Mood, Sexuality, Hormones, and the Menstrual

Cycle. II. Hormone Levels and Their Relationship

to the Premenstrual Syndrome

TORBJORN BACKSTROM,1 M D , DlANA SANDERS, P H D , ROSEMARY LEASK,

DAVID DAVIDSON, PAMELA WARNER, B S C , AND JOHN BANCROFT, MD

In women with premenstrual syndrome, negative changes start soon after ovulation gradually
increasing as the corpus luteum develops, and reach a maximum during the last 5 days of the
luteal phase. They decline rapidly once menstruation starts, disappearing within one or two
days of ovarian steroids reaching baseline levels. Positive moods are at maximum when
preovulatory estradiol reaches its peak. A comparison of hormone levels in women with high
and low degrees of cyclical mood change showed no difference in progesterone, estradiol,
testosterone, or androstenedione.

The etiology of mood change during In this paper, we report the temporal
the menstrual cycle, and in particular, the relationships between hormonal change
premenstrual worsening of mood and and subjective change, and compare hor-
physical well-being i.e., the premenstrual mone levels at different stages of the cycle
syndrome is not understood. in women with high and low degrees of
The role of estradiol in causing the cyclical mood change.
premenstrual syndrome (PMS) was first
suggested by Frank (1). Relative proges-
terone deficiency during the luteal phase RESULTS
has been proposed as a rationale for pro-
gesterone therapy (2). Reports of lower This relationship was examined in all
plasma progesterone concentrations dur- those women from the "high-change"
ing the luteal phase in patients with PMS groups who had given daily blood sam-
compared to controls have been published ples during the luteal half of the cycle
(3-8). Other studies have failed to repli- (n = 12). For this purpose, the raw scores
cate this pattern (9-11). from the daily ratings of depression, ir-
ritability, cheerfulness, "energetic," feel-
ing of body swelling, and breast tender-
'Present address: Department of Obstetrics and ness were standardized for each subject
Gynaecology, University of Umea, Sweden.
From the Medical Research Council Reproductive
(i.e., into scores, when z = (Xi—x/SD and
Biology Unit and Department of Obstetrics and the mean daily z scores for the 12 women
Gynaecology, University of Edinburgh, 37 Chalmers were used. In Figure 1, the mean moods
Street. Scotland.
Address reprint requests to: John Bancroft, MD,
and physical symptoms are centered
Medical Research Council, Centre for Reproductive around the day of the LH peak and around
Biology, 37 Chalmers Street, Edinburgh EH3 9EW. the first day of menses, The pattern for
Scotland,
Received for publication March 16, 1982; final
body swelling was very similar to that for
revision received December 17, 1982. breast tenderness. These findings show

Psychosomatic Medicine Vol. 45, No. 6 (December 1983) 503

Copyright © 1983 by the American Psychosomatic Society, Inc

Published by Elsevier Science Publishing C o , Inc
52 Vanderbilt Ave., New York, NY 10017 OO33-3174V83/S3.00
 MENSTRUAL CYCLE. II

negative effects occurs during the last five

premenstrual days with a rapid decline as
soon as menstruation begins, The data
were also centered around the day when
estradiol and progesterone reach the
"follicle-phase baseline" (below 1 ng/ml).
The results showed that negative mood
scores were at the maximum 2 or 3 days
earlier and at the minimum 2 to 3 days
after this hormonal point.
Positive moods showed an opposite
pattern with their peaks coinciding with
the preovulatory estradiol peak and de-
clining during the luteal phase.
Mean concentrations of estradiol, pro-
gesterone, androstenedione, and testos-
terone for each of the six cycle phases
were compared in the original three
groups (clinic PMS, nonclinic PMS, and
no PMS) and also in the three re-
categorized groups (high, medium, and
low mood change). No significant dif-
ferences emerged that could be related to
the degree of mood change. (For details,
see Sanders (12)).
In order to make a more precise com-
parison of the luteal stage hormone levels
in the high and low mood change groups,
Figure 1. Mean z scores for mood and physical
levels on particular days rather than mean
symptoms and hormonal data during the
menstrual cycle in patients with PMS (n levels for cycle phases were then com-
= 12). Part A: The data are synchronized pared. Women from these two groups
around the day of LH surge. Part B: The (high and low change) were paired ac-
data are synchronized around the first cording to whether they had blood sam-
day of menstrual bleeding. Day 0 = the
first day of menstrual bleeding.
ples taken on the same day of the luteal
half of the cycle. Several subjects were
studied for more than one cycle but the
results presented so far have all been
that irritability and depression and the based on the first cycle studied in each
two physical symptoms start to increase woman. In order to increase the number of
within two or three days after the LH women who could be precisely paired in
surge and increase initially in parallel this way, either the first or second cycles
with the luteal rise in progesterone and were used. This allowed 18 pairs of high
estradiol, but continue to increase for 2 to and low mood change women with each
4 days after the ovarian steroids have member of a pair having blood samples on
started to fall. The maximum level of their the same day within each of the following

