| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Letters to the Editor |
Departments of Medicine and Physiology University of California, San Francisco San Francisco, California 94115 Dean K. Jenkins Metra Biosystems, Inc. Mountain View, California 94043
We have noted some disturbing inconsistencies and inaccuracies and found the breadth of analysis lacking in the recent report by Rosen et al. (1). The authors compared the responses of various biochemical markers of bone metabolism in a study of hormone replacement therapy (HRT) in postmenopausal women. The study reported results for N-telopeptides of type I collagen (NTx), deoxypyridinoline (Dpd), osteocalcin (OC), and bone alkaline phosphatase (BAP) in 236 women randomized to receive either calcium alone (CTL) or calcium plus HRT. The study was conducted over the course of one year, and bone mineral density of the lumbar spine and hip were measured by dual-energy x-ray absorptiometry at baseline, 3, 6, and 12 months. Markers were measured at baseline, 1, 3, 6, and 12 months. Almost all of the analyses had been previously reported by Chesnut et al. (2), with the exception of the addition of the Dpd measurements.
The authors reported in the Results section of their paper that a significant decrease from baseline (P < 0.0001) was seen in the HRT group at 1 month with NTx, Dpd, and OC. However, in the Discussion, they stated that the earliest and most significant change from baseline was noted for NTx and OC. These two statements do not appear compatible in view of the statistics presented. We are concerned that this conclusion was drawn on the basis of the relative percent changes observed for the different markers, -28%, -10%, and -15%, respectively. Percent change, however, is not a statistical measure and ignores the contribution of the variability inherent in each of the different biochemical markers.
Rosen et al. (1) reported a lack of predictive value of baseline Dpd for HRT-induced increases in spine BMD on the basis of the fact that there was no difference between first and fourth quartiles of the marker. However, third quartile Dpd values were considerably higher than those observed in all other quartiles, and fourth quartile results were higher than first and second. This suggests the possibility that above-median Dpd values might be associated with significantly higher spine BMD changes than below-median Dpd values, and possibly that the trend of increasing quartiles might also be significant. If either of these were so, the predictive value, or lack thereof, would depend upon the method of analysis and would be open to interpretation.
In their receiver operating characteristic (ROC) analysis, the authors presented data for the change in marker levels at 6 months compared with one year BMD change. They described in Results that NTx provided a greater discrimination between gain and loss of BMD than the other markers. However, the authors report no statistical comparison of the areas under the ROC curves that leads to this conclusion. NTx, Dpd, and OC were reported to decrease significantly at 1 and 12 months, and presumably at 3 and 6 months (not stated) as the magnitudes of percent change are greater at 3 and 6 than at 1 month, with comparable standard error. However, the authors did not offer any comparisons using ROC analysis at those time points other than 6 months.
The analysis of marker values in CTL was misleading as the authors chose to average all on-therapy values for their analysis rather than determine predictive value of baseline measurements, as was done for the HRT group. As that would be a desirable finding, we conclude that there was no baseline predictive value of the markers, at least for NTx, the commercial product of the study’s corporate sponsor. Thus, the reported data offers no insight on the ability of any of the 4 markers to predict BMD changes over 1 yr in postmenopausal women receiving calcium.
As with spine BMD, hip BMD was reported to be increased significantly at 12 months in the HRT group [and decreased in the CTL group in the other published report of this study (2)]. However, no associations between any of the biochemical variables and hip BMD were reported. We presume there were none, or at least none favorable to the sponsor’s product.
In their Discussion, Rosen et al. reported that a 20–25% CV for NTx and Dpd in an individual is not unexpected and is similar to reports in other studies. The studies cited to support this statement make no report of intraindividual variability of Dpd measured by the assay they used. In fact, there are no published reports of intraindividual variability as high as the authors have stated for the Dpd assay; intraindividual CV for the Dpd assay used is reported to average between 10–16% (3, 4). This lower figure affects the conclusions the authors have drawn regarding the relative balance between treatment response and variability for the two markers.
In summary, we are concerned about the reporting of data and review of the literature by Rosen et al. and by the imprecise conclusions they have made in their paper. Their report fails to fully elucidate associations that exist between the various markers and BMD in HRT- or calcium-treated postmenopausal women. Such information is needed to enable physicians to appropriately select tools to aid in patient management.
Footnotes
Address correspondence to: Claude D. Arnaud, Departments of Medicine and Physiology, University of California, San Francisco, 1710 Scott Street, Third Floor, San Francisco, California 94115.
Received October 23, 1997.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
