0021-972x/96/$03.00/0
Journal
of Clinical
Endocrinology
and Metabobsm
Copyright
0 1996 by The Endocrine
Society
Vol. 81, No 2
Prmted in U.S.A.
Bone Mass and Subtle Abnormalities
Function
in Healthy
Women*
KIRSTEN
WALLER,
JENNIFER
BABETTE
BRUMBACK,
GAYLE
AND ROBERT MARCUS
REIM, LAURA
C. WINDHAM,
in Ovulatory
FENSTER,
SHANNA
H. SWAN,
BILL LASLEY,
BRUCE ETTINGER,
Reproductive
Epidemiology
Section, California
Department
of Health Services (K. W., L.F., S.H.S.,
B.B., G.C. W.), Berkeley, California
00000; Musculoskeletal
Research Laboratory,
Geriatric Research,
Education,
and Clinical Center, Veterans Affairs Medical Center (J.R., R.M.), Palo Alto, California
94304; the Department
of Medicine, Stanford
University (J.R., R.M.), Stanford,
California
94304; the
Institute for Toxicology and Environmental
Health, University of California
(B.L.), Davis, California
95616; and the Division of Research, Kaiser Permanente
Medical Care Program
(B.E.), Oakland,
California
94611
ABSTRACT
Women with occasional anovulatory or short luteal phase menstrual cycles have been reported to lose bone mineral density (BMD)
at a greatly accelerated rate compared to women without such abnormalities. To investigate this association, we performed a longitudinal study of BMD in a group of healthy premenopausal women
enrolled in a comprehensive study of ovulatory function. Subjects had
collected daily urine samples that were analyzed for estrone and
progesterone metabolites by enzyme-linked immunoassay. The 53
participants collected urine for an average of 4.1 cycles. Computer
algorithms identified 7 (13.2%) women with luteal phase abnormalities (>l anovulatory cycle or cycle with luteal phase length 510 days)
and 17 (32.1%) women with other menstrual abnormalities. Area1
BMD (grams per cm’) was measured at the lumbar spine, hip, and
whole body using dual energy x-ray absorptiometry; BMD was measured 2-3 times over an average observation period of 17.5 months.
At baseline, women with luteal abnormalities had mean BMD similar
to those of the 29 women with no abnormal cycles: lumbar spine, 1.06
us. 1.09 g/cm2; total hip, 0.95 vs. 0.94 g/cm2; whole body, 1.15 vs. 1.11
g/cm2 (P > 0.10; adjusted for age and weight at baseline, parity,
physical activity level, and calcium intake). When compared at follow-up to women with no abnormal cycles, women with luteal abnormalities tended to gain BMD at the spine and hip (P > 0.10). On whole
body measurement, women with luteal abnormalities tended to lose
BMD compared to women with no abnormal cycles (- l.l%/yr vs.
O%/yr; P = 0.08); however, the magnitude of loss was not unusual for
women in this age range and was within the coefficient of variation
for replicate measurements. Neither mean luteal phase length, percent time in luteal phase, nor average daily excretion of progesterone
metabolites was associated with baseline BMD or percent annual
change in BMD at any measurement site. Thus, we did not confirm
a relationship between luteal abnormalities and accelerated bone loss
in this population of healthy premenopausal women. (J Clin Endocrinol Metab
81:663-668,
1996)
T
HE CRITICAL
relationship
between reproductive
hormone status and skeletal integrity
is widely
recognized. Most work in this area has focused on the skeletal
consequences
of menopause,
but it is now clear that even
temporary
interruption
of menses during
adolescence
impairs the last phase of bone acquisition
and during young
adult life leads to bone loss. Skeletal deficits have been reported in women with amenorrhea
due to rigorous exercise
(l-3) or PRL-secreting
pituitary
tumors (4, 5) and in those
treated with GnRH analogs (6, 7). In athletes, amenorrhea
is
associated with reductions
in bone mineral density (BMD)
despite high skeletal loading. The observed skeletal deficits
in such women are of sufficient magnitude
that risk of injury
and fracture are substantial
(8).
However,
the role of more subtle menstrual
abnormalities
in the maintenance
of bone mass is less clear. Prior et al. (9)
reported that among a group of healthy young women with
regular menses, those who experienced
one or more anovulatory or more than one short luteal phase cycle per yr lost
considerable
spinal trabecular
density, as determined
by
quantitative
computed
tomography
(QCT; mean annual loss
of 5.0 mg/cm3 compared
to a mean gain of 1.5 mg/cm3 in
women
with all normal cycles; overall mean spinal bone
density was 154.1 mg/cm3).
To determine
whether
subtle hormonal
abnormalities
of
the menstrual
cycle were associated with BMD loss, we evaluated changes in bone mineral status in a group of premenopausal women who had previously
been enrolled in a comprehensive
study of ovulatory
function,
the Women’s
Reproductive
Health Study (WRHS).
