| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
LETTER TO THE EDITOR |
University of Wisconsin-Madison
Department of Nutritional Sciences
Madison, WI 53706
27305 Presley Street
Sun City, CA 92586
E-mail: sherry{at}nutrisci.wisc.edu
Dear Sir:
We have concerns about the article by Verhoef and West (1), which claims that the results obtained with the relative-dose-response (RDR) and the modified-RDR (MRDR) tests are due to random variations in serum retinol concentrations. Unfortunately, for the RDR, they misinterpreted the meaning of the 5-h postdosing value (R5), which approximates homeostatically regulated serum retinol in vitamin A-replete individuals only and not in individuals with inadequate liver reserves. For the latter group, R5 reflects the magnitude of accumulated retinol binding protein (RBP) in the liver and not the homeostatic value (2). We question the underlying assumption for model 1, for which values are assigned to serum retinol at baseline (R0) and R5, as measured in the RDR test. Does the resulting mathematical plot reflect the reality of the test when applied to a population? The scatter and frequency of negative values shown are not biologically plausible if the RDR test is properly applied. Negative values have occurred among replete individuals or when the smaller dose (450 µg retinyl ester) and capillary blood were used. Venipuncture and the 1-mg test dose reduce the variability and frequency of negative RDR values (3). Also, RDR values >0.4 are rarely seen, yet the graphs show otherwise. Thus, reality and biological plausibility are not reflected in these graphs.
The association shown between the MRDR and serum retinol can only be obtained from large studies that include both severely vitamin A-depleted and -adequate individuals. The MRDR test can be done on a subsample to determine group vitamin A status. In many groups, serum retinol did not change, but the MRDR significantly improved (4-6), thus better reflecting intervention response. Serum retinol can also improve in placebo groups because of seasonality or deworming, but the MRDR will still respond to the intervention (7). An intervention provided to vitamin A-depleted individuals, in whom homeostatic control is no longer maintained, can result in increased serum retinol concentrations, but the MRDR may not become normal, ie, liver stores may not become fully replete. Thus, the MRDR offers more information than does the serum retinol concentration alone. The graph presented in the article by Verhoef and West suggests that this association will occur in any group. However, when vitamin A stores are adequate, false-positive results are rare because RBP does not accumulate in the liver; thus, little response to treatment with 3,4-didehydroretinol occurs.
The MRDR is normally distributed in a vitamin A-replete population. Does this discount its use in a population that is not vitamin A replete when the distribution is no longer normal? We believe that an inspection of the scatter in the distribution helps to delineate the actual vitamin A status. When looking at MRDR test distributions from 2 groups of American children (8, 9), one distribution is normal, which reflects normal status, and one distribution has a tail toward positive MRDR values, which represents a marginally depleted group.
3,4-Didehydroretinol was first used as a tracer to determine total body stores. However, it became clear that the association with liver reserves can occur earlier than equilibrium and that 3,4-didehydroretinol is analogous to the rapid release of holo-RBP in vitamin A-depleted animals. The ratio of 3,4-didehydroretinol to retinol was chosen as a standardized measure or indicator for global application. We ascertained that this ratio is more accurate than is the concentration of any one analyte between laboratories. Moreover, in severely vitamin A-deficient individuals, 3,4-didehydroretinol may be muted because it is taken from the chylomicra to meet immediate tissue needs.
Verhoef and West state that more blood indicators of liver vitamin A stores are needed. Stable and radioactive isotope methods have existed for decades and are gaining in popularity, even though they are technically challenging. In contrast, the MRDR only requires HPLC, gives more information than does serum retinol, and requires only a single blood sample. Validation data in a piglet model for infants showed that a mean ratio of 3,4-didehydroretinol to retinol of 0.09 reflected a liver reserve of 18 µg/g (vitamin A deficiency) and of 0.04 reflected a liver reserve of 27 µg/g (uncertain vitamin A status) (10). When liver reserves were below a critical 17 µg/g, the MRDR was invariably positive. Thus, the consensus from the study by Wieringa et al (11) was that the Indonesian infants studied were severely vitamin A deficient.
We caution readers to look carefully at the published data. The dose response tests have frequently been inappropriately applied or the results have been misinterpreted. For example, to truly reflect liver reserves, a postintervention washout period is essential. In comparisons of intervention groups, serum retinol may increase but the RDR or MRDR may not change. A certain percentage of serum retinol reflects recent dietary intake or physiologic status (eg, parasitic loads) regardless of vitamin A status. We recommend a postintervention period of 
10 d to allow mixing of the intervention vitamin A with the total body pool of vitamin A (7) and reaccumulation of RBP if liver reserves are still depleted.
If the 3,4-didehydroretinol response is a random effect, as has been suggested, then 2 groups with similar serum retinol concentrations would have the same mean MRDR value. However, groups of children with serum retinol concentrations of 0.7 µmol/L can have a mean MRDR value of 0.06, 0.03, or even 0.02 before or after an intervention (3-5, 7). This is not a random effect but rather a reflection of vitamin A status. When serum retinol changes marginally in response to an intervention, the MRDR change is even more remarkable (12). More intriguing is that a mean serum retinol concentration <0.7 µmol/L does not necessarily reflect vitamin A deficiency. A marginal depletion of vitamin A was noted in South African children (7). Although the MRDR test is somewhat dependent on serum retinol concentrations between populations, hence 2 cutoffs, this dependency does not account for all of the variability.
The RDR and MRDR tests are categorical indicators of vitamin A status and do not give quantitative estimates of liver reserves. We have never claimed that dose-response tests are "gold standards." Even the accuracy of isotope methods is limited to individuals whose liver reserves are adequate but not excessive. Liver reserves are also subject to sampling differences. Indirect methods are just that—indirect. Albeit not perfect, the dose-response tests offer more information than does serum retinol alone and are more sensitive to liver reserve changes.
Thus, we respectfully disagree with the mathematical representations put forward by Verhoef and West. The advantage of the RDR and MRDR tests is that they glean more information about vitamin A status. Moreover, when the MRDR test is used to measure an intervention effect in a population in whom serum retinol has not changed (5) or has increased in a placebo group (7), it will reflect changes in vitamin A status if the population has shifted from a marginal or depleted vitamin A status to adequate vitamin A reserves (13).
ACKNOWLEDGMENTS
The 3,4-didehydroretinyl acetate used as the test dose in the MRDR test is not commercially available. SAT supplies the compound as a service to researchers at or below costs, depending on the circumstance or collaboration. All monies received go directly to the University of Wisconsin-Madison and not to SAT.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
