likelihood of strict baseline comparability can be improved by randomization within strata on factors that relate to the outcome variables and that vary substantially among units to be randomized. Adequate stratification increases the precision of estimates, but must be weighed against the added field work that may result.

If strict comparability at baseline is not achieved in unstratified studies, adjustments are still possible during data analysis either by stratification or by regression analyses. This is possible, of course, only if the appropriate data have been collected at baseline.

The subcommittee recognizes the value of non-experimental studies and believes that they should be conducted where experiments are not feasible. Case-control studies, in particular, seem appropriate for studies of marginal vitamin-A status and risk of morbidity.

Case definition for such studies may present problems, however. If a case is defined as a child with diarrhoea at a given time and a control as a child without, the definition of a case will be weak as a result of the ephemeral nature of diarrhoea. Case definition could be improved by collecting histories of past infections; recall biases should be dealt with on the basis of the type of data. Another possible approach would be to select cases on the basis of marginal vitamin-A status, identify matched controls, and measure outcome through longitudinal follow-up. Such a design raises ethical issues related to the failure to treat those with identified clinical conditions. The subcommittee concurs with Feachem [22], who holds that such studies are unethical.

A major problem in case-control studies is the difficulty of satisfactorily matching for characteristics known to affect both the outcomes and vitamin-A status in the same direction. Case-control studies carried out to date have been equivocal because of poor matching. Variables to be matched should include age, sex, anthropometric and socio-economic status, nutritional status (anthropometric indicators of protein-energy malnutrition, indicators of iron and zinc deficiency, and other indicators where appropriate), environmental sanitation, and access to health care.

Content of the proposal

Below are listed issues which should be addressed in the design of proposals for future studies of vitamin-A supplementation and its relation to morbidity.

Specify the hypothesis

Research proposals should clearly state whether the hypothesis is to test the effect of marginal vitamin-A status on morbidity or to test the effect of vitamin-A supplementation on morbidity outcomes. If the hypothesis concerns testing the effect of marginal vitamin-A status, quantitative measures of depletion of vitamin A will be needed. The stated hypothesis should also indicate whether there is any intent to test the mechanisms by which vitamin-A deficiency or supplementation may lead to changes in morbidity - for example, by changes in immunocompetence. It should also state whether the intent is to study only morbidity per se or also severity of illness.

Specify sampling considerations and population description

• What type of population will be studied? What information is available that may be significant for vitamin-A studies?
• Is randomization to be done on the basis of the village or on an individual level?
• If randomization is on the level of village or cluster, how will cluster sampling affect estimates of sample size?

Specify expected effects

• What magnitude of effect on incidence or duration of morbidity is expected from vitamin-A supplementation? Is there any experimental basis for this estimate?
• How would different assumptions about the expected magnitude of the impact on morbidity affect sample size?

Specify methods

• Morbidity experience (for example, how will the frequency and duration of diarrhoea or respiratory disease be quantified?). Will attempts be made to identify causative agents?
• Data-collection plans.
• Indicators to be used for vitamin-A status (biochemical, clinical, dietary).
• Nutritional status (including feeding history, dietary history, and protein-energy, zinc, and iron nutritional status).
• Socio-economic status (for example, income, occupation, possessions, housing, education, exposure to infections, and access to services).
• Immunization status for immunizable diseases (especially measles).
• Intervening mechanisms such as immunocompetence.

Describe quality-control and data-management procedures

• What quality-control procedures will be implemented to ensure adequate coverage of the targeted population and satisfactory data quality?
• What data-management procedures will be used to ensure the availability of adequate information for analysis?

Specify the analysis plan

This should detail the specific methods to be used to test the study hypotheses and should, in addition, address the following issues:

• How will the analysis adjust for possible confounding factors such as nutritional status and socioeconomic status?
• What indicators will be used to quantify the key variables in the analysis (for example, vitamin-A status, nutritional status, socio-economic status, morbidity status)?
• Which indicators will be treated as dichotomous variables and which as continuous variables?
• If cluster sampling is used, how will adjustments be made for design effects when calculating confidence intervals?
• How will the analysis deal with multiple data points collected on the same individual? For example, if height and weight data are collected at several points, which measurements or combinations of measurements will be used to define nutritional status?
• What assumptions will be made in the statistical analysis, and what will be the effect of likely deviations from these assumptions?

This report is intended to provide guidance for developing programmes to assess the impact of vitamin A on child morbidity. It also gives an overview of the research priorities in this area, focusing on field studies of the impact of vitamin A on infectious diseases.

Subcommittee on Vitamin A Deficiency

Prevention and Control

Reynaldo Martorell (Chairman), Food Research Institute, Stanford University, Stanford, California

Abdelmonem A. Afifi, School of Public Health, University of California, Los Angeles, California

Guillermo Arroyave, School of Family and Consumer Sciences, San Diego State University, San Diego, California

Ranjit Kumar Chandra, Directory of Immunology, Memorial University of Newfoundland and Janeway Child Health Centre, St. John's, Newfoundland, Canada

Frank Chytil, Department of Biochemistry, School of Medicine, Vanderbilt University, Nashville, Tennessee

Samuel Preston, Population Studies Center, University of Pennsylvania, Philadelphia, Pennsylvania

Mervyn W. Susser, Columbia University, New York, New York

Frederick Trowbridge, Division of Nutrition, Center for Health Promotion and Education, Centers for Disease Control, Atlanta, Georgia

Barbara A. Underwood, National Eye Institute, National Institutes of Health, Bethesda, Maryland

Virginia H. Laukaran, Staff Officer, Food and Nutrition Board

Susan Berkow, Staff Officer, Food and Nutrition Board

Frances Peter, Deputy Director, Food and Nutrition Board

Jean Shirhall, Editor

References

1. World Health Organization. Control of vitamin A deficiency and xerophthalmia. Report of a joint WHO/ UNICEF/USAID/Helen Keller International/IVACG meeting. WHO technical report series, no. 672. Geneva: World Health Organization, 1982.
2. National Research Council-National Academy of Sciences, Food and Nutrition Board. Vitamin A supplementation: methodologies for field trials. Report of the Subcommittee on Vitamin A Deficiency Prevention and Control. Washington, D.C.: National Academy of Sciences, 1987.
3. Beisel WR. Single nutrients and immunity. Am J Clin Nutr 1982;35:417-68.
4. Chandra RK, ed. Nutrition and immunology. New York: Alan R. Liss, 1988.
5. Keusch G. Nutrition and infection. In: Remington JS, Schwartz MN, eds. Current clinical topics in infectious diseases. New York: McGraw-Hill, 1984:106-23
6. Vyas D, Chandra RK. Vitamin A and immunocompetence. In: Watson RR, ed. Nutrition, disease resistance and immune function. New York: Marcel Dekker, 1984:325-44.
7. Arroyave G. Chichester CO, Flores H. et al. Biochemical methodology for the assessment of vitamin A status. Washington, D.C: Vitamin A Consultative Group, Nutrition Foundation, 1982.
8. Pilch SM, ed. Assessment of vitamin A nutritional status of the US population based on data collected in the Health and Nutrition Examination Surveys. Prepared for the Center for Food Safety and Applied Nutrition. US Food and Drug Administration Contract No. FAD 233-84-2059. Bethesda, Md, USA: Life Sciences Reseach Office, Federation of American Societies for Experimental Biology, 1985.
9. Pilch SM, ed. Analysis of vitamin A data from the Health and Nutrition Examination Surveys. J Nutr 1987;117:636-40.
10. Looker AC, Johnson CL, Woteki CE, Yetley EA, Underwood BA. Ethnic and racial differences in serum vitamin A levels of children aged 4-11 years. Am J Clin Nutr 1988;47:247-52.
11. Looker AC, Johnson CL, Underwood BA. Serum retinal levels of persons ages 4-74 years from three Hispanic groups. Am J Clin Nutr 1988;48: 1490-96.
12. Flores H. Campos F. Araujo CRC, Underwood BA. Assessment of marginal vitamin A deficiency in Brazilian children using the relative dose-response procedure. Am J Clin Nutr 1984;40:1281-89.
13. Loerch ID, Underwood BA, Lewis KC. Response of plasma levels of vitamin A to a dose of vitamin A as an indicator of hepatic vitamin A reserves in rats. J Nutr 1979;109:778-86,
14. Amédée-Manesme O. Mourey MS, Hanck A, Therasse J. Vitamin A relative dose response test: validation by intravenous injection in children with liver disease. Am J Clin Nutr 1987;46:286-89.
15. Amédée-Manesme O. Anderson D, Olson JA. Relation of the relative dose response to liver concentrations of vitamin A in generally well-nourished surgical patients. Am J Clin Nutr 1984;39:898-902.
16. Tanumihardjo SA, Olson JA. A modified relative dose-response assay employing 3,4-didehydroretinol (vitamin A2) in rats. J Nutr 1988;118:598-603.
17. Olson IA. Indicators of vitamin A status. Bull Xerophthalmia Club 1988;38:3.
18. Natadisastra G. Wittpenn JR, West KP, Muhilal, Sommer A. Impression cytology for detection of vitamin A deficiency. Arch Ophthalmol 1987;105:1224-28.
19. Amédée-Manesme O. Luzeau R. Wittpenn IR, Hanck A, Sommer A. Impression cytology detects subclinical vitamin A deficiency. Am J Clin Nutr 1988;47:875-78.
20. Gadomski AM, Kjolhede CL, Wittpenn J. Bulux 1, Rosas AR, Forman MR. Conjunctival impression cytology (CIC) to detect subclinical vitamin A deficiency: comparison of CIC with biochemical assessments. Am J Clin Nutr 1989;49:495-500.
21. Kjolhede CL, Gadomski AM, Wittpenn J. et al. Conjunctival impression cytology (CIC): feasibility of a field trial to detect subclinical vitamin A deficiency. Am J Clin Nutr 1989;49:490-94.
22. Feachem RG. Vitamin A deficiency and diarrhoea: a review of interrelationships and their implications for the control of xerophthalmia and diarrhoea. Trop Dis Bull 1987;84:RI-R16.

Vitamin A, immunocompetence, and infection

Ranjit K. Chandra and Devhuti Vyas

The interactions between nutrition, immunity, and infection have been the focus of much recent work. It has been observed that protein-energy malnutrition (PEM) is associated with impaired immunocompetence [1-3], including depressed cell-mediated immunity, phagocyte dysfunction, decreased levels of complement components, reduced mucosal secretory antibody responses, and lower antibody affinity. At the same time, the complexity of clinical malnutrition is recognized. This has led to the examination of immunological effects produced by single nutrient deficiencies.

Among other nutrients, vitamin A has a profound effect on immune responses in man and laboratory animals. In humans, vitamin-A deficiency seldom occurs in isolation; it is usually associated with varying degrees of PEM. In many countries there is a high prevalence of xerophthalmia and night blindness among infants and children suffering from PEM. However, the numerical association between PEM and xerophthalmia varies enormously in different geographical regions of the world, from 75% in Indonesia to 32% in India, 1%-6% in Latin America, and 1%-2% in Lebanon and Uganda. However, it is possible that patient selection and subjective diagnostic criteria have influenced the reported low prevalences. Besides low dietary intake, reduced absorption of vitamin A in PEM can further aggravate its deficiency. Oral water-miscible vitamin A is absorbed in an erratic fashion in children with kwashiorkar. A study carried out in South Carolina showed that 50% of underprivileged preschool children of a poor income group had either low or marginal levels of serum vitamin A [4]. Thus, vitamin-A deficiency is not completely restricted to underprivileged or developing countries but has worldwide prevalence and is the third most frequent nutritional problem after

PEM and iron deficiency. It is the most frequent cause of preventable blindness. This has led to international efforts at prevention by periodic massive doses using oral or injectable vitamin A. The rapidity with which vitamin A is absorbed is one of the determinants of early recovery from night blindness and xerophthalmia.

Vitamin A and infection

High death rates among PEM children parallel their vitamin-A status: children dying of PEM have lower levels of vitamin A than the survivors [5]. Most of the deaths are due to respiratory and gastrointestinal infections. It was observed in one study that the overall mortality rate in victims of PEM was 15%, whereas it was as high as 80% if malnutrition was associated with xerophthalmia [6]. The association between hypovitaminosis-A and bacterial and viral infections in humans and various species of animals has been reviewed [7-9]. Many pathological mechanisms probably contribute to the increased risk of infection in vitamin-A deficiency, including tissue changes and altered specific and non-specific immunity. Furthermore, infection itself can aggravate vitamin-A deficiency.

Other than its well-recognized role in rhodopsin synthesis and vision, vitamin A also regulates differentiation of epithelial tissues and inhibits keratinization. In hypovitaminosis-A, the secretory epithelia of respiratory tract and salivary and prostate glands show keratinization; this change offers less resistance against penetration of infectious agents and makes the individual susceptible to infections. Our recent studies indicate that bacteria bind to respiratory epithelial cells in greater numbers in vitamin-A-deficient subjects than in healthy controls (table 1) [10]. Increased mortality has been noted in children with mild vitamin-A deficiency [11]. Vitamin-A deficiency and infections aggravate each other, as the deficiency predisposes the host to infection, which in turn decreases the intestinal absorption of the vitamin [12]. Infection can even precipitate the symptoms of deficiency in an individual with marginal levels of the vitamin. In partially vitamin-A-deficient rats, even a subclinical load of malarial parasites has been shown to precipitate deficiency symptoms at a faster rate and resulted in more severe parasitaemia [13]. Vitamin-A-deficient patients sometimes develop xerophthalmia in only one eye, possibly because of the presence of infection in one site. Nasal mucosa of vitamin-A-deficient chicks keratinized only in areas of virus infection [14]. From these observations, one can deduce that the effects of deficiency are compounded if there is an accompanying infection. In germ-free environments, vitamin-A-deficient rats live for a relatively long time; it is believed that the vitamin requirement is low in the absence of stress of infection [15].

TABLE 1. Clinical data and bacterial binding to epithelial cells

Group	Number			Age (months)	Serum retinol (m mol/L)
	Total	Boys	Girls		Mean	Range
I	14	8	6	22 ± 3	2.234 ± 0.28a	1.413-3.351
II	10	6	4	24 ± 4	1.117 ± 0 14b	0.733-1.397
III	12	7	5	20 ± 3	0.419 ± 0.10c	0.139-0.698

 

Group	Dietary Vitamin A (retinol equivalents)	Ocular signs (number of subjects)			Weight for height (% of standard)	Bacteria per epithelial cell
		Xerophthalmia	Bitot's spots	Corneal opacity
I	321±44a	0	0	0	81±4	4.8±0.6a
II	201±29b	4	2	1	77±3	7.9±1.0b
III	186±22b	9	6	4	74±5	10.3±0.8c

Data are given as mean ± SE.
Values in the same column designated by different superscript letters differ significantly from each other at a probability level of .01 or less.

