- Email: [email protected]
Pineal and adrenal function before and after refeeding in anorexia nervosa
216
BIOL PSYCHIATRY 1991;30:2|6-224
Pineal and Adrenal Function Before and After Refeeding in Anorexia Nervosa Sidney H. Kennedy, Gregory M. Brown, Gail McVey, and Paul E. Garfinkel
Nine female subjects with anorexia nervosa fAN) were studied when emaciated (mean 72% of average body weight), and after refeeding (mean 85% of average body weight). They were compared to 9 individuali.v age-matched female control subjects. On each occasion blood was sampled for serum melatonin and plasma cortisol through the night, and urine was collected over 24 hr to measure su~atoxy melatonin levels. The AN group did not differ in their level of depression before and after weight gain. There were no significant differences in serum melatonin values ~nong the patient group before or after weight gain ~nd the control group. Levels of urinary sulfatoxy melatonin were also significantly higher in nighttime compared to daytime samples both before and after weight gain. Plasma cortisol values were significantly elevated in the emaciated state and this was accounted for by higher cortisol levels at 9, lO, I l, and 12 eu and at 6 compared with the weight restored state and to controls. This study suggests that pineal activit3, in patients with AN is not altered by chronic changes in weight, and is not closely associated with changes in cortisol. Underweight patients with anorexia nervosa (AN) display a variety of psychological characteristics, including disturbances in attitudes and feelings about their bodies and in self-esteem. They also demonstrate a variety of neuroendocrine abnormalities, including amenorrhea. A deficiency in the amount or secretion pattern of gonadotropin-releasing hormone and an increased output of corticotropin-releasing factor, ~socia~d wiLh hypercortisolism, have both been implicated in this endocrine dysregulation (Brown et al 1977; Gold et al 1986). Melatonin is secreted from the pineal gland primarily in darkness, under direct sympathetic noradrenergic control, and appears to influence seasonal breeding patterns in mammals (Tamarkin et al 1985). However, its relationship to the human reproductive cycle is less clear (Wetterberg et al 1976; Brzezinski et al 1988). Also, in humans, melatonin levels are highest prepubertally and decline with aging (Waldhauser and Steger 1986). Given the similarities between underweight patients with AN and prepubertal individuals, it might be expected that nocturnal melatonin levels would also be elevated in AN. This idea is strengthened by findings that starvation in animals is associated with
From the Department of Psychiatry. Toronto General Hospital, Toronto, Ontario, Canada. Address reprint requests to Dr. S. H Kennedy, Department of Psychiatry, Toronto General Hospital. 200 Elizabeth Street. /'oronto, Ontario M5G 2C4, Canada. Received June 19, 1990; revised December 20, 1990. © 1991 Society of Biological Psychiatry
0006-3223/91/503.50
Pineal and Adrenal Function in Anorexia
Bt~ ~YCmATR't'
217
elevated melatonin levels (Chik et al 1987). On the other hand~ because undemumtion is associated with reduced activity in the sympathetic ner'¢ous system (Kaplan et M 1989: Kennedy and Heslegrave 1989), reduced nocturnal pineal activity ~ g h t be predicted in AN. Though there is general agreement that underweight AN sub~cts display elevated cortisol levels (Walsh et al 1978: Doerr et al 1980), there is less consensus about ~latonin levels in AN. Both normal (Dalery et a~ 1985: B ~ o w s k a et al 1986: Kennedy et M 1989a) and elevated (Brambilla et al 1988) levels of serum melatonin have been reI~rted. in the only study using a repeated measure design, B e n et M (1988) fourM ~ change in the urinary metabolite of melatonin, sulfatoxy mehtonm (aMT6s), following refeeding. Low nocturnal levels of melatonin have also been reported in patients with major depression (Beck-Friis et al 1985: Brown et al 1985L and ~is has been ~stulated as an index of decreased noradrenergic activity (Lewy 1984). In two previous studies ~Kennedy et al 1989a, l ~ ) , there were no significant differences in nocturnal serum melatonin or aMT6s levels between AN patients and control subjects. However, the wide intersubject variability of melatonin levels (Arato et al 1985) may have obscured abnormal findings caused by low weight or by ~ e presence of AN itself. A subgroup of these AN patients completed the same overnight sampling protocol after a period of refeeding and weight gain. This lo~gitudinal strategy using intrasubject comparisons was designed to permit a better controlled examination of the influence of weight gaip on nocturnal serum melatonin, plasma comsol, and u r i n ~ sulfatoxy melatonin. An age-matched female control group was included for comparison. We hypothesized that underweight AN patients would show elevated noctumM levels of serum melatonin, urinary sulfatoxy melatonin, and plasma cortisoi; all of which would decrease following weight gain.
Methods
Subjects Nine female subjects who met DSM-III-R criteria for AN and provided written mformed cnn~rtt nartieinated in the study on two occasions, before and after weight l~ain. They were inpatients at the Programme for Eating Disorders, The Toronto Hospital, who were treated in a controlled environment in which meals were supervised, calories were adjusted weekly to promote a weight gain of I kg/week, and no exercise was permitted d,lring the admission (Kerr et al 1987). Control subjects were 9 female volunteers, individually matched for age to within 5 years, who were free of major psychiatric illness according to the same structured clinical interview and who gave written informed consent. A detailed description of patient and control characteristics is presented in Table i. Two of the 9 AN patients h ,,a k,~,,, actively involved in binge eating and purging prior to admission. No subjects admitted to laxative or diuretic misuse and 4 met criteria concurrently for major depression (MD) both before and after refeeding. Sampling was carried out in Lhe first instance after 7 days of a standard 1500 kcal/day hospital diet, when the acute effects of nutritional deprivation were felt to be reversed. The time of the second sampling procedure was 4-12 weeks later and in all cases occurred during the last 7 days of admission. On this occasion, no subjects were believed to be binge eating, purging, or exercising excessively. None of the 9 subw v . . w w . . =
~
.
