Thyroid axis function during the menstrual cycle in women with premenstrual syndrome

Pergamon

Psychoneuroendocrinology, Vol. 20, No. 4, pp. 395-403, 1995 Copyright © 1995 Elsevier Science Ltd Printed in the USA. All rights reserved 0306-4530/95 $9.50 + .00

0306-4530(94)00068-9

THYROID AXIS F U N C T I O N DURING THE M E N S T R U A L CYCLE IN WOMEN WITH PREMENSTRUAL S Y N D R O M E SUSAN S. GIRDLER, CORT A . PEDERSEN, a n d K A T H L E E N C. L I G H T Department of Psychiatry, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA

(Received 23 February 1994; in final form 22 August 1994)

SUMMARY Fifteen women with premenstrual syndrome (PMS) and 15 control women were tested twice for thyroid axis measures, once during the follicular and once during the luteal phase of the menstrual cycle. While PMS and control women did not differ in mean hormone values during either phase of the menstrual cycle, PMS women showed significantly greater variability in hormone measures including TSH, T3 uptake, T4, and FTI relative to controls. These findings are consistent with the conceptualization that, for a subset of women with PMS, thyroid axis abnormalities may contribute to their disorder. Additionally, when analyzed in the sample as a whole, the menstrual cycle exerted a significant effect on reverse T3 with greater levels observed in the luteal relative to the follicular phase of the cycle. Keywords--Premenstrual syndrome; Menstrual cycle; Thyroid hormones.

INTRODUCTION PREMENSTRUAL SYNDROME (PMS) is a disorder characterized by the cyclic recurrence of a constellation of emotional, behavioral, and physical symptoms during the iuteal (premenstrual) phase of the menstrual cycle. These symptoms are of sufficient severity to interfere with normal activity and/or interpersonal relationships, thus distinguishing this syndrome from the premenstrual correlates experienced by most women. Some of the more common symptoms associated with this disorder include lethargy, fatigue, depression, irritability, and anxiety. Although these symptoms are clearly time-linked to the menstrual cycle, studies comparing PMS women with controls have failed to obtain evidence for a direct role of reproductive hormones in producing the symptoms (Rubinow et al., 1988; Schmidt et al., 1991). Thus, investigation into other neuroendocrine factors continues. Hypothalamic-pituitary-thyroid (HPT) axis dysregulation is among the factors that continue to be examined. There are numerous commonalities between the physical and affective symptoms associated with PMS and those associated with thyroid disorders (Denicoff et al., 1990). Additionally, the evidence for a high concordance rate between

Address correspondence and reprint requests to: Susan S. Girdler, PhD, Department of Psychiatry, CB 7175, University of North Carolina, Chapel Hill, NC 27599, USA. 395

396

S . S . GIRDLER et al.

PMS and major affective disorders, especially depression (Halbreich & Endicott, 1985), combined with the substantial literature on thyroid dysfunction in major affective illness (e.g., Gold et al., 1981; Kirkegaard & Faber, 1991; Linnoila et al., 1983; Maes et al., 1990; Prange et al., 1984; Stern & Prange, 1993), suggest that HPT axis impairment may play a role in PMS. Despite these observations, there have been relatively few studies to date which have examined thyroid hormones in prospectively confirmed PMS women as compared with controls. In those that have, with one exception (Nikolai et al., 1990), measures have typically been limited to basal thyroid stimulating hormone (TSH) and/ or TSH responses to thyrotropin-releasing hormone (TRH; Casper et al., 1989; Roy-Byrne et al., 1987; Schmidt et al., 1993). These studies, collectively, found no significant group differences in either mean TSH levels or mean TSH response to TRH during either the follicular or luteal phase of the menstrual cycle. Taken together, these data do not suggest that underlying abnormalities in thyroid function characterize women with PMS in any global fashion. There remains the possibility that, for a subgroup of women with PMS, abnormalities in HPT axis function may contribute to their disorder. For example, it has been reported that abnormalities of thyroid function appear with higher than expected frequency in women with prospectively confirmed PMS (Schmidt et al., 1987). Additionally, Roy-Byrne and colleagues (1987) observed significantly greater variability in TSH responses to TRH (both blunted and augmented) in PMS women who were symptomatic at the time of testing compared with controls. This was true during both the follicular and luteal phase of the menstrual cycle. Similarly, Schmidt and associates (1993) observed that in those women with normal basal TSH, 30% of the PMS women showed either blunted or exaggerated TSH responses relative to the controls. A recent case report (Schmidt et al., 1990) described abnormal basal TSH and free thyroxine (T4) levels in a PMS patient. In addition to normalizing basal thyroid hormone levels, thyroid hormone supplementation in this patient resulted in complete remission of her affective premenstrual symptoms. Thus, while these studies do not indicate that PMS is simply a manifestation of a thyroid disorder, they do suggest that for some PMS women thyroid dysfunction may contribute to their premenstrual emotional and/or physical distress, thus warranting further investigation of this issue. The purpose of the present study was to extend the findings of others by examining a broader array of thyroid measures at more than one point in the menstrual cycle in women with prospectively confirmed PMS as compared with control women. First, we hypothesized that if HPT axis dysfunction characterizes a subset of PMS women, then in addition to greater variability of TSH (Roy-Byrne et al., 1987; Schmidt et al., 1993), we would observe greater variability in other HPT axis indices as well in women with PMS. Second, studies investigating changes in thyroid hormone levels across the menstrual cycle in healthy, euthyroid women have been scant. Given the far reaching biological effects of thyroid hormones, we thought it of interest to examine the extent to which the menstrual cycle influences HPT axis variables in both healthy controls as well as in PMS women. METHODS