Materials
and Methods
Study design
Received August 7, 1995. Revision received September 8, 1995.
Accepted October 3, 1995.
Address all correspondence and requests for reprints to: Robert
Marcus, M.D., Geriatric Research, Education, and Clinical Center
182-B, Veterans Affairs Medical Center, 3801 Miranda Avenue, Palo
Alto, California 94304.
* This work was supported in part by the Research Service of the
Department of Veteran Affairs.
The WRHS, conducted by the California Department of Health Services in collaboration with the Kaiser Permanente Medical Care Program, Division of Research, was designed to explore the relationship of
selected environmental exposures to reproductive function in healthy
married women, aged 18-39 yr. Participants were recruited by telephone from the membership list of the Santa Clara clinic of the Kaiser
Permanente Health Plan, a large prepaid health organization.
Women
663
WALLER
were not eligible if they had conditions
that might interfere
with their
becoming
pregnant,
i.e. if they were currently
pregnant;
had not had a
menstrual
period in 6 weeks; had been surgically
sterilized;
were using
oral contraceptives,
intrauterine
devices, or hormonal
medications;
or
had been having unprotected
intercourse
for more than 3 months without becoming
pregnant.
A history
of irregular
menses was not an exclusion criterion.
A total of 411 women
completed
a baseline telephone
interview
and collected
daily urine samples for at least 2 months.
We designed
a substudy
in which
WRHS participants
would
be
invited
to undergo
sequential
measurements
of BMD at entry into the
substudy
and again after 9 and 18 months.
The study protocol
was
approved
by the institutional
review
boards of Stanford
University,
Kaiser Permanente,
and the California
Department
of Health Services,
and subjects provided
written
consent. WRHS subjects who had completed their urine collection
by July 31, 1991 (n = 3611, were invited
to
participate.
Seventy-nine
women
expressed
interest,
of whom 12 were
excluded
because of recent pregnancy,
systemic illness, treatment
with
reproductive
hormones,
or failure
to keep appointments.
Of the 67
participants,
14 withdrew
after their first bone density
examination
because of pregnancy
or loss of interest. A total of 53 women underwent
serial bone density measurements.
Hormone
analyses
Each WRHS subject had been instructed
to collect and freeze a sample
of first morning
urine every day for six consecutive
cycles or until she
had missed two consecutive
menstrual
periods.
Subjects were asked to
flag the first sample collected after the beginning
of a menstrual
period.
Frozen urine samules were renularlv
retrieved
bv WRHS field workers
and shipped
to the University
of C’alifornia-Davis.
Each urine sample
was analyzed
for creatinine
and metabolites
of sex steroid hormones
[estrone conjugates
(ElC), and pregnanediol-3-glucuronide
(PdG)J by
enzyme-linked
immunoassay
(10). Although
individuals
can vary
widely
in the pathways
by which their hormones
are metabolized,
metabolite
excretion
patterns
are characteristic
enough to permit accurate assignment
of the day of ovulation
and estimation
of luteal phase
length
(11). All cycles with ovulatory
abnormalities
(see definitions
below) were reassayed
to confirm
the abnormality.
1
Determination
of menstrual
parameters
Urine samples in which the creatinine
level was less than 0.2 mg/mL
were considered
too dilute to yield accurate
measurements;
thus, for
these samples, the hormone
values were designated
as missing.
In adequately concentrated
samples, hormone
values below the assays’ minimum detection limits (PdG, co.15 n.g/mL;
ElC, <7.8 ng/mL)
were set
at the minimum
detection
limits. Extremely
high values (PdG, >25
pg/mL;
EIC, >250 ng/mL)
were designated
as missing. Daily hormone
values were divided
by the urinary
creatinine
measurement
to adjust for
variations
in urine volume.
Eight cycles with 30% or more of daily urine
data missing (excluding
the first and last 5 days) were considered
unusable.
Ovulatory
status was determined
for all usable cycles (modelled
after
the approach
in Ref. 12). The lowest 5-day average of creatinine-adjusted
PdG (APdG) was defined as the cycle baseline APdG level. The baseline
was used to calculate
an anovulation
threshold,
which was defined
as
the baseline plus the square root of the baseline. Cycles that failed to have
any 5-day period in which at least 3 days had APdG values exceeding
the threshold
were defined as anovulatory;
the remainder
were defined
as ovulatory.
One cycle had its ovulatory
status reassigned
after visual
inspection
of graphed
data.
In ovulatory
cycles, the day of ovulation
was estimated
using a
modification
of an algorithm
created by Baird et al. (13). The Baird
algorithm
usually
selects the day after the cycle day with the highest
ElC/PdG
ratio as the urobable
dav of ovulation.