.
.
.
.
E
.
.
.
.
.
.
.
.
218
BIOL PSYCHIATRY 1991;30:216-224
S.H. Kennedy et al
Table I. Demographic, Weight, and Mood Variables in Pa,.ients with Anorexia Nervosa Before and After Weight Gain and in Age-Matched Control Subjects*
Age (yr) I. 22 b (22) 2. 32 ~37) 3. 29b (28) 4. 24 (25) 5. 31 (26) 6. 27 (27) 7. 31 (26) 8. 22 (20) 9. 26 (23) 27.2 (28) ~
Duration of illness
Duration of ameba
Major depression
%MPMW
(yr)
(yr)
(DSM-III-R)
72.3(100.9) 96. ! 58.7 (100.3) 76.9 81.5 (100.9) 86.9 85.8 (87.6) 95.5 75.1 (97.7) 79.4 70.6 (93.8) 91.4 65.0(!16.0) 68.3 59.2 (113.3) 80.0 79.4 (!23.0) 90.0 72.0 (103.7) 85.0
6
9
6
7
4
4
3
3
15
5.5
14
3.5
15
4
7
3
4
3
8.2
4.6
Yes Yes No No No No Yes Yes No Yes Yes No No No No Nc Yes Yes
HRSD
BDI
33 (00) I/7 19 (06) l0 i l (02) O0 2777 (IX)) 29 !7 (05) 20 13 (01) 07 08 (03) !! 08 (03) 09 21 (01) 24 1777.4(2.3) 14.1
38 (00) 2777 31 (00) 06 - - (00) 0777 34 (01) 14 36 (0777) 26 09 (00) 08 00 (01) 03 02 (00) 00 32 (04) 3777 22.8 (I.4) 14.2
%MPMW is percent of matched population mean weight (Metropolitan Life 1983L ~Details of age-matched control subjects are presented in parentheses. ~oncurrent binge eating and vomiting within previous 3 months. ' Mean values.
jects had resumed menses by the second sampling point and no pharmacological treatments had been prescribed.
Measures T&hsa~, ~ ~. ~le*¢n*-~, ~ l'ql-'.,l,,.nl • ~'lgltl~tblgll~ltli~L/~ ~.ilill~.O.l •IIlLGIVIGW "" " ~" t' ."t'o") . . . . :.... for "~"'" t.,o~-m-r~
(Spi tzer et al 1 ~ ) was administered on both occasions by a trained research clinician before the sampling procedure. In addition, depressive symptoms were assessed using the 2 l-item Hamilton Rating Scale for Depression (HRSD) (Hamilton 1960), and the Beck Depression Inventory (BDI) (Beck et al 1961).
Assays Urinary aMT6s and serum melatonin (MT) were assayed using commercially available techniques (CIDtech Research Inc., Hamilton, Ont., Canada). The radioimmunoassay (RIA) technique for aMT6s was developed by A!dhous and Arendt (1988). Inter- and intra-assay coefficients of variation were both less than 10% at 10 pg/ml. The assay limit of detection was 2 pg per tube. The RIA for serum MT was developed by Brown et al (1983). Intra- and interassay
Pineal and Adrenal Function in Anorexia
atOL PSYCHt~V
219
Table 2. Plasma Cortisol Values at Different, Time Points for Anorexia Nervosa Control Groups Hour
Time 1 {mean ___ SD)
Time 2 {mean _ SD)
Coatmt (mean -,- SD}
8 PM 9 PM l0 ~ t I I ~t 12 ~ 1 AM 2 AM 3 AM 4 AM 5 AM 6 AM
185. I 194 ! 246.0 17502 130.9 |04.| 103.7 161.8 231.7 290.8 382.3
94°4 9(}.5 97. | 7'0.7 63.3 6&2 |0S.6 |59.5 186.9 251.4 321.0
1 |2°5 (35°8) ! |7~5 (55.3} t t-L9 (~o9) ~ . 9 (4!.8~ 86.4 (36.0} 83~6 (49.3) [ ~ . | (35.2) [52.6 (88.4) I7l,.4 (8[.6) |86.8 (86.2) ~ 8 . 4 (95.0)
{ 131.6) {! 16.2) {155.8~ ~9&5~ {63.5) {51o0) {37. !) {7708~ (160.5) ~131.9) (i05.9)
(29.9) (40.8) (53.2) (36°6} (29°6} {58.1) (6[ .9} {98.3) (113.2~ (|06.5) (~.[)
coefficients of variation were both less than 10%, with 5 pg/ml being tbe lowest limit of detection. Plasma cortisol was measured using an in-house PdA (Wong et al 1979).
Procedure On each occasion, the study was carried out at the Clinical Investigation Unit of the Toronto Hospital. An intravenous catheter was inserted into an antecubital vein, which was kept pate at with intravenous saline. Subjects were placed on a van~lyl mandelic acid-free diet for 48 hr prior to each sampling procedure, h e y were given 2 hr to acclimatize, and hourly blood samples were drawn by a research nurse between 8 PM and 6 AM. Two separate urine collections were obtained from 6 PM t o 6 AM and from 6 AM tO 6 PM the following day. On each occasion, subjects wore an overnight flight mask from 10 PM tO 6 AM tO ensure consistent light conditions.
Data Analysis For plasma cortisol and serum melatonin there were I l hourly sampling times and for urinary aMT6s there were two sampling times (day and night). An analysis of variance with repeated measures for patients at two time poims and for matched controls (using time and weight as within-subject variables) was carried out on plasma cortisol and serum MT data. For urinary aMT6s values, multiple t-tests for day/night and between group comparisons were carded out with Bonferroni Correction.
