| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Letters to the Editor |
New Britain General Hospital New Britain, CT 06050 Bill L. Lasley University of California Davis, CA 95616
We are pleased that Drs. Petit, Prior, and collegues have taken interest in our paper titled "Bone health is not affected by luteal phase abnormalities and decreased ovarian progesterone production in female runners" (1). We strongly agree that the issue of bone mass and ovulatory status is an interesting subject of inquiry.
Our paper supports data recently reported by other investigators of bone mass and ovulatory status, including Winters et al. (2) and Waller et al. (3). Our conclusion, similar to theirs (2, 3), is that luteal phase abnormalities associated with decreased progesterone production was not associated with bone loss, whether in exercising women (2) or in sedentary women (3). These three sets of data are, however, in sharp contrast to that presented by Prior et al. (4).
Since we have previously (1) discussed the serious methodological limitations of the 1990 data published by Prior et al. (4), we find it somewhat redundant to reiterate these limitations; however, we shall focus our discussion of this issue to the following points:
1) The current state-of-the-art methodology for assessing menstrual and ovulatory status is the measurement of daily hormone levels of the ovarian steroids in serum or urine (5, 6). We utilized the mean values of daily urine samples analyzed for E1C and PdG during three consecutive menstrual cycles in 33 women. In contrast, Prior et al. (4) utilized basal body temperature (BBT) monitoring to document ovulatory status and corpus luteum function. Daily urinary hormone profiles provide much more than the detection of the end of the follicular phase and the duration of the luteal phase. The follicular phase estrone conjugate dynamics permits the characterization of the time course of follicular development, maturation, and collapse. The pregnanediol-3-glucuronide profile confirms ovulation and provides a quantitative estimate of the relative strength of the luteal phase. The midcycle gonadotropin surge provides an exact demarcation of the periovulatory period as a precise and objective indicator of cycle phase lengths. Together with the menstrual calenders, these measurements provide a comprehensive and complete evaluation of the menstrual cycle that could not be obtained by other means. Clearly, a direct quantification of ovarian steroid excretion patterns, as done in this study (1) and the two previous studies (2, 3), provides a more direct and objective evaluation of both follicular and luteal phase ovarian hormone production than do BBT shift patterns (7, 8, 9, 10, 11). While BBT shifts may be somewhat accurate in identifying abnormal luteal phases, it is not clear that BBT clearly identifies the relative normality of the follicular and luteal phases of the menstrual cycle (7, 8, 9, 10, 11). In fact, BBT is limited to detecting only luteal phase hormone changes and provides little, if any, information regarding absolute hormone production in the follicular phase or the luteal phase (7, 8, 9, 10, 11). Thus, in studies in which ovarian hormones are evaluated only by BBT, it would be very difficult to rule out concomitant abnormalities in the follicular phase. In addition, the use of BBT for documenting ovulation and luteal phase function has been criticized elsewhere (7, 8, 9, 10, 11).
Our study (1) was the third study to use daily hormone measurements, and it confirms the findings of the previous two investigations (2, 3). That is, these studies do not support luteal phase defects as a primary cause of bone loss. Our study supports Winters et al. (2) in identifying a time period of reduced estrogen excretion associated with long follicular phases and bone loss. Despite the relatively small numbers and short time interval of the studies, the results from these three investigations reach the same conclusion. More importantly, this conclusion is not inconsistent with a relationship of bone loss to luteal phase defects but points to follicular phase abnormalities as the more important phase of the menstrual cycle, which cannot be accomplished with the use of BBT monitoring.
Of course, it is possible that progesterone may play a role in bone metabolism; however, most data reported to date do not support such a role. In fact, we have been unable to find any studies that have reproduced or supported the findings of Prior et al. (4), whereas in the last two years, our study (1) and two others (2, 3) have concluded that luteal phase progesterone does not impact bone health. It should also be pointed out that some of the conclusions of Prior et al. (4) have been drawn from correlational findings. We would expect a correlation of abnormal luteal phases to bone loss because there is a functional relationship between an abnormal follicular phase and an abnormal luteal phase. These events cannot be completely separated from one another except through experimental manipulations.
