点击显示 收起
1 From the Department of Nutrition, The Pennsylvania State University, University Park.
See corresponding article on page1170.
Stephensen and Gildengorin (1) are to be commended for their creative use of publicly available data from the third National Health and Nutrition Examination Survey (NHANES III) to assess the relation between C-reactive protein (CRP), an indicator of an active acute phase response, and serum retinol concentration. Their analysis follows observations made in several epidemiologic, clinical, and experimental studies that infection or inflammation and the resulting acute phase response are associated with low concentrations of serum retinol. The acute phase response to infection involves changes in hepatic protein synthesis, including increased synthesis of CRP and several other positive acute phase proteins, and reduced synthesis of several nutrient-transport proteins, including retinol binding protein (RBP) and transthyretin as well as transferrin, ceruloplasmin, and albumin (2). Previous examinations that probed for a relation between low serum retinol and acute phase proteins were conducted in population subgroups such as children living in underdeveloped regions of the world and during times when children had specific infections such as malaria, measles, or HIV infection. In this regard, research by Thurnham and Singkamani (3) led the way to an understanding, now confirmed and expanded upon, that low serum retinol might be a consequence of infection-induced changes in nutrient transport rather than of a true deficiency of vitamin A. Mechanistically, infection-induced hyporetinolemia is now understood to result from impaired synthesis or release of the retinol transport proteins RBP and transthyretin, or both, from the liver during infection (4, 5).
Although the unique resources provided by NHANES III are generally well known in the nutrition community, it nonetheless warrants mention that NHANES data represent samples of the noninstitutionalized US population living in households; that NHANES involves collection of extensive sociologic, anthropometric, medical, and nutritional data; and that NHANES data are publicly available for secondary reanalysis. Therefore, NHANES III provides a rich, extensive resource for creative probing. The data set used by Stephensen and Gildengorin included serum retinol and CRP measurements from >22000 individuals aged 4 y who participated in NHANES III (conducted from 1988 to 1994). The authors selected concentrations of CRP >10 mg/L as indicating a positive acute phase response and concentrations of serum retinol < 1.05 µmol/L as indicating a lack of vitamin A reserves. On the basis of these criteria, serum retinol was significantly lower in males with elevated CRP concentrations from the first through the eighth decades and in females from the fourth through the ninth decades. Moreover, for all age and sex groups, the fraction of subjects with low serum retinol was significantly higher when CRP concentrations were >10 mg/L, ranging from nearly 67% in those aged <10 y to 4.2% and 6.6% in all men and women, respectively.
The authors also used several conditions reported in NHANES III as potential indicators of infection and chronic disease (eg, arthritis and respiratory disease) to calculate age-adjusted odds ratios for elevated CRP concentrations. Several of these results were statistically significant. It is interesting that even though a "physician's diagnosis of possible active infection" yielded a higher odds ratio than did "self-report of an acute infection in past few days," the odds ratios for both of these conditions were significantly >1.0 for males and for females. The robustness of these data suggests that self-reporting may be a reliable source of information concerning acute infection, at least as indicated by elevated CRP concentrations. The odds ratio for elevated CRP concentrations was >1.0 for conditions of chronic inflammatory disease (eg, arthritis and gout), for smoking and emphysema in men and women, and for increased body mass index in women. The authors' extensive analysis provides strong support for the potential misclassification of vitamin A status in persons with infection or inflammation, even in the US population. Although the association of low serum retinol with acute infection is becoming accepted, the influence of chronic diseases on serum retinol has not received due attention. It is clear that future studies of vitamin A nutritional status will need to consider both acute infection and chronic disease status. It is also possible that additional new insights may be gained from retrospective data analyses, as were shown in the study by Stephensen and Gildengorin.
Three points deserve further emphasis. First, the use of a CRP concentration >10 mg/L as an indicator of the acute phase response and a serum retinol concentration <1.05 µmol/L as an indicator of low vitamin A reserves are both conventions. As mentioned by Stephensen and Gildengorin, different CRP values have been used by other investigators to indicate an acute phase response. Moreover, because serum retinol concentrations are age dependent, the criterion used to indicate low serum retinol should be adjusted for age. Thus, further exploration of the association between elevated CRP and low serum retinol with the use of cutoff values different from those used by Stephensen and Gildengorin would be of interest. Second, it is important to understand which of the acute phase proteins provides the most reliable information about acute and chronic infections (2, 6). It is known that the half-lives of different acute phase proteins differ, and even the direction of change may vary with the type of infection, such as in malaria (7). Therefore, measurement of several acute phase reactants may help to further define or refine the association of the acute phase response with serum retinol. Third, it should be emphasized that the implications of this work extend beyond the assessment of vitamin A status. As mentioned (1), the hepatic synthesis of other nutrient-transport proteins besides RBP and transthyretin is altered in the acute phase response. Therefore, might the nutritional status of iron, copper, and zinc also be misclassified during acute infection or chronic inflammation because the synthesis of transferrin, ceruloplasmin, and albumin, respectively, also is depressed? The importance of addressing this issue is obvious with respect to designing optimal treatments for improving low nutrient status.
In the mid-1980s, I participated in an ad hoc panel convened by the Life Science Research Office of the Federation of Societies for Experimental Biology to assess and interpret the vitamin A data collected in NHANES I (1971–1974), NHANES II (1976–1980), and the Hispanic HANES (1982–1984) (8, 9). Part of our charge was to provide interpretive criteria for assessing the prevalence of vitamin A deficiency and toxicity on the basis of the serum retinol data from these surveys. The panel reviewed the available literature and attempted to define, to the best of our knowledge at the time, ranges of serum vitamin A concentrations for which an increase in the consumption of vitamin A would be likely to yield improvement in vitamin A status. I do not recall, nor do the reports document (8, 9), that the concept of infection or inflammation as a modulator of serum retinol concentrations entered significantly into our discussion. Thus, the results of Stephensen and Gildengorin are a reminder that new understandings can be gleaned from revisiting older data. Their article also is another excellent demonstration of the strength of national survey data for exploring important questions in the nutritional sciences.
REFERENCES