Conditioning exercise decreases premenstrual symptoms: a prospective, controlled 6-month trial*

FERTILITY AND STERILITY Copyright 1987 The American Fertility Society Printed in UBA. Conditioning exercise decreases premenstrual symptoms: a prospective, controlled 6-month trial* Jerilynn C. Prior, M.D.t+ Yvette Vigna, B.A., R.N.t Danielle Sciarretta, M.Sc. t Nenita Alojado, R.N. t Michael Schulzer, M.D., Ph.D. Faculty of Medicine, University of British Columbia, Vancouver, British Columbia, Canada Six months of exercise training was associated with decreased premenstrual symptoms in two groups of women. There was no change in symptoms in nontraining women. Eight sedentary (ST) women increased running from to 76.± 26 kmlcycle (mean ± standard deviation) over 6 months and seven runners (MT) trained for a marathon (422 km). Six normally active, nontraining (C-NT) women kept their activity constant. Each subject completed monthly intensity-graded questionnaires or kept daily symptoms diaries concerning premenstrual symptoms. All monitored basal body temperature, weight, and exercise. Gonadal steroids were measured in ST women. For ST subjects, breast (P =.5), fluid (P =.1), and personal stress (P =.25) decreased. MT women experienced decreased fluid (P =.34) and depression (P =.14). Anxiety tended to decrease (P =.87). ST and MT subjects experienced decreases in premenstrual symptoms without documented hormonal, menstrual cycle, or weight changes. These symptom changes appear to be the earliest evidence of the effects of conditioning exercise on the reproductive system. Fertil SteriI47:42, 1987 Conditioning exercise is known to be associated with changes in the menstrual cycle. 1-3 Adaptive changes to the reproductive system with exercise were first postulated in 1939. 4 The hypothesis that the hypothalamus orchestrated the "conditioning" of reproduction 5 has recently gained support. Luteinizing hormone pulse frequency studies of regularly menstruating women run- Received June 12, 1986; revised and accepted November 24, 1986. *Supported in part by the British Columbia Health Care Research Foundation. tdepartment of Medicine. :j:reprint requests: Jerilynn C. Prior, M.D., Faculty of Medicine, University of British Columbia, Suite 3318-91 West loth Avenue, Vancouver, British Columbia, Canada V5Z 1M9. Department of Statistics. 42 Prior et al. Exercise decreases premenstrual symptoms ners have documented an alteration of gonadotropin-releasing hormone pulsatile pattern,6 possibly related to increases in secretion of neurotransmitters such as beta endorphin. 7 The normal luteal phase is associated with a number of physical and emotional symptoms, termed molimina. These mild symptoms can predict the ovulatory cycle, as documented by basal body temperature (BBT) and progesterone (P) criteria.s We tested the hypothesis that molimina will decrease during a program of increasing conditioning exercise (training) over 6 months and that this decrease will be independent of the original level of fitness. Three months of exercise training in sedentary women has been shown to decrease breast and fluid-related symptoms in addition to lessening overall molimina. 9

SUBJECTS MATERIALS AND METHODS The participants were documented on two consecutive biphasic oral (BBT) records to have ovulatory menstrual cycles. None had used exogenous steroids within the preceding 6 months, and all had normal gynecologic histories. Three wom ~n in the control, nontraining (C-NT) group (sub Jects 9-11) had well-controlled insulin-dependent diabetes mellitus without effect on their daily lives or functioning. All subjects gave informed consent. Eight sedentary (ST) women volunteered to begin a 6-month conditioning exercise program. Seven women runners who were training for a marathon (MT) volunteered to record training and menstrual cycle data over 6 months. A final group of six normally active women was recrui~ed. These women were doing some occasional exercise (two ran, and 1 bicycled) but were not training and therefore became a C-NT group. Initial and 3-month training data on the eight ST women have been reported. 