Dieting causes menstrual irregularities in normal weight young women through impairment of episodic luteinizing hormone secretion

Transcription

1 FERTILITY AND STERILITY Copyright<> 1989 The American Fertility Society Printed in U.S.A. Dieting causes menstrual irregularities in normal weight young women through impairment of episodic luteinizing hormone secretion Karl M. Pirke, M.D.* Ulrich Schweiger, M.D. Thomas Strowitzki, M.D. ReinhardJ. Tuschl, Ph.D. Reinhold G. Laessle, Ph.D. Andreas Broocks, M.D. Brigitte Huber, B.S. Ralf Middendorf, B.S. Max-Planck-Institut fur Psychiatrie, Division of Psychoneuroendocrinology, Munich, West Germany Thirteen healthy, normal weight young women were studied throughout a control cycle and a diet cycle, during which they lost 1 kg per week on a vegetarian kcal diet. Blood was sampled daily in the morning, and at weekly intervals, collected at 10-minute intervals for 6 hours. Follicle growth was monitored by ultrasonic measurement. All subjects showed normal cyclic gonadal function during the control cycle. Cyclic gonadal function remained unaltered in two subjects during the diet cycle. No dominant follicle developed in seven others, while another four showed apparently normal follicular development but impaired progesterone secretion by the corpus luteum. Comparison of both cycles revealed that episodic luteinizing hormone (LH) secretion during the follicular phase was altered by dieting. Average LH concentrations and the frequency of episodic secretions were significantly reduced during the follicular phase but not during the luteal phase. Follicle-stimulating hormone was unaltered. Fertil Steril51:263, 1989 Severe weight loss causes amenorrhea, as has been demonstrated in patients with anorexia nervosa and other types of malnutrition. 1 Frisch and Revelle 2 have claimed that a weight reduction of more than 13% of the ideal body weight (IBW) will cause amenorrhea in most women. Bates et al. 3 found that smaller weight deficits from weight reduction diets ( -10%) also may impair fertility. Amenorrhea, but more frequently anovulatory cycles and luteal phase defects, may be responsible for infertility. After weight normalization, the majority of such infertile patients become pregnant without further treatment. In a series of studies, 4-6 we recently showed that mild weight reduction diets will cause menstrual irregularities in the ma- Received December 11, 1987; revised and accepted September 29, * Reprint requests: Karl M. Pirke, M.D., Division of Psychoneuroendocrinology, Max-Planck-Institut fiir Psychiatrie, Kraepelinstr 10, D-0 Miinchen 40, West Germany. jority of normal weight young women, even when their body weight does not fall below 100% IBW. Either follicular development becomes disturbed so that no ovulation can occur, or luteal phase defects develop. We found that the likelihood of menstrual irregularities is greater the more weight subjects lose and the younger they are. Vegetarian diets affect the cycle more than a nonvegetarian diet, when both cause the same weight loss. In the present experiment, we studied the episodic secretion of luteinizing hormone (LH) and follicle-stimulating hormone (FSH) at weekly intervals throughout a control and a diet cycle, during which subjects lost about 1 kg body weight per week. Subjects MATERIALS AND METHODS Thirteen healthy, normal weight young women aged 23.5 ± 0. 7 years volunteered for the study after Pirke et al. LH secretion is impaired during dieting 263

2 having given their written informed consent. The age at menarche was 12.8 ± 0.8 years. All women reported regular menstrual cycles of 22 to 33 days length. The individual variability was less than 4 days during the past year. None had taken medication during the past half year. Thorough physical and psychiatric examinations including laboratory screening, electrocardiogram, and chest x ray were carried out. Women with clinical or subclinical eating disorders were excluded, as were vegetarians and subjects with other food idiosyncrasies. Women with high levels of exercise who trained for or participated in sport events also were excluded from the study. Experimental Design Each control and diet period started with the first day of a new menstrual cycle. Blood was