Subjects Fifteen women with prospectively confirmed PMS and 15 non-PMS controls served as subjects (see diagnostic criteria below). Subjects read a detailed description of the

THYROID AXlS AND PMS

397

protocol, approved by the University of North Carolina Committee on the Protection of the Rights of Human Subjects, and provided informed consent. The PMS women did not differ from the control women in average age (31 and 32 years, respectively), height (65.6 and 65.7 inches, respectively), or weight (134.5 and 134.1 Ib, respectively). In addition, PMS and controls did not differ in average menstrual cycle length (30 and 29 days, respectively). None of the women had taken oral contraceptives for at least 3 mo before testing, all were medication-free, and all agreed to refrain from taking any nonprescription medications for at least 48 h before testing. Additionally, none of the women tested reported themselves to have ever been diagnosed with any thyroid disorder nor had they ever taken any thyroid medication. Finally, although not assessed through structured diagnostic interviews, only women who reported no past or present psychiatric condition were eligable for participation.

Assessment of PMS During an initial screening interview, the PRISM calendar (Reid, 1985) was explained to each subject by a female researcher. This form allows for quantification of the severity of both physical and psychological symptoms that occur over the course of a menstrual cycle. For each symptom experienced, subjects assign a 1 if mild (noticeable but not troublesome), a 2 if moderate (interfering with normal activity), or a 3 if severe (temporarily incapacitating). In addition, the PRISM calendar incorporates measures of life-style impact together with information on life events and the use of medications that may modify symptomatology. Subjects were asked to complete the calendar on a daily basis for two menstrual cycles. To discourage retrospective reporting, subjects mailed back the calendars on a weekly basis. To classify women as having PMS, we included the guidelines set forth by the National Institute of Mental Health Premenstrual Syndrome Workship (1983) together with the American Psychiatric Association's (APA) criteria used for diagnosis of late luteal phase dysphoric disorder as described in the Diagnostic and Statistical Manual of Mental Disorders (DSM-II1-R, APA, 1987). Therefore, to be classified as having PMS, the women in the present study were required to meet all of the following criteria: (a) a 30% increase in symptom severity during the 6 days preceding menses (luteal phase or premenstrual period) compared with Days 5-10 of the cycle; (b) rating of symptoms as moderate or severe (as opposed to mild) during the premenstrual period; (c) moderate to severe symptoms on at least 3 of the 6 premenstrual days; (d) a total of five or more symptoms premenstrually; (e) at least one moderate to severe psychological symptom; and (f) meeting criteria (a)-(e) on two consecutive menstrual cycles. Despite meeting these criteria, women were excluded from participation (n = 3) if visual inspection of a woman's PRISM calendar indicated persistent affective symptomatology throughout the menstrual cycle. Of the 54 women presenting with PMS, only 28% (n = 15) actually met these criteria. The women classified as controls met the following criteria: (a) not completely asymptomatic during the premenstrual period and (b) having mild or moderate symptoms (as opposed to severe) during the premenstrual period with moderate symptoms, however, occurring on fewer than 3 of the 6 premenstrual days. Each woman was tested twice, once during the early follicular phase (Days 4-9) and once during the late luteal phase (Days 24-28, adjusted for individual differences in cycle length) of her menstrual cycle. Cycle phase at first testing was counterbalanced within the groups.