Ec:ults
Weight and Depressive Symptoms For AN patients, there was a significant increase in weight between sampling points (73 _+ 9.0%-84 _+ 19.7%) using the Metropolitan Life Insurance (1983) tables as the reference point (t = - 4 . 0 df = 8, p < 0.01). Using SCID criteria I patient switched from positive to negative majcr depression status and I switched in the reverse direction
220
BIOL PSYCHIATRY !~1;30:216-224
S.H. Kennedy et al
Table 3. Melatonin Values at Different Time Points for Experimental and Control Groups of Subjects Hour
8 9 10 II 12 ! 2 3 4 5 6
Time I (mean ± SD)
PM i'M PM PM Phi AM AM AM AM AM AM
24.4 24.5 30.~ 45.7 62.3 55.5 60.4 60.3 60.3 50.4 46.9
(30.2) (21.5) (24.1) (50.1) (61.3) (49.9) (55.6) (38,7) (38.7) (37.5) (38.I)
( ~
Time 2 "4" SD)
22.3 (15.6) 27.7 (19.8) 49.6 (36. I) 54.9 (26.5) 49.4 (21.3) 44.2 (21.O~ 45.8 (21.3) 62.8 (42,5) 66.9 (35.2) 53.8 (23,5) 48.2 (22,9)
Control (mean _ SD)
8.0 (5.6) 14.8 (!0.5) 22.4 (18.8) 36~2 (22.1) 42.7 (19.6) 51.6 (32.4) 48,7 (27,8) 50.8 (31,6) 51.3 (33.9) 43,5 (31,8) 35,0 (22.4)
between sampling times. The remaining 7 subjects did not change with regard to depression status and the mean HRSD score.s were 17.4 and 14. !, respectively, before and after refeeding. Patients also were at a significantly lower weight than controls on both admission (t = 4.72, df = 8, p < 0.01) and discharge (t = 3.53, df = 8 , p < 0.01), and displayed significantly higher levels of depression as measured by HRSD and BDI [HRSD t = 8.94, df = 8, p < 0.001; BDI t = 15.8, df = 7 , p < 0.0001 (before refeeding); HRSD t = 9.98, df = 8, p < 0.001; BDI t = 12.8, df = 8, p < 0.0001 (a~ter refeeding)].
Plasma Cortisol Levels There were significant main effects for time IF = 19.86, df = (2,14), p < 0.0001] and group [F = 5.26, df = (2,14), p < 0.02], with no significant interaction effect IF = 1.49, df = (2,14), NS]. Planned comparisons revealed significant elevations of plasma cortisol at 9, 10, 11, and 12 PM and at 6 AM in the low weight condition compared with the weight-restored condition and to the contro ~-group (p < 0.05). Mean hourly values for plasma cortisol between 8 PM and 6 AM are illustrated in Figure I.
Serum Melatonin Levels There was a significant main effect for time [F = 8.12, df = (2,14), p < 0.0001] but not for group ( F = 0.9, df = 2, p < 0.77), and no significant interaction effect between time and group ( F = 1.20, df = 10, p < 0.30). Mean values for serum melatonin are illustrated in Figure 2.
Urinary aMT6s Levels There was a significant difference between daytime and nighttime output of aMT6s, both before refeeding (day = 8.44 I~g -+ 6.5 p.g versus night = 32.4 p,g _+ 22.8 Ixg, t = - 3 . 7 3 , df = 8, p < 0.005) and afterwards (day = 9.73 lxg -+ 7.7 lxg versus night = 31.6 Ixg -+ 24.7 ~g, t = - 3 . 3 0 , df = 8, p < 0.01). However, no differences in daytime
Pineal and Adrenal Function in Anorexia
a~
~¥Ch'IA~V
~l
---*- C ~ ¢ ~ S
Cortisol Levels 40O
9 J
300-
.
~
.
J
II
1003-
-
-
-o-
-
-
-<~
~
~
~ ~:~"
20
21
/"
22
---
-C-
-
. . --
-- "-C~
24
23
J'
"#
at oa Hours
~
~
05
o6
* s~n~.w~y ainemntat i ~ . ~
Figure 1. Plasma cordsol from 8 PM to 6 AM.
or nighttime aMT6s levels were observed between anorexic and control subjec~ (control day = 12.44 p.g _ 10.3 IJ.g versus night = 20.19 izg ~ 9.9 v-g).
Discussion The three principal findings in this study are (1) that MT outpm does not differ between emaciated and weight-restored AN patients or (2) between AN patients and matched controls, and (3) in contrast, co~isol levels are elevated m the emaciated AN condition compared with both the weight-restored AN state and the control subjects. The first two conclusions are in agreement with Beam et al (1988) who reposed no change in aMT6s
brae 1
-3-
time 2
~
controls
~-.~.erurnMT Levels 70 60
/
l
/z
j"--~_-
~
+
o 20
21
22
23
24
01
02
03
04
Hours
Figure 2. Serum melatonin output from 8 PM
tO
6 AM.
05
06
322
BIOL PSYCHIATRY 1991:30:216~224
S.H. Kennedy et ai
excretion after weight gain in patients with AN, but not with Brambilla et al (1988L who reported elevated MT levels in emaciated anorexic patients compared with controls. However, it is of interest to note in our study that 3 of the 4 subjects with the lowest percent of average weights when initially tested had the highest values for both nighttime serum MT and for urinary aMT6s (subjects 5, 6, and 7). Several explanations are possible: ( I ) elevation in MT secretion only occurs beyond a certain degree of weight loss; (2) the rate of weight loss may influence MT output, and only in those individuals with the most rapid weight loss will elevated levels occur; (3) the effect of starvation may be m o ~ important than the actual weight loss, so that only those subjects who have failed to reverse the most acute effects of starvation will show elevated levels. It is atse apparent that our subjects showed a moderate level of depression, as assessed by the HRSD and the BDI. Several investigators have reported reduced levels of MT in depression (Wetterbere et al 1979; Brown et a! 1985; Miles and Philbrick 1988) and a lower level of MT in the depressive phase compared with the manic phase of bipolar affective disorder (Kennedy et al 1989b). It is therefore possible that the effect of depression in our subjects to s ~ , ~ c~,~e~t exerted ~n effect in the opposite direction and obscured any effect from weight loss. The third finding of an elevation in cortisol levels during the earlier part of the night at low weight is in agreement with previous reports (Boyar et al 1977; Doerr et al 1980) and reflects a reduced rate of cortisol b~eakdown as the body conserves energy OValsh et al 1978). Treasure et al (1985) also reported a linear relationship between the fall in basal cortisol levels and percentage weight gain in anorexics during refeeding. Thus, reports of a reciprocal relationship between MT and cortisol secretion, and a potential relationship between amenorrhea and abnormal elevation of MT output in AN are not supported. Further studies are indicated in patients with AN to evaluate the acute effects of starvation rather than chronic low weight on MT and aMT6s ou .tput. Measurement of additional urinary metabolites of MT would address the speculation that AN patients may excrete MT and its metabolites in altered proportions.