This method has been
found to correlate
wei with methods
based on urinary
LH levels (14).
Because our data set contained
more very low PdG values than the data
set used to develoo
the Baird algorithm,
we used the ratio of ElC/(PdG
+ 1) instead. This modification
ritains the general shape of the ElC/I’dG
curve while minimizing
the influence
of small fluctuations
in PdG near
the detection
limit. Seven cycles had their assigned
day of ovulation
changed by 2 or more days after visual inspection
of graphed
data. Our
algorithms
were validated
using an independent
data set of 69 menstrual
JCE & M . 1996
Vol81
. No 2
ET AL..
cycles that included
serum and/or
urinary
LH measurements.
A day of
ovulation
that was within
tl day of the LH peak was selected by our
modified
algorithm
in 84% of the cycles and by the unmodified
Baird
algorithm
in 77% of the cycles.
Follicular
phase length was defined as the number of days from the
first cycle day up to and including
the day of ovulation.
Luteal phase
length was calculated
as the difference
between
the cycle length and
follicular
phase length. Each cycle was evaluated
for the presence of four
empirically
defined abnormalities
in addition
to anovulation:
short cycle
length (<24 days), long cycle length (>34 days), long follicular
phase
leneth (>20 davs, ovulatorv
cvcles onlv), and short luteal phase length
(<l”o days, ovulatory
cycles only). Mean cycle length, mean folliciar
and luteal phase lengths, percent
time in luteal phase (sum of luteal
phase lengths/sum
of cycle lengths), and mean daily creatinine-adjusted
ElC and PdG excretion
were calculated
for each woman.
For the purposes of this paper, a luteal abnormality
was defined as a short or absent
(i.e. anovulatory)
luteal phase.
Measurement
of BMD
Subjects underwent
their first bone density examination
an average
of 5.1 months
(range, 1.1-15.7 months)
after completion
of urine collection. Area1 BMD (grams per cm*) was measured
at the lumbar spine
(L2-L4),
right total hip, and whole body using dual energy x-ray absorptiometry
(DXA)
(QDRlOOOW,
Hologic,
Waltham
MA). In our laboratory,
the coefficients
of variation
of replicate
measurements
for
women in this age group are 0.5% for the lumbar spine, 0.9% for the total
hip, and 1.2% for the whole body (14). Valid lumbar spine measurements
could not be obtained
for 2 women
with histories
of spinal surgery.
Subjects were followed
with serial BMD measurements
for an average
of 17.5 months (range, 8.9-22.8 months).
Ten (18.9%) women underwent
2 BMD examinations;
the remaining
43 women
were scanned 3 times.
Simple linear regression
was used to calculate
the percent
annual
change in BMD at each measurement
site for each woman.
We defined
6 bone parameters
for each woman:
baseline BMD at the lumbar spine,
total hip, and whole body; and percent annual change in BMD at the
lumbar
spine, total hip, and whole body.
Determination
of other covariates
Body weight
and height were measured
at each visit to the bone
densitometry
laboratory
using a balance beam and wall-mounted
stadiometer,
respectively.
Using software
supplied
by Hologic, the percent
body fat was calculated
directly
from the baseline whole body scan.
Nutrient
intake was determined
from the Block food frequency
questionnaire
(15,161, which was given to the participants
at their initial visit
to the bone density laboratory.
All but two subjects completed
the food
frequency
questionnaire.
Information
on demographics,
previous
reproductive
history, smoking, and physical
activity
was obtained
from the baseline
interview
completed
by all WRHS subjects upon enrollment.
Women
who reported exercising
at least twice weekly were asked to specify each type
of exercise and approximately
how many hours per week they spent
doing each. Physical
exercise level was quantified
by assigning
all exercises a score based on the typical rate of energy expenditure
(17) and
the amount of time spent doing the activity.
Scores were summed across
all categories
of exercises to attain a final value, known as the MET score.
Women who exercised less than twice a week were assigned a MET score
of zero.
Statistical
analysis
All analyses were performed
using SAS statistical
software,
version
6.08 (SAS Institute,
Cary, NC). Mean values of BMD and change in BMD
for women with no abnormal
cycles were compared
to those in women
with abnormal
cycles using t tests. Adjusted
means for BMD were
calculated
using the least squares method
in PROC GLM. No adjustments were made for multiple
comparisons
(18). Covariates
used in the
calculation
of adjusted
means and in the multiple
regressions
were
selected on the basis of a review
of the literature
(19) and associations
observed
in our data with menstrual
and/or
bone parameters.
These
covariates
included
age and weight
at entry into the bone substudy,
BONE MASS AND OVULATORY
parity
intake,
(0 vs. ~1 previous
live births),
MET score
and, for change in BMD, baseline BMD.