398

S . S . GmDLER et al.

Thyroid Measures Basal thyroid measures included serum levels of the following: 1) thyroid stimulating hormone (TSH), which is released from the pituitary and acts to increase thyroid gland production of thyroid hormones; 2) total thyroxine (TT4), the major secretory product of the thyroid gland and precursor of triiodothyronine (T3); 3) free thyroxine (FT4), of total T4, that which is not bound to proteins; 4) free thyroxine index (FTI), an estimate of free T4; 5) total triiodothyronine (TT3), the metabolically active thyroid hormone and the vast majority of which arises from peripheral deiodination of free T4; 6) free triiodothyronine (FT3), of total T3, that which is not bound to proteins; 7) 3,3o',5crtriiodothyronine (reverse T3, rT3), the metabolically inactive metabolite of free T4 conversion; and 8) T3 uptake (T3U), reflecting the unsaturated thyroid hormone binding sites on proteins. Procedures Blood samples were collected about 30 min following an hour-long psychophysiological stress protocol. Briefly, the stress protocol involved assessment of cardiovascular responses to five standardized laboratory stressors (e.g., mental arithmetic). The battery of stressors was identical for each subject (PMS and controls) and for each test session. See Girdler et al. (1993) for further description of the stress testing procedure and for comparison of cardiovascular responses between PMS women and controls. Owing to the time constraints associated with the cardiovascular/stress protocol, all blood samples were collected between 1800-2000h, with time of day held constant for each subject. Serum TT4, FT4, TT3, FT3, and T3U were measured by radioimmunoassay (RIA) procedures using commercial kits from Becton-Dickinson. The intra- and interassay coefficients of variation were, respectively, 6% and 8% for TT4, 5% and 8% for FT4, 5% and 7% for TT3, 4% and 7% for FT3, and 3% and 4% for T3U. Reverse T3 was measured by RIA using commercial kits from Serono with intra- and interassay coefficients of variation of 8% and 10%, respectively. TSH was measured by immunoradiometric assay using kits from Becton-Dickinson with intra- and interassay coefficients of variation of 3% and 7%, respectively. For each thyroid measure, all samples (i.e., from both groups and for both phases) were randomly assigned positions within a single assay thereby removing any systematic influence that inter-assay variability may exert on the values obtained. Finally, FTI was calculated as the product of TT4 and T3U. Serum progesterone levels were also determined by RIA using commercial kits from ICN Biomedicals, Inc. These were used to confirm that the phase at Days 24-28 was luteal and that ovulation had previously occurred. Women who exhibited serum progesterone levels of ---3.0 ng/ml during Days 24-28 were considered to have ovulated during that cycle. Design and Data Analysis Progesterone analyses revealed that one PMS woman and four control women failed to ovulate. Hence, analyses were conducted on a final sample of 14 PMS women and 11 control women. For the 25 ovulatory women, lack of sufficient serum volume prevented us from analyzing the total 50 possible samples for each HPT-axis measure. Thus, for the various thyroid-axis measures, the total number of samples assayed to determine the respective mean values ranged between 45-50, with loss of assay unsystematically related to group or phase of cycle. Our first analytical strategy was to examine group differences in variability of HPT axis parameters. Owing to our specific a priori hypothe-

THYROID AXlS AND P M S

TABLE I.

MEANS

AXIS FUNCTION WOMEN

(±

VARIANCE)

VARIABLES DURING

FOR SERUM

THYROID

IN PMS AND CONTROL

BOTH PHASES

MENSTRUAL

399

OF THE

CYCLE

Follicular

Luteal

PMS

Controls

PMS

Controls

TSH

2.1 (2.5)*

1.8 (0.5)

2.3 (3.0)*

1.8 (0.6)

TT4

7.4 (0.8)

7.4 (0.9)

7. I ( 1.8)t

7.1 (0.7)

FT4

1.6 (0.04)

1.6 (0.05)

1.6 (0.04)

1.5 (0.05)

FTI

2.7 (0.09)

2.6 (0.1)

2.6 (0.2)+

2.6 (0.09)

TT3

100 (198)

92 (97)

95 (109)

94 (66)

FT3

4.1 (0.2)

3.9 (0.2)

4.0 (0.3)