Mucosal immunity

Secretory antibody response, mucosal cell-mediated immunity, and non-specific barrier functions are important host defences. The disastrous effect of infection in vitamin-A-deficiency state may be due to the increased pathogenicity of infectious agents and/ or reduced immunocompetence of the host. Hypovitaminosis impairs tissue integrity by permitting keratinization of mucociliary linings. In the intestine, a reduction in the number of goblet cells and in mucus production disrupts non-specific defence mechanisms of the gastrointestinal tract. Impaired barrier function increases the systemic spread of infectious agents. By exposing vitamin-A-deficient rats to the nematode Angiostrongylus cantonensis, it was shown that the penetration power of the larvae increased and the animals had a higher worm burden and shorter survival period [16]. The observed effect was not due to inanition, as the animals were tube fed with a vitamin-A-deficient diet. Vitamin-A-deficient chicks after Newcastle disease virus infection had 100 times more virus in the throat swabs than normal non-deficient chicks [17]. The high frequencies of respiratory and gastrointestinal tract infections in PEM may in part be due to tissue changes resulting from associated vitamin-A deficiency.

Reduced levels of secretory IgA (slgA) have been observed in the saliva, nasopharyngeal secretions, duodenal fluid, and tears in malnourished children [18-21]. slgA antibody response to polio virus and measles vaccines is impaired [18]. Intraepithelial Iymphocytes and slgA-secreting plasma cells have been shown to be numerically lower in malnourished patients [22]. Also, levels of slgA in the intestinal fluid of rats on vitamin-A-deficient diets were reported to be very low compared with those in normal animals: rats given 5 mg retinoate per gram of diet showed a maximum of 180.2 mg of slgA per millilitre of intestinal fluid, and withdrawal of retinoic acid resulted in a low level of 46.0 mg of slgA per millilitre [23]. Moreover, it was also shown that supplementation to the deficient animals of 500 mg of retinyl palmitate per day, for just two days, increased the levels of slgA to 148.5 mg per millilitre of intestinal fluid. Intestinal cells of the deficient animals did not show any decrease in immunofluorescence for secretory component in the initial period of deficiency, but, after eight days and more, decreased intensity of fluorescence staining was observed. In addition, vitamin-A-deficient animals showed poor anti-dinitrophenyl response to 2,4-dinitrophenylated bovine gamma globulin (50 mol DNP per mole BOG) fed through tap water after initial priming with an intraperitoneal injection. In fact, only one of seven deficient rats showed a detectable anti-DNP activity, but three of four control animals responded to the antigen. The only deficient rat that responded to the antigen had significantly low levels of antibody (7.4 pmol antibody combining site per millilitre of intestinal fluid) compared with that of controls (18.9 pmol antibody combining site per millilitre).

There are few data on cell traffic in vitamin-A deficiency. Mesenteric Iymph node lymphocytes (MLNL) are precursors of intestinal IgA-secreting plasma cells and have a tendency to localize selectively in the gut mucosa. Protein-deprived rodents show reduced localization of MLNL in the gut. McDermott et al. [24] have shown that PEM and vitamin-A deficiency alter the capacity of MLNL to localize in the gut. Weanling rats were fed a low-vitamin-A diet until growth ceased, but there was no evidence of xerophthalmia or alteration in intestinal mucosa. Labelled Iymph node cells of normal or vitamin-A-deficient animals were injected into normal or vitamin-A-deficient recipient animals. Localization to the gut was decreased irrespective of the nutritional status of the recipients. However, the separate effects of vitamin A and PEM were not clearly dissected out in this study [25]. Reduced homing of MLNL in PEM and vitamin-A deficiency might be one of the reasons for reduced sIgA levels in this nutrient deficiency.

Increased keratinization, low levels of mucus secretion, and decreased numbers of goblet cells in the intestine in vitamin-A deficiency, together with reduced slgA secretion and reduced local antibody response, act synergistically to make the deficient individual vulnerable to a variety of gastrointestinal and respiratory infections. In addition, because of compromised systemic immunocompetence, such an individual may succumb to repeated and chronic infections [26; 27].

Phagocytes

Once the infectious agent crosses the anatomic barriers of skin and mucous membrane and enters the body, phagocytes deal in a non-specific way with certain bacteria and fungi before antigen-specific cellular and humoral immune mechanisms come into play. Large doses of vitamin A given to normal mice have been shown to offer non-specific resistance to grampositive and gram-negative bacterial and fungal infection [28]. A decreased mortality rate and low levels of bacteraemia were reported after challenge with Pseudomonas aeruginosa, Listeria monocytogenes, and Candida albicans. It was concluded that this increase in resistance is non-specific and might result from an augmented capacity of phagocytes (macrophages) to deal with these agents. However, there was no change in the uptake of colloidal carbon or aggregated human serum albumin by the reticuloendothelial system. The effect could not be due to changes in the organisms themselves, since vitamin A did not alter in vitro growth rates of the three microorganisms.

Hof and Emmerling [29] also observed a 100-fold increase in resistance to infection with L. monocytogenes. This was seen during the early stage of infection and was therefore attributed to increased functional capacity of the phagocytic system in rats treated with vitamin A. However, it should be noted that very large toxic doses (4 x 15,000 IU) of retinoic acid had to be used ro achieve such protective effects. Vitamin-A-deficient and PEM rats were shown to have a reduced number of glass-adhering cells (macrophages), and, moreover, these phagocytes had decreased ability to clear infection with a malarial parasite, Plasmodium berghei. Oral supplementation of vitamin A to deficient rats after infection has been shown to augment the recovery process [14]. Survival time after infection was shorter in vitamin-A-deficient animals than in pair-fed animals, but in control animals no death occurred during the experimental period of five weeks.

Lymphoid tissue

An impaired immune system is one of the factors that add to the vulnerability of vitamin-A-deficient individuals to infection. Usually vitamin-A deficiency occurs concomitantly with PEM, and in experimental animals vitamin-A deficiency causes inanition. There is ample evidence of reduced immunocompetence in PEM [1-3]. If PEM is accompanied by vitamin-A deficiency, the deficiency adds significantly to susceptibility to infection because of its additional detrimental effects on Iymphoid tissues and organs [30-35].

A diet deficient in vitamin A has been shown to cause atrophic changes in the thymus and spleen in rats [30]. Pair-fed rats that developed PEM also showed atrophic changes in bath tissues, but the magnitude of change was much greater in the vitamin-A-deficient group. The cortical region of the thymus of vitamin-A-deficient animals becomes devoid of Iymphocytes. The spleen also shows atrophy of germinal centers. However, Chandra and Au [31] observed only a slight difference in weight of the thymus and spleen from vitamin-A-deficient and pair-fed control animals, though the weights were significantly different Mom those of control animals fed ad libitum (table 2). These results suggest that the observed effect is mainly due to anorexia and malabsorption associated with vitamin-A deficiency. Similarly, Nauss et al. [36] also observed that weights of the spleen and thymus of vitamin-A-deficient and pair-fed controls were comparable but were lower than those from controls fed ad libitum.

TABLE 2. Organ weights

	No.	Thymus	Spleen
Group	of	weight	weight
	rats	(mg)	(mg)
Vitamin-A-deficient	6	181 ± 39	317 ± 41
Pair-fed control	6	208 ± 28	339 ± 47
Ad libitum control	6	281 ± 40	387 ± 59

Source: Ref. 31.

Involution of the thymus and bursa of Fabricius have been observed in vitamin-A-deficient chicks [33]. Chickens on a vitamin-A-deficient diet from the time of hatching showed atrophic changes in Iymphoid tissues. After 30 days on a vitamin-A-deficient diet, the epithelium of the bursa of Fabricius became pseudo-stratified, or cystic, and was disorganized compared with normal tissue. Follicles also displayed irregularity and fibrosis. The medulla showed invasion by heterophil cells, which replaced Iymphocytes. Panda and Combs [37] also observed a decrease in weight of bursae in vitamin-A-deficient chicks. Infection of deficient chicks by Newcastle disease virus caused further regression of bursae, interfollicular fibrosis, epithelial metaplasia, and keratinization. Follicles showed loss of lymphocytes and invasion of polymorphonuclear leukocytes and eosinophils five days after infection, and the medulla showed debris of disintegrating cell nuclei. These atrophic changes were progressive, and after nine days of infection, bursae of vitamin-A-deficient chicks were significantly different from those of vitamin-A-deficient non-infected controls. The thymus of vitamin-A-deprived chicks showed invasion of polymorphonuclear cells and loss of cortical width. After one day of virus infection, thymic corpuscles increased in size and number, and the cortex decreased in size and showed cell debris. These changes were progressive: Three days after infection a complete loss of cortex became evident, as shown by a reticular epithelial mesh, and thymic corpuscles showed dilation and degeneration. After seven days the organ appeared devoid of thymocytes and showed thymic corpuscles, degenerating cells, thick-walled vessels, and vacuoles containing polymorphonuclear cells. Atrophic changes were modest in the noninfective vitamin-deficient state [33]. Thus, vitamin deficiency and infection had synergistic effects on the Iymphoid tissue, thereby increasing the severity of the course of infection.

It is evident that vitamin-A deficiency causes atrophic changes in Iymphoid tissue. At the same time, a high dose of retinoic acid (1,000 fig per mouse per day for seven days) also resulted in reduction in body weight and spleen weight; it affected cell populations in the spleen and thymus, but the bone marrow was more or less resistant to this toxic effect. The thymus and spleen showed 95% and 50% reduction in cellularity respectively when high doses of retinoic acid were used [34].

T cells

In clinical practice, it is difficult to sort out the individual effects of vitamin-A deficiency from those of other nutrient deficiencies. The frequent presence of PEM confounds the picture. Thus, the reported reduction in the number of T cells [35] may well be due to concomitant PEM rather than to vitamin-A deficiency itself. In rodents fed a low-vitamin-A diet, the level of serum thymic factor is unaltered [32]. A decreased total leucocyte number has been observed in vitamin-A deficiency; the differential count revealed a relative increase in neutrophils and a decrease in the number of Iymphocytes. The mitogenic response of splenic Iymphocytes of vitamin-A-deficient rats to phytohaemagglutinin, concavalin A, and Escherichia coli lipopolysaccharide is reported to be less than one-third that of control and pair-fed animals [31; 36], although the mitogenic response of thymic Iymphocytes to concavalin A was comparable in vitamin-A-deficient, pair-fed, and control animals studied by 3H incorporation. Vitamin-A supplementation for three days was shown to increase circulating Iymphocytes and to restore the mitogenic response of splenic Iymphocytes to normal levels [36]. Decreased 3H incorporation by thymacytes and splenocytes was also reported in vitamin-A-deficient and PEM rats [30].

Decreased nucleic acid synthesis and mitogenic response (table 3) may be due to defective synthesis of membrane receptors. Marked changes in glycoprotein synthesis have been observed in vitamin-A deficiency [38; 39], and similar changes in membrane glycoproteins of lymphocytes (R. K. Chandra and V. Kumar, unpublished data) may contribute to impaired cell-mediated immunity seen in vitamin-A deficiency. Alternatively, if changes in T cells in vitamin-A deficiency are accompanied by increased suppressor T cells, then decreased mitogenic response would also result.

TABLE 3. Mitogen stimulation response

Stimulation index
PHA (µg)a	Vitamin-A- deficient rats (N = 6)	Pair-fed controls(N = 6)	P value
0.2	20.7 ± 8.8	87.8 ± 39.5	<.05
0.5	37.5 ± 19.0	104.8 ±36.4	< .01
2.0	38.7 ± 31.9	121.8 ±57.5	<.01

Source: Ref. 31. a. Amount of phytohaemagglutinin (PHA) in each well of microtitre plate.

In human volunteers, a modest supplement of beta-carotene is associated with increase in the number and function of CD4+ helper T cells. If this translates into enhanced resistance, it would have immense clinical significance. Vitamin-A supplementation has been shown to augment cell-mediated immune response, and injections of 150 IU per gram per day for five days accelerated the onset and decreased the duration of skin-graft rejection in mice [40]. Mice treated with 150 mg of vitamin A and sheep red-blood cells (SRBC) simultaneously and challenged 3-21 days later with the same antigen by subcutaneous injection into the pinna showed increased cell-mediated immune response measured as an increase in the thickness of the ear [41]. Conversely, it has been reported [34] that mice treated with 25 and 50 mg of retinoic acid, primed with SRBC, and then challenged later with SRBC into the footpad did not show any increase in levels of cell-mediated immune response measured as an increase in footpad diameter. Moreover, 100 and 300 mg of retinoic acid slightly suppressed the delayed hypersensitivity reaction. The discrepancy in the results of the two groups of investigators might be due to differences in the design of the experiment, particularly the dosage of retinoic acid and the time of its administration. In one study [41], vitamin A and SRBC were injected simultaneously, and, in the other [34], retinoic acid was injected for 1-6 days and the animals were primed with SRBC on day 10. Retinoic acid was shown to have no stimulatory effect on the mixed lymphocyte reaction in mice [34].

B cells

Vitamin A also modified the humoral response, especially to T-dependent antigens. The number of plaque-forming cells in vitamin-A-deficient rats was lower than in pair-fed animals (table 4) [31]. Haemagglutinin titre against diphtheria and tetanus antigens were found to be reduced in vitamin-A-deficient rats as compared with pair-fed controls [30]. Another study showed increased haemagglutinin titre to SRBC in mice after treatment with 600 IU of vitamin per gram of body weight per day for five days, either before or shortly after sensitization with antigen [40]. The response to vitamin-A treatment was highly significant: the relative reciprocal titre of haemagglutininin the controls was 64 and in the vitamin-A-treated animals was in the range 1,024-4,096.

TABLE 4. Plaque-forming cell response

Direct PFC per spleen (10 -3)	Vitamin-A- deficient	Pair-fed controls	P value
rats (N = 6)	(N = 6)
Background	0.39 ± 0.08	0.24 ± 0.09	<.05
Day 5 after	17.5 ± 4.8	56.4 ± 9.0	<.01

Source: Ref. 31.

Cohen and Cohen [42] have also reported increased plaque-forming cells to SRBC in vitamin-A-treated mice. They showed that intraperitoneal administration of 1,000 IU of vitamin A per day for four days increased plaque-forming cells. The maximum effect (a fivefold increase) was observed with a dose of 300 IU per day for four days; the toxic dose of 9,000 IU per day did not increase antibody-forming cells in the spleen of mice. Increased antibody production to 2,4-dinitrophenyl conjugate of ovalbumin after two subcutaneous injections of either 2,000 or 5,000 IU of vitamin A was also observed in mice [42].

Partial vitamin-A deficiency in chicks reduces agglutinin titre against Salmonella pullorum antigen [35]. Uhr et al. [43] have reported an immunosuppressive effect of vitamin A in guinea pigs. They did not find any increase either in the clearance of bacteriophage ox174 from the circulation or in the production of antibodies to it. However, they observed prolonged production of 19S antibodies if vitamin A was injected simultaneously with antigen [44]. Dennert and Lotan [34], in studies in vitro, observed 80% reduction in plaque-forming cell response of spleen cells from retinoic-acid-treated mice, and 10^-5-10^-9 M concentration of retinoic acid during sensitization of spleen cells to SRBC was shown to completely suppress the induction of plaque-forming cells. In viva studies showed no effect of 100, 300, and 1,000 fig of retinoic acid for five days and sensitization with SRBC on day 4 on the number of plaque-forming cells in the spleen after eight days.