9 Subjects 1 and 2 in the training study (ST) were used in the C-NT during a nontraining period 2 years later. Subjects 9 through 11 performed the correlation study of menstrual cycle questionnaire (MCQ) and daily symptom diary (DSD) methods. The three groups did not differ in height (164.6 ± 6 cm) (mean ± standard deviation), weight (59. ± 6.8 kg), or body mass index (21. 7 ± 1.5). The ST women were younger than the MT women (3 ± 6 versus 36 ± 2 years, P <.5) but did not differ from C-NT women (35 ± 6 years). The overall mean age for the three groups was 33.4 ± 5.2 years. The women who began the program were told that the purpose of the study was to document hormonal and menstrual cycle changes with increased conditioning exercise. They were blind to the hypothesis that molimina would decrease with exercise training. METHODS All of the subjects kept continuous records of BBT, weekly body weight, and one or both measures of menstrual symptoms (MCQ, DSD). Women who were exercising recorded distance run exercise heart rates, and basal heart rates. Th; C-NT group reported the same data, except that exercise heart rate data are available for only one subject. The data are presented from the last sedentary cycle () and the sixth (6) exercise cycle for ST subjects. The first (1) and sixth cycle (6) are presented for MT and C-NT subjects.. Cycle and luteal phase length were determined with the use of mean temperature analysis of BBT.1 Temperature data were recorded by the subjects in list form and graphed by researchers, to prevent transcribing.errors. The luteal phase was accepted as the number of days after the intercept of the actual temperature curve with the mean temperature line. All of the ST subjects and three of the C-NT women answered a ten-item intensity-graded menstrual symptom questionnaire (scored 1 to 9 with 5 indicating no change) during days 2 to 5 of each cycle, relating to the preceding 2 weeks. 9 Each subject served as her own control. The nurse explaining the questionnaire was careful not to suggest that any change in symptoms was anticipated. The MT and C-NT women all recorded symptoms on a-dsd with a scale of to 4 for symptoms such as breast tenderness, which may be absent. A scale of S through W, centering around U, for usual, was employed for symptoms that have no true O,suchas breast size (Fig. 1). Each ST woman was weighed on a beam balance (in kilograms) and her height measured (in centimeters) at the beginning and end of 6 months oftraining. Body mass index (weight, in kilograms, divided by height, in meters squared) was calculated for all subjects on the initial and sixth cycle. All of the ST subjects had midluteal phase (days 18 to 26) serum P and estradiol measurements. Two resting midafternoon samples, drawn after 24 hours without exercise with the subjects in the postprandial state, were pooled for analysis. The time of day was standardized for each subject. All specimens for a single subject were analyzed in the same assay with standard radioimmunoassay methods. 11 Exercise was quantitated according to three methods: estimated distance (km/cycle), average training distance (average length of run) and training impulses (TRIMPS), an integrated'measure of exercise time and average exercise heart rate with respect to basal heart rate and maximal heart rates.12 The maximal heart rate obtained during a test of aerobic threshold, was dsed in the calculation of TRIMPS, as previously described. 9 Prior et al. Exercise decreases premenstrual symptoms 43