3.9 (0.7)

rT3$

152 (6816)

111 (3986)

190 (6415)

162 (7732)

T3U

36 (0.8)*

36 (0.3)

36 (1.1)*

36 (0.4)

TSH = thyroid stimulating hormone (~U/ml); TT4 = total thyroxine (/xg/dl); FT4 = free thyroxine (ng/dl); FTI = free thyroxine index (arbitrary units); TT3 = total triiodothyronine (ng/dl); FT3 = free triiodothyronine (pg/ml): rT3 = reverse triiodothyronine (pg/ml); T3U = triiodothyronine uptake (%). * Variance PMS > Controls, p < .05; t Variance PMS > Controls, p < .10: ~+Luteal > Follicular, p < .01.

ses regarding the unidirectional nature of greater variability (i.e., greater variability limited to the PMS women), for each HPT axis measure and for each phase of the menstrual cycle differences in sample variance were analyzed using a one-tailed F-test. Second, to examine group differences in the extent to which the menstrual cycle influences HPT axis function, for each dependent measure a 2(Group) x 2(Phase) within-subject repeated measures analysis of variance (ANOVA) was performed. Where significant Fratios involving interactions were obtained, the Bonferroni posthoc test comparing the group and phase means was employed to examine the source of the interaction. RESULTS

Means and associated variances for each HPT axis measure are presented in Table I separately for the PMS and control women during both the follicular and luteal phases of the menstrual cycle.

400

S . S . GIRDLER et al.

Variability Mean values did not differ significantly in the PMS women compared with controls for any HPT axis measure. However, PMS women showed greater variability in several HPT indices during the menstrual cycle. TSH variance was significantly greater in PMS women relative to controls during both the follicular [F(13, 10) = 5.39, p < .01] and luteal [F(13, 10) = 4.52, p < .025] phases of the cycle. Similarly, variance in T3U was significantly greater in PMS women during both follicular and luteal phases of the cycle IF(13, I0) = 2.96, p < .05; F(13, 10) = 2.94, p < .05, respectively]. Additionally, there was a tendency for luteal phase levels of TT4 and FTI to show greater variability in the PMS women vs. the controls [F(13, 10) = 2.66, p < .10; F(13, 10) = 2.28, p < .10, respectively]. Variability in the other measures did not differ between PMS and controls during either phase of the cycle (p > . 10).

Menstrual Cycle Effects The menstrual cycle exerted a significant effect on rT3 levels in both the PMS and control women [main effect of Phase: F(1, 23) = 12.13, p < .01]; rT3 levels were significantly greater in the luteal compared with the follicular phase (Table I). This menstrual cycle effect on rT3 was significant despite the large degree of between subject variance seen in this measure because, unlike the other thyroid measures, the variability was not different for PMS and control women. In addition to this main effect of cycle phase on rT3, we also obtained a significant Phase × Group interaction IF(l, 23) = 5.67, p < .05]. Subsequent post hoc comparisons of mean rT3 values failed, however, to reveal any significant differences between groups at either phase or between phases within either group. Thus, rT3 values increased significantly from the follicular to the luteal phase of the menstrual cycle when analyzed in the sample as a whole. Additionally, for FT4, although the controls showed a nonsignificant decrease in the mean FT4 level from their follicular to luteal phase while the PMS women showed no change (yielding a marginally significant Group x Phase interaction: F(1, 23) = 4.24, p = .051), the subsequent post hoc tests of means were uniformly nonsignificant. The menstrual cycle did not significantly influence any other thyroid measure in either group of women (p > . 10).

Serum Progesterone Levels For the 11 controls and 14 PMS women exhibiting ovulatory cycles, mean progesterone follicular phase levels, ranges, and standard errors were 0.86 ng/ml (range = 0.24-4.40) -+ 0.37 and 0.53 ng/ml (range = 0.20-1.0) +- 0.9 for the control and PMS women, respectively. During the luteal phase, these mean levels rose to 14.46 ng/ml (range = 3.0-32.0) -+ 3.0 and 15.21 ng/ml (range = 4.40-49.0) _+ 3.4 for the control and PMS women, respectively. Groups did not significantly differ in average serum progesterone levels during either phase of the menstrual cycle (p > . 10). DISCUSSION Consistent with our a priori hypothesis, women with PMS showed significantly greater variability in multiple HPT axis measures relative to control women. In keeping with other reports (Roy-Byrne et al., 1987; Schmidt et al., 1993), while PMS and control women did not differ in mean TSH levels during either phase of the menstrual cycle, we found greater variability in TSH in the PMS women compared with controls during both the follicular and luteal phase of the cycle. Our findings extend those of previous