References Aidhous ME, Arendt J (1988): Radioimmunoassay for 6-sulphatoxymelatonin in urine using an iodinated tracer. Ann Ctin Biochem 25:298-303. Arato M, Grof E, Grof P, Laszlo 1, Brown GM (1985): Reproducibility of the overnight melatonin secretion pattern in healthy men. In Brown GM, Wainwright SD (eds), The Pineal Gland: Endocrine Aspects. Toronto: Pergamon Press, pp 277-282. Baranowska B, Soszynski P, Mlsiorowski W, Doroket W, Witkowski M (1986): Circadian melatonin rhythm in women with hypothalamic amenorrhea. Neuroendocrinol Lett 8:295-300. Beam J, Treasure J, Murphy M, et al (1988): A study of sulphatoxymelatonin excretion and gon=dotropin status during weight gain in anorexia nervosa. Br J Psychiato, 152:372-376. Beck AT, Ward CH, Mendcison M, Mock J, Erbaugh J (1961): An inventory for measuring depression. Arch Gen P.D'chiatrb.' 4:561-57 I. Beck-Friis J, Kjellman BF, Aperia B, et al (1985): Serum melatonin in relation to clinical variables in patients w!rh major depressive disorder and a hypothesis of a low melatonin syndrome. Acta Ps3'chiatr Scand 7 ! :319-330. Boyar RM, Hellman LD, Roffwang H, et al (1977): Cortisol secretion and metabolism in anorexia nervosa. N Engl J Med 296: i90-193.
Pineal and Adrenal Function in Anorexia
s~oL ~VCa~a'mV 199|~.~:216-224
223
Brarnbilla F, Fraschini F, Esposti G, et al (1988): Melatonin circadian rhythm in anorexia netwosa and obesity. Psychiat~, Res 23:267-276. Brown GM, Garlinkel PE, Jeuniewic N, Moldofsky H, Stancer HC (1977): ~ r i n e anorexia nervosa. In Vigersky RA (ed)o Anorexia Nervosa~ New, York: Raven ~ , 135.
profiles m ~ [23-
Brown GM, Gram l.J, Pulido O, Bums TG, Ni|es LP, Snieckus V (1983): Application of immunologic techniques to the study of pineal indo|ea|kylamines. In Reiter RJ (ed), Pineal Research Reviews !. New York: Alan R. Liss, pp 207-246. Brown R, Kocsis JH, Caroff S, et al (1985): Differences in nocturnal melatonin secretion between melancholic depressed patients and control subjects. Am J Ps'chiatry 142:81 I--816. Brzezinski A, Lynch HJ, Seibel MM, et al (1988): The circadian rhythm of plasma melatonin during the normal menstrual cycle and in amenorrheic women. J Ctin Endocrinol Metab 66:891895. Chik CL, Ha AK, Brown GM (1987): Effect of food restriction on 24 hour serum and pineal melatonin content in male rats. Acta Endocrinol i 15:507-513. Dalery J, Claustrat B, Brun J, de Villard LE (1985): Plasma melatonin and cortisol levels in patients with anorexia nervo~. Nueroendocrinol Len 7:159-164. Doerr P, Fichter M, Pirke KM, Lund R (1980): Relationship between weight gain and hypothalarnic pituitary adrenal function in patients with anorexia nervosa. J Steroid Biachem | 3:529-537. Gold P, Loriaux D, Roy A, et al (1986): The corticotropin releasing factor ~st: Implications for the diagnosis and pathophysiology of hypercortisolism in depression and C u s ~ g ' s disease. N Engl J Med 314: i 330-1335. Hamilton MA (1960): A rating scale for depression. J Neural Neurosurg Psychiatry 23:46-62. Kaplan AS, Garfinkel PE, Brown GM (1989): The DST and TRH test in bulimia nervosa. Br J Psychiatry 154:86-92. Kennedy SH, Heslegrave ILl (1989): Heart regulation in bulimia nervosa. J Psychiatr Res 23:267273. Kennedy SH, Garfinkel PE, Parienti V, Costa D, Brown GM (1989a): Changes in melatomn levels but not cortisol levels are associated with depression in patients with eating disorders. Arch Gen Psychiatry 46:73-78. Kennedy SH, Tighe S, MeVey G, Brown GM (1989b): Melatonin and cortisol "switches" d ~ g mania, depression, and euthynfia in a drug-free bipolar patient. J Nerv Merit Dis 177:300-303. Kennedy SH, Brown GM, Garfinkel PE, McVey G, Costa D, Parienti V (1990): Sulphatoxy melatonin: An index of depression in anorexia nervosa and b u ~ a nervosa. Psychiatry Res 32:221-227. Kerr A, Kennedy SH, Stem D (1987): Inpatient treatment for anorexia nervosa and bulimia Br Rev Anorexia Bulimia. 2:5-16. Lewy AJ (1984): Human melatonin secretion It: A marker for the c i r c ~ system and the affects of light. In Post RM, Ballenger JC (eds), Neurobiology of Mood Disorders. Balt~ore: Williams & Wilkins, pp 2 ! 5-226. Metropolitan Life Insurance Company (1983): 1983 height and weight tables. Stat Bull 64:3-9. Miles A, Philbrick DRS (1988): Melatonin and psychiatry: A review. Biol Psychiatry 23:405--425. Spitzer RL, Williams JBW, Gibbon M, First MB (1988): Structured Clinical Interview for DSMIll-R-Patient Version. Biometrics Research Department, New York State P s y c ~ c Institute, New York, NY. Tamarkin L, Baird CJ, Alraeida OFS (1985): Melatonin: A coordinating signal for reproduction? Science 227:714-720. Treasure J, Wheeler MJ, Salieh B, Russell GFM (1985): Anorexia nervosa and the adrenal: The effect of weight gain. J Psychiatr Res 19:221-225.
