点击显示 收起
Creighton University, Osteoporosis Research Center, 601 North 30th Street, Suite 4841, Omaha, NE 68131, E-mail: rheaney{at}creighton.edu
Dear Sir:
I thank Kerstetter et al for their comments on my article (1), in which I showed no apparent effect of wide variations in intakes of protein and phosphorus on calcium absorption efficiency. I find myself in substantial agreement with their conclusions about the importance of an adequate protein intake for bone health. It may even be that the optimal intake for the elderly is somewhat higher than the current recommended dietary allowance (2). As Kerstetter et al note, high protein intakes tend to be associated with higher bone mineral densities, and protein supplements were shown in randomized controlled trials to improve recovery from hip fracture. The presumed negative effect of protein, related to its calciuric action, presents a problem only at low calcium intakes, in which case the adaptive increase in 1,25-dihydroxyvitamin D synthesis is not adequate to produce an absorptive increment equal to the excretory increase.
The safety and probable value of a high protein intake is supported by recent evidence indicating that our paleolithic ancestors, and hunter-gatherers studied extensively in the past century, had protein intakes typically twice as high as those provided by contemporary diets (3–5) and that their skeletal remains are generally robust.
Generally, I agree that within-subject designs are more sensitive than are between-subject designs and hence can detect small differences. However, if an effect is to have nutritional importance, it ultimately must be detectable at a population level. The point of my article was to communicate that I found no such effect in a population that consumed a habitual diet typical of what US women generally ingest.
Kerstetter et al stress, repeatedly, the rigor of their studies, for which they are to be commended. However, I reject any implication that my studies, conducted under full metabolic balance conditions in persons consuming diets designed to match intakes derived from 7-d diet records, were any less rigorous.
More substantively, Kerstetter et al suggest that my controlling for the effect of dietary calcium on intestinal calcium absorption may have "obscured the effect of protein." They call attention to the known association between calcium and protein in dairy sources. However, the median calcium intake in my subjects was equivalent to <1.5 servings of dairy products daily. In fact, many of my subjects had no dairy intake at all and most of their protein intake was derived from nondairy sources. Hence, it is unlikely that the suggested obscuring effect occurred. However, their suggestion prompted me to reanalyze my data by stepwise linear regression, using the actually measured values for the calcium absorption fraction. This reanalysis confirmed the foregoing suspicion. When dietary calcium was included in the model, dietary protein did not enter the model. The only variables in the model were dietary calcium and estrogen status. Only when dietary calcium was excluded was protein intake significantly associated with absorption, and the regression coefficient was negative. This finding was the opposite of the effect suggested by Kerstetter et al.
However, because the relation between calcium absorption and calcium intake is nonlinear, a linear model may not have been the most sensitive approach. Hence, I log-transformed the calcium intake values, which linearized the relation, and reran the model. Once again, calcium intake dominated—even more strongly than when raw intakes were used. Estrogen status was significantly related to calcium absorption, and total energy intake entered the model as well. Protein intake entered the model only if estrogen status was eliminated, but even then the regression coefficient for the protein term was small (0.00076). This indicates that an elevation in protein intake of 10 g is associated with a rise in the absorption fraction of only 0.0076; the SE of this estimate is 0.0033. This finding appears to be far too small to evoke the changes in calcitropic hormones described earlier by Kerstetter et al for a difference in intake of this magnitude (6) and attributed by them to reduced calcium absorption. When the model was run separately for estrogen-replete and estrogen-deprived women, there was no hint of a protein effect. The appearance of a relation when both groups were combined (ie, estrogen status excluded) was because estrogen-replete and estrogen-deprived women had slightly different protein intakes and absorption fractions.
Thus, I am unable to resolve the discrepancy between our respective studies. At the same time, it is unclear why the increase in serum 1,25-dihydroxyvitamin D concentrations with low protein intakes that Kerstetter et al observed did not evoke a corresponding increase in calcium absorption efficiency. One possible explanation is that mucosal mass decreased as food intake decreased. (As noted above, energy intake was positively associated with absorption in our study.) This seems unlikely given the controlled intakes in Kerstetter et al's study; however, it could be a factor in free-living subjects in whom protein and energy intakes are positively correlated. Finally, if the relatively abrupt change in calcitropic hormones at a protein intake of 0.8 g/kg (6) cannot be explained by a correspondingly large change in absorption efficiency at that threshold, it remains a puzzling observation.
REFERENCES