THYROID AXIS AND PMS

401

investigators, however, in that we obtained evidence for significantly greater variability in other measures as well. For example, during both phases of the menstrual cycle we observed significantly greater variability in T3U, and a tendency towards greater variability in TT4 and FTI during the luteal phase of the cycle. This greater variability in multiple thyroid function measures further supports Roy-Byrne and colleagues (1987) suggestion that, while PMS women as a group cannot be characterized by thyroid disease, there may be a subset of PMS women (e.g., those outliers contributing to the enhanced variance) for whom HPT axis abnormalities underlie their condition. Consistent with other disorders, it is likely that PMS is also one of heterogeneous etiology. Thus, while the existing literature on thyroid function in PMS does not warrant routine thyroid treatment for women afflicted with this disorder, the existing data do suggest that routine thyroid screening may be an appropriate course of action in women presenting with PMS. Further confirmation of the hypothesis that the greater variability in thyroid measures observed in PMS women reflects a subgroup with HPT axis dysfunction would come from studies demonstrating consistency in thyroid measures over time in the PMS women showing abnormal or outlying values. With respect to menstrual cycle effects on HPT axis measures, significant effects were limited to the metabolite reverse T3, for which we observed greater levels in the luteal compared with the follicular phase of the cycle when analyzed in the sample as a whole. It is not clear why rT3 levels would shift across the menstrual cycle. One possibility, although speculative, is that the female gonadal steroids alter the metabolic pathway involved in the conversion of T4, such that elevated levels (e.g., during the luteal days of testing in the present study) preferentially increase the conversion of T4 to rT3 as opposed to T3. This interpretation is limited, however, by our lack of finding for a significant effect of the menstrual cycle on T3 measures, although, at least for the PMS women, the means are in the appropriate direction li.e., lower levels luteally). To the best of our knowledge only one other study has assessed an array of thyroid measures in normally cycling, euthyroid women during the menstrual cycle (Hegedus et al., 1986). These investigators did not, however, measure rT3. In that study, Hegedus and associates (1986) reported that the menstrual cycle did not alter any of the thyroid function variables that they assessed, although it did significantly influence thyroid size. Before we conclude that the menstrual cycle exerts minimal or no effect on thyroid hormone measures, it should be kept in mind that both our study and that of Hegedus and colleagues (1986) are limited by their relatively small sample size and inherent loss of statistical power. Our findings for a significant effect of cycle phase on rT3 levels, combined with that for increased thyroid gland size in the luteal phase (Hegedus et al., 1986), and the well documented increase in thyroid binding globulin (TBG) that occurs with pregnancy, estrogen replacement therapy, and oral contraceptives use (Wilson & Foster, 1991), suggest that menstrual cycle physiology may impact HPT-axis function. Thus, future studies investigating HPT-axis variables in larger samples of healthy, euthyroid women, at numerous points throughout the cycle, and examining multiple measures of thyroid function, including TBG, are certainly still warranted as any potential cycle phase effects may have relevance for thyroid panel interpretation in women. In summary, we obtained evidence for greater variability of multiple thyroid-axis measures in women with prospectively confirmed PMS relative to control women. These data are consistent with previous reports suggesting that for a subset of women with PMS,

402

S. S. GIRDLERet

al.

underlying thyroid axis abnormalities m a y contribute to their disorder and underscore the need to conceptualize PMS as a disorder of heterogeneous etiology. Additionally, we obtained evidence that, for all w o m e n , the luteal phase of the menstrual cycle m a y be associated with elevated levels of reverse T3.

Acknowledgments:The authors are grateful to Dr. Arthur J. Prange, Jr. for his critical review of the manuscript and The Endocrine Assay Laboratory of the NIMH funded Mental Health Clinical Research Center, University of North Carolina, Department of Psychiatry for performing all thyroid hormone assays. This research was in part supported by National Institute of Mental Health Grants MH09885 and MH3312715 and by National Instititues of Health Grant RO1 HL31533.