224
BIOL PSYCHIATRY 1991:30:216-224
S.H. Kennedy et al
Waldhauser F, Steger H (1986): Changes in melatonin secretion with age and pubescence. J Neural Transm 21(Suppl): 183-197. Walsh TT, Katz JL, Levin J, et al (1978): Adrenal activity in anorexia nervosa. Psychosom Med 40:499-506. Wetterberg L, Arendt J, Paunier L. Sizonenko PC, Donselaar W, Heyden T (1976): Human serum melatonin changes during the menstrual cycle. J Clin Endocrinol Metab 42:185-188, Wetterberg L, Beck-Friis J, Aperia B, Petterson V (1979): Melatonin: Cortisol ratio in depression. Lancet 2:1361. Wong PY, Mee AV, Ho FFK (1979): A direct radioimmunoassay of serum cortisol with in-house 125 l-tracer and preconjugated double antibody. Clin Chem 25:914-917.
BIOL PSYCHIATRY 1991;30:2|6-224
Pineal and Adrenal Function Before and After Refeeding in Anorexia Nervosa Sidney H. Kennedy, Gregory M. Brown, Gail McVey, and Paul E. Garfinkel
Nine female subjects with anorexia nervosa fAN) were studied when emaciated (mean 72% of average body weight), and after refeeding (mean 85% of average body weight). They were compared to 9 individuali.v age-matched female control subjects. On each occasion blood was sampled for serum melatonin and plasma cortisol through the night, and urine was collected over 24 hr to measure su~atoxy melatonin levels. The AN group did not differ in their level of depression before and after weight gain. There were no significant differences in serum melatonin values ~nong the patient group before or after weight gain ~nd the control group. Levels of urinary sulfatoxy melatonin were also significantly higher in nighttime compared to daytime samples both before and after weight gain. Plasma cortisol values were significantly elevated in the emaciated state and this was accounted for by higher cortisol levels at 9, lO, I l, and 12 eu and at 6 compared with the weight restored state and to controls. This study suggests that pineal activit3, in patients with AN is not altered by chronic changes in weight, and is not closely associated with changes in cortisol. Underweight patients with anorexia nervosa (AN) display a variety of psychological characteristics, including disturbances in attitudes and feelings about their bodies and in self-esteem. They also demonstrate a variety of neuroendocrine abnormalities, including amenorrhea. A deficiency in the amount or secretion pattern of gonadotropin-releasing hormone and an increased output of corticotropin-releasing factor, ~socia~d wiLh hypercortisolism, have both been implicated in this endocrine dysregulation (Brown et al 1977; Gold et al 1986). Melatonin is secreted from the pineal gland primarily in darkness, under direct sympathetic noradrenergic control, and appears to influence seasonal breeding patterns in mammals (Tamarkin et al 1985). However, its relationship to the human reproductive cycle is less clear (Wetterberg et al 1976; Brzezinski et al 1988). Also, in humans, melatonin levels are highest prepubertally and decline with aging (Waldhauser and Steger 1986). Given the similarities between underweight patients with AN and prepubertal individuals, it might be expected that nocturnal melatonin levels would also be elevated in AN. This idea is strengthened by findings that starvation in animals is associated with
From the Department of Psychiatry. Toronto General Hospital, Toronto, Ontario, Canada. Address reprint requests to Dr. S. H Kennedy, Department of Psychiatry, Toronto General Hospital. 200 Elizabeth Street. /'oronto, Ontario M5G 2C4, Canada. Received June 19, 1990; revised December 20, 1990. © 1991 Society of Biological Psychiatry
0006-3223/91/503.50
Pineal and Adrenal Function in Anorexia
Bt~ ~YCmATR't'
217
elevated melatonin levels (Chik et al 1987). On the other hand~ because undemumtion is associated with reduced activity in the sympathetic ner'¢ous system (Kaplan et M 1989: Kennedy and Heslegrave 1989), reduced nocturnal pineal activity ~ g h t be predicted in AN. Though there is general agreement that underweight AN sub~cts display elevated cortisol levels (Walsh et al 1978: Doerr et al 1980), there is less consensus about ~latonin levels in AN. Both normal (Dalery et a~ 1985: B ~ o w s k a et al 1986: Kennedy et M 1989a) and elevated (Brambilla et al 1988) levels of serum melatonin have been reI~rted. in the only study using a repeated measure design, B e n et M (1988) fourM ~ change in the urinary metabolite of melatonin, sulfatoxy mehtonm (aMT6s), following refeeding. Low nocturnal levels of melatonin have also been reported in patients with major depression (Beck-Friis et al 1985: Brown et al 1985L and ~is has been ~stulated as an index of decreased noradrenergic activity (Lewy 1984). In two previous studies ~Kennedy et al 1989a, l ~ ) , there were no significant differences in nocturnal serum melatonin or aMT6s levels between AN patients and control subjects. However, the wide intersubject variability of melatonin levels (Arato et al 1985) may have obscured abnormal findings caused by low weight or by ~ e presence of AN itself. A subgroup of these AN patients completed the same overnight sampling protocol after a period of refeeding and weight gain. This lo~gitudinal strategy using intrasubject comparisons was designed to permit a better controlled examination of the influence of weight gaip on nocturnal serum melatonin, plasma comsol, and u r i n ~ sulfatoxy melatonin. An age-matched female control group was included for comparison. We hypothesized that underweight AN patients would show elevated noctumM levels of serum melatonin, urinary sulfatoxy melatonin, and plasma cortisoi; all of which would decrease following weight gain.