DAILY SYMPTOM DIARY SYMPTOMS. Menst. Flow - amt. Nwnber of pads/campons Nam.: Month: year: Please mark each day. Designated groups of symptoms ~111 have their own rating scale. Please grade the symptoms as directed. Please return form to: Dr. J. Prior, 3318-91 WlOth Ave., Van B.C. V5Z IM9. * Cycle Day ** Oate I. 1 2 ) 4 5 6 7 8 9 1 11 12 13 14 15 16 17 18 19\ 2 21 22 2) 24 25 26 27 28 29 ) Jl J2 I. Record if none. 1 - m1n1mal. 2 - moderate, ) - mod l.ntcnsc. - ~ramp. uvuj.atory Discomfort. Drease Tend-'l"nc Fluid Recent..at ~onsc: on He.adache Sleep Dlstrub. Feeling ansel'/ frustraiton F.el1na of deoresslon Feellns of anxiety II. Record 5 1f much le T -. A celte Bre t s1&. lnter e ~n X F.~~ln& ot Selfwort.h Fecltng of enerrv Stress -,",ork, home. money Comments l1ttle: les., U - -. usua.l, V ve.ry ntense l1.ttle: increased, \J much - increased I Symptoms this cycle compared to prevl,ous cycle (overall): Figure 1 Form for daily recording and grading of menstrual cycle-related and other symptoms, feelings, and events.! The ST subjects began a gradually increasing program of running after receiving instruction in stretching, injury prevention, footwear, and a two to four run-per-week schedule that alternated intense and less intense sessions. The MT and C-NT subjects recorded exercise as described above. The MT subjects began a self-chosen marathon training program. Statistical analysis of the measured data was carried out using standard nonparametric methods. DSD scores for the last 14 days of the cycle were averaged for analysis. The MCQ and DSD scores were compared with transformation of the o to 4 scale to a 1 to 9 scale for analysis with the Spearman Rank Correlation test (rs). Significance testing of the mean ranks was done with Lehman's modification of the Wilcoxon test. 13 The within-group questionnaire and DSD scores were compared with a Wilcoxon Paired Signed Rank test. Baseline and sixth cycle scores for nontraining and training women were compared with a Mann-Whitney U test. All of the tests were two-sided, unless otherwise noted. 44 Prior et al. Exercise decreases premenstrual symptoms RESULTS The daily and monthly questionnaire methods of documentation of molimina showed good correlation for breast and fluid symptoms (rs =.85 and.7; P values.2 and.9, respectively; data not shown). Global symptom and external stress scores were omitted from the version of the DSD used for these comparisons, therefore they cannot be compared. The ST and MT women increased their running distance (in kilometers) and the distance of their average run (training distance in kilometers) (Table 1). The ST subjects increased TRIMPS by 32 ± 8 and the MT subjects by 26 ± 16. Although those who began 6 months of training as sedentary women were running far less, their change in exercise equalled that of the marathon runners. None of the groups experienced a significant change in mean body mass index (data not shown). Body weight did not change in either training group (ST from 6.9 ± 8.1 to 6.1 ± 7.1 and MT from 55.6 ± 5.9 to 55.2 ± 65.5 kg).

Table 1. Luteal Phase Length and Exercise Data Over 6 Months Subject Distance/cycle Training distance TRIMPS Luteallengtb 6 6 6 6 km km km km days days ST 1 6.6 6.7 23 14 13 2 94.7 6.7 4 13 12 3 64.8 5.9 35 11 1 4 66.4 4.4 31 13 14 5 72. 7.2 44 9 8 6 76.8 8.9 32 11 1 7 131. 4.5 2 16 8 43.8 7.2 32 13 12 Mean 76.3 6.4 32 12.5 9.9 SD 26.3 1.4 8 2.1 4.4 P <.1 <.1 <.1 <.5 MT 12 142.5 29.2 11. 15.2 45 66 1 12 13 142.5 363.2 15.9 24.2 45 77 13 11 14 141.1 274. 11.7 19.5 38 86 12 9 15 192.9 265.9 11.9 2.4 42 7 12 1 16 113.5 133.9 11.3 14.9 43 64 12 1 17 116.7 12.6 11.6 8.4 35 31 12 11 18 372.9 366.4 14.9 22.9 42 77 11 7 Mean 174.6 256.6 12.6 17.9 41 67 11.7 1. SD 91.2 13. 1.9 5.4 4 18 1. 1.6 P <.5 <.2 <.5 =.53 C-NT 19 1 14 1 14 13 2 27 18 3.7 2.9 2 16 12 12 9 12 14 1 58 a 113 a 2.9 a 4.5 a 16 11 11 36 3.2 1 11 Mean 2 22 1.6 1.2 5 3 12. 12. SD 24 45 1.8 2. 1 7 2. 