504 Psychosomatic Medicine Vol. 45, No. 6 (December 1983)

BACKSTROMetal.

four time periods: Days 12 to 10, 9 to 7, 6 same woman, there were substantial in-
to 4, and 3 to 0 before menstruation, Day creases in estradiol without any rise in
0 being the first day of menstrual bleed- progesterone. The mood and physical
ing. The levels of estradiol and proges- symptom ratings in these five cycles were
terone were then compared for the two analyzed by dividing each cycle into six
groups after logarithmic transformation, equal parts. An analysis of variance for
using analysis of variance for repeated each variable across these six phases
measures. The results are shown in Figure showed no premenstrual decline in mood
2. There were no significant differences but a significant premenstrual increase in
between groups for levels of either hor- breast tenderness (p < 0.01) and body
mone. swelling (p < 0.01). The breast changes,
though statistically significant, were
slight in degree. Further evidence from
Anovular Cycles anovular cycles is required to distinguish
It is obviously of considerable rele- those with and without follicular de-
vance to know whether PMS occurs dur- velopment.
ing anovular cycles. There is a lack of
clear evidence on this point in the litera-
ture. In this study, there were five anovu- DISCUSSION
lar cycles studied in three women (none
in the clinic group). In two cycles from the Results of this study show that the
cyclically occurring negative mood
changes in women with premenstrual
Estradiol Progesterone syndrome are closely linked to the luteal
phase of the hormonal cycle. They start
soon after ovulation, gradually increase as
the corpus luteum develops, and reach a
maximum during the last five days of the
luteal phase. These negative experiences
decline rapidly once menstruation starts
and usually disappear within a day or. two
100 of the estradiol and progesterone reaching
early follicular "baseline" levels. It is also
clear that positive moods are at their
maximum when the preovulatory es-
tradiol reaches its peak.
12-10 6-4 12-K) 6-4
9-7 3-0 9-7 3-0 This study has served to emphasize
Days before menstruation
some of the methodological issues men-
tioned in Part I. Studying the whole cycle
Figure 2. Comparison of estradiol and progesterone encourages us to look for cyclical
levels in 18 high-change and 18 "low- mechanisms operating throughout the
change" women at some stage prior to
menstruation. There was no significant
month and not just in the premenstrual
difference between the two groups week. The contrast between the late fol-
• • High Change. O O, Low licular and late luteal phase in many of
Change. these women is striking. Not only does

Psychosomatic Medicine Vol. 45, No. 6 (December 1983) 505

 BACKSTROM et al.