In mice, the simultaneous administration of vitamin and tetanus toxoid resulted in an enhanced antitoxin response. The effect of three different doses of vitamin A (3,000, 25,000, and 30,000 IU) in mice on the antitoxin response to tetanus toxoid indicated a direct relation of the response to the dose of vitamin [45]. Comparable doses of vitamin A given to children would have had undesirable side effects. At the dose level of 30,000 IU, mice showed signs of toxicity. However, these investigators did not observe any significant effect of 200,000 IU of vitamin A on the production of antitoxin to tetanus toxoid in a field trial involving Bangladeshi children.

Complement

In PEM, in addition to a decrease in cell-mediated and humoral immune responses, haemolytic complement activity of the serum is depressed. Vitamin-A deficiency has been shown to aggravate the disastrous effects of PEM on immunocompetence. However, in rats, vitamin-A deficiency did not have any additive effect on serum complement levels in PEM. Contrary to the case with PEM, in which complement levels decrease [1-31, vitamin-A deficiency has been shown to enhance complement levels [46]. In addition to this, Azar and Good [47] have shown suppression of serum complement haemolytic activity 24 hours after a large dose of vitamin A is administered.

Adjuvant effects

Vitamin A acts as an adjuvant at non-toxic doses and enhances cell-mediated and humoral immune responses. Injections of vitamin A increased the cellularity of redonal Iymph nodes.[48]. The vitamin has also been shown to stimulate antibody production to bovine gamma globulin, which would otherwise have resulted in immunological paralysis [49]. The adjuvant property of vitamin A was thought to be due to its membrane-labilizing effect on Iysosomes. Lysosomal membrane labilization can induce Iymphoid cell proliferation [50]. Vitamin A has been shown to enhance cell-mediated immune response when administered simultaneously with or close to the antigen challenge. Moreover, injection of vitamin A at a site remote from that of the antigen was shown to be ineffective in enhancing cellular immunity [40; 41], suggesting that the draining Iymph node might be the site of adjuvant action of vitamin A. Taub et al. [48] reported on the adjuvanticity of vitamin A. Two days after injection of 0.5 mg of vitamin A in liquid paraffin in the right footpad, the popliteal Iymph node showed enlargement and hypercellularity of paracortical areas. After six days, the germinal centres were observed in the cortex, and the medullary region showed slight expansion and a few place cells. Vitamin A was shown to work as a classic adjuvant without any antigenicity. The formation of germinal centres after vitamin-A administration was not as pronounced as with other adjuvants, such as alum-precipitated bovine gamma globulin, Freund's complete adjuvant, and pertussis coccobacillus. Enlargement of the Iymph node and increased Iymphocyte traffic after vitamin-A treatment close to the time of antigen treatment may help to augment immune response by increasing contact between lymphocytes and antigens. Allison and Davies [51] have also shown cellular proliferation and blast transformation in thymus-dependent areas of a draining Iymph node after vitamin-A administration.

Concluding remarks

The world-wide prevalence of isolated vitamin-A deficiency and its occurrence in association with other nutritional deficiencies, especially PEM, and with infection has stimulated considerable recent work on the role of vitamin A in host resistence. Vitamin A subserves a number of important physiological functions that contribute to effective immunocompetence. Moderately large doses of beta-carotene have an immunostimulatory effect and can reverse the suppression produced by pharmacological agents such as cortisone. It acts as an adjuvant and influences both cell-membrane and intracellular composition and function. This applies to Iymphocytes and monocytes. Thus, it is reasonable to expect that vitamin-A deficiency is associated with an increased incidence of infection and that its prevention or treatment will result in decreased illness. This may be due to epithelial changes and improved cell-mediated and mucosal immunity. Recent work [26; 27; 52-55] has affirmed the critical role of beta-carotene and vitamin A in optimum immunocompetence. At the same time, massive doses of vitamin A if given for prolonged periods may have a deleterious effect.

Acknowledgements

Our work has been supported by Health and Welfare Canada, the Medical Research Council, Ross Laboratories, Sandoz Nutrition, and the Carnation Company.

The present article is based on an earlier review article l27].

References

1. Chandra RK Nutrition, immunity and infection: present knowledge and future directions. Lancet 1983; 1 :688-91.
2. Chandra RK, Newberne PM. Nutrition, immunity and infection: mechanisms of interaction. New York: Plenum, 1977.
3. Chandra RK, ed. Nutrition and immunology. New York: Alan R. Liss, 1988.
4. High EG. Some aspects of nutritional vitamin A levels in preschool children of Beaufort County, North Carolina. Am J Clin Nutr 1969;22:1129-32.
5. McLaren DS, Shirajian E, Loshkajian H. Shadarevian S. Short-term prognosis in protein calorie malnutrition. Am J Clin Nutr 1969;22:863-70.
6. McLaren DS. Xerophthalmia: a neglected problem. Nutr Rev 1964;22:289-91.
7. Scrimshaw NS, Taylor CE, Gordon JE. Interactions of nutrition and infection. WHO monograph series, no. 57. Geneva: WHO, 1968.
8. Gross RL, Newberne PM. Role of nutrition on immunologic function. Physiol Rev 1980;60:188-302.
9. Beisel WR. Single nutrients and immunity. Am J Clin Nutr 1972;3S:417-68.
10. Chandra RK. Increased bacterial binding to respiratory epithelial cells in vitamin A deficiency. Brit Med J 1988;297:834-35.
11. Sommer A, Tarwatjo I, Hussaini G. Susanto D. Increased mortality in children with mild vitamin A deficiency. Lancet 1983;2:585-88.
12. WHO. Control of vitamin A deficiency and xerophthalmia. Report of a joint WHO/UNICEF/USAID/Helen Keller International/lVACG meeting. WHO technical report series, no. 672. Geneva: WHO, 1982.
13. Krishnan S. Krishnan AD, Mustafa AS, Talwar GP, Ramalingaswami V. Effect of vitamin A and undernutrition on the susceptibility of rodents to a malarial parasite Plasmodium berghei. J Nutr 1976;106:784-91.
14. Bang BG, Bang FB. Replacement of virus-destroyed epithelium by keratinized squamous cells in vitamin A deprived chickens. J Exp Biol Med 1969;132:50-54.
15. Rogers WE, Bieri JO, McDaniel EG. Vitamin A deficiency in the germfree state. Fed Proc 1971 ;30: 1773-78.
16. Darip MD, Sirisinha S. Lamb AL. Effect of vitamin A deficiency on susceptibility of rats to Angiostrongylus cantonensis. Proc Soc Exp Biol Med 1979;161:600-04.
17. Bang FB, Foard MA. The effect of acute vitamin A deficiency on the susceptibility of chicks to Newcastle disease and influenza virus. Johns Hopkins Med J 1971;129:100-09.
18. Chandra RK. Reduced secretory antibody response to live attenuated measles and poliovirus vaccines in malnourished children. Br Med J 1975;2:583.
19. Sirisinha S. Suskind R. Edelman R. Asvapaka C, Olson RE. Secretory and serum IgA in children with proteincalorie malnutrition. Pediatrics 1975;55:166.
20. Reddy V, Raghuramulu N. Bhaskaram C. Secretory IgA in protein-calorie malnutrition. Arch Dis Child 1976;51 :871-74.
21. Watson R. Reyes MA, McMurray DN. Influence of malnutrition on the concentration of IgA, Iysezyme, amylase and aminopeptidase in children's tears. Proc Soc Exp Biol Med 1978;157:215.
22. Chandra RK. Nutrition and gastrointestinal tract. In: Kelley VC, ed. Brennemants Practice of pediatrics. New York: Harper & Row, 1979.
23. Sirisinha S. Darip MD, Moongkamdi P. Ongsakul M, Lamb AJ. Impaired local immune response in vitamin A deficient rats. Clin Exp Immunol 1980;40:127-35.
24. McDermott MR, Mark DA, Befus AD, Baligo BS, Suskind RM, Bienenstock J. Impaired intestinal localization of mesenteric lymphoblasts associated with vitamin A deficiency and protein-calorie malnutrition. Immunol 1982;45: 1-5.
25. Lymphocyte traffic in nutritional deficiency. Nutr Rev 1982;40:340-42.
26. Chandra RK, Bendich A, eds. Micronutrient effects on immune functions. New York: New York Academy of Sciences (in press).
27. Vyas D, Chandra RK. Vitamin A and immunocompetence. In: Watson RR, ed. Nutrition, disease resistance and immune system. New York: Marcel Dekker, 1984:325-43.
28. Cohen BE, Elin RJ. Vitamin A-induced nonspecific resistance to infection. J Infect Dis 1974;129:597-600.
29. Hof H. Emmerling B. Stimulation of cell-mediated resistance in mice to infection with Listeria monocytogens by vitamin A. Ann Immunol 1979;130C:587-94.
30. Krishnan S. Bhuyan UN, Talwar GP, Ramalingaswami V. Effect of vitamin A and protein-calorie undernutrition on immune responses. Immunol 1974;27:383-92.
31. Chandra RK, Au B. Single nutrient deficiency and cell-mediated immune responses: Ill. Vitamin A. Nutrition RES 1981;1:181-85.
32. Chandra RK, Heresi G. Au B. Serum thymic factor activity in deficiencies of calories, zinc, vitamin acid pyridoxine. Clin Exp Immunol 1980;43:332-35.
33. Bang BG, Foard MA, Bang FB. The effect of vitamin A deficiency and Newcastle disease on Iymphoid cell system in chickens. Proc Soc Exp Biol Med 1973;143: 1140 46.
34. Dennert G. Lotan R. Effect of retinoic acid on the immune system: stimulation of T-killer cell induction. Eur J Immunol 1978;8:23-29.
35. Bhaskaram C, Reddy V. Cell-mediated immunity in iron and vitamin-deficient children. Br Med J 1975; 3:522.
36. Nauss KM, Mark DA, Suskind RM. The effect of vitamin A deficiency on the in vitro cellular immune response of rats. J Nutr 1979;109: 1815-23.
37. Panda B. Combs GF. Impaired antibody production in chicks fed diets low in vitamin A, pantothenic acid or riboflavin. Proc Soc Exp Biol Med 1963;113:530-34.
38. Kim YL, Wolf G. Vitamin A deficiency and the glycoproteins of rat corneal epithelium. J Nutr 1974; 104:17-718.
39. Bonanni F. Levinson SS, Wolf G. De Luca L. Glycoproteins from the hamster respiratory tract and their response to vitamin A. Biochim Biophys Acta 1973;297:441-51.
40. Jurin M, Tannock IF. Influence of vitamin A on immunological response. Immunol 1972;23:283-87.
41. Athanassiades TJ. Adjuvant effect of vitamin A palmitate and analogs on cell-mediated immunity. JCCI 1981 ;67:1153-56.
42. Cohen BE, Cohen IK. Vitamin A: adjuvant and steroid antagonist in the immune response. J Immunol 1973; 111:1376-80
43. Uhr J. Weissmann G. Thomas L. Acute hypervitaminosis A in guinea pigs: Il. Effects on delayed type hypersensitivity. Proc Soc Exp Biol Med 1963 ;112:287.
44. Uhr J. Weissmann G. Intracellular distribution of bacteriophage in mammalian tissue. J Immunol 1965 ;94:544
45. Brown KH, Rayan MM, Chakraborty J. Aziz KMA. Failure of a large dose of vitamin A to enhance the antibody response to tetanus toxoid in children. Am J Clin Nutr 1980;33:212-17.
46. Madjid B. Sirisinha S. Lamb A. The effect of vitamin A and protein deficiency on complement levels in rats. Proc Soc Exp Biol Med 1978;158:92-95.
47. Azar MM, Good RA. The inhibitory effect of vitamin A on complement levels and tolerance production. J Immunol 1971;106:241-45.
48. Taub RN, Krantz AR, Dresser DW. The effect of localized injection of adjuvant material on the draining Iymph node: 1. Histology. Immunol 1970;118:171-86.
49. Dresser DW. Adjuvanticity of vitamin A. Nature (London) 1968;217:527-29.
50. Allison AC, Mallucci L. Lysosomes in dividing cells with special reference to Iymphocytes. Lancet 1964; 2:1371.
51. Allison AC, Davies AJS. Requirement of thymus dependent Iymphocytes for potentiation by adjuvants of antibody formation. Nature 1971;233:330-32.
52. Gensler HL. Reduced immunosuppression in UV-irradiated mice by dietary retinyl palmitate. Carcinogenesis 1989;10:203-07.
53. Lippman SM. Vitamin A derivatives in the prevention and treatment of human cancer. J Am Coll Nutr 1988;7:269-84.
54. Lavasa S. Early immune response in vitamin A deficiency: an experimental study. Indian J Exp Med 1988; 26:431 -35.
55. Kastl PR. Topical vitamin A ointment increased healing of cataract incisions. Ann Ophthalmol 1989;5:56469.

Effects of vitamin-A status on hamster tracheal epithelium in viva in vitro

Luigi M. De Luca and Elizabeth M. McDowell

This paper highlights what is known about how vitamin A and the retinoids control epithelial morphology and function. The system of choice is the tracheal epithelium of the Syrian golden hamster. This species was originally selected by Saffiotti et al. [1] in their studies of chemical carcinogenesis of the respiratory tract because it is relatively resistant to respiratory infection. The system was then rendered more amenable to in vitro investigation by Sporn et al. [2], who defined conditions for maintaining the trachea in organ explant culture, as well as the tissue's requirement for retinoic acid (RA) in the maintenance of normal mucociliary differentiation. Following in viva observations [3; 4], McDowell et al. have recently used epithelial cell culture techniques to define the mucous cell as the target of vitamin A [5].

In vivo effect of vitamin-A status on the hamster tracheal epithelium

The final effect of nutritional deficiency of vitamin A on the tracheal epithelium is the replacement of ciliated cells and normal mucous cells (fig. 1) by squamoid cells (altered mucous cells), which normally characterize the epidermoid type of differentiation. Figure 2 shows the extent of such replacement, which results in the near occlusion of the tracheal lumen. Accumulation of bacteria and other external material influences survival of the animal, which eventually succumbs to infection.

The terminal stage is preceded by primary effects of the deficient diet. Careful monitoring of the changes in body weight and serum retinal levels allowed the definition of a stage of "minimal morphological change," which preceded the loss of body weight [3; 4; 6]. Measurement of the cell-division rates in the mucous and basal cells during the development of vitamin-A deficiency revealed that the rates (mitotic rates) were lowered in a non-uniform manner. The replication of mucous cells was profoundly reduced compared with that of the basal cells during the stage of "minimal morphological change." This is clearly shown in table 1.

This work leads to the conclusion that vitamin-A deficiency depresses epithelial cell division before epidermoid metaplasia formation is evident. If vitamin-A deficiency continues, however, the altered mucous cells regain the capacity to divide and the epithelium is replaced by flat "squamoid cells." These cells generally arise as a consequence of cell injury, whether caused by carcinogen exposure, mechanical injury, or, as in this case, nutritional deficiency (fig. 3). It is clear, then, that squamoid cells (but not columnar mucous cells) can survive and multiply in the absence of vitamin A. Reversal of squamous metaplasia to the normal mucociliary phenotype of the tracheal epithelium is only possible in the presence of vitamin A or one of its biologically active analogues.