Methods
Subjects Nine female subjects who met DSM-III-R criteria for AN and provided written mformed cnn~rtt nartieinated in the study on two occasions, before and after weight l~ain. They were inpatients at the Programme for Eating Disorders, The Toronto Hospital, who were treated in a controlled environment in which meals were supervised, calories were adjusted weekly to promote a weight gain of I kg/week, and no exercise was permitted d,lring the admission (Kerr et al 1987). Control subjects were 9 female volunteers, individually matched for age to within 5 years, who were free of major psychiatric illness according to the same structured clinical interview and who gave written informed consent. A detailed description of patient and control characteristics is presented in Table i. Two of the 9 AN patients h ,,a k,~,,, actively involved in binge eating and purging prior to admission. No subjects admitted to laxative or diuretic misuse and 4 met criteria concurrently for major depression (MD) both before and after refeeding. Sampling was carried out in Lhe first instance after 7 days of a standard 1500 kcal/day hospital diet, when the acute effects of nutritional deprivation were felt to be reversed. The time of the second sampling procedure was 4-12 weeks later and in all cases occurred during the last 7 days of admission. On this occasion, no subjects were believed to be binge eating, purging, or exercising excessively. None of the 9 subw v . . w w . . =
~
.
.
.
.
.
E
.
.
.
.
.
.
.
.
218
BIOL PSYCHIATRY 1991;30:216-224
S.H. Kennedy et al
Table I. Demographic, Weight, and Mood Variables in Pa,.ients with Anorexia Nervosa Before and After Weight Gain and in Age-Matched Control Subjects*
Age (yr) I. 22 b (22) 2. 32 ~37) 3. 29b (28) 4. 24 (25) 5. 31 (26) 6. 27 (27) 7. 31 (26) 8. 22 (20) 9. 26 (23) 27.2 (28) ~
Duration of illness
Duration of ameba
Major depression
%MPMW
(yr)
(yr)
(DSM-III-R)
72.3(100.9) 96. ! 58.7 (100.3) 76.9 81.5 (100.9) 86.9 85.8 (87.6) 95.5 75.1 (97.7) 79.4 70.6 (93.8) 91.4 65.0(!16.0) 68.3 59.2 (113.3) 80.0 79.4 (!23.0) 90.0 72.0 (103.7) 85.0
6
9
6
7
4
4
3
3
15
5.5
14
3.5
15
4
7
3
4
3
8.2
4.6
Yes Yes No No No No Yes Yes No Yes Yes No No No No Nc Yes Yes
HRSD
BDI
33 (00) I/7 19 (06) l0 i l (02) O0 2777 (IX)) 29 !7 (05) 20 13 (01) 07 08 (03) !! 08 (03) 09 21 (01) 24 1777.4(2.3) 14.1
38 (00) 2777 31 (00) 06 - - (00) 0777 34 (01) 14 36 (0777) 26 09 (00) 08 00 (01) 03 02 (00) 00 32 (04) 3777 22.8 (I.4) 14.2
%MPMW is percent of matched population mean weight (Metropolitan Life 1983L ~Details of age-matched control subjects are presented in parentheses. ~oncurrent binge eating and vomiting within previous 3 months. ' Mean values.
jects had resumed menses by the second sampling point and no pharmacological treatments had been prescribed.
Measures T&hsa~, ~ ~. ~le*¢n*-~, ~ l'ql-'.,l,,.nl • ~'lgltl~tblgll~ltli~L/~ ~.ilill~.O.l •IIlLGIVIGW "" " ~" t' ."t'o") . . . . :.... for "~"'" t.,o~-m-r~
(Spi tzer et al 1 ~ ) was administered on both occasions by a trained research clinician before the sampling procedure. In addition, depressive symptoms were assessed using the 2 l-item Hamilton Rating Scale for Depression (HRSD) (Hamilton 1960), and the Beck Depression Inventory (BDI) (Beck et al 1961).
Assays Urinary aMT6s and serum melatonin (MT) were assayed using commercially available techniques (CIDtech Research Inc., Hamilton, Ont., Canada). The radioimmunoassay (RIA) technique for aMT6s was developed by A!dhous and Arendt (1988). Inter- and intra-assay coefficients of variation were both less than 10% at 10 pg/ml. The assay limit of detection was 2 pg per tube. The RIA for serum MT was developed by Brown et al (1983). Intra- and interassay
Pineal and Adrenal Function in Anorexia
atOL PSYCHt~V
219
Table 2. Plasma Cortisol Values at Different, Time Points for Anorexia Nervosa Control Groups Hour
Time 1 {mean ___ SD)
Time 2 {mean _ SD)
Coatmt (mean -,- SD}
8 PM 9 PM l0 ~ t I I ~t 12 ~ 1 AM 2 AM 3 AM 4 AM 5 AM 6 AM
185. I 194 ! 246.0 17502 130.9 |04.| 103.7 161.8 231.7 290.8 382.3
94°4 9(}.5 97. | 7'0.7 63.3 6&2 |0S.6 |59.5 186.9 251.4 321.0
1 |2°5 (35°8) ! |7~5 (55.3} t t-L9 (~o9) ~ . 9 (4!.8~ 86.4 (36.0} 83~6 (49.3) [ ~ . | (35.2) [52.6 (88.4) I7l,.4 (8[.6) |86.8 (86.2) ~ 8 . 4 (95.0)
{ 131.6) {! 16.2) {155.8~ ~9&5~ {63.5) {51o0) {37. !) {7708~ (160.5) ~131.9) (i05.9)
(29.9) (40.8) (53.2) (36°6} (29°6} {58.1) (6[ .9} {98.3) (113.2~ (|06.5) (~.[)
coefficients of variation were both less than 10%, with 5 pg/ml being tbe lowest limit of detection. Plasma cortisol was measured using an in-house PdA (Wong et al 1979).