this help us to see that there is a good as for the total picture of PMS as the negative
well as a bad part of the cycle, but allow- symptoms are well established before the
ances for some contrast effect, leading to hormonal fall begins.
amplification of differences must also be Our findings suggest that a marked dif-
made (i.e., the follicular phase may be ference between women in cyclical mood
experienced and reported as particularly change can occur in the absence of obvi-
good, partly as a result of the contrast with ous differences in circulating ovarian
a really bad premenstrual phase and vice steroid levels. Other methods of investiga-
versa). tion will be required to identify the un-
The importance of repeated hormonal derlying mechanism of this important but
assessment in studies of this kind is enigmatic pattern. As sensitivity of the
shown by the proportion of anovulatory central nervous system to circulating
and otherwise abnormal cycles identified. steroids may vary between, women, mea-
The very clear temporal relationship be- surement of circulating levels may be of
tween hormonal and psychological only partial relevance. But, as yet, no
changes does not, of course, indicate a satisfactory methodology for measuring
causal link but increases the likelihood of such target organ sensivitiy has been de-
there being some common factor. The veloped. The greater incidence of postpar-
recent demonstration that such cyclical tum mood disturbance and adverse reac-
patterns continue after removal of the tion to oral contraceptives in our clinic
uterus (13) indicates that the pattern is not group may be relevant. Sensitivity of the
explainable as a psychological or physical hypothalamopituitary system to negative
reaction to menstruation and increases the feedback, as reflected in FSH levels, may
likelihood that an ovarian or hypothalam- provide a useful lead in this respect (15).
ic factor is involved. Alternative research strategies should in-
In this study, we looked for relatively volve the study of mood change in the
crude indicators of hormonal differences presence of clearly abnormal hormonal
between women with high and low mood cycles (e.g., anovular cycles) and during
changes and found none. We have been times of change in cyclicity, such as
unable to replicate earlier evidence of adolescence, the postpartum period fol-
relative progesterone deficiency in PMS. lowing lactation, and the perimenopausal
It is clear that severe PMS can occur in the period. Also, comparison in the same
presence of an apparently normal luteal woman of hormonal cycles with and
phase. The premenstrual fall in plasma without cyclical mood change may be
hormone concentrations has been sug- fruitful. Methods of manipulating the ova-
gested as a cause of negative mood change rian cycle without administering exogen-
(1) comparable to the postpartum situa- ous steroids (e.g., with LHRH analogs)
tion (14). This clearly could not account may also be worth pursuing.

REFERENCES

1. Frank RT: The hormonal causes of PMT. Arch Neurol Psychiatry 26:1053-1057, 1931
2. Dalton K. Premenstrual Syndrome and Progesterone Therapy. London: W. Heineman, 1977
3. Backstrom T, Carstensen H: Estrogen and progesterone in plasma in relation to premenstrual tension. J
Steroid Biochem 5:257-260. 1974

506 Psychosomatic Medicine Vol. 45, No. 6 (December 1983)

MENSTRUAL CYCLE.

4. Backstrom T, Wide L, Sodergard R, Carstensen H: FSH, LH TeBG-capacity, estrogen, and progesterone

in women with premenstrual tension during the luteal phase. J Steroid Biochem 1976; 7:473-476
5. Smith SL: Mood and the menstrual cycle. In Sachar EJ, Ed., Topics in Psychoendocrinology. NewYork.
Grune & Stratton, 1975: pp. 19-58.
6. Sitruk-Ware LR, Sterkers N. Mowszowicz I, Mauvais-Jarvis P: Inadequate corpus luteal function in
women with benign breast diseases. J Clin Endocr Metab 44:771-774, 1977
7. Munday MR, Brush MG, Taylor RW: Correlations between progesterone, estradiol, and aldosterone
levels in the premenstrual syndrome. Clin Endocrin (London) 14:1-9, 1981
8. Kicovic PM, Cortes-Prieto J, Franchi F, Luisi M: Hormonal profile of the premenstrual tension
syndrome. International Symposium of Progesterone and Progestms, Pans: May 7 - 9 , 1981
9; Taylor JW: Plasma progesterone, estradiol 17B and premenstrual symptoms. Acta Psychiat Scand
60:76-86, 1979
10. Andersch B, Abrahamsson L, Wendestam C, Ohman R, Kahn L: Hormone profile in PMT: Effects of
bromocriptine and diuretics. Clin Endocrin (London) 11:657-664, 1979
11. O'Brien PMS, Selby C, Symonds EM: Progesterone, fluid, and electrolytes in premenstrual syndrome.
BrMedJ. i:1161-1163, 1980
12. Sanders DJ: Hormones and behaviour during the menstrual cycle [Ph.D Thesis|. Edinburgh, Scotland:
University of Edinburgh, 1981
13. Backstrom CT, Boyle H, Baird DT: Persistence of symptoms of premenstrual tension in hysterectomized
women. Br J Obst Gynaecol 88:530-536. 1981
14. McNair DM, Lorr M: An analysis of mood in neurotics. J Abnorm Soc Psychol 69:620-627, 1964
15. Backstrom T: in preparation.

Psychosomatic Medicine Vol. 45, No. 6 (December 1983) 507