sampled between 8:00 and 10:00 A.M. daily during each cycle. No more than 3 days per cycle were missed by any subjects. Every Wednesday, blood was drawn after an overnight fast in order to measure!3-hydroxybutyric acid (i3hba). Between the third and the fifth day of the cycle, an Abbocath (Deutsche Abbott GmbH, D-6200 Wiesbaden, Federal Republic of Germany} needle was inserted into a forearm vein at 6:00P.M. and kept open by a slow-drip saline infusion. Blood then was sampled at 10-minute intervals for 6 hours. The 6-hour blood sampling was repeated at weekly intervals throughout the control and diet cycles. During each cycle, subjects recorded physical activity, mood, and subjective feeling of stress in a standardized diary. Follicular Growth Determination Follicular growth was determined by serial ultrasonographic examinations starting at day 11 of the menstrual cycle. Ultrasound was continued at 2- day intervals until ovulation occurred or signs of an anovulatory cycle were detected, shown by char..: acteristic changes of the endometrium. Number of growing follicles, maximum size of the dominant follicle, and hormone-dependent alterations of the endometrium were evaluated. All scans were performed by the same examiner, using a transabdominal sector scan DRF400, (Sonotron, Garching, Federal Republic of Germany), or an Ultramark 4, (Kranzbuehler, Solingen, Federal Republic of Germany}. Diet Subjects were instructed to eat an average of kcal per day, and adjusted caloric intake in order to lose 1 kg body weight per week. They were asked to avoid meat and fish and to eat only small amounts of milk products. Weight was measured three times a week. All subjects kept nutrition diaries, which were evaluated according to Souci et al. 7 In addition, mood and physical activity were recorded in a standardized diary, as described earlier. Analysis and Statistics Plasma LH and FSH were measured using the MAlA Clone radioimmunoassay (RIA; Serono, Freiburg, Federal Republic of Germany). All analyses were performed under statistical quality control. Intra-assay variability was 7.6% for LH at an average concentration of 4 ng/ml, and 8.2% for FSH at an average concentration of 6.2 ng. Estradiol (E 2) and progesterone (P) were measured by RIA after ether extraction. Radioimmunoassay kits were obtained from Travenol, Munich, Germany. Intra-assay variability was 7.1% at an average concentration of 3.4 ng/ml P, and 8.2% at an average concentration of 120 pg/ml E 2 The follicular phase was defined as the time from the beginning of the menstrual bleeding until -2 days of the midcycle. Midcycle was defined as the day of the E 2 maximum. The luteal phase included all days from +2 days of the midcycle to the end of the cycle. Longitudinal means were calculated for the follicular and the luteal phase for LH, FSH, E 2, and P. When no E 2 maximum >150 pg occurred, when no dominant follicle was detected, and when P values never exceeded 2 ng, as occurred in subjects during the diet period, the longitudinal means over the whole observation period were calculated. Episodic Gonadotropin Secretion The average concentrations during the 6-hour blood sampling periods were calculated. Number and size of the episodic secretion were evaluated according to Santen and Bardin. 8 When gonadotropin values rise by more than three standard deviations, a significant increase was assumed. The variability of the method was measured in each assay by analyzing at least 12 quality control samples. Statistical Evaluation Nonparametric tests were used throughout. Data during the control and the dieting period were compared using the Wilcoxon test for paired data. 264 Pirke et al. LH secretion is impaired during dieting Fertility and Sterility