Procedure On each occasion, the study was carried out at the Clinical Investigation Unit of the Toronto Hospital. An intravenous catheter was inserted into an antecubital vein, which was kept pate at with intravenous saline. Subjects were placed on a van~lyl mandelic acid-free diet for 48 hr prior to each sampling procedure, h e y were given 2 hr to acclimatize, and hourly blood samples were drawn by a research nurse between 8 PM and 6 AM. Two separate urine collections were obtained from 6 PM t o 6 AM and from 6 AM tO 6 PM the following day. On each occasion, subjects wore an overnight flight mask from 10 PM tO 6 AM tO ensure consistent light conditions.
Data Analysis For plasma cortisol and serum melatonin there were I l hourly sampling times and for urinary aMT6s there were two sampling times (day and night). An analysis of variance with repeated measures for patients at two time poims and for matched controls (using time and weight as within-subject variables) was carried out on plasma cortisol and serum MT data. For urinary aMT6s values, multiple t-tests for day/night and between group comparisons were carded out with Bonferroni Correction.
Ec:ults
Weight and Depressive Symptoms For AN patients, there was a significant increase in weight between sampling points (73 _+ 9.0%-84 _+ 19.7%) using the Metropolitan Life Insurance (1983) tables as the reference point (t = - 4 . 0 df = 8, p < 0.01). Using SCID criteria I patient switched from positive to negative majcr depression status and I switched in the reverse direction
220
BIOL PSYCHIATRY !~1;30:216-224
S.H. Kennedy et al
Table 3. Melatonin Values at Different Time Points for Experimental and Control Groups of Subjects Hour
8 9 10 II 12 ! 2 3 4 5 6
Time I (mean ± SD)
PM i'M PM PM Phi AM AM AM AM AM AM
24.4 24.5 30.~ 45.7 62.3 55.5 60.4 60.3 60.3 50.4 46.9
(30.2) (21.5) (24.1) (50.1) (61.3) (49.9) (55.6) (38,7) (38.7) (37.5) (38.I)
( ~
Time 2 "4" SD)
22.3 (15.6) 27.7 (19.8) 49.6 (36. I) 54.9 (26.5) 49.4 (21.3) 44.2 (21.O~ 45.8 (21.3) 62.8 (42,5) 66.9 (35.2) 53.8 (23,5) 48.2 (22,9)
Control (mean _ SD)
8.0 (5.6) 14.8 (!0.5) 22.4 (18.8) 36~2 (22.1) 42.7 (19.6) 51.6 (32.4) 48,7 (27,8) 50.8 (31,6) 51.3 (33.9) 43,5 (31,8) 35,0 (22.4)
between sampling times. The remaining 7 subjects did not change with regard to depression status and the mean HRSD score.s were 17.4 and 14. !, respectively, before and after refeeding. Patients also were at a significantly lower weight than controls on both admission (t = 4.72, df = 8, p < 0.01) and discharge (t = 3.53, df = 8 , p < 0.01), and displayed significantly higher levels of depression as measured by HRSD and BDI [HRSD t = 8.94, df = 8, p < 0.001; BDI t = 15.8, df = 7 , p < 0.0001 (before refeeding); HRSD t = 9.98, df = 8, p < 0.001; BDI t = 12.8, df = 8, p < 0.0001 (a~ter refeeding)].
Plasma Cortisol Levels There were significant main effects for time IF = 19.86, df = (2,14), p < 0.0001] and group [F = 5.26, df = (2,14), p < 0.02], with no significant interaction effect IF = 1.49, df = (2,14), NS]. Planned comparisons revealed significant elevations of plasma cortisol at 9, 10, 11, and 12 PM and at 6 AM in the low weight condition compared with the weight-restored condition and to the contro ~-group (p < 0.05). Mean hourly values for plasma cortisol between 8 PM and 6 AM are illustrated in Figure I.
Serum Melatonin Levels There was a significant main effect for time [F = 8.12, df = (2,14), p < 0.0001] but not for group ( F = 0.9, df = 2, p < 0.77), and no significant interaction effect between time and group ( F = 1.20, df = 10, p < 0.30). Mean values for serum melatonin are illustrated in Figure 2.
Urinary aMT6s Levels There was a significant difference between daytime and nighttime output of aMT6s, both before refeeding (day = 8.44 I~g -+ 6.5 p.g versus night = 32.4 p,g _+ 22.8 Ixg, t = - 3 . 7 3 , df = 8, p < 0.005) and afterwards (day = 9.73 lxg -+ 7.7 lxg versus night = 31.6 Ixg -+ 24.7 ~g, t = - 3 . 3 0 , df = 8, p < 0.01). However, no differences in daytime
Pineal and Adrenal Function in Anorexia
a~
~¥Ch'IA~V
~l
---*- C ~ ¢ ~ S
Cortisol Levels 40O
9 J
300-
.
~
.
J
II
1003-
-
-
-o-
-
-
-<~
~
~
~ ~:~"
20
21
/"
22
---
-C-
-
. . --
-- "-C~
24
23
J'
"#
at oa Hours
~
~
05
o6
* s~n~.w~y ainemntat i ~ . ~
Figure 1. Plasma cordsol from 8 PM to 6 AM.
or nighttime aMT6s levels were observed between anorexic and control subjec~ (control day = 12.44 p.g _ 10.3 IJ.g versus night = 20.19 izg ~ 9.9 v-g).