(0 vs. Al),
a greater fiber intake (15.3 DS.12.4 g/day), but these differenceswere not significant at the 0.10 level (by t test). There
were no apparent differences between these groups in mean
age or intake of calcium, protein, or total calories (data not
shown).
calcium
Results
Characteristics of the study sample are presented in Table
1. Seven women (13.2%)had body massindex measurements
of 30 kg/m2 or more, and 23 women (43.3%) exercised less
than 2 times a week (MET score of zero). Fourteen women
(26.4%) were nulliparous. None of the subjectswere current
smokers. Most of the women were non-Hispanic white
(n = 43; 81.l%); the others identified themselves as Hispanic
(n = 5), Asian (n = 3), or other ethnicity (n = 2).
Menstrual
BMD
characteristics
1. Characteristics
of 53 women
enrolled
Characteristic
Age (yr)
Wt at entry (kg)
% Body fat
MET score=
Calcium
intake
(mg/day)
Fiber intake
(g/day)
Protein
intake
(g/day)
Total calories/day
Mean cycle length (days)
Mean fillicular-phase-length
(days)
Mean luteal
phase length (days)
% of time in luteal phase
Urinary
ElC excretion
(kg/mg
. day)
Urinary
PdG excretion
(ng/mg
. day)
Baseline
BMD (&rn’)
Lumbar
spine(L2-4)
Total hip
Whole body
Change
in BMD (%/yr)
Lumbar
spine (L2-4)
Total hip
Whole body
a See definition
in text.
in the bone
density
measurements
BMD measurements were within normative values established by previous studies of premenopausal women in our
laboratory and also by the manufacturer’s reference data
base(Hologic). The overall mean change in BMD at each site
was very small (Table 1). Within each woman, baselineBMD
at the lumbar spine, total hip, and whole body were correlated Pearson correlation coefficients, 0.54-0.72; P = 0.0001).
However, the change in BMD showed no within-woman
correlation (Pearsoncorrelation coefficients: lumbar spineVS.
total hip, r = 0.03 and P = 0.81; whole body us.lumbar spine,
r = -0.08 and P = 0.60; whole body us. total hip, r = -0.12
and P = 0.40). At the hip and whole body, women with high
baseline BMDs tended to have more negative changein BMD
(total hip, r = -0.13 and P = 0.36; whole body, r = -0.32 and
P = 0.02), but this trend was not seenat the lumbar spine (r
= 0.02 and P = 0.88).
In all, 217 cycles were usable, representing an average of
4.1 cycles/woman. Of these, 182 (83.9%) met the definition
of a normal cycle, i.e. they were 24-34 days long and were
ovulatory with a follicular phase less than 20 days and a
luteal phasemore than 10 days long. For 29 (54.7%) women,
all cycles were classified as normal. Seven (13.2%) women
had at least one cycle with a luteal phase abnormality: 1 had
a single anovulatory cycle, 3 had a single short luteal phase
cycle, and 3 women had 2 or more short luteal phase cycles.
Other abnormalities observed among thesewomen included
long cycles (1 woman) and long follicular phases (4). Seventeen (32.1%) women had abnormalities other than anovulatory or short luteal phase cycles. Of these, 11 had long
cycles, 4 had short cycles, and 12 had cycles with long follicular phases(women could have more than 1 abnormality).
Selected examples of normal and abnormal cycles are displayed in Fig. 1.
Compared to women with all normal cycles, women with
luteal phase abnormalities were heavier (mean weight, 82.5
VS.65.5 kg; P = 0.01) and had more body fat (36.4 us. 29.4%;
P = 0.06). Women with luteal phase abnormalities were also
slightly less active (mean MET score, 18.1 ZX. 23.5) and had
TABLE
FUNCTION
Menstrual
characteristics
and BMD
We divided the women into three groups: those with all
normal cycles (n = 27), those with luteal phase abnormalities
(21 anovulatory or short luteal phase cycles; n = 7), and
those with other cycle abnormalities (n = 17). BaselineBMD
measurements were similar among menstrual groups, although women with luteal abnormalities had a slightly
higher unadjusted mean whole body baseline BMD than
normal women (Table 2). Adjustment for covariates changed
the results only slightly, and trends were all in the same
direction. Table 3 describes the relationships observed with
substudy
MeaIl
33.4
67.3
30.6
17.9
699
13.0
60.6
1481
29.6
16.6
12.9
44.1
40.9
2.7
1.08
0.92
1.11
+0.06
+0.34
-0.04
SO
4.3
15.2
8.5
29.2
380
6.0
21.4
511
4.2
4.3
1.3
6.3
12.7
1.1
0.11
0.10
0.07
1.54
1.24
1.35
Range
24-39
46.8-115.0
17.4-53.6
o-147
158-2201
1.7-29.0
21.5-101.9
555-2793
23.6-47.0
11.5-38.0
9.0-15.0
19.1-54.1
16.3-74.6
1.0-5.0
0.89-1.33
0.73-1.19
0.97-1.30
-6.07-3.36
-2.03-3.76
-3.64-3.23
WALLER
666
ET
JCE & M . 1996
Volt31 . No 2
AL.