3 Table 1 Body Mass and Nutrition During Control and Diet Cycle for all13 Subjects Control cycle Diet cycle p Body mass index (kg/sqm) 23.3 ± ± 1.1 <0.001 {J-Hydroxybutyric-acid (mmol/1) ± ± 0.13 <0.01 Triiodothyronine 1.1 ± ± 0.4 <0.01 Caloric intake (kcal) 2153 ± ±304 <0.001 Percent protein 13 ± 2 15 ± 304 NS Percent carbohydrate 44±6 58± 10 <0.001 Percent fat 39±4 26±8 <0.001 Percent alcohol 4±3 1±3 <0.02 RESULTS Table 1 summarizes the basal data of the subjects. The body mass index decreased from 22.3 to 20.5 (kg/sqm), which is equivalent to an average weight loss of 4.9 ± 0. 7 kg. The body mass index at the end of the dieting period is equivalent to 99% of IBW according to the Metropolitan Life Insurance Company. 9 Triiodothyronine (T 3) values decreased and,bhba increased significantly during the diet period. Caloric intake decreased to an average of 880 ± 304 kcal. The vegetarian diet resulted in an increase in percent carbohydrate and in a decrease in percent fat, while protein content was not significantly changed. Physical activity as recorded in the diaries did not change during the diet period. Figures 1 and 2 present two typical examples of cycles before and during dieting. Figure 1 shows a normal hormonal pattern in the control cycle. On day 12, ultrasonic evaluation revealed a dominant follicle of 19 X 15 mm, which had grown to 22 X 16 mm by day 15. On day 19-1 day after the E 2 maximum-the follicle had ruptured and could no longer be detected by ultrasonic measurement. During the diet cycle, a small follicle (diameter 10 mm) was detected on day 14; by day it had grown to 19 X 15 mm. The third ultrasonic measurement on day 19 showed no follicle. Note that the P increase during the luteal phase was slight and that P levels were low. This pattern-hormonally and ultrasonically normal follicle growth followed by impaired P secretion during the luteal phase-was seen in 4 of the 13 subjects during the diet cycle. Figure 2 shows another example. During the control cycle, a dominant follicle was observed on day 13. Three days later, the E 2 maximum was reached. On day, the follicle had ruptured and could not be detected. During the diet cycle, ultrasonic examination revealed multiple small cysts ( <10 mm) on days 14 and, but no follicles on day 26. In this case, no dominant follicle developed. Estradiol never exceeded 150 pg/ml, and no luteal phase occurred. This pattern-no development of a dominant follicle, no ovulation, and no luteal phasewas observed in 7 of the 13 subjects during the diet period. All 13 subjects had ovulatory cycles during the control period, and two subjects maintained ovulatory cycles with normal luteal phases during the diet period as well. Table 2 shows the E 2 maxima and the longitudinal means of E 2, P, LH, and FSH during both control and diet cycles. When no dominant follicle developed (n = 7), the longitudinal means over the whole observation period were considered in the calculation of the average longitudinal means of follicular and luteal phase. Maximal E 2 values did not differ significantly between cycles. In the seven cases of anovulatory cycles, they were decreased during the diet cycle. In three of the four cases of an impaired luteal phase during the diet cycle, they (pg/ml) -E GOO (pg/ml) 100 Control cycle c24,4 p -4,1 24 veara t (days) clh FSH P (mu/~ii(mu{~il ln!j.mll t (days) alh FSH P f) 10 Figure 1 Hormone concentrations in plasma during a control (upper part) and a diet cycle. Daily E 2 values over the whole cycle and P values during the luteal phase are plotted Pirke et al. LH secretion is impaired during dieting 265

4 (pg/mll -~ a D103,3 a F -4,5kg 24 years = :_"J~YM 50 :... -./"..iv ~ (days) D14,5 -E2 al.h FSH p, (pg/ml) ~\ ~J a\.. eoo rj"-1 v~\ ' r lv"'l. v t (dayo) Figure 2 Hormone concentration in plasma during a control and a dieting cycle. During the diet phase, no dominant follicle developed (see text). The permanently low P values indicate that no corpus luteum developed. were higher in the diet than in the control cycle. Both subjects whose menstrual cycle was not affected by dieting also showed slightly higher E 2 maxima during the diet cycle. Longitudinal means of the hormone values during the follicular phase were not significantly affected by dieting. During the luteal phase, however, the longitudinal means of E 2 and P were significantly decreased. Table 3 shows the results of the LH values measured at 10-minute intervals over 6 hours during the first, second, and third studies in the follicular phase (F 1, F 2, F 3) and the first study and second study in the luteal phase (L 1, and L 2 ). Folliclestimulating hormone secretion was not affected significantly by dieting at any point, and is not discussed here. Luteinizing hormone secretion, however, was impaired in all