BoneKEy-Osteovision | Commentary
Vitamin a and hip fracture: Should we be concerned?
DOI:10.1138/2002014
Until quite recently, the focus of nutrition research and policy has been on preventing deficiency. Both in the US and elsewhere, tremendous progress against nutrient deficiencies has been made through the enrichment and fortification of foods. Far less attention has been given to the possible negative effects of high intakes of vitamins or minerals. In a recent issue of the Journal of the American Medical Association, Feskanich and colleagues showed that relatively high usual intake of vitamin A, as retinol, was associated with increased risk of hip fracture ().
Among more than 70,000 postmenopausal women participating in the Nurses Health Study, those with vitamin A intakes greater than 3000 µg of retinol equivalents per day were 1.5 times more likely to suffer a hip fracture during 18 years of follow-up than were those consuming less than 1250 µg (p for trend <.01). More specifically, those consuming more than 2000 vs. < 500 µg of retinol were 1.9 times more likely to have a hip fracture (p < .001). A similar, though less significant, association remained for the subset of non-supplement users who consumed greater than 1000 vs < 400 µg of retinol from food (OR=1.7). No association was seen with intake of beta-carotene, the precurser of vitamin A found largely in fruits and vegetables.
The recent report on dietary reference intakes, from the Institute of Medicine (IOM), set the new RDA for vitamin A at 900 µg for men and 700 µg for women, with an upper limit of 3000 µg per day of preformed vitamin A (). The committee based this upper limit on evidence of teratogenicity with high vitamin A intakes. Although they reviewed existing evidence on vitamin A and bone status, they concluded that existing studies showed conflicting results and were, therefore, not useful in setting an upper limit.
The Feskanich et al. study was not available during the deliberations of the IOM committee. It supports the findings of another recent study () that found that vitamin A intakes above 1500 mg were significantly associated with both lower bone mineral density and greater risk of hip fracture in Swedish women. On the other hand, three studies of US women () did not find associations between vitamin A intake and bone status and/or fracture. Furthermore, a recent cross sectional analysis of the Third National Health and Nutrition Examination Survey (NHANES III) found no association between serum retinyl esters and bone mineral density at the femoral neck, trochanter, intertrochanter, or total hip in 5790 adults ().
The Feskanich study has several advantages. First, it is the largest study available on this topic to date, with 603 incident hip fractures from low-moderate trauma among more than 70,000 women. Multiple dietary assessments improved estimation of long term exposure prior to fracture. Although no dietary assessment is without error, the use of food frequency questionnaire is particularly appropriate for vitamin A, given the large day-to-day variability in intake associated with few concentrated food sources.
However, observational studies cannot prove causality. Incomplete control of confounding is of particular concern in studies of nutrient intake, due to collinearity of nutrients in the diet and to their interactive effects. Feskanich et al. included intake of calcium, vitamin D, vitamin K, protein, alcohol and caffeine in their models of vitamin A and fracture. However, the shape of the associations of several nutrients on bone status and hip fracture remain inconclusive and adjustment is, therefore, likely to be imperfect. Some, including alcohol and protein, appear to have negative effects on bone at both the lower and upper end of their intake distributions. Nutrient supplements are also problematic as most subjects take multivitamin preparations from which the retinol cannot easily be separated from other nutrients in the same preparation. It is a strength of this study that relatively consistent results were seen for retinol from food + supplements and from food alone.
The strength of the association was moderate, and appeared variable for many of the sub-analyses, but was greatest (OR=2.5 (1.5-4.3)) among non-postmenopausal hormone users with retinol intakes ≥2000 relative to < 500 µg. In contrast, associations were not significant among hormone users, suggesting a protective effect of hormones in the face of high retinol intake. This interaction is an important finding and raises questions about the mechanism of protection by hormone use.
Numerous animal and in vitro studies support the biological plausibility of a causal link between retinol and bone, as recently reviewed by Binkley and Krueger (). In several animal species, high retinol has been shown to stimulate bone resorption and to inhibit bone formation. It has also been shown to antagonize the action of vitamin D in rats (). A limitation of the study by Feskanich et al. was the absence of biological markers, such as serum retinol or bone mineral density measures to support the association between intake and fracture. The lack of consistency in the existing literature, including the finding of no association between elevated serum retinyl esters and low bone mineral density in the NHANES III () suggests that further research is needed.
In the meantime, the suggestion by this well conducted longitudinal study that intakes as low as 2000 µg of retinol—an amount consumed by a large segment of the population—may be associated with increased risk of fracture is important. These intake levels are well below the recently set upper limit of 3000 µg per day of preformed vitamin A and raise concern regarding levels of food fortification and the use of retinol supplements by US adults. Indeed, with more and more fortified foods on the market and larger segments of the population using nutrient supplements, considerably more attention to the risks of high nutrient intakes should become a priority.
This work is licensed under a
Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.