TABLE 1. Proportions and mitotic rates of tracheal basal, mucous, and ciliated cells of control and vitamin-A-deprived hamsters

Cell type	Control	Vitamin-A-deprived	P value
Proportion of total (%)
Basal	28.7 ± 2.5	39.7 ± 2.7	<.0001
Mucous	59.3 ± 2.6	53.1 ± 1.9	<.001
Ciliated	11.0 ± 3.1	6.8 ± 1.8	<.05
Mitotic rate (% of total)a
Basal	0 61 ± 0.13	0.29 ± 0.35	<.062
Mucous	2.42 ± 1.19	0.15 ± 0.10	<.01
Ciliatedb	0	0
Mitotic rate (% of own cell type)c
Basal	2.14 ± 0.48	0.72 ± 0.91	<.01
Mucous	4.06 ± 1.97	0.29 ± 0.2	<.01
Ciliatedb	0	0

Source: Ref. 4 Reproduced by permission.

The data for vitamin-A-deprived cells were derived from epithelia showing minimal changes after five weeks on diet. Foci of stratification and/or epidermoid metaplasia (about 5% of all epithelial cells) were excluded from the analysis.
a. Mean percentage of total number of epithelial cells counted in cross-section of two tracheal rings per hamster (about 1,600 cells). Each hamster received 3H-thymidine and colchicine six hours before sacrifice.
b. Ciliated cells do not divide.
c. Mean percentage of total number of basal cells or mucous cells.

Administration of retinyl acetate to vitamin-A-deficient hamsters, again, primarily affects cell division of the mucous cells (fig. 4). Within three days their cell-division rate returned to normal levels. The number of preciliated cells, which are progeny of replicating mucous cells, was restored to normal levels, but vitamin-A repletion had no effect on the replication of the basal cells.

The first unequivocal conclusion from this in viva work is that vitamin A is necessary to maintain normal rates of mucous cell division in the tracheal epithelium.

FIG. 4. Relationship between the proportion of. preciliated cells and the mitotic rates (M.R.) of mucous cells and basal cells All data represent mean percentages of the total number of epithelial cells counted. Control values were 0.39% for the proportion of preciliated cells and 0.61% and 2.2% for the mitotic rate of basal cells and mucous cells respectively.

In vitro cell culture work

Recent work is] has permitted observation of the effects of vitamin-A depletion in cultured epithelial cells from hamster tracheas. The cells recapitulated the development of normal epithelium during seven days of culture. In the presence of retinoic acid, the mucous cells divided at a high rate and the progeny rapidly matured to fully differentiated mucous and ciliated cells (fig. 5). Smaller cells were also visible in the culture and probably represented basal cells found in viva. The study compared the morphology and cell-division rates of the cells in the presence and absence of retinoic acid. It is clear from figure 6 that deficiency of vitamin A markedly lowered the ability of the larger mucous cells to divide.

FIG. 6. The effect retinoic-acid (RA) depletion on ³H-thymidine-labelled nuclei of small (basal) and large (secretory) tracheal epithelial cells - percentages of each kind of cell. The large cells were very flat in RA- cultures on days 2 and 4, and so it was possible to discriminate between the nuclei of the two kind of cells

When cells grew in the absence of retinoic acid on collagen substrate, they failed to mature into normal columnar mucous cells and, instead, showed the squamous type of differentiation characteristic of vitamin-A deficiency in vivo and of organ-cultured epithelium (fig7).

This in vitro work confirms the notation that retinoic acid in required for cell division and differentation of mucous cells.

The control of mucus production by vitamin A

In conjunction with the lowered rate of mucous cell division, vitamin-A deficiency also causes a decrease in periodic-acid Schiff base (PAS)-positive cells, which indicates less production of mucus. Whether this precedes or follows the decrease in cell division is unclear, but it certainly appears to precede the enhancement of keratin production in the trachea.

Squamous metaplasia is not observed in the small intestine, even in severe vitamin-A deficiency [7]; however, mucin biosynthesis, as measured by the incorporation of 3H-glucosamine, is decreased [8]. We were able to raise an antibody to the purified goblet cell glycoprotein [9]. Indirect immunofluorescence studies clearly indicated the presence of a cross-reactive antigen in a variety of rat epithelial tissues. Vitamin-A deficiency caused a marked drop in the amount of cross-reacting antigen in a variety of epithelial tissues, including the trachea [10]. Therefore, we can conclude that, in addition to cell division of mucous cells, vitamin A also controls mucus production.

Vitamin-A control of keratin gone expression

A later event during the progression of vitamin-A deficiency is keratin gene activation and consequent keratin production. A co-ordinated expression of acidic and basic keratins takes place in vitamin-A deficient hamster tracheas [11]. These keratins are not readily detectable in hamsters fed vitamin A or in tracheas cultured in the presence of retinoic acid. Notwithstanding the morphological similarity between the epidermis and the squamoid epidermoid tracheas, keratin gene products expressed in vitamin-A-deficient tracheas are not the same as in the skin. One outstanding difference is that keratin 1 (67,000 daltons), a prominent epidermal keratin, is not produced in vitamin-A-deficient tracheas [11].

Carcinogenesis

Exposure to chemical carcinogens, such as benzo[a]pyrene and 7,12-dimethylbenz[a]anthracene, causes similar squamous metaplastic lesions in the trachea [12]. When tracheas from four-week-old hamsters fed a normal diet were cultured in the presence of either benzo[a]pyrene or 7,12-dimethylbenz[a]anthracene for two weeks without retinoic acid, they developed squamoid metaplastic lesions. These lesions were not visible when retinoic acid was included in the medium containing the carcinogens (F. L. Huang et al., in preparation). Thus, clearly, retinoic acid is capable of repairing the squamoid metaplasia caused by carcinogen exposure. The squamoid lesions caused by the presence of the carcinogen were also positive by immunofluorescence with keratin antibodies (F. L. Huang et al., in preparation).

Importance of nutritional status in histogenesis: The concept of exotrophism

Organisms capable of synthesizing various essential nutrients are said to be prototrophic for those nutrients. Neurospora is an organism that is prototrophic for pantothenic acid, among other nutrients. When Neurospora is treated with a mutagenic dose of ultraviolet radiation, the result may be the establishment of a requirement for pantothenic acid in the offspring, which is said to have become a "pantothenic-acid auxotroph." Utilizing this approach, Beadle and Tatum [13] have elucidated the steps involved in the biosynthesis of various essential nutrients. Thus, auxotrophism is defined by the need for the exogenous supply of a nutrient in the presence of a mutated phenotype [14].

The mucociliary tracheas, or other epithelia, need vitamin A to maintain their differentiation. The tracheal squamoid cell, which prevails under conditions of vitamin-A deficiency, however, apparently does not need the vitamin. We propose to call this squamous cell "exotrophic" for vitamin A, that is, as having escaped the vitamin-A-requiring status normally characteristic of the columnar epithelium of the trachea. In the presence of vitamin A, this cell is replaced by the normal columnar cells. If, however, the supply of the vitamin is scarce, a conditional "vitamin-A exotroph," that is the squamoid cell, may persist at the site. This exotroph may become permanent by the action of a mutagenic agent, which would fix the exotrophic state, as shown schematically in figure 8. It is then possible to postulate that, when subject to the action of carcinogens and/or tumour-promoting substances, fixed exotrophic cells might divide and contribute to a tumour.

Figure 8

What is the advantage of the exotrophic state to the tumour cell? Quite simply that it has bypassed the requirement for the essential substance and thus the stringency of normal growth control. A cell can become more autonomous as it bypasses the requirements for more essential nutrients. The end result would be a cell that has been liberated from the usual constraints of regulatory substances and that may eventually prevail over other, normally regulated cells.

We and others have examined hepatoma cells and have found that, in general, they contain very little if any retinyl palmitate, whereas the surrounding hepatic host tissue contains normal concentrations of retinyl palmitate [15]. Transplanted hepatomas, whether minimally or maximally deviated from normal, are devoid of retinyl palmitate, whereas the host rat liver tissue and regenerating liver contain normal levels of the vitamin (table 2). Naturally, the lack of detectable retinyl palmitate in hepatoma tissue may be due to a variety of reasons, which are not mutually exclusive, among which are the following:

• The hepatoma cell may destroy vitamin A as it comes in contact with it, as suggested by the findings of Muto et al. [16] that some hepatomas may form anhydroretinol from retinol. - The deficiency may be the result of defective transport for retinol.
• A hepatic retinol exotroph may be the stem cell of hepatomas.

TABLE 2. Retinyl palmitate of hepatoma, host rat and regenerating liver postnuclear membranes (nanograms per milligram of protein)

	Liver	Tumour
20-1-1	580 ± 30	<1.6
16-2-1	513 ± 52	<1.6
7787-1-1	400 ± 21	<1.6
9618A-1-1	942 ± 2.2	<1.6
44-1-2	547 ± 21	<1.6
5123D-1-1	150 ± 3.8	<1.6
3924A-1-1	177 ± 59	<1.6
7800-1-1	73 ± 1.8	<1.6
5123 tc 1-2	363 ± 6.7	<1.6
7777-2-1	243 ± 38	<1.6
Primary cystic tumour	302 ±28	<1.6
Regenerating liver
24 hours	113 ± 7
48 hours	100 ±10

Summary and conclusions

In this paper we have suggested the new concept of exotrophic cells, i.e. cells that have conditionally escaped the need for an essential nutrient, such as vitamin A. These exotrophs might become fixed by a mutation and eventually contribute to the tumorigenic phenotype.

The discovery of the retinoic acid receptor (RAR) has opened up new horizons in the search for the mechanism of action of retinoic acid [17; 18]. It is intriguing that a second retinoic acid receptor, RARE, is abundantly expressed in hepatoma tissue and not in normal liver; Benbrook et al. [191 suggest that the erroneous expression of the RARE might contribute to tumour development in liver. How and whether these findings relate to the vitamin-A-deficient status of hepatoma cells remains to be understood.

References

1. Saffiotti U. Montesano R. Sellakumar AR, Borg SA. Experimental cancer of the lung: inhibition by vitamin A of the induction of tracheobronchial squamous metaplasia and squamous cell tumors. Cancer 1967;20:85764.
2. Sporn MB, Dunlop NM, Newton DL, Smith JM. Prevention of chemical carcinogenesis by vitamin A and its synthetic analogs (retinoids). Fed Proc 1976;35: 1332-38.
3. McDowell EM, Keenan KP, Huang M. Effects of vitamin A-deprivation on hamster tracheal epithelium: a quantitative morphologic study. Virchows Arch B 1984;45:197-219.
4. McDowell EM, Keenan KP, Huang M. Restoration of mucociliary tracheal epithelium following deprivation of vitamin A: a quantitative morphologic study. Virchows Arch B 1984;45:221-40.
5. McDowell EM, Ben T. Coleman B. Chang S. Newkirk C, De Luca LM. Effects of retinoic acid on the growth and morphology of hamster tracheal epithelial cells in primary culture. Virchows Arch B 1987;54:38-51.
6. Strum JM, Latham PS, Schmidt ML, McDowell EM. Vitamin A deprivation in hamsters: correlations between tracheal epithelial morphology and serum/tissue levels of vitamin A. Virchows Arch B 1985;50:43-57.
7. De Luca LM, Little EP, Wolf G. Vitamin A and protein synthesis by rat intestinal mucosa. J Biol Chem 1969; 244:701-08.
8. De Luca LM, Schumacher M, Wolf C. Biosynthesis of a fucose-containing glycopeptide from rat small intestine in normal and vitamin A-deficient conditions. J Biol Chem 1970;245:4551-58.
9. De Luca LM, Schumacher M, Nelson DP. Localization of the retinol-dependent fucose-glycopeptide in the goblet cell of the rat small intestine. J Biol Chem 1971 ;246:5762-65.
10. De Luca LM, Maestri N. Bonanni F. Nelson DP. Maintenance of epithelial cell differentiation: the mode of action of vitamin A. Cancer 1972;30:1326-31.
11. Huang FL, Roop DR, De Luca LM. Vitamin A deficiency and keratin biosynthesis in cultured hamster trachea. In Vitro 1986;22:223-30.
12. Harris CC, Sporn MB, Kaufman DG, Smith JM, Jackson FE, Saffiotti U. Histogenesis of squamous metaplasia in the hamster tracheal epithelium caused by vitamin A deficiency or benzo [a]pyreneferric oxide. J Natl Cancer Inst 1972;48:743-61.
13. Beadle GW, Tatum EL. Genetic control of biochemical reactions in Neurospora. Proc Natl Acad Sci 1941; 27:499-506.
14. Ayala FJ, Kiger JA Jr. Modem genetics. 2nd ed. Menlo Park, Calif, USA: Benjamin/Cummings, 1984.
15. De Luca LM, Brugh M, Silverman-Jones CS. Retinyl palmitate, retinyl phosphate, and dolichyl phosphate of postnuclear membrane fraction from hepatoma, host liver, and regenerating liver: marginal vitamin A status of hepatoma tissue. Cancer Res 1984;44:224-32.
16. Muto Y. Moriwaki H. Antitumor activity of vitamin A and its derivatives. J Natl Cancer Inst 1984;73:1389-93.
17. Petkovich M, Brand NJ, Krust A, Chambon P. A human retinoic acid receptor which belongs to the family of nuclear receptors. Nature 1987;330:444-50.
18. Giguere V, Ong ES, Segui P. Evans RM. Identification of a receptor for the morphogen retinoic acid. Nature 1987;330:624-29.
19. Benbrock D, Lernhardt E, Pfahl M. A new retinoic acid receptor identified from a hepatocellular carcinoma. Nature 1988;333:669-72.

Research priorities and strategies for investigation of the influence of vitamin-A supplementation on morbidity

Michele R. Forman

This paper reviews recent clinical and field research in vitamin A and morbidity and examines, primarily, the epidemiologic evidence for the association between vitamin-A deficiency and morbidity. Since I have been requested to discuss the findings from the most significant work, this review is not exhaustive and is based on an epidemiologist's view of significance. From the available literature, most vitamin-A research focuses on the following morbidities: respiratory disease, diarrhoeal disease, and measles complications, of which respiratory and diarrhoeal infections and corneal ulcers are the most common in less-developed communities. The research was designed as either longitudinal cohort studies of pre-school-aged children or hospital-based trials.

For this review, several criteria were used to evaluate individual studies, including the following:

• the strength of the association or effect;
• whether important factors related to vitamin-A status and to morbidity were controlled in the analysis;
• the identification of positive or negative findings and whether any systematic bias could have altered those findings, which are usually computed as an estimate of effect, such as a relative risk; and
• the appropriateness of, as well as the ability to implement, the study design.

A summary of study findings and recommendations for future research follow the review of recent research.

Review of recent research

Prospective cohort studies

Prospective cohort studies of vitamin-A deficiency and respiratory and diarrhoeal disease have been con ducted in central Java, Indonesia, and Hyderabad, India. Neither study, however, was designed as a vitamin-A and morbidity study. In an 18-month longitudinal cohort study of pre-school-aged central Javanese children (N = 4,600), ocular and other clinical data were collected at baseline and every three months for evidence of xerophthalmia and respiratory and diarrhoeal disease; anthropometric data were collected to evaluate nutritional status. The child's ocular status was assessed by an ophthalmologist at each three-month examination. The morbidity data were based on clinical signs of current respiratory disease, diagnosed by a paediatrician at the examination, and the diarrhoeal disease data were based on maternal recall of the child's experiencing four or more loose stools per day at any time during the month prior to the examination.