three studies during the follicular phase. The average concentration over 6 hours was significantly decreased during the diet cycle at F 1, F 2, and F 3, but the pulse frequency of LH peaks was significantly decreased only at F 2 When the varying length of the follicular phase was taken into account and measurements were assigned the categories early (EF), mid-(mf), or late (LF) follicular phase, a similar pattern emerged. In the diet phase, average concentration of LH was decreased during MF (2.6 ± 1.5 ng/ml versus 3.9 ± 2.1, diet versus control P < 0.01) and LF (3.2 ± 1.8 versus 5.6 ± 2.3, diet versus control P < 0.01). The number of LH pulses over 6 hours was decreased during LF (2.9 ± 1.4 versus 4.6 ± 2.0, diet versus control, P < 0.05). DISCUSSION We have shown previously that dieting may lead to menstrual disturbances in normal weight, healthy young women. 4-6 This effect was apparent when the weight loss was as low as 1 kg body weight per week over one menstrual cycle. Two different forms of menstrual disturbances develop. Measurement of E 2 and P throughout the cycle indicated either a lack of follicular development or, after obviously normal follicular development, a. disturbed luteal phase. The effect was stronger in younger women and correlated with the extent of weight loss. 6 In order to study the hypothesis that impaired gonadotropin secretion is responsible for the diet-induced disturbances in normal weight women, we chose for this experiment conditions under which we could expect a high incidence of diet-induced disturbances. The subjects were younger than 24 years (range, to 24 years), they lost at least 4 kg body weight, and they had been asked to eat a vegetarian reduction diet. We have chosen a vegetarian diet because we have demonstrated earlier that a vegetarian diet is more effec- Table 2 E 2 Maximum and Hormonal Longitudinal Means During the Follicular and Luteal Phase Control cycle Diet cycle p E2 maximum (pg/ml) 319 ± ±284 NS 4 Follicular phase E2 (pg/ml) 66± 19 65±29 NS P (ng/ml) 0.77± ±0.25 NS LH (ng/ml) 6.2 ± ±2.7 NS FSH (ng/ml) 6.9 ± ± 1.6 NS Luteal phase E2 (pg/ml) 130 ±54 93±88 <0.05 P (ng/ml) 8.03 ± ± 5.48 <0.025 LH (ng/ml) 6.0± ± 3.0 NS FSH (ng/ml) 4.4 ± ± 2.3 NS a NS, not significant. 266 Pirke et al. LH secretion is impaired during dieting Fertility and Sterility

5 Table 3 Pulsatile LH Secretion During the Follicular Phase (F 1, F 2, F 3) and During the Luteal Phase (L 1, L 2) F, F Fa Average Average Average No. of size of No. of size of No. of size of i peaks peaks i peaks peaks i peaks peaks RIJ!ml RIJ/ml RIJ/ml Control cycle 2.9 ± ± ± 1.2 n= ± ± ±2.0 n= ± ± ± 1.2 n=5 Diet cycle 2.0± ± ± 1.1 n= ± ± ± 1.6 n= ± ± ± 1.4 n= 11 p 0.01 NS" NS NS 0.02 NS NS Control cycle 5.3 ± ± ± 6.0 n = ± ± ± 2.5 n = 5 Diet cycle 3.1 ± ± ± 5.4 n = 4 n = 0 P NS NS NS NS, not significant. tive in disturbing the cycle than a mixed diet when both diets cause the same weight loss. As predicted, we observed a high percentage of cycle disturbances. Four of 13 developed impaired P secretion during the luteal phase after apparently normal follicular development during the first half of the cycle. Seven of 13 subjects showed impaired follicular development. No dominant follicle was detected by ultrasonic measurement and E 2 did not increase to a preovulatory maximum. The episodic secretion of FSH was not affected by dieting at any time of the cycle. Episodic secretion of LH, however, was impaired. During all three 6-hour studies in the follicular phase, the average LH concentration was significantly reduced. During the second follicular study, the frequency of LH peaks also was reduced, as in the third follicular study, though here not significantly. However, it must be considered that the follicular phases were of different lengths, so that during the control cycle, a third follicular 6-hour study was possible in only five of 13 subjects. Similar observations were made by Cumming et al. 10 in eumenorrheic runners. As a consequence of disturbed follicular development, the number of 6- hour studies during the luteal phase of the diet cycle was small (n = 4). No conclusions are possible, therefore, regarding the role of LH secretion in the impairment of the luteal phase. In an earlier study, we found no changes in episodic LH secretion during dieting. 