2) We agree that there are differences between quantitative computed tomography (QCT) and dual-energy x-ray absorptiometry (DXA) when assessing bone mineral density, and we acknowledge the limitations of DXA when assessing trabecular bone (DXA measurements also include cortical bone). QCT, on the other hand, has the ability to measure purely trabecular bone, which is the metabolically active component. Thus, changes in bone metabolism resulting from therapeutic intervention or a changing hormonal milieu will be observed first in trabecular bone, rather than in the cortical envelope. Thus, QCT may be more sensitive in finding changes in bone mass during longitudinal studies. However, as stated in the response by Waller et al. (12), trabecular bone comprises about 35–50% of the total lumbar vertebral bone, so that a change in BMD restricted to the trabecular envelope would be detectable by DXA as an apparent change that is about one third to one half as great.
A major point that needs to be addressed is that the purpose of our investigation was not to assess BMD changes over time, but rather to accurately document menstrual status and measure BMD cross-sectionally. We used very strict criteria to categorize the subjects into both exercise and menstrual status groups. Also critically important, the reproductive hormone data is the mean of three menstrual cycles of data; thus, although our study was cross-sectional with respect to BMD measurement, all hormonal data are representative of three months of longitudinal data. It was our contention that our subject categorizations were representative of both exercise habit and menstrual status of the subjects, not only at the time of the study, but for the year before study participation. We agree with the comment by Petit et al. that cross-sectional assessment of BMD does not have adequate sensitivity to indicate the contribution of the current hormonal milieu to the existing BMD. Because we originally recognized the limitations of DXA assessment in our study, we also sought to assess bone turnover at the time of study participation. We included two markers of bone formation and three markers of bone resorption. Our results were unremarkable. Again of critical importance, the bone marker data is a mean of three menstrual cycles of data; thus, although our study was cross-sectional, all bone marker data are representative of 3 months of longitudinal data. There appeared to be no influence of lower progesterone production during the luteal phase on bone markers. Our bone marker findings were curiously excluded as part of the criticisms discussed by Petit and colleagues.
3) In the first paragraph of number 3 in the letter above, Petit and collegues state that the exercising women with short luteal phase lengths (ExLPD) in our study tended to have lower whole body BMD than exercising women with consistent ovulation (ExOvul). First, the subjects in the ExLPD group were not categorized only by luteal phase length, but also by low PdG excretion during the luteal phase. Second, we have taken notice of the similarity in BMD between the SedOvul and ExLPD groups, and the tendency for the ExOvul women to have higher BMD’s. The ExLPD group, however, did have significantly lower E1C excretion during days 2–5 of the follicular phase, and we found that E1C during days 2–5 was related to proximal femur and total body BMD. That is, a lower E1C excretion on days 2–5 was related to a lower BMD. Interestingly, we did not find any correlations between our luteal phase parameters and BMD. Petit and collegues comment that our statements regarding the significant relationship between E1C excretion on days 2–5 and BMD was an oversight. Additionally, they feel that it is inappropriate to state a nonsignificant result as a primary conclusion. We strongly disagree with Dr. Petit and collegues with respect to this issue. As researchers investigating physiological mechanisms and responses, we do not believe that it is appropriate to disregard P values at the 0.056 or 0.061 level. All statistical procedures have limitations and are based on assumptions. Thus, should we as researchers work only within the statistical boundaries of P < 0.05 and completely abort physiological significance of outcome parameters and their relationships to each other? In contrast, we felt it would be inappropriate to ignore physiological findings of statistical importance, especially when considering our sample size.
4) Generally, power function analyses are performed at the time of
study design. It is not appropriate nor generally acceptable to
calculate sample size effects after the study has been completed. And,
if it is necessary to have over 200 subjects to find such a difference,
low BMD cannot be that prevalent in that population, and it would be
questionable if such a difference, although statistical, has any
physiologic importance. We included enough women in our study such that
a meaningful or physiological difference in BMD (i.e., that
would increase risk of fractures) would not be missed. Prior et
al. are profoundly incorrect when stating that "...generally
healthy, premenopausal women commonly report differences between groups
in the order of 2–5%, not 13%." In fact, we report a summary of
several studies (13, 14, 15, 16, 17) comparing the BMD of eumenorrheic and
amenorrheic athletes, assessed by DEXA or DPA, and the average of the
studies listed in Table 1
is 13.1% for the spine.
|
Footnotes
Address correspondence to: Mary Jane De Souza, Ph.D., Division of Reproductive Endocrinology, Bone Health Osteoporosis, New Britain General Hospital, 100 Grand Street, New Britain, Connecticut 06050.
Received June 30, 1998.
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 |