SUBJECT A
FIG. 1. Examples
of sex steroid
profiles. The vertical lines topped by an oval
mark
the calculated
probable
day of
ovulation
within
each cycle. Subject A
had no abnormal
cycles. For subject B,
cycles 1, 2, and 3 were categorized
as
having
a short luteal phase (9, 10, and
10 days, respectively).
In addition,
cycles 1 and 3 had long follicular
phases
(both 22 days; in the first cycle, her last
menstrual
period preceded
data collection).
TABLE
2. Mean
baseline
BMD
SUBJECT B
in 53 women,
by menstrual
group
All cycles normal
Unadjusted
baseline
BMD (g/cm’)
Lumbar
spine (L2-4)
Total hip
Whole body
Adjusted”
baseline
BMD (g/cm?
Lumbar
spine (L2-4)
Total hip
Whole body
No. of women
2 1 anovulatory
luteal
phase
~1 cycle
cycle
1.04 t 0.12 (0.41)
0.93 + 0.05 (0.81)
1.17 2 0.07 (0.02)
1.08 2 0.10
0.92 -t 0.10
1.10 -e 0.07
1.06
0.95
1.15
1.09
0.94
1.11
29
Values are the mean i SD. P values, in parentheses,
BMD among women
with all normal
cycles.
a Least squares
means adjusted
for age and weight
calcium
intake.
or short
were
determined
at entry,
parity
change in BMD. At the lumbar spine and total hip, women
with luteal abnormalities showed no greater loss in BMD
than women with no abnormal cycles (P > 0.10). Women
with luteal abnormalities tended to lose more whole body
BMD than women with all normal cycles, but the difference
was small (-1.07% US.+0.02%; P = 0.08), was of a magnitude
not atypical for women in this age range (20), and was within
the coefficient of variation for replicate measurements.Only
three women in our study lost 3% or more of their BMD/yr
at any site. One of these women had a single short luteal
phase cycle; the other two had no abnormal cycles.
We alsoexplored relationshipsbetween BMD and three continuous measuresof luteal activity: per woman mean luteal
phaselength, percent time spent in luteal phase,and average
daily PdG excretion (Table 4). In no instancedid baselineBMD
or changein BMD decreasesignificantly with decreasingluteal
activity. A similar analysiswas performed comparing the bone
by
t test
with
abnormality
1.10 2 0.10 (0.45)
0.91 2 O.ll(O.68)
1.09 t 0.07 (0.69)
(0.50)
(0.73)
(0.13)
7
1.13
0.93
1.12
of the hypothesis
(0 vs. 2 1 live births),
other
MET
that
the BMD
in this
(0.28)
(0.92)
(0.65)
17
group
score (0 vs. 2 1, see definition
is equal
in text),
to the
and
parameters to the per woman mean cycle length, mean follicular phase length, and average daily ElC excretion (data not
shown). A slight, but significant, decreasein baselinewhole
body BMD with increasing mean cycle length was noted;
otherwise, no significant trends were seen.
Discussion
It has been postulated that progesterone has an anabolic
effect on bone (21, 22). If that hypothesis is correct, women
with luteal phase abnormalities, such asanovulatory or short
luteal phase cycles, might be expected to have a lesserbone
massor greater bone loss than women without such abnormalities. One would also expect to seerelationships between
other measuresof luteal activity (mean luteal phase length,
percent time in luteal phase, or average daily PdG excretion)
and bone density.
BONE MASS AND OWLATORY
TABLE
3. Mean
All
cycles
change
in BMD
in 53 women,
normal
by menstrual
2 1 anovulatory
-0.20
+0.08
+0.01
luteal
were
at entry,
TABLE
4. Regression
coefficients
for three progesterone-related
and percent
change in BMD at three sites, in 53 women
cycle
21
cycle
-t 1.67
2 1.32
2 1.17
luteal
phase
with
+0.55
+0.86
-1.13
other
abnormality
2 1.64 (0.30)
2 1.47 (0.18)
-c 1.69 (0.04)
+0.66
+0.38
- 1.07
+0.24
+0.58
+0.33
(0.38)
(0.68)
(0.08)
t 1.29 (0.37)
k 0.91 (0.17)
2 1.36 (0.40)
+0.27
+0.58
-0.18
(0.60)
(0.32)
(0.65)
7
determined
baseline
by t test of the hypothesis
BMD,
menstrual
Menstrual
Mean
Baseline
BMD
Lumbar
spine
Total hip
Whole body
% Change in BMD
Lumbar
spine
Total hip
Whole body
phase
-0.05
+0.14
+0.02
29
Values are the mean 2 SD. P values, in parentheses,
BMD among women with all.normal
cycles.
a Least squares means adjusted
for age and weight
in text), and calcium
intake.