1. P NS NS NS NS acycling distance (km) divided by 3 to approximate running. The menstrual cycle interval did not change in any group (initial, 27.5 ± 2.2 days; 6 months 28.6 ± 4.1 days). The luteal lengths (Table 1) significantly shortened (ST, 12.5 ± 2.1 to 9.9 ± 4.4 with the anovulatory subject included, P <.5; MT, 11.7 ± 1. to 1. ± 1.6, P =.53). The mean midluteal estradiol (124.4 ± 59.1 to 126.3 ± 57.5 pg/ml) and P (12 ± 5 to 9 ± 5 ng/ml) levels were unchanged in the ST subjects. The women who were training (ST and MT) experienced significant decreases in several premenstrual symptoms at 6 months (Figs. 2 and 3). The nontraining controls had no change in moliminal symptoms. The ST women reported global premenstrual symptoms decreased from 6.5 ± 1.8 to 3.25 ± 1.2 (P <.1). Breast awareness or tenderness decreased from 8.25 ±.9 to 4. ± 1.6 (P <.5). Fluid symptoms or perceived puffiness and bloating decreased from 7.3 ± 1.8 to 4.5 ± 1.8 (P <.1). There were no significant changes in depressive symptoms or anxiety. Perceived external causes of stress showed a significant decrease (P <.25). When subject 7, who became anovulatory, is excluded, the changes remain significant. The MT women had lower breast symptom scores (.4 ±.7, P =.19) during the initial cycle than the C-NT women (1.3 ± 1.) (Fig. 3). Because of wide standard deviations, MT decreases in breast symptoms did not reach significance. Fluid-related symptoms decreased (.7 ±.4 to.2 ±.3, P <.5). Luteal phase feelings of depression also decreased (.5 ±.4 to.1 ±.2, P <.2). Feelings of anxiety tended to de- Prior et al. Exercise decreases premenstrual symptoms 45

9 8 '" 7 ~ 6 o ~5 4 CII ~3 2 1 o EXERCISE T SD vs ~ * p =.25 ~ 6th month t p =.1 i p =.5 Global Personal Breast Fluid Dysphoria Stress PREMENSTRUAL SYMPTOMS n = 8 Anxiety Figure 2 Mean ± standard deviation ofmcq scores (reported days 2 to 5 of the cycle) during the last sedentary and sixth exercise training cycle in eight women serving as their own controls. The data were analyzed with the Wilcoxon Paired Signed Ranks test. crease, but not significantly (P =.87). Both groups of women who were training, whether they began as sedentary or athletic, experienced decreases in physical and emotional symptoms over time. There were no changes in the nontraining group. DISCUSSION Eight normal ST women began a running exercise program of gradually increasing intensity over a 6-month period. They reported decreases in overall premenstrual symptoms, specifically those related to breast tenderness and fluid retention. Seven ovulatory women who were running and intensifying their training over 6 months before a marathon also experienced significant decreases in molimina. The non training control women, who did not differ in demographic or menstrual cycle characteristics from the groups intensifying exercise, had no moliminal changes over 6 months. The ST women, who served as their own controls and were monitoring premenstrual symptoms with a questionnaire rather than daily scores, were younger than the MT women (P =.37). The two training groups do not need to be 46 Prior et al. Exercise decreases premenstrual symptoms compared and are parallel experiments. However, the correlation of the questionnaire and daily symptom methods for the predominant physical premenstrual symptoms, although it was in nontraining subjects, suggests that both groups showed the same types of changes. The exercise-training groups showed effects of their running through decreased luteal phase lengths. Only one ST subject became anovulatory, and this was in association with major personal stress as well as weight loss. Despite the minimal reproductive system alterations, significant increases occurred in km/cycle, average training distance, and TRIMPS as a measure of energy output relative to level of fitness. The measured mean midluteal phase levels of the gonadal steroids, estradiol, and P were unchanged in ST women. The MT women tended to have lower scores on all parameters, compared with C-NT women; but TRAINING n = 7 1. vs 6 = cycles 3. * p<.5 ** p<.2., ~ v en ~ 2. 1. ** ~ E NON-TRAINING n = 6 a. 3 E >- en >- 2 ' I o Depression External Stress Figure 3 Mean ± standard deviation of DSD scores during the 14 days preceding menstrual flow, with the first and sixth consecutive menstrual cycle in seven women runners who intensified exercise before a marathon race and in six non training women. The data were analyzed with the Wilcoxon Paired Signed Ranks test. Intergroup analysis (Mann-Whitney U test) showed that training women had significantly lower breast symptoms during cycle 1 (P =.19). At cycle 6, the two groups differed in depression (P =.45), fluid (P =.19), and breast symptoms (P =.6). For external stress, the data were available on only six training and three nontraining women. * Breast Fluid Anxiety