Two separate sets of analyses have been published from this study [1; 2]. In one set [1] the incidence rates of respiratory disease and the prevalence rates of diarrhoeal disease at the end of each three-month interval were determined for children with and without mild xerophthalmia (Bitot's spots and/or night blindness) at both the onset and end of the cycle. Compared with healthy children without xerophthalmia, those children with mild xerophthalmia at the start and end of a three-month cycle had an almost two and threefold risk of respiratory and diarrhoeal disease respectively (table 1). Age-specific relative risks were not appreciably different for diarrhoeal disease and were varied for respiratory disease. These findings were similar within weight-for-height groups (based on the Waterlow classification).

In the second set of analyses [2], all children without xerophthalmia at the beginning of a cycle were stratified by the presence or absence of respiratory or diarrhoeal disease, and the prevalence of mild xerophthalmia was determined at the following examination three months later. Those children with either respiratory or diarrhoeal disease at the onset of a cycle had 2.2 and 2.5 times as great a relative risk respectively of developing mild xerophthalmia by the " end of the cycle as healthy (non-respiratory/non-diarrhoeal disease) controls (table 2). The excess risk of xerophthalmia among respiratory disease cases was restricted to children three years old and over, and the relative risks of xerophthalmia among diarrhoe al cases versus children without diarrhoeal diseases varied by age. These findings were similar within weight-for-height groups.

TABLE 1. Age-specific incidence of respiratory and diarrhoeal disease among children with and without xerophthalmia

Age (years)	Number of child-intervals		Cases of disease		Rate per 1,000		Relative risk (N-N: X-X)	p (2-tailed)
	N-N	X-X	N-N	X-X	N-N	X-X
Respiratory disease
L 1	5,484	42	470	8	86	191	1:2.2	<.05
2	2 993	143	257	31	86	217	1:2.3	<.001
3	3 051	188	176	21	58	112	1:1.9	<.01
4	3,031	164	100	13	33	79	1:2.4	<.001
3 5	3,644	191	89	5	24	26	1:1.1	NS
Total	18,203	728	1,092	78	60	107	1:1.8	<.001
Diarrhoea
L 1	5,425	36	421	9	78	250	1:3.2	<.001
2	3,014	135	202	31	67	230	1:3.4	<.001
3	3,018	160	151	27	50	169	1:3.4	<.001
4	2,958	147	93	14	31	95	1:3.1	<.001
3 5	3,624	183	87	14	24	77	1:3.2	<.001
Total	18,039	661	954	95	53	144	1 :2.7	<.001

Source: Ref. 1.

N-N: Children with normal eyes at both the start and the end of the three-month observational interval.
X-X: Children with mild xerophthalmia (night blindness and/or Bitot's spots) at both the start and the end of the interval.

TABLE 2. Age-specific incidence of xerophthalmia for children with and without respiratory disease and for children with and without diarrhoea

Age (years)	Child-intervals		Developed xerophthalmiaa		Rate per 1,000		Relative risk ( -: + )
	-	+	-	+	-	+
Respiratory disease
L 1	5,533	595	6	0	1.1	0	-
2	3001	417	8	1	2.7	2.4	1:0.9
3	3061	257	10	3	3.3	11.7	1:3.6
4	3042	170	11	2	3.6	11.8	1:3.3
3 5	3657	137	13	3	36	219	1:6.2
Total	18,294	1,576	48	9	2 6	5.7	1:2.2b
Diarrhoea
L 1	4,990	465	4	2	0.8	4.3	1:5.4
2	3038	289	8	1	2 6	3.5	1:1.3
3	3,045	172	11	1	3.6	5.8	1:1.6
4	2979	151	11	2	3.7	13.2	1:3.6
3 5	3644	93	14	2	3.8	21.5	1:5.7
Total	17,696	1,170	48	8	2.7	6.8	1:2.5b

Source: Ref. 2.

Minus and plus signs indicate the absence or presence respectively of respiratory disease at the examination initiating a three-month interval, or of a history of four or more loose stools a day within the month preceding the examination.

a. Xerophthalmia present at the examination terminating the three-month interval.
b. p < .05,2-tailed.

TABLE 3. Incidence of respiratory disease and diarrhoea for children under five years of age with and without mild xerophthalmia-India and Indonesia

Country and age (years)	Child-intervalsa		Cases of disease		Rate per 1,000		Relative risk (+ X /-X)	p (2-tailed)
	-X	+X	-X	+X	-X	+X
Respiratory disease
India
L 1	1,540	3	153	0	99	0	0.0
2	724	11	49	3	68	273	4.0
3	756	21	36	2	48	95	2.0
4	600	27	29	4	48	148	3.1
<5	3,620	62	267	9	74	145	2.0	.06
Indonesia	14,559	537	1,003	73	69	136	2.0	< 001
Diarrhoea
India
L 1	1.540	3	961	1	624	333	0.5
2	724	11	363	5	501	455	0.9
3	756	21	269	11	356	524	1.5
4	600	27	175	10	292	370	1.3
<5	3,620	62	1,763	27	488	435	0.9	NS
Indonesia	14,415	478	867	81	60	169	2.8	<.001

Source: Ref. 3.

-X: without xerophthalmia. +X: with xeropthalmia. See text for definitions.
a. Six-month intervals for India; three-month intervals for Indonesia.

A prospective community-based study of all preschool-aged children living in the slums of Hyderabad, India (N = 1,544, based on a population registry), was conducted to estimate the incidence of post-measles corneal disease and its relationship to nutritional status [3]. At baseline, all children were examined by medical officers for signs of vitamin-A deficiency and for clinical signs of lower-respiratory infection, diarrhoea, measles, kwashiorkor, and marasmus; weights were also measured. A diarrhoeal-disease episode was defined as a day with three or more loose stools. This baseline examination was repeated at the next two six-month intervals. After the baseline examination, a surveillance system was established whereby field workers determined morbidity status from weekly home visits.

The method of analysis followed that of the central Java data set [1], but classification as vitamin-A deficient for a six-month interval was restricted to the child's xerophthalmia status at the onset of the interval. For children who were mildly xerophthalmic at the onset of a six-month cycle, the risk of lower-respiratory infection, relative to that for non-xerophthalmic children, was 2 (p = .06), and the relative risk of diarrhoeal disease was close to 3 (table 3). These results were not adjusted for potential covariates.

All incidence episodes during a six-month interval were counted in estimating the relative risk. This cumulative episodic approach to the estimation of relative risk and broader definition of a diarrhoeal case than in the central Java study should increase the relative risk for diarrhoeal disease, assuming that mildly vitamin-A-deficient children would have more episodes of diarrhoeal disease during an interval than non-vitamin-A-deficient children. The rates of diarrhoeal disease among the xerophthalmic and non-xerophthalmic were not, however, significantly different, with a relative risk of 0.90.

In sum, the central Java analyses demonstrated a two-way association between vitamin-A deficiency and diarrhoeal or respiratory infection. The Indian analyses supported an association between mild vitamin-A deficiency and an increased risk of respiratory disease, but they did not demonstrate any association between vitamin-A deficiency and diarrhoeal disease. 95% confidence intervals around each relative risk were not provided in the published reports, it is difficult to determine the extent to which the relative risks overlap.

In both studies, children with corneal xerophthalmia were hospitalized, treated with 200,000 IU of vitamin A, and dropped from the study. Children with severe systemic disease were referred to a health facility. These exclusions for ethical reasons may have reduced the relative risks of xerophthalmia and of infection respectively.

Let us examine the ability to detect an association from the data-collection techniques of these studies. In the central Java study, a child's xerophthalmia status was assessed at an examination every three months, and the diarrhoeal data were based on maternal recall for the month preceding the examination. In the data analysis, this cross-sectionally or retrospectively based classification was considered to be stable across the three-month interval. That is, if a child was diagnosed with mild xerophthalmia at the onset and end of a three-month cycle, then he or she was classified as a xerophthalmia if a mother recalled at least one day during the month preceding the exam in which the child had four or more loose stools, then the child was classified as a diarrhoeal-disease case. However, if a mother with a xerophthalmia child selectively recalled more diarrhoeal episodes than mothers of children without mild xerophthalmia, then this systematic bias could alter the estimate of the relative risk. A crucial question for these investigators is indeed whether the mothers of xerophthalmia children recalled more diarrhoeal disease than mothers of children without xerophthalmia. This analysis would include all children in the study (including those with severe systemic disease who were sent to the hospital) rather than only those with or without mild xerophthalmia at both the onset and end of a three-month period.

In the Indian data analysis, a child's xerophthalmia status at the beginning of a six-month period determined her or his classification for that interval. Given the seasonality of vitamin-A-containing foods in many communities and of vitamin-A deficiency 14]. a healthy child could develop xerophthalmia and vice versa over six months. Thus, children could be misclassified among the healthy and among the xerophthalmics. Such bidirectional misclassification of the exposure variable would bias the estimate of the relative risk of morbidity toward the null value and may have indeed done so [5]. With their weekly surveillance data, the investigators might also be able to analyse the potential association between vitamin-A deficiency and severity and duration of respiratory infection, which would be well worth examining.

An additional effort to clarity the association between vitamin-A deficiency and morbidity requires the collection and analysis of data on potential covariates, such as socio-economic status and nutritional status. For example, poverty may be associated with risk of vitamin-A deficiency and of diarrhoeal disease. Adjustment for socio-economic status may appreciably reduce any association between vitamin-A deficiency and respiratory or diarrhoeal disease [6]. Low weight for a given height may be associated with an acute illness [7] and vitamin-A status. In the central Java study, one might want to analyse those ill with diarrhoeal disease at the examination within levels of weight for height to reduce the potential bias from data collected by maternal recall.

In summary, there is a potentially strong crude association between mild xerophthalmia and an increased risk of a new episode of respiratory disease, but few covariates, if any, have been adjusted in the published reports. The association between vitamin-A deficiency and severity of respiratory disease is unknown. The association between vitamin-A deficiency and the risk and severity of diarrhoeal disease is unclear. The directionality of these associations is also unclear. Since the objective of the above-mentioned studies was to examine either blindness or post-measles corneal ulcers, there is a great need for research on vitamin-A deficiency and morbidity.

Clinical trials

Clinical trials conducted in the United States and Australia are noteworthy because of the consistency of their findings regarding the relationship between vitamin-A status and risk of new respiratory infection [8; 9]. In both trials, the investigators presumed that infants and children at high risk of respiratory infection due to lung injury from either neonatal pulmonary insults from hyaline-membrane disease or from lower-respiratory disease early in childhood would benefit from prophylactic doses of vitamin A, as evidence by reduced incidence of subsequent respiratory disease.

The US trial was a randomized, double-blind, hospital-based study of 40 very low birthweight (<1,300 g) neonates to examine the effect of vitamin-A supplementation on the risk of broncho-pulmonary dysplasia (BPD) [hi. At enrolment, the neonates were clinically and anthropometrically comparable, with mean serum retinal levels of 20 fig per decilitre per group. An intramuscular injection of retinyl palmitate (2,000 IU) or of saline solution was administered to each group every other day for 28 days. During the trial, the vitamin-A-supplemented group experienced increased levels of serum retinal and serum-retinol-binding protein (RBP) following the initial dose of vitamin A, but the placebo group did not show a significant change in the serum retinal and RBP levels (fig. 1). Thus, the change in the serum retinal and RBP levels in the vitamin-A group indicated higher rates of mobilization of vitamin A. No evidence of toxicity from hypervitaminosis-A was identified in the vitamin-A group.

FIG. 1. Plasma concentrations of vitamin A and RBP in infants. Mean values of plasma vitamin A and RBP on postnatal day 4 reflect pre-intervention values and were comparable in the two study groups. In all subsequent determinations, mean values were significantly higher the vitamin-A group. (Source: Ref. 9)

The incidence of BPD and clinical signs associated with it were significantly reduced among the vitamin-A-supplemented group as compared with the placebo group at the end of the trial (45% versus 85%, p < .008) (table 4). The number of infants requiring mechanical ventilation by the end of the trial was significantly lower in the vitamin-A group than in the placebo group (21% versus 55%). The vitamin-A-supplemented group required significantly fewer days of intensive care than the placebo group (63 versus 79, p < .02). However, during the first six months of life, four infants in the vitamin-A group and none in the control group died. Probable causes of death were related either to severe hyaline-membrane disease before enrolment in the trial, which led to BPD in an infant who died on day 13 of the trial, or to postnatal infections acquired two or more months after completion of the trial.

The Australian trial was an offshoot of a pneumo caccal vaccine trial (N = 1,273) [8]. Children one to four years old with a history of 15± days of cough or three separate episodes of respiratory illness during the three months preceding the vitamin-A trial were enrolled (N = 47). Children were randomly allocated to a vitamin-A dose of 3,828 IU or a look- and taste-alike placebo without vitamin A. The treatments were administered three times weekly for six months. Mean serum retinal levels at the onset and completion of the trial were 49 and 50 fig per decilitre respectively, in the two groups (table 5). Mothers kept daily records of signs and symptoms of respiratory illness over the trial period.

Children in the vitamin-A supplemented group who received over 50% of the dosage during the trial were included in the analysis. The vitamin-A-supplemented group had approximately 20% fewer respiratory episodes on average than the placebo group (p = .049) (table 6). Among those with a history of acute or chronic lower-respiratory infection (i.e., prior history of bronchitis, pneumonia, croup, whooping cough, or persistent cough) before the trial, the supplemented group had a 25% lower average rate of respiratory disease episodes than the placebo group (p < .05) (table 7).

Thus, in both trials the vitamin-A-supplemented group experienced a significantly reduced risk of a new episode of respiratory infection compared with the placebo group. These trials were conducted among infants and children who were at high risk of severe respiratory infection and who had different vitamin-A status, neonates with marginal serum vitamin-A levels and children with normal serum vitamin-A levels.

A cautionary note about each trial is in order. In the BPD trial, the increased mortality among the vitamin-A-supplemented group is problematic. An four deaths were attributed to BPD. One possible explanation for this problem focuses on uncontrollable post-trial factors, such as distance from the subject's home to a health-care facility. Although this explanation is conjectural, future research should, as suggested by the authors, maintain surveillance of the diet, morbidity, and vitamin-A status of the study subjects beyond the end of the trial.

The findings of the Australian trial are quite exciting, but because of methodological shortcomings they must be viewed with caution. Notably, the analysis compared the rates of respiratory episodes among the placebo group and those who had a 50% or more compliance with the treatment schedule among the vitamin-A group. Such an analysis destroys the randomization component of the study design. The investigators need to reanalyse their data using all subjects in the vitamin-A group or adjust the current respiratory infection rates of the vitamin-A group to those factors distinguishing the non-compliants from the compliants. Moreover, it is not clear why there was a 20% difference in the rates of respiratory episodes between the vitamin-A and placebo groups but a much smaller, insignificant difference in the mean number of days with specific respiratory symptoms. Perhaps the investigators can describe the missing data about the respiratory-disease episodes that determine the significant difference between the supplemented group and the placebo group. Finally, given the small cell sizes, it may always be uncertain whether these findings are real or due to chance, and, therefore, the trial requires replication.