667
group
or short
Unadjusted
change in BMD (%/yr)
Lumbar
spine
Total hip
Whole body
Adjusted”
change in BMD (%/yr)
Lumbar
spine
Total hip
Whole body
No. of women
FUNCTION
length
parity
describing
regression
% of time
the BMD
(0 us. 2 1 live births),
parameters,
parameter
17
that
in luteal
-0.02
-0.02
-0.01
to.111
(0.18)
(0.17)
0.0 (0.71)
0.0 (0.74)
0.0 (0.85)
+0.12
+0.15
+0.25
(0.59)
(0.34)
(0.14)
0.0 (0.99)
0.0 (0.89)
+0.02 (0.56)
coeffkient
phase
their
MET
in this
group
is equal
to the
score (0 us. 2 1; see definition
relationship
to baseline
BMD
(P valueY
Mean
daily
PdG excretion
+0.02 (0.19)
-0.01(0.43)
0.0 (0.91)
+0.35
-0.29
+0.04
(0.25)
(0.17)
(0.87)
a Each regression
coefficient
represents
a separate
multiple
linear regression
with the baseline
BMD or change in BMD as the dependent
variable,
and the menstrual
parameter,
age, weight at entry, parity (0 US. ~1 live births),
MET score (0 us. ~1; see definition
in text), and calcium
intake
as the independent
variables.
The regression
coefficient
is equivalent
to the slope of the regression
line that describes
the relationship
between
the menstrual
parameter
and the baseline
BMD or change in BMD.
In the present study, the few women with luteal phase
abnormalities (primarily short luteal phases) did not h; e
lower baselineBMD or greater lossin BMD than women with
normal cycles. Furthermore, no decreasewas seenin BMD or
change in BMD with decreasing luteal activity, as measured
by the indexes described above. Although the rate of participation in the study was low (53of 361invited), the reasons
for lack of participation were nonspecific and were unlikely
to have biased the results.
In contrast, Prior et al. (9) reported increasing rates of
spinal trabecular bone loss for women with one short luteal
phase, more than 1 short luteal phase, and anovulatory cycles, respectively. Prior et al. (9) also found a strong inverse
relationship between percent time in luteal phaseand spinal
bone loss. Although we had only seven women with luteal
abnormalities in our study population, post-hocpower calculations indicated that we had 80% power to detect (at the
0.05 level) a percent change in lumbar spine BMD of -2.4%
among these women; thus, we should have been able to
detect bone lossof the relatively large magnitude reported by
Prior et al. Some of this difference could be related to the
technique used to estimate BMD. Although the interassay
precision of QCT, the technique used by Prior et al., is not as
good as that of DXA, QCT does detect changes in purely
trabecular bone with greater sensitivity. As vertebral bone is
approximately 50% trabecular, a 4% annual decreasein trabecular bone would register as a 2% loss by DXA, which
approximates the power of our study.
A remarkablefeature of the report of Prior et al. (9) was a high
prevalence of luteal phaseabnormalities,with 29%of all cycles
abnormal and 80.3%of subjectshaving at leastoneanovulatory
and/or short luteal phase cycle (9). In our data only 5.1% of
cyclesand 13.2%of women demonstrated luteal abnormalities.
Thesedifferences may be due to differences in study populations. Two thirds of the women studied by Prior et al. were
regular runners. Not surprisingly, their subjectswere also substantially leanerthan our subjects(meanweight, 58.2us.67.3kg;
mean body fat, 19.6ZIS.30.6%).Although Prior et al. stated that
their “runners . . . did not have menstrual-cycle changes” relative to the nonrunners, other investigators (23-27) have suggestedthat the prevalence of luteal phaseabnormalitiesissomewhat greater in women who exerciseintensively.
As the women reported here were participants in the
Women’s Reproductive Health Study, they had satisfied criteria that would exclude some women with abnormal fertility. Although such exclusion might cause us to underestimate the prevalence of luteal abnormalities in the general
population, someof these women would have failed to meet
this criterion becauseof infertile husbandsor immunological
or other systemic factors, rather than luteal dysfunction. The
disparity in prevalence of luteal abnormalities may also reflect differences in study methods. Numerous investigators
have reported that the basal body temperature method does
not accurately or reliably predict ovulation (28-33), and its use
by Prior et al. may have led to an overestimateof the prevalence
of luteal abnormalities. Other investigators have not observed
a similarly high prevalence of luteal abnormalitiesin regularly
cycling women (34). Conversely, our shorter period of ovula-
WALLER ET AL
668
tory assessment
(4.1 cycles VS.12 months) may have led us to
underestimatethe proportion of women with luteal abnormalities. Such an underestimate would weaken an associationbetween luteal abnormalitiesand bone loss,but would not affect
the overall prevalence of bone loss.In fact, only three (5.7%)of
our cohort exhibited BMD loss of 3%/yr or more at any site.