only breast tenderness was significantly lower. Although the change in training was associated with decreased symptoms, there is probably some effect of lean body weight and regular exercise in preventing intense premenstrual symptoms. These studies confirm the earlier 3-month training results in sedentary women 9 by showing continued decreases in molimina with 6 months of training in the same subjects. This report extends that work by showing the same types of premenstrual symptom improvement over 6 months in women who were initially athletes. The concept that reproduction is altered by the training or conditioning process itself, not the absolute level of exercise, was initially suggested by Selye4 and is further supported by longitudinal studies. 1-3 Only two cross-sectional and poorly documented studies have been done to show that exercise decreases premenstrual symptoms.14. 16 Both of these studies were of students who were away from home, in a stressful environment and likely to be anovulatory.1 No documentation of menstrual cycle interval, phase lengths, or gonadal steroids was presented. The MT group showed a significant decrease in feelings of depression between cycles 1 and 6. This is the first documentation of menstrual cycle-related mood symptoms being altered by conditioning exercise. Anxiety tended to decrease (P =.87) but did not reach significance. The significant decreases in fluid and breast symptoms in the ST women serving as their own controls and the MT women suggest that these are physiologic changes, rather than cultural expectations. It is likely, if exercise were having a placebo effect, that mood symptoms would improve first and be the major change recorded. Moliminal symptoms decreased in association with increasing exercise without the development of anovulatory menstrual cycles, loss of body weight, or measured decreases in gonadal steroids. Given the sample size and the hormonesampling frequency, these parameters may have changed without detection. One woman, who experienced severe stress, became anovulatory in the sixth cycle. We reanalyzed for this subject the moliminal data, substituting data from the fifth month of training, which had a luteal phase> 1 days. The statistical evaluation was unchanged. It has previously been documented that exercise training increases exercise-stimulated beta endorphin,7 decreases gonadotropin pulse fre- quency 7 and gonadal steroids,1, 16 and changes menstrual cycle characteristics.1-3 The relationships between neurotransmitters, luteinizing hormone pulse frequency, gonadal steroid levels, menstrual cycle characteristics, and molimina are unclear. It was recently suggested that premenstrual phase beta endorphin levels are lower in women defined as having premenstrual syndrome than in apparently asymptomatic controls (though ovulatory cycles were not documented in any of the subjects)p Cyclic physical and mood symptoms (molimina) are normal and diagnostic of ovulatory cycles.8 Because of the cyclicity of symptoms, questionnaires are not ideal tools for studying molimina.18 The ST women answered their questionnaires on cycle-specific days, thus eliminating one. problem with this method. However, a new tool was developed for daily recording of symptoms (Fig. 1). The correlation of the MCQ and the DSD diary scores for fluid and breast symptoms suggests that, at least in non training women, the two instruments are comparable. 