TABLE 4. Clinical outcome of infants during

	Vitamin-A supplemented	Control	p
Incidence of broncho pulmonary dysplasia
N	9/20	17/20
%	45	85	<.008
Need for mechanical ventilation on study day 28a
N	4/19	11/20
%	21	55	<.029
Ventilatory requirements on study day 28a
FiO2	783 ± 381	895 ± 241	<.040
Ventilator rate	223 ± 530	313 ± 338	NS
peak inspiratory pressure	112 ± 224	231 ± 215	NS
positive end-expiratory pressure	29 ± 43	66 ± 49	<.020
mean airway pressure	59 ± 107	101 ± 76	<.040
oxygenation index	0.63 ± 0.36	0.69 ± 0.15	< 030
Sepsis
episodes per infantb	2.5 ± 1.0	3.1 ± 1.7	NS
airway infectionc
N	4/19	11/20
%	21	55	<.029
Retinopathy of prematurity
N	5/19	12/20	<.034
%	26	60

Source: Ref. 9.

Plus/minus values are mean ±SD. NS: not significant.
a. Area-under-curve values obtained by plotting multiple readings of variable in 24-hour period against time. Patient 1, who died on postnatal day 13 is excluded from this analysis.
b. Clinically suspected sepsis with or without confirmation by microbiologic culture resulting in initiation of antimicrobial therapy.
c. Confirmed by positive microbiologic cultures of airway secretions.

Table 5. Mean plasma vitamin-A concentrations in supplemented and unsupplemented groups

Retinol (1g 100ml)
Initial			Final
Mean	SE M	Mean	SE M	N
Placebo	48.9	2.0	49.9	1.6	43
Supplement	50.5	1.8	50.2	1.7	47

Source: Ref. 8.

In sum, compared with the placebo group, the vitamin-A-supplemented group had a reduced incidence of new episodes of (BPD) respiratory disease. The Australian trial data were stratified by history of allergy and of lower-respiratory infection, but no other adjustments for potential covariates were made. These trials were conducted among infants and children at high risk of respiratory disease, but this effect has not been demonstrated among pre-school-aged children who are not at risk of severe respiratory infections.

Vitamin-A deficiency and measles complications

Now let us turn to the potential relationship between vitamin-A deficiency and measles complications, including respiratory and diarhoeal disease, corneal ulcers, and mortality. Measles is a highly contagious disease that infects all susceptibles when exposed. Infection with this virus is characterized by impaired host immunity, which may increase the risk of complications among the already malnourished [10; 11]. Measles reduces energy intake, utilization, and absorption and damages the epithelial cells of the respiratory and gastrointestinal tracts [12; 13]. All these factors may act individually or in combination to increase the risk of complications.

TABLE 6. Mean respiratory symptoms in vitamin-A-supplemented and placebo groups

	Placebo (N=54)	Supplement (N= 53)	Difference (%)	Significance (p)a
	Mean SE M	Mean SE M
Episodes of all symptoms	8.0 0.57	6.5 0.45	-19	.049
Days of all symptoms	72.7 5.7	72.7 8.7	0	NS
Nose (days)	62.5 5.2	54.7 7.4	-12	NS
Cough (days)	28.3 3.6	32.2 5.3	-12	NS
Chest (days)	15.1 3.3	13.7 5.0	-9	NS

Source: Ref. 8.
a. Student's t-test. NS: not significant.

TABLE 7. Percentage differences between supplemented- (S) and placebo-group (P) means (P - S) stratified by allergy and lowerrespiratory-illness (LRI) history

	History of allergy		History of LRI		Total population
	No	Yes	No	Yes
N	63	44	39	68	107
Episodes of all symptoms	-11 (NS)	-25 (NS)	0 (NS)	-25(P < .05)	- 19 (P < .05)
Days of all symptoms	17 (NS)	-16 (NS)	4 (NS)	-1 (NS)	0

Source: Ref. 8.

Vitamin A has an important role in cell reproduction and differentiation of the respiratory and gastrointestinal tracts, mucosa of other organs, immune system, and conjunctiva/cornea of the eye [14-21]. Given the physiological functions of vitamin A, its deficiency has the potential to act synergistically with measles infection in the incidence of measles complications. Indeed, measles and vitamin-A deficiency have similar complications, such as diarrhoea, acute respiratory disease, and xerophthalmia [20; 22-27].

There are two recent publications on this subject. In the 15-month community-based study of post-measles complications and nutritional status in a slum area of Hyderabad, India, that was discussed earlier [3], the measles attack rate among pre-school-aged children was 23% (N = 318/1,544) [28]. Among the measles cases, the incidence rate for bronchopneumonia was 34%, for diarrhoea 37%, and for corneal lesions 3% (table 8). Children between one and two years of age suffered from the highest rates of measles and measles complications.

TABLE 8. Incidence of complications during measles

Age (years)	No. of measles cases	With broncho- pneumonia	With diarrhoea	With corneal lesions
<1	69	24	29	3
1+	105	40	45	4
2+	62	21	22	0
3+	47	16	11	0
4+	35	8	11	3
0-4+	318	109	118	10
(100%)	(34 3%)	(37.1%)	(3 1%)

Source: Ref. 28.

The mean serum retinal levels of the Indian children were always below 20 fig per decilitre regardless of measles status (table 9). During measles, the mean serum retinal levels of children with and without secondary infections or corneal lesions were not significantly different (table 10). During fever, serum retinal levels are spuriously lowered, and in a small subsample of 32 measles cases the serum retinal levels were lower during the acute phase than during the pre- and post-measles period (table 11). Malnourished measles cases (based on Gomez classification) had significantly lower mean serum retinol and albumin levels than their better nourished counter-parts (table 12).

In a Tanzanian randomized trial of hospitalized measles cases who received either a high dose of vitamin A (200,000 IU) on two consecutive days (N=88) or standard therapy without vitamin a (N=92), the case fatality rate during the month following measles was 7% (N=6/88) in the vitamin-A-supplemented group (p=.13) (table 13) [29]. A significant difference was observed among children under two years old, with a 2 % versus 17% case fatality rate (p<.05). Additionally, marasmic children had a fourfold higher case fatality rate than better nourished children regardless of treatment allocation. These findings should be considered suggestive because the total number of deaths (N=18) was small and trial was not double-blind inasmuch as the paedriatricians knew each subject’s treatment group.

TABLE 9. Biochemical parameters (cross-sectional data)

Serum levels
N		Albumin (g/dL)	Retinol (µg/dL)	RBp (mg/dL)
Control (C)	117	3.6 ± 0.03	17.9 ± 0.81	2.5 ± 0.06
Measles during infection
(Ma)	153	3.4 ± 0.03	11.5 ± 0.44	2.1 ± 0.07
after recovery (M2)	108	3.5 ± 0.05	19.8 ± 0.69	2.5 ± 0.06

Source: Ref. 28. Values are mean ± SE
Statistical significance -
C vs. M₁ p < .001 for albumin for retinol and for RBP.
M₁ vs. M₂: NS for albumin; p< .001 for RBP.

TABLE 10. Biochemical parameters (longitudinal data on TABLE 13. Mortality of children admitted with measles 32 children)

	Serum levels
	Albumin (g/dL)	Retinol (µg/dL)	RBP (mg/dL)
Premeasles (PM)	3.6 ± 0.07	16.5 ± 1.75	2.4 ± 0.14
Measles ( M, )	3.4 ± 0.08	11.1 ± 1.1	2.2 ± 0.17
Postmeasles (M2)	3.7 ± 0.07	19.2 ± 1.68	2.3 ± 0.15

Source: Ref. 28.
Statistical significance -
PM vs. M₁: p<.05 for albumin; p< .02 for retinal; NS for RBP.
M₁ vs. M₂: p < .01 for albumin, p < .001 for retinol: NS for RBP

TABLE 11. Serum vitamin-A levels in relation to measles and corneal lesions

Corneal changes in measles	N	Albumin (g/dL)	Retinol (µg/dL)
No change	47	3.3 ± 0.09	13.7 ± 1.37
Coarse keratitis	19	3.2 ± 0.09	10.8 ± 1.08
Fine keratitis	6	3.2 ± 0.14	11.6 ± 1.39
Corneal xerosis	20	3.3 ± 0.07	11.5 ± 1.63

Source: Ref. 28.

TABLE 12. Biochemical parameters in measles according to nutritional status

Nutrition Grade	No. of children	Albumin (g/dL)	Retinol (µg/dL)	RBP (mg/dL)
Normal+
Grade I	58	3.4 ± 0.05	12.5 ± 0.80	2.0 ± 0.80
Grade II	74	3.4 ± 0.05	11.3 ± 0.60	2.2 ± 0.09
Grade III	21	3.2 ± 0.07a	8.3 ±0.75b	2.2 ± 0.15

Source: Ref. 28.
a. p < .05 compared with Normal + grade I.
b. p < .02 compared with Normal + grade I.

TABLE 13. Mortality of children admitted with measles

Age (months)	Children admitted		Children who died
	Given vitamin A	Controls	Given vitamin A	Controls
<9	14	9	0	2(22)
9 - 11	12	10	0	2(20)
12-23	20	23	1(5)	3(13)
24-35	11	16	3(27)	2(13)
36-47	11	13	1(9)	1(8)
48- 59	8	6	1(13)	0
³	12	15	0	2(13)
Total	88	92	6(7)	12(13)

Source: Ref. 29.
Principle data are numbers of children. Figures in parentheses are percentages.

TABLE 14. Measles complications and associated mortality

Complication	Children with complications		Children who died
	Given vitamin A	Controls	Given vitamin A	Controls
Pneumonia	38 (43)	47 (51)	3 (8)	7 (15)
Otitis media	19 (22)	20 (22)	1 (5)	3 (15)
Group or laryngo tracheobronchitis	8 (9)	13 (14)	1 (50)	4 (31)
Dysentery	2 (2)	6 (7)	1 (4)	3 (50)
Haemorrhagic rash	28 (32)	34 (37)	1 (11)	4 (12)
Oral candidiasis	9 (10)	5 (5)		1 (20)

Source: Ref. 29.
Principle data are numbers of children. Figures in parentheses are percentages.

Based on the two measles studies presented above, vitamin-A deficiency and measles complications do co-occur but are not necessarily associated. Nevertheless, the pathophysiology is sufficiently similar in measles and vitamin-A deficiency to suggest the need for future research in this area.

The evidence from the hospital-based trial leads us back to the potential association between vitamin-A deficiency and respiratory disease inasmuch as respiratory-specific mortalities were reduced among the vitamin-A-supplemented versus the control group (table 14). Given the findings of respiratory-disease-specific mortality differences in the Tanzanian trial, a follow-up study might be designed as a randomized double-blind trial of vitamin-A supplementation (versus a placebo) administered to measles and non-measles respiratory-disease cases to compare their risk of mortality.

A curious feature of the Tanzanian trial is the absence of any diarrhoeal-disease deaths. In the Indian study [28|, 34% of the measles cases suffered from diarrhoeal-disease complications during the acute stage of measles, and 76% suffered from diarrhoeal disease during the six months following measles. Thus, the clear absence of any diarrhoeal deaths in the Tanzanian trial is noteworthy.

Before leaving the subject of vitamin-A deficiency and measles complications, I would like to discuss current World Health Organization recommendations 130] briefly. The WHO and the United Nations Children's Fund recommend that a high-potency preparation of 100,000 and 200,000 IU of vitamin A be administered to measles cases under and above one year old respectively. They state that this therapeutic regimen should be followed in communities with measles fatality rates of 1% or higher or in communities with recognized vitamin-A deficiency. This recommendation is based on the above-mentioned hospital-based trial in Tanzania, but the dosages differ. Moreover, this recommendation is presented for a broader range of communities than can be extrapolated from the trial: vitamin-A distribution to measles cases is recommended for any community with a measles-case fatality rate of 1% or more regardless of its vitamin-A status.

Before such a recommendation is implemented, several studies might be conducted, including a trial of vitamin-A supplementation (versus a placebo) before the measles season to examine the risk of the incidence and duration of measles complications. Another trial might examine the impact of a high-potency preparation of vitamin A (versus a placebo) administered to non-hospitalized measles cases during the acute stage on the incidence and duration of complications. These trials would address the preventive effect of the administration of vitamin A compared with that of a placebo prior to and during the measles season.

Summary and research recommendations

Let me briefly summarize this review and discuss several issues regarding future research. First, there is a potentially strong association between vitamin-A status and the risk of respiratory infection. Second, it is more difficult to state the potential for the association between vitamin-A deficiency and diarrhoeal disease based on the techniques of morbidity assessment or the analytic approach to data taken in published articles. It is noteworthy that the studies presented were not designed to examine the association between vitamin-A deficiency and morbidity. Third, vitamin-A deficiency and measles complications co-occur, and indeed, vitamin-A supplementation during measles may reduce respiratory-specific causes of mortality. These statements refer primarily to crude associations that require adjustment for potential covariates in future research.

One cannot address future research in vitamin-A deficiency without examining methods of vitamin-A assessment and detection of morbidity. Essentially, researchers must be able to determine subclinical and clinical vitamin-A deficiency in order (a) to compute an unbiased estimate of the risk of morbidity among these two groups in comparison with those who have adequate vitamin-A status and (b) to examine the potential threshold effect of vitamin-A status on morbidity. Various methods of vitamin-A assessment exist, but ideally they should be evaluated for their field applicability [31; 321, their sensitivity and specificity across the spectrum of vitamin-A deficiency, and their reproducibility before research on vitamin-A deficiency and morbidity is conducted.

Another complicating issue in vitamin-A assessment is determining the vitamin-A status of febrile children, notably children with acute infections, because fever spuriously lowers serum retinal levels. This issue needs to be addressed in virtually all vitamin-A and morbidity studies that assess the vitamin-A status at the time of diagnosis of an infection. One plausible method of correcting for the effect of fever on serum retinal levels is to measure the levels of serum acute-phase reactant protein, which is inversely associated with serum retinal levels under stressful conditions (such as after surgery). Thus, the changes in C-reactive protein levels [33] can be used to adjust the serum retinal levels to pre-infection status.

The ability to detect an association between vitamin-A deficiency and diarrhoeal and respiratory infections, as well as the direction of that association, is contingent on the quality of morbidity data collection. The information required for diagnosis and determination of the severity of infection includes standardized clinical validation and, preferably, microbiologic analysis. In a proposed University of Gadjah Mada and Johns Hopkins University vitamin-A and morbidity field trial [34], my colleagues and I are developing a morbidity surveillance system of twice-weekly home contact with the child's caretaker in order to identify symptoms and signs of infection. Within 24 hours of the identification of a potentially ill child, a physician will *sit the home to validate the diagnosis clinically and take a specimen for microbiologic analysis. With this information, we will be able to distinguish single from multiple episodes, the duration of each episode, and acute from chronic episodes. With such data one can examine whether vitamin-A deficiency increases the risk of a new episode and/or the severity of an infection and, indeed, the association between vitamin-A deficiency and etiologic-specific agents of diarrhoeal and/or respiratory disease.