REFERENCES American Psychiatric Association (1987) Diagnostic and Statistical Manual of Mental Disorders, 3rd Edition, Revised. APA, Washington, DC. Casper RF, Ptel-Christopher A, Powell A (1989) Thyrotropin and prolactin responses to thyrotropin-releasing hormone in premenstrual syndrome. J Clin Endocrinol Metab 68:608-612. Denicoff KD, Joffe RT, Lakshmanan MC, Robbins J, Rubinow DR (1990) Neuropsychiatric manifestations of altered thyroid state. Am J Psychiatry 147(1):94-99. Girdler SS, Pedersen CA, Stern RA, Light KC (1993) Menstrual cycle and premenstrual syndrome: Modifiers of cardiovascular reactivity in women. Health Psychol 12(3): 180-192. Gold MS, Pottash ALC, Extein I (1981) Hypothyroidism and depression: Evidence from complete thyroid function evaluation. JAMA 245(19): 1919-1922. Halbreich U, Endicott J (1985) Relationship of dysphoric premenstrual changes to depressive disorders. Acta Psychiatrica Scand 71:331-338. Hegedus L, Karstrup S, Rasmussen N (1986) Evidence of cycle alterations of thyroid size during the menstrual cycle in healthy women. Am J Obstet Gynecol 155:142-145. Kirkegaard C, Faber J (1991) Free thyroxine and 3,3o-,5o--triiodothyronine levels in cerebrospinal fluid in patients with endogenous depression. Acta Endocrinol (Copenh) 124:166-172. Linnoila M, Cowdry R, Lamberg B-A, Makinen T, Rubinow D (1983) CSF triiodothyronine (rT3) levels in patients with affective disorders. Biol Psychiatry 18(12):1489-1492. Maes M, Vandewoude M, Schotte C, Martin M, Blockx P (1990) Suprresive effects of dexamethasone on hypothalamic-pituitary-thyroid axis function in depressed patients. J Affect Disord 20:55-61. National Institute of Mental Health (1983) Proceedings of the National Institute of Mental Health Premenstrual Syndrome Workshop. NIMH, Bethesda, MD. Nikolai TF, Mulligan GM, Gribble RK, Harkins PG, Meier PR, Roberts RC (1990) Thyroid function and treatment in premenstrual syndrome. J Clin Endocrinol Metab 70:1108-1113. Prange AJ Jr, Loosen PT, Wilson IC, Lipton MA (1984) The therapeutic use of hormones of the thyroid axis in depression. In: Post RM, Ballenger JC (Eds) Neurobiology of Mood Disorders. Williams & Wilkins, Baltimore/London, pp 311-322. Reid RL (1985) Premenstrual syndrome. In: Leventhal JM, Hoffman JJ, Keith LG, Taylor PJ (Eds) Obstetrics, Gynecology and Fertility. Yearbook Medical Publishers, Chicago, pp 4-57. Roy-Byrne PP, Rubinow DR, Hoban MC, Grover GN, Blank D (I 987) TSH and prolactin responses to TRH in patients with premenstrual syndrome. Am J Psychiatry 144:480-484. Rubinow DR, Hoban C, Grover GN, Galloway DS, Roy-Byrne P, Andersen R, Merriam GR (1988) Changes in plasma hormones across the menstrual cycle in patients with menstrually related mood disorders and in control subjects. Am J Obstet Gynecol 158:5-11. Schmidt PJ, Grover GN, Roy-Byrne PP, Rubinow DR (1993) Thryoid function in women with premenstrual syndrome. J Clin Endocrinol Metab 76:671-674. Schmidt P J, Khan RA, Rubinow DR (1987) Thyroid function in premenstrual syndrome (correspondence). New Engl J Med 317(24):1537-1538. Schmidt PJ, Nieman LK, Grover GN, Muller KL, Merriam GR, Rubinow DR (1991) Lack of effect of induced menses on symptoms in owmen with premenstrual syndrome. New Engl J Med 324:1174-1179.

THYROID AXIS AND PMS

403

Schmidt PJ, Rosenfeld D, Muller KL, Grover GN, Rubinow DR (1990) A case report of autoimmune thyroiditis presenting as menstrual related mood disorder. J Clin Psychiatry 51:434-436. Stern RA, Prange AJ Jr (1993) Neuropsychiatric aspects of endocrine disorders. In: Kaplan HI, Sadock BJ (Eds) Comprehensive Textbook of Psychiatry, 6th Edition. Williams & Wilkins, Baltimore/London, pp 663-677. Wilson JD, Foster DW (Eds) (1991) Williams' Textbook of Endocrinology, 8th Edition. Saunders, Philadelphia.