This suggeststhat our results were not biased toward the null
by underascertainmentof abnormal luteal activity.
In our study, the period of time during which ovulatory
patterns were assessedpreceded the bone density assessment period by several months. Thus, we cannot exclude the
possibility that a woman who experienced anovulatory or
short luteal phase cycles during the ovulatory assessment
had a transient episode of accelerated bone loss that had
resolved by the time she underwent skeletal measurements.
However, we would expect that many women with a history
of luteal abnormalities would continue to have similar problems during the period of bone assessment,yet we did not
find increased rates of bone loss in these women. Moreover,
if women with luteal abnormalities have greatly increased
rates of bone loss, they should eventually manifest reductions in baseline BMD. We did not observe such reductions.
In conclusion, we monitored BMD over an average of 17.5
months in an observational study of healthy premenopausal
women. Overall, changesin BMD were small and consistent
with findings reported in other longitudinal studies (20). We
did not confirm a relationship between luteal phaseabnormalities or luteal hormonal activity and acceleratedbone loss.Although it is possiblethat the impact of luteal abnormalities is
greater in highly active women, our results suggest that in
women with body weights and activity levels more representative of the general population, luteal activity has little influenceon bone density. Further work is neededto identify wh
if any, menstrual and/or hormonal factors influence bone density in normal postadolescentpremenopausalwomen.
Acknowledgments
We wish to acknowledge
the
Soora Wi of the Kaiser Permanente
and the WRHS field staff, and
Barbara
Lewis-Wipfler
of the
V.A. Medical
Center (Palo Alto,
contributions
of Dr. Cathy Schaefer and
Division
of Research, David Epstein
Costanza
Cocchilovo,
Kyla Kent, and
Musculoskeletal
Research
Laboratory,
CA).
References
III CH, et al. 1984 Bone mineral content
athletes. N Engl J Med. 311:277-281.
2. Marcus
R, Cann C, Madvig
P, et al. 1985 Menstrual
function
and bone mass
in elite women distance runners. Endocrine and metabolic features. Ann Intern
Med. 102~158-163.
1. Drinkwater
BL, Milson
of amenorrheic
3. Myburgh
KH,
K, Chesnut
and eumenorrheic
Bachrach
LK,
Lewis
B, Kent
mineral density at axial and appendicular
Sci Sports Exer. 25~1197-1202.
4. Klibanski
A, Neer
1980 Decreased
303:1511-1514.
5. Schlechte
JA, Sherman
with and without
6. Scharia
RM,
Beitins
bone density
SH, Minne
IZ, Ridgway
EC, Zervas
in hyperprolactinemic
B, Martin
R. 1983 Bone density
hyperprolactinemia.
HW,
K, Marcus
Waibel-Trebel
R. 1993
sites in amenorrheic
J Clin Endocrinol
Low
bone
athletes. Med
NT, McArthur
women.
JW.
N Engl J Med.
in amenorrheic women
Metab. 56:1120-1123.
S, et al. 1990 Bone
mass
reduction
after estrogen deprivation
by long-acting gonadotropin-releasing
hormone
agonists and its relation to pretreatment
serum concentrations
of 1,25-dihydroxyvitamin
D,. J Clin Endocrinol Metab. 70:1055-1061.
7. Surrey
ES, Judd HL. 1992 Reduction of vasomotor
symptoms and bone min-
JCE & BI . 1996
Vol 81 . No 2
eral density loss with combined norethindrone
and long acting gonadotropinreleasing hormone agonist therapy of symptomatic endometriosis: a prospective randomized
trial. J Clin Endocrinol Metab. 75:558-563.
8. Cann CE, Martin
MC, Genant
HK, laffe RB. 1984 Decreased
suinal mineral
content in amenorrheic women. JAMA. 251:626-629.
9 Prior JC, Vigna YM, Schechter
MT, Burgess
AE. 1990 Spinal bone loss and
ovulatory disturbances. N Engl J Med. 323:1221-1227.
10 Munro
CJ, Stabenfeldt
GH, Cragun
JR, Addiego
LA, Overstreet
JW, Lasley
BL. 1991 Relationship
of serum estradiol and progesterone concentrations to
the excretion profiles of their major urinary metabolites as measured by enzyme immunoassay
and radioimmunoassay.
Clin Chem. 37838-844.
11 Collins
WP, Branch CM, Collins
PO. 1981 Ovulation
prediction
and detection
by the measurement of steroid glucuronides.