4 In that study, we had measured episodic LH secretion at only one point before dieting started, as we had expected to find strong suppression of LH to an infantile or a pubertal pattern, as observed in anorexia nervosa or under conditions of severe weight loss The current study shows that mild dieting brings about moderate suppression of LH secretion, although the values remain above those seen in infants and pubertal girls during the daytime. The rather small but significant changes in LH secretion raise the question of whether the diet-induced effects on follicular development and corpus luteum function are only centrally mediated or whether there exist other mechanisms influencing ovarian function directly during diet. Functional tests of ovarian function will help to resolve this question. The data presented in this paper certainly do not by themselves allow one to discriminate between factors such as diet composition, caloric deprivation, weight loss, or altered body composition as possible causes of the observed alterations of LH secretion. It can be inferred from previous studies that caloric deprivation and diet composition combine to induce menstrual disturbance. Caloric deprivation seems to be effective by itself 14 ; altered diet composition may be effective only in combination with caloric restriction. 5 The alterations of body weight and body composition are very small at the first appearance of altered endocrine functions. Whether they contribute to the observed disturbance remains doubtful. On the basis of the thresholds postulated by the critical body weight hypothesis, the absence of endocrine alterations would have been expected. In conclusion, these data indicate that mild weight reduction diets can disturb ovarian function and fertility in healthy, normal weight young Pirke et al. LH secretion is impaired during dieting 267

6 women even when weight does not fall below ideal body weight. Impaired LH secretion is probably one of the mechanisms responsible. REFERENCES 1. Pirke KM, Ploog D: Biology of human starvation. In Handbook of Eating Disorder, Part 1; Edited by PJV Beumont, GD Burrows, RC Casper, Amsterdam, Elsevier Science Publishers, 1987, p Frisch RE, Revelle R: Height and weight at menarche and a hypothesis of critical body weights and adolescent events. Science 169:397, Bates GW, Bates SR, Whitworth NS: Reproductive failure in women who practice weight control. Fertil Steril 37:373, Pirke KM, Schweiger U, Lemmel W, Krieg JC, Berger M: The influence of dieting on the menstrual cycle of healthy young women. J Clin Endocrinol Metab 60:1174, Pirke KM, Schweiger U, Laessle R, Dickhaut B, Schweiger M, W aechtler M: Dieting influences the menstrual cycle: Vegetarian versus nonvegetarian diet. Fertil Steril46:1083, Schweiger U, Laessle R, Pfister H, Roehl C, Schwingenschloegel M, Schweiger M, Pirke KM: Diet-induced men- strual irregularities: effects of age and weight loss. Fertil Steril48:746, Souci SW, Fachmann W, KrautH: Die Zusammensetzung der Lebensmittel. Stuttgart, Wissenschaftliche Verlagsgesellschaft, 1981/1982 p 1 8. Santen RJ, Bardin CW: Episodic luteinizing hormone secretion in man. Pulse analysis, clinical interpretation, physiologic mechanisms. J Clin Invest, 52:2617, Metropolitan Life Insurance Company. Statistical Bulletin 40:1, Cumming DC, Vickovic MM, Wall SR, Fluker MR: Defects in pulsatile LH release in normally menstruating runners. J Clin Endocrinol Metab, 60:810, Boyar RM, Katz J, Finkelstein JW, Kapen S, Weiner H, Weitzman ED, Hellman L: Anorexia nervosa: immaturity of the 24-hour luteinizing hormone secretory pattern. N Engl J Med 291:861, Pirke KM, Fichter MM, Lund R, Doerr P: Twenty-four hour sleep-wake pattern of plasma LH in patients with anorexia nervosa. Acta Endocrinol (Copenh) 92:193, Vigersky RA, Andersen AE, Thompson RH, Loriaux DL: Hypothalamic dysfunction in secondary amenorrhea associated with simple weight loss. N Engl J Med 297:1141, Fichter MM, Pirke KM: Hypothalamic pituitary function in starving healthy subjects. In The Psychobiology of Anorexia Nervosa, Edited by K. M. Pirke, D. Ploog. Berlin, Heidelberg, Springer Verlag, 1984, p Pirke et al. LH secretion is impaired during dieting Fertility and Sterility