If future studies of vitamin A and morbidity are designed as clinical of field trials, then some attempt should be made to validate the proper dosage of vitamin A to be administered in the trials. Two hundred thousand IU of vitamin A is the high-potency preparation of vitamin A that will be administered in all the ongoing mortality trials and the proposed morbidity trials. This dosage is based on one study of Brazilian children under 10 years old who had a relative-dose-response test at baseline, followed by the administration of a dose of 200,000 IU of vitamin A. At 30, 120, and 180 days after the dosing, the relative-dose-response test was repeated to determine the time interval until return to the baseline serum retinal levels 1351. The impact of dosages lower than 200,000 IU on the risk of morbidity or mortality has not been examined, but the ethical objectives of current field trials should be the administration of the lowest dosage of vitamin A that potentially reduces the risk of morbidity.

All research on vitamin A and morbidity should be conducted in communities with evidence of clinical and/or subclinical vitamin-A deficiency to ensure that there will be beneficial effects of a trial by treating children with xerophthalmia. Thus, recent data on vitamin-A deficiency rates within the past five years are required before the onset of the trial.

Finally, covariates that may potentially influence the association between vitamin A and morbidity include season; dietary intake of vitamin A, protein, and fats; anthropometric status, including weight and height; intestinal parasites that may reduce the ability to absorb vitamin A; socio-economic status; access to and utilization of health care; and child's age and sex. Information about these variables should be collected during a study, and the variables should be examined as potential interaction terms, effect modifiers, and/or confounding variables with respect to the association between vitamin-A deficiency and morbidity.

References

1. Sommer A, Katz J. Tarwatjo I. Increased risk of respiratory disease and diarrhea in children with preexisting mild vitamin A deficiency. Am J Clin Nutr 1984; 40:1090 95.
2. Sommer A, Tarwotjo I, Katz J. Increased risk of xerophthalmia following diarrhea and respiratory disease. Am J Clin Nutr 1987;45:977-80.
3. Milton RC, Reddy V, Naidu AN. Mild vitamin A deficiency and childhood morbidity: an Indian experience. Am J Clin Nutr 1987;46:827-29.
4. Oomen HAPC, Doesschate JT. The periodicity of xerophthalmia in South and East Asia. Ecol Food Nutr 1973;2:207-17.
5. Copeland KT, Checkoway H. McMichael AJ, Holbrook RH. Bias due to misclassification in the estimation of relative risk. Am J Epidemiol 1977;105:488-95.
6. Feachem RG. Vitamin A deficiency and diarrhoea: a review of interrelationships and their implications for the control of xerophthalmia and diarrhoea. Trop Dis Bull 1987;84:R1-R16.
7. WHO Working Group. Use and interpretation of anthropometric indicators of nutritional status. Bull WHO 1986;64:929-41.
8. Pinnock CB, Douglas RM, Badcock NR. Vitamin A status in children who are prone to respiratory tract infections. Aust Paediatr 1 1986;22:95-99.
9. Shenai JP, Kennedy KA, Chytil F. Stahlman MT. Clinical trial of vitamin A supplementation in infants susceptible to bronchopulmonary dysplasia. J Pediatr 1987;111:269-77.
10. Chen LC, Chowdhury AKMA, Huffman SL. Anthropometric assessment of energy-protein malnutrition and subsequent risk of mortality among preschool aged children. Am J Clin Nutr 1980;33: 1836.
11. Enders JF. Measles virus: historical review, isolation and behavior in various systems. Am J Dis Child 1962;103:282.
12. Duggan MB, Alwar J. Milner RDG. The nutritional cost of measles in Africa. Arch Dis Child 19S6;61:61.
13. Duggan MB, Milner RDG. The maintenance energy requirement for children: an estimate based on a study of children with infection associated underfeeding. Am J Clin Nutr 1986;43:870-78.
14. Brown KH, Rajan MM, Chakraborty J. Aziz KMA. Failure of a large dose of vitamin A to enhance the antibody response to tetanus toxoid in children. Am J Clin Nutr 1980;33:212-17.
15. Fusi S. Kupper TS, Green RD, Ariyan S. Reversal of postburn immunosuppression by the administration of vitamin A. Surgery 1984;96:330-35.
16. Jetten AM, Smits H. Regulation of differentiation of tracheal epithelial cells by retinoids. In: Nugent J. Clark S. eds. Retinoids, differentiation and disease. Ciba Foundation Symposium 113. London: Pittman, 1985:6167.
17. Moriguchi S. Werner L, Watson RR. High dietary vitamin A (retinyl palmitate) and cellular immune functions in mice. Immunology 1985;56:169-77.
18. Olson JA. The biological role of vitamin A in maintaining epithelial tissues. Isr J Med Sci 1972;8: 1170- 78.
19. Shapiro SS. Retinoids and epithelial differentiation. In: Retinoids and cell differentiation. Boca Raton, Fla, USA: CRC Press, 1986:29-59.
20. Sommer A. Nutritional blindness: xerophthalmia and keratomalacia. New York: Oxford University Press, 1982.
21. Underwood BA. Vitamin A in animal and human nutrition. In: Sporn MB, Roberts AB, Goodman DS, eds. The retinoids. Orlando, Fla, USA: Academic Press, 1984:vol.1 :281-392.
22. Bhaskaram P. Reddy V, Raj S. Bhatnagar RC. Effect of measles on the nutritional status of preschool children. J Trop Med Hyg 1984;8:21-25.
23. Blackfan KD, Wolbach SB. Vitamin A deficiency in infants: clinical and pathological study. J Pediatr 1933; 33:679-706.
24. Koster FT, Curlin GC, Aziz KMA, Haque A. Synergistic impact of measles and diarrhoea on nutrition and mortality in Bangladesh. Bull WHO 1981;59:901-08.
25. Pongrithsukda V, Phonboon K, Manunpichu K. Measles-associated diarrhoea in northeastern Thailand. Southeast Asian J Trop Med Public Health 1986;17:4347.
26. Sanford-Smith JH, Whittle HC. Corneal ulceration following measles in Nigerian children. Br J Ophthamol 1979;63:720-24.
27. Shahid NS, Caluquin P. Shaikh K, Zimicki S. Long-term complication of measles in rural Bangladesh. J Trop Med Hyg 1983;86:77-80.
28. Reddy V, Bhaskaram P. Raghuramulu N. et al. Relationship between measles, malnutrition, and blindness: a prospective study in Indian children. Am J Clin Nutr 1986;44:924-30.
29. Barclay AJ, Foster A, Sommer A. Vitamin A supplements and mortality related to measles: a randomized clinical trial. Br Med J 1987;294:294-96.
30. Vitamin A for measles Joint WHO/UNICEF statement. Wkly Epidem Rec 1987;19:133-34.
31. Underwood BA. Effect of protein quantity and quality on plasma response to an oral dose of vitamin A as an indicator of hepatic vitamin A reserves in rats. J Nutr 1980;110:1635-40.
32. Wittpenn JR, Tseng SCG, Sommer A. Detection of early xerophthalmia by impression cytology. Arch Ophthalmol 1986;104:237-39.
33. Labadarios D, Brink PA, Weich FH, et al. Plasma vitamin A, E, C and B6 levels in myocardial infarction. S Afr Med J 1987;71:561-63.
34. Forman MR, Brown KH, Black RE, Sack RB, Steinoff M. Proposal to assess the impact of vitamin A supplementation on the reduction of the incidence and duration of episodes of morbidity among children aged 6-60 months in Indonesia. Presented at the National Academy of Sciences, 17 Aug 1986. Institute of International Programs and Department of International Health, School of Hygiene and Public Health, Johns Hopkins University, in collaboration with the University of Gadjah Mada, Yogyakarta, Indonesia, 1988.
35. Flores H, Campos F, Araujo CRC, Underwood BA. Assessment of marginal vitamin A deficiency in Brazilian children using the relative dose response procedure. Am J Clin Nutr 1984;40:1281-89.

Ethical issues related to nutrition field trials

Joan P. Porter

This paper reviews several fundamental ethical principles, provisions of the US federal regulations for the protection of human subjects, and past deliberations of the Subcommittee on Vitamin A Deficiency Prevention and Control regarding ethical concerns in vitamin-A studies. The fundamental ethical principles were set forth by the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research in its Belmont Report |l]. These principles were the basis for the regulations of the Department of Health and Human Services (DHHS) for the "Protection of Human Subjects" [2], and they guide the design, review, and conduct of any research involving human subjects - they are the foundation for our rules and norms.

Ethical principles

The first ethical principle is respect for persons. Two corollaries of this principle are that individuals should be treated as autonomous agents and that persons with diminished autonomy are entitled to protection. Kant stated the principle in these words: "So act as to treat humanity, whether in thine own person or in that of any other, in every case as an end withal, never as a means only`' [3]. Obtaining informed consent is an important activity in research that derives, in part, from this principle.

The second principle is beneficence. Persons are treated in an ethical way by making efforts to serve their well-being. Beneficent actions involve two rules: "do not harm"; maximize possible benefit and minimize possible harms. Claude Bernard indicated that one should not injure even one person involved in research regardless of the benefit that might come to others 13]. This is not always an easy principle to app ly, because avoiding what is harmful requires learning what is harmful, and that sometimes means exposing people to risks of harm. Also, researchers have to act to benefit their subjects based on their best judgement - and at times there is no ready consensus or scientific information about what is in the best interest of human beings. These imperatives, then, require us to weigh possible risks and possible benefits and to determine what is too great a risk in pursuit of a possible benefit. Research designers and groups who review research seek to minimize the risks and to maximize the benefits by using procedures, processes, and designs that clearly weight the benefits side of the scale - for the individuals involved and also for individuals to follow and for society in general.

Some ethicists who have considered beneficence have said there are obligations that derive from this principle. In Frankena's schema [cited in ref. 3, p. ill], for example, these are, in order of strength: (l) One ought not to inflict evil or harm. (2) One ought to prevent evil or harm. (3) One ought to remove evil. (4) One ought to do or promote good. I will return to the balancing of risks and benefits below.

The third principle is justice. Justice requires that we treat persons fairly and that we give each person what he or she is due. The National Commission [1] was concerned, in large part, with "distributive" justice, which involves the distribution of scarce benefits when there is competition for them. It was also concerned with the distribution of burdens, particularly with regard to imposing burdens on fewer than all members of a class of persons. lust distribution of burdens and benefits is an important concept underlying the selection of research subjects. As the National Commission notes, researchers have to decide who should receive the benefits and who should bear the burdens. The selection of subjects must be carefully reviewed to determine whether some classes are being systematically selected primarily because of their easy availability or manipulability. If the research pays off, one concept of justice requires that those who are disadvantaged have access to the benefits of research and that those who have borne the burdens be considered with high priority in distributing the benefits [l].

In writing about research and treatment of acquired immunodeficiency syndrome (AIDS), Leroy Walters [4] of the Joseph and Rose Kennedy Institute of Ethics has suggested a still nebulous, fourth ethical principle: "community," or "mutuality," or "solidarity." This principle acknowledges that, in view of the complexities of major health epidemics or problems, we must include this value with others to guide actions. Some of the vitamin-A studies in question here must accommodate the importance of local authority and of community reassurance and participation in decision making. These values appear to be related to Dr. Walters's emergent principle.

Risks and benefits

Risks and benefits can be physical, psychological, social, and economic. In this regard, we tend to think of the vitamin-A studies as trade-offs of physical risks and benefits, but the culture of the research subjects may sometimes create subtle benefits or risks - community pressure about participating or not participating in a study or risks of excessive expectation. Possible psychological risks might derive from one's having improved health by virtue of receiving an intervention or accompanying medical attention and then having that benefit withdrawn, sometimes rather rapidly, after the completion of a trial. The mere inconvenience of an intrusion into one's daily activities may also be a risk.

In the best of situations, those who are designing the research would provide quantitative estimates of the probability and magnitude of risks and benefits based on empirical data, but that is not an easy task. There may not be agreement, for example, on how to weigh a risk or benefit, or agreement on the relevance or accuracy of previous findings.

Several points about benefits should be considered when vitamin-A studies are contemplated. First, anticipated benefits really have to be weighed in terms of expected duration. Also, research designers must consider what happens if the benefit of an experimental health intervention is provided in a community that would not otherwise have access to it. Is there any possibility of continuation through other governmental or private support once an efficacious research intervention ceases?

Second, in areas of the world where individuals have no access to basic health care, one sometimes hears the argument that persons in the control group of a randomized trial, who receive no intervention or who receive a placebo, are "no worse off" for the re search. They did not have access to a vitamin or other nutrient before the trials, so why should we be concerned if they have none now in the trial? Are the risks any greater, considering probability and magnitude, than those they ordinarily encounter in daily life [2]? Some may argue that this rationale seems to contradict the obligation to maximize benefits - especially if there are relatively basic medical interventions basic medical interventions that can be provided to study subjects.

Third, there exists a dilemma if agreement is lacking on what works and what does not or on what previous research means or requires us to do to avoid harm and to promote benefit. Widespread application of tentative or unproven interventions is not ethical - it prevents us from finding the best treatment, raises hopes inappropriately, and exposes those who receive the unproven intervention to unknown risks. Further, unsound health interventions waste resources in a world where they are scarce.

Fourth, the concept of benefit is also closely linked to selection of subjects, as noted above. Those who stand to gain the most from the research results are those who might first be asked to assume the risks. In discussing maximization of benefits, Robert Levine [3] of Yale University suggests that ethical codes and regulations forcefully prohibit causing death or injury, but obligations to promote good are based on good scientific design and good balance of harms and benefits. If the benefit of promoting health is based on the obligation to avoid harm, then every subject - even those in a control group - should have the best proven diagnostic and therapeutic method. Witholding an effective therapy for a disease that if untreated may produce death or disability is not acceptable according to Levine [3, P 45].

A corollary to justifying risks in terms of benefits is the obligation to minimize risks. Sometimes minimization takes the form of eliminating non-essential procedures, such as drawing extra blood. Another approach is monitoring, through data and safety monitoring boards, in blinded trials to detect unanticipated statistical trends that reveal problems or results that are so dramatic that justification to continue the controlled trial is unwarranted. Setting clear end points and building consensus on the meaning of the data flow from the monitoring concept.

Prescreening is another way to minimize risk. For example, in the vitamin-A studies, children with clear signs of ophthalmological symptoms were not entered into the trials or were withrawn from the test groups and treated. In selecting subjects, research designers should consider involving the least vulnerable, least at-risk persons to obtain the data needed. Some may say that this takes the form of involving persons with marginal deficiencies or only mild illnesses rather than those at later stages of deprivation or illness, if possible. Other things being equal, one looks for the least vulnerable representatives of populations as subjects.

Some of the studies the subcommittee reviewed in 19S6 involved the randomized clinical trial, in which one group receives the intervention and a control group receives a placebo or nothing [5]. Many ethical problems are peculiar to this type of study design. Some researchers believe that we have relied on randomized clinical trials too much. D. L. Sackett [cited in ref. 3, p. 136] maintains that the objectives of clinical trials are validity, generalizability, and efficiency - the first objective, validity, being the mandatory one. Sackett believes that problems arise when the three objectives are out of balance or given the wrong priority. For example, efficiency might require that high-risk persons be enrolled so that dramatic results can be achieved, but validity and generalizability may be compromised.