In: Cartes-Prieto J, Campos da
Paz A, Neves-e-Castro
M, eds. Research on fertility and sterility. London: MTP
Press; 19-33.
12 Kassam A, Overstreet
JW, Gold EB, Lasley BL. A simple algorithm
using daily
urinary pregnanediol-3-glucuronide
to evaluate luteal phase progesterone
production [Abstract 311. I’roc of the 41st Annual Meet of the Sot for Gynecol
Invest. 1994; 210.
13 Baird DD, Weinberg
CR, Wilcoxx
AJ, McConnaughey
DR. 1991 Using the
ratio of urinary oestrogen and progesterone
metabolites to estimate day of
ovulation. Stat Med. 10~244-266.
14 Snow-Harter
C, Bouxsein
M, Lewis 8, Charette
S, Weinstein
P, Marcus
R.
1990 Muscle strength as a predictor of bone mineral density in young women.
J Bone Miner Res 5:589-595..
15 Block G, Hartman
AM, Dresser
CM, Carroll
MD, Gannon
J, Gardner
L. 1986
A data-based approach to diet questionnaire
design and testing. Am J Epidemiol. 124x453-469.
16 Block G, Woods
M, Potosky
A, Clifford
C. 1990 Validation of a self-administered diet history questionnaire using multiple diet records. J Clin Epidemiol.
43:1327-1335.
17 Ainsworthy
BE, Haskell
WL, Leon AS, et al. 1993 Compendium
of physical
activities: classification of energy costs of human physical activities. Med Sci
Sports Exer. 25:71-80.
18 Rothman
KJ. 1990 No adjustments
are needed for multiple comparisons.
Epidemiology.
1:43-46.
19 Sowers MR, Galuska
DA. 1993 Epidemiology
of bone mass in premenopausal
women. Epidemiol Rev. 15374-398.
20. Gallagher
JC, Goldgar
D, Moy D. 1987 Total bone calcium
in normal women:
effect of age and menopause status. J Bone Miner Res 2:491-496.
21. Prior JC. 1990 Progesterone as a bone-trophic hormone. Endocr Rev. 1:386-398.
22. Snow CR, Anderson
C. 1986 The effects of 17 beta estradiol and progestagen
on trabecular bone remodeling in oophorectomized
dogs. C&if Tissue Int.
39:198-205.
23. Bullen
BA, Skrinar
GS, Beitins
IZ, van Mering
G, Tumbull
BA, McArthur
JW. 1985 Induction of menstrual disorders by strenuous exercise in untrained
women. N Engl J Med. 312:X349-1353.
24. Ronkainen
H, Pakarinen
A, Kirkinen
P, Kauppila
A. 1985 Physical exerciseinduced changes and season associated differences in the pituitary-ovarian
function of runners and joggers. J Clin Endocrinol Metab. 60:416-422.
25. Prior JC, Cameron
K, Yuen GH, Thomas J. 1982 Menstrual
cycle changes with
marathon Training: anovulation
and short luteal phase. Can J Appl Sport Sci.
7:173-177.
26. Shangold
M, Freeman
R, Thysen
B, Gatz M. 1979 The relationship
between
long-distance running, plasma progesterone,
and luteal phase length. Fertil
Steril. 31:130-133.
27. Hetlund
ML, Haarbo
J, Christiansen
C, Larsen
T. 1993 Running induces
menstrual disturbances but bone mass is unaffected, except in amenorrheic
women. Am J Med. 95:53-60.
28. Bauman
JE. 1981 Basal body temperature:
unreliable method of ovulation
detection. Fertil Steril. 36:729-733.
29. Gross BA. 1987 Natural family planning indicators of ovulation. Clin Reprod
Fertil. 5:91-117.
30. Luciana
AA, Peluso J, Koch EL, Maier D, Kuslis S, Davison
E. 1990 Temporal
relationship
and reliability of the clinical, hormonal, and ultrasonographic
indices of ovulation in infertile women. Obstet Gynecol. 75:412-416.
31. McCarthy
Jr JJ, Rockette
HE. 1986 Prediction
of ovulation
with basal body
temperature.
J Reprod Med. 31:742-747.
32. Shoupe
D, Mishell
Jr DR, Lacarra
M, et al. 1989 Correlation
of endometrial
maturation with four methods of estimating day of ovulation. Obstet Gynecol.
73:88-92.
33. Templeton
AA, Penney
GC, Lees MM. 1982 Relation between the luteinizing
hormone peak, the nadir of the basal body temperature, and the cervical MUCUS
score. Br J Obstet Gynaecol. 89:985-988.
34. Nagata I, Kato K, Seki K, Furuya K. 1986 Ovulatory
disturbances:
causative
factors among Japanese student nurses in a dormitory.
J Adolesc Health Care.
71-5.