9 The MT women, who were running 174.6 ± 91.2 km/cycle when they volunteered for the study, had less breast symptoms than nontraining controls. This is not surprising, given the previously documented effect of 3 months' training on decreasing breast symptoms. Initially, the MT and C-NT groups did not differ on any other symptoms. However, by 6 months of training, the two groups differed in feelings of depression (P =.45) and breast (P =.6) and fluid (P =.19) symptoms. Reported personal stress did not change in the two groups over time, but anxiety-related symptoms tended to decrease in the training women (P =.59). These controlled studies on exercise training indicate specific ways in which exercise may make a woman feel better. Perhaps the perceived benefit derived from exercise training is the decrease in premenstrual symptoms. These diminished moliminal symptoms occur with very little or no menstrual cycle change and may reflect physiologic changes in neurotransmitter levels, representing hypothalamic adaptation to conditioning exercise. Acknowledgment. We thank Mr. Lou Reimer for patient assistance in data management. Prior et al. Exercise decreases premenstrual symptoms 47

REFERENCES 1. Bullen BA, Skrinar GS, Beitins IZ, von Mering G, Turnball BA, McArthur JW: Induction of menstrual disorders by strenuous exercise in untrained women. N Engl J Med 312:1349, 1985 2. Prior JC, Ho Yuen B, Clement P, Bowie L, Thomas J: Reversible luteal phase changes and infertility associated with marathon training. Lancet 1:269, 1982 3. Prior JC, Cameron K, Ho Yuen, Thomas J: Menstrual cycle changes with marathon training: anovulation and short luteal phase. Can J Appl Sport Sci 7:173, 1982 4. Selye H: The effect of adaptation to various damaging agents on the female sex organs in the rat. Endocrinology 25:615, 1939 5. Prior JC: Endocrine "conditioning" with exercise: a preliminary review. Can J Appl Sport Sci 7:149, 1982 6. Cumming DC, Vickovic MM, Wall SR, Fluker MR: Defects in pulsatile LH release in normally menstruating runners. J Endocrinol Metab 6:81, 1985 7. Carr DB, Bullen BA, Skrinar GS, Arnold MA, Rosenblatt M, Beitins IZ, Martins JB, McArthur JW: Physical conditioning facilitates exercise-induced secretion of betaendorphin and beta-lipotrophin in women. N Engl J Med 35:56, 1981 8. Magyar DM, Boyers SP, Marshall JR, Abraham GE: Regular menstrual cycles and premenstrual molimina as indicators of ovulation. Obstet Gynecol 53:411, 1979 9. Prior JC, Vigna Y, Alojado N: Conditioning exercise decreases premenstrual symptoms: a prospective controlled three month trial. Eur J Appl Physiol 55:349, 1986 1. Vollman RF: The menstrual cycle. In Majo~ Problems in Obstetrics and Gynecology, Vol 7, Edited by EA Freeman. Toronto, W. B. Saunders, 1977 11. Ho Yuen B, McComb P, Sy L, Lewis J, Cannon W: Plasma prolactin, human chorionic gonadotrophin, estradiol, testosterone and progesterone in ovarian hyperstimulation syndrome. Am J Obstet Gynecol 133:316, 1979 12. Banister EW, Hamilton CL: Variations in iron status with fatigue modelled from training female distance runners. Eur J Appl Physiol 54:16, 1985 13. Lehman SY: Exact and approximate distributions for the Wilcoxon statistic with ties. Trans Am Stat Assoc 56:293 1961 ' 14. Timonen S, Procope BJ: Premenstrual syndrome and physical exercise. Acta Obstet Gynecol Scand 5:331 1971 ' 15. Israel. SL, Sutton M, O'Brien P: Effects of aerobic training on primary dysmenorrhea symptomatology in college females. J Am ColI Health 33:241, 1985 16. Boyden TW, Pamenter RW, Stanforth P, Rotkis T, Wilmore JH: Sex steroids and endurance running in women.. Fertil Steril 39:629, 1983 17. Chuong CJ, Coulam CB, Kao PC, Bergstralh EJ, Go VLW: Neuropeptide levels in premenstrual syndrome. Fertil Steril 44:76, 1985 18. Sampson GA, Prescott P: The assessment ofthe symptoms of premenstrual syndrome and their response to therapy. Br J Psychiatry 138:399, 1981 48 Prior et ai. Exercise decreases premenstrual symptoms