Use of historical controls is one alternative to randomized clinical trials. But, in the face of many confounding factors, this design is often criticized. Richard Feachem has suggested that it is unethical to follow prospectively children with any signs or symptoms of vitamin-A deficiency without providing full vitamin-A therapy. He also suggested use of the case-control method in diarrhoeal and vitamin-A studies as an alternative to randomized clinical trials [6, p. 13].

US federal regulations

The ethical principles that have been set forth are not always easy to apply. Nor is it always evident how to apply them. The DHHS regulations for the protection of human subjects [2l provide some rules and processes for application of the principles. Two processes required by the regulations provide a means to put the principles into practice: use of an institutional review board (IRB) and obtaining informed consent. Although some of ethical codes, such as those of Helsinki or the Council of International Organizations of Medical Societies of the World Health Organization, recommend general courses of action for the conduct of ethical research, the DHHS regulations are more explicit and precribe procedures.

(Not addressed here is subpart D of the regulations, "Additional Protections for Children Involved as Subjects in Research." Note that the regulations provide several additional risk/benefit categories for an IRB to examine. For example, if, in a research protocol intended for support, the risks are relatively high and the knowledge to be gained and benefits are quite important but far removed from the children who are subjects, the Secretary of Health and Human Services would convene a second review to consider the research and make recommendations about whether it can go forward.)

Institutional review boards

The regulations call for a local committee of specific composition, an institutional review beard, with sufficient authority and an independent perspective to assess the balance of the possible risks and possible benefits of research. The IRB assesses the minimization of risks, the adequacy of the informed-consent process, and protection of vulnerable subjects. Along with research experts, the IRB must include persons unaffiliated with the institution sponsoring the research, persons who can represent community attitudes and values, and persons who are not scientists [2l. Although it is important to involve government officials in the planning of field studies, sometimes the perspective of those with other than political agendas can help sort out what are acceptable risks and benefits.

In negotiating assurances of compliance with the regulations for the protection of human subjects for research that is sponsored by a domestic institution in a foreign country, the Office for Protection from Research Risks, with few exceptions, asks for at least two IRB reviews - one by the IRB in the domestic institution and one in the country where the research is to be conducted. Local review, i.e., in the community where the research will take place, is sought. Whereas most IRBs in the United States are attached to academic or research organizations, there are many variations abroad. Quite often an existing government review committee, supplemented to meet the requirements of the regulations, is utilized. Because the governments often do not do research, reviews can be complicated by other goals. In most of the trials, the national governments are involved at some point, and they must be for the success of the effort and for ensuring continued commitment if the results of the research are favourable.

Obtaining informed consent

The other process required by the regulations is obtaining informed consent. The basic elements of informed consent include considerations a reasonable person would want to know. The informed-consent process respects the value of autonomy of an individual. In this process, persons must be told (a) that the intervention is research, (b) what alternative treatments there are, if any, and (c) the foreseeable risks and foreseeable benefits to themselves and to others from their participation as subjects. They are informed about the extent of confidentiality and, if the research is of more than minimal risk, of any availability of compensation in the event of an injury. They are told that their participation is voluntary, that they may withdraw at any time, and that, if they decide not to participate in the research, they will not be denied any benefits to which they might otherwise be entitled. They are also given the name of someone of whom they may ask additional questions [2].

Informed consent, verbal or written, must be obtained; but, if it is to be meaningful, information must be explained, and it is no easy task to explain alternative methods, long-term risks and discomforts, and possible benefits, particularly to potential subjects in developing countries [7].

To hand out consent forms to illiterate people is not sufficient. Also, in some areas, signing any papers is considered an act of self-incrimination or makes possible subject quite uneasy. One might also need to seek consent from the traditional head of an area, a parent, or the chief teacher for schoolchildren. Husbands must also consent for their wives in some areas

In the United States, we place a high value on autonomy and individual rights. In other cultures, this value is modulated somewhat by values of community solidarity and community authority. In such cultures, it is important to convince local tribal and village elders and religious leaders of the acceptability of the research to be done. This requires more lead time to prepare for a study, but it helps to ensure a high degree of participation and compliance. Obtaining the local leader's consent is necessary, but is not sufficient to guarantee participation. Dr. Keith West of the Johns Hopkins Medical Institution cautions that sensitivity to superstitions, local norms, local behavior - indicators that tell a researcher whether someone is consenting or refusing - is essential (personal communication, 1988). De Maar et al. [7] and others writing about the management of clinical trials in developing countries have advised that constituting permanent national or international advisory committees to provide broad support can be helpful (e.g., scientific advice and data monitoring), but they can be costly. Also, the co-operation and understanding of local health dignitaries and the traditional medicine man or the local dispenser should be sought. Their dissatisfaction could jeopardize a study [7].

In developing countries in which field trials are to be conducted, researchers and IRBs need to consider, then, the constraints on obtaining permission to conduct trials and on obtaining informed consent. E. Ekamen [8] of Nigeria, in discussing methodological constraints and limitations in developing countries, cited illiteracy as the major drawback to the conduct of research. Populations are often ill-informed and have little understanding of the value and objectives of research projects; they do not participate unless some clear benefit is offered. Language barriers should not be underestimated; sometimes there are no local concepts for the technical terms involved in research. Concepts of time and causality are culturally defined. Placebo-controlled trials and blinded and double-blinded studies are not well understood [9]. Also, it should be remembered that selecting persons as potential subjects who are in schools or other institutions in developing countries may lead to a biased sample of the general population.

Previous subcommittee deliberations

The deliberations of the subcommittee in August 1986, as reported in Vitamin A Supplementation: Methodologies for Field Trials [5], identified the following problems:

• the lack of data about the effects of vitamin A on mortality among people with marginal vitamin-A status, i.e. without clinical signs;
• inconclusive animal data concerning the relationship of vitamin A and resistance to infection;
• lack of consensus on the meaning of Sommer's [10] clinical trial in Indonesia, and the lack of consensus on the need for additional controlled trials to demonstrate with scientific rigour the effects of vitamin A on morbidity.

There were minority opinions among the subcommittee. Some, for example, noted that persons with marginal vitamin-A status are at increased risk of developing severe deficiency. Their risk remains if they are in the control group receiving a placebo. Children who will eventually become xerophthalmic because of vitamin-A deficiency are already compromised, but the ways to measure and diagnose this are not so direct. Because some subcommittee members do not consider vitamin A an "experimental agent," they urged studies without placebo groups in areas of high morbidity and xerophthalmia. Others were concerned that high risks were borne by the control population, whereas benefits would be universal.

A suggestion was that an alternative to randomized clinical trials that would shift or minimize risks more appropriately might be retrospective case control studies of mortality based on who accepts and who refuses vitamin-A supplements. Other possibilities suggested were time-staggered introduction of capsules and other services into planned vitamin-A health care delivery and intensive vitamin-A distribution to selected populations, and regular distribution of government supplements to other populations.

Ethical safeguards considered by the subcommittee included the following:

• careful assessment of risk based on the data available;
• identification of benefits that could accrue to both participating and control groups in the trial;
• building into studies US and foreign reviews for the protection of human subjects;
• establishing data and safety monitoring boards; - identification of the need for informed consent;
• extension of the benefits of the research to the control group as soon possible after the trial.

It appears that tensions surround the questions of how much we already know and of the extent to which subcommittee members agree on the validity and generalizability of the research data about vitamin-A studies. Ethical deliberations are not undertaken within a neat framework. There is always some degree of conflict about the nature and weights of values, benefits, and risks, about the primacy of principles, and about interpretation of previous studies. There are also conflicts about deciding on research priorities and about balancing research versus other immediate health treatment necessities in the face of scarce resources.

The subcommittee needs to address these points in considering future studies. Although there are no easy answers, this paper should help to refresh memories about past discussions and to guide future

References

1. National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research. The Belmont report: ethical principles and guidelines for the protection of human subjects of research. US Department of Health, Education and Welfare. DHEW publication no. (05)78~012. Washington, D.C.: US Government Printing Office, 1978.
2. US Department of Health and Human Services (DHHS). Protection of human subjects. Title 45, Code of Federal Regulations, part 46. Office for Protection from Research Risks. OPPR reports. Washington, D.C.: National Institutes of Health, Public Health Service, 1983.
3. Levine R1. Ethics and regulation of clinical research. 2nd ed. Baltimore, Md, USA: Urban & Schwarzenberg, 1981.
4. Walters L. Ethical issues in the prevention and treatment of HIV infection and AIDS. Science 1988;239:597603.
5. National Research Council. Vitamin A supplementation: methodologies for field trials. Subcommittee on Vitamin A Deficiency Prevention and Control, Committee on International Nutrition Programs, Food and Nutrition Board, Commission on Life Sciences. Washington, D.C.: National Academy Press, 1987.
6. Feachem RG. Vitamin A deficiency and diarrhoea: a review of interrelationships and their implications for the control of xerophthalmia and diarrhoea. Trop Dis Bull 1987;84:R1-R16.
7. De Maar EWJ, Chaudhury RR, Ekue JMK, Granata F. Walker AN. Management of clinical trials in developing countries. J Int Med Res 1983;11:1-5.
8. Ekamen EE. Field epidemiology: methodological constraints and limitations in the developing world. Public Health 1985;99:33-36.
9. Halsey N. Responsibilities in international AIDS research. Regional workshop on the protection of human subjects: new trends and issues. 12-13 May 1988. Sponsored by the University of Maryland Graduate School, the National Institutes of Health, Office for Protection from Research Risks, and the Food and Drug Administration, Public Health Service. Baltimore, Md, USA: Department of Health and Human Services, 1988.
10. Sommer A, Tarwotjo 1, Djunaedi E, et al. Impact of vitamin A supplementation on childhood mortality. Lancet 1986;2:1169-73.

Commentary: Underpinning vitamin-A deficiency prevention and control programmes

Barbara A. Underwood

If you give a hungry man a fish, he is fed for one day but is dependent upon you for continued sustenance. If you teach a hungry man to fish, he is independent for life.

The analogy between this well-known saying and the approaches to the prevention and control of vitamin-A deficiency is obvious: providing children with high-dose capsules of vitamin A saves many from developing clinical symptoms and perhaps reduces mortality and morbidity, as long as the dose can be delivered repeatedly at specified intervals. If the system fails or the individual child is not reached, the problem recurs. Approaches to prevention that foster practical solutions attainable through better utilization of available food and other resources are more difficult to implement and take longer to bring about the needed behavioural changes in child-rearing practices. But they can be permanent and address health and nutrition issues that commonly coexist with vitamin-A deficiency.

Most vitamin-A intervention programmes recognize these facts and include an "educational" component. In practice, however, the educational component takes a back seat to efforts required for the delivery and monitoring of the high-dose capsule. The personnel responsible for capsule delivery frequently inform recipients of what the capsule is for and of foods they should eat that contain the vitamin, but fail to communicate the message in a locally appropriate, meaningful way that changes behaviours: such communication may be perceived as taking too much time. This fact is illustrated by the evaluation report of the Bangladesh vitamin-A distribution programme described in an earlier issue of the Food and Nutrition Bulletin [1].

Clearly there is need to rethink strategies for vitamin-A-deficiency prevention and control. The high-dose medical approach is appropriate under circumstances where a public health problem exists and alternatives are not feasible, e.g. where water, transportation, and food-storage facilities are in short supply or non-existent. Often, however, these circumstances are regionally clustered and not applicable to an entire country. But even under these circumstances, strategies that combine the short-term medical approach with programmes addressing underlying conditions that contribute to high rates of infections - e.g. programmer to improve personal and environmental sanitation and increase immunization coverage - can have beneficial spin-off effects on the vitamin-A deficiency problem. As the evaluation of the Bangladesh programme by Darnton-Hill et al. [1] illustrates, the efficacy of the medical model is limited by the inefficient delivery system. There is no doubt that the programme has saved the sight and lives of many Bengali children, but, as the authors note, it has not reduced the overall prevalence of the problem - even after 14 years. In addition, the struggle to improve the delivery system and its monitoring is consuming much of the national human and economic resource pool.

During the 14 years of the Bangladesh programme, some evidence indicates that diets not only have not improved nutritionally with respect to vitamin A but have deteriorated, and that little change has occurred in personal and environmental health practices. After 23 rounds of vitamin-A-capsule distribution, limited knowledge about the programme exists: 34%-60% of mothers did not know what the capsule was for, 15%21% had not seen the educational materials, and 51%75% could not name a vitamin-A-rich food. It is precisely this kind of evaluation data that frequently is used by opponents to illustrate that educational approaches don't work! But can we blame this failure on the educational approach, or should we admit that we have been ineffective communicators in the educational component of the currently operational high-dose programmes? Often we ask overburdened, unmotivated, and minimally trained delivery personnel to get the message out. Or we determine that only those who have higher education have sufficient knowledge to effectively compose and communicate the message, whereas those to whom we want most to relate are underprivileged and often lack formal education and access to other social programmes. But they are survivors. As survivors they have had to make choices - choices that include which of the many messages they hear and programmes forced upon them they will choose to act upon in the use of their limited resources, both of time and of money. Choice, however limited, is valued irrespective of socio-economic status.

People change practices when they are convinced that the change is to their benefit and they choose to change. Choice is too frequently left out of approaches to solving public health problems, including vitamin-A intervention strategies. Most universal capsule distribution programmes do not entertain choice as an option, yet targeted recipients for such programmes choose not to participate in increasing numbers in successive rounds, as evaluations of the Bangladesh and other national programmes illustrate. Indeed, proponents of fortification programmes proclaim the lack of choice as the major advantage of a fortification strategy. But, as occurred with the sugar fortification programme in Guatemala, as effective as the programme was shown to be while operational, the situation deteriorated rapidly when it was disrupted by internal political and economic changes. No demand for continuation of the programme had been created among the passive recipients.

How can the concept of choice be introduced into strategies for the prevention and control of vitamin-A deficiency? Just as with any other programme, there is not likely to be a universally applicable answer. Each situation has to be evaluated at the national, community, and family levels. The important point is that choices usually do exist if imagination and innovative thinking are applied, and these choices could be made available when considering strategies at each level of intervention. In some instances where clinical deficiency is rare, a national programme to improve the intra-country preservation, storage, and year-round availability of vitamin-A-containing foods, combined with an effective programme to improve consumption, might be an appropriate alternative to a high-dose programme. Elsewhere, a community-based feeding programme, a community- or family-level income-generation programme to provide economic resources to permit a choice of appropriate foods, or a kitchen/community garden may be alternatives - and these programmes are not mutually exclusive. Until we create a demand for a programme or a product, i.e. convert programme recipients into programme consumers, whether for a high-dose capsule, lower-cost green leafy vegetables, or better means of preparing and preserving vitamin-A-rich foods for feeding young children, it is difficult to conceive of achieving the effective sustained behavioural change that must occur to eradicate and control vitamin-A deficiency as a public health problem.

Reference

1. Daraton-Hill 1, Sibanda F. Mitra M, et al. Distribution of vitamin-A capsules for the prevention and control of vitamin-A deficiency in Bangladesh. Food Nutr Bull 1988;10(3):60-70.