Discussion The three principal findings in this study are (1) that MT outpm does not differ between emaciated and weight-restored AN patients or (2) between AN patients and matched controls, and (3) in contrast, co~isol levels are elevated m the emaciated AN condition compared with both the weight-restored AN state and the control subjects. The first two conclusions are in agreement with Beam et al (1988) who reposed no change in aMT6s
brae 1
-3-
time 2
~
controls
~-.~.erurnMT Levels 70 60
/
l
/z
j"--~_-
~
+
o 20
21
22
23
24
01
02
03
04
Hours
Figure 2. Serum melatonin output from 8 PM
tO
6 AM.
05
06
322
BIOL PSYCHIATRY 1991:30:216~224
S.H. Kennedy et ai
excretion after weight gain in patients with AN, but not with Brambilla et al (1988L who reported elevated MT levels in emaciated anorexic patients compared with controls. However, it is of interest to note in our study that 3 of the 4 subjects with the lowest percent of average weights when initially tested had the highest values for both nighttime serum MT and for urinary aMT6s (subjects 5, 6, and 7). Several explanations are possible: ( I ) elevation in MT secretion only occurs beyond a certain degree of weight loss; (2) the rate of weight loss may influence MT output, and only in those individuals with the most rapid weight loss will elevated levels occur; (3) the effect of starvation may be m o ~ important than the actual weight loss, so that only those subjects who have failed to reverse the most acute effects of starvation will show elevated levels. It is atse apparent that our subjects showed a moderate level of depression, as assessed by the HRSD and the BDI. Several investigators have reported reduced levels of MT in depression (Wetterbere et al 1979; Brown et a! 1985; Miles and Philbrick 1988) and a lower level of MT in the depressive phase compared with the manic phase of bipolar affective disorder (Kennedy et al 1989b). It is therefore possible that the effect of depression in our subjects to s ~ , ~ c~,~e~t exerted ~n effect in the opposite direction and obscured any effect from weight loss. The third finding of an elevation in cortisol levels during the earlier part of the night at low weight is in agreement with previous reports (Boyar et al 1977; Doerr et al 1980) and reflects a reduced rate of cortisol b~eakdown as the body conserves energy OValsh et al 1978). Treasure et al (1985) also reported a linear relationship between the fall in basal cortisol levels and percentage weight gain in anorexics during refeeding. Thus, reports of a reciprocal relationship between MT and cortisol secretion, and a potential relationship between amenorrhea and abnormal elevation of MT output in AN are not supported. Further studies are indicated in patients with AN to evaluate the acute effects of starvation rather than chronic low weight on MT and aMT6s ou .tput. Measurement of additional urinary metabolites of MT would address the speculation that AN patients may excrete MT and its metabolites in altered proportions.
References Aidhous ME, Arendt J (1988): Radioimmunoassay for 6-sulphatoxymelatonin in urine using an iodinated tracer. Ann Ctin Biochem 25:298-303. Arato M, Grof E, Grof P, Laszlo 1, Brown GM (1985): Reproducibility of the overnight melatonin secretion pattern in healthy men. In Brown GM, Wainwright SD (eds), The Pineal Gland: Endocrine Aspects. Toronto: Pergamon Press, pp 277-282. Baranowska B, Soszynski P, Mlsiorowski W, Doroket W, Witkowski M (1986): Circadian melatonin rhythm in women with hypothalamic amenorrhea. Neuroendocrinol Lett 8:295-300. Beam J, Treasure J, Murphy M, et al (1988): A study of sulphatoxymelatonin excretion and gon=dotropin status during weight gain in anorexia nervosa. Br J Psychiato, 152:372-376. Beck AT, Ward CH, Mendcison M, Mock J, Erbaugh J (1961): An inventory for measuring depression. Arch Gen P.D'chiatrb.' 4:561-57 I. Beck-Friis J, Kjellman BF, Aperia B, et al (1985): Serum melatonin in relation to clinical variables in patients w!rh major depressive disorder and a hypothesis of a low melatonin syndrome. Acta Ps3'chiatr Scand 7 ! :319-330. Boyar RM, Hellman LD, Roffwang H, et al (1977): Cortisol secretion and metabolism in anorexia nervosa. N Engl J Med 296: i90-193.
Pineal and Adrenal Function in Anorexia
s~oL ~VCa~a'mV 199|~.~:216-224
223
Brarnbilla F, Fraschini F, Esposti G, et al (1988): Melatonin circadian rhythm in anorexia netwosa and obesity. Psychiat~, Res 23:267-276. Brown GM, Garlinkel PE, Jeuniewic N, Moldofsky H, Stancer HC (1977): ~ r i n e anorexia nervosa. In Vigersky RA (ed)o Anorexia Nervosa~ New, York: Raven ~ , 135.
profiles m ~ [23-
Brown GM, Gram l.J, Pulido O, Bums TG, Ni|es LP, Snieckus V (1983): Application of immunologic techniques to the study of pineal indo|ea|kylamines. In Reiter RJ (ed), Pineal Research Reviews !. New York: Alan R. Liss, pp 207-246. Brown R, Kocsis JH, Caroff S, et al (1985): Differences in nocturnal melatonin secretion between melancholic depressed patients and control subjects. Am J Ps'chiatry 142:81 I--816. Brzezinski A, Lynch HJ, Seibel MM, et al (1988): The circadian rhythm of plasma melatonin during the normal menstrual cycle and in amenorrheic women. J Ctin Endocrinol Metab 66:891895. Chik CL, Ha AK, Brown GM (1987): Effect of food restriction on 24 hour serum and pineal melatonin content in male rats. Acta Endocrinol i 15:507-513. Dalery J, Claustrat B, Brun J, de Villard LE (1985): Plasma melatonin and cortisol levels in patients with anorexia nervo~. Nueroendocrinol Len 7:159-164. Doerr P, Fichter M, Pirke KM, Lund R (1980): Relationship between weight gain and hypothalarnic pituitary adrenal function in patients with anorexia nervosa. J Steroid Biachem | 3:529-537. Gold P, Loriaux D, Roy A, et al (1986): The corticotropin releasing factor ~st: Implications for the diagnosis and pathophysiology of hypercortisolism in depression and C u s ~ g ' s disease. N Engl J Med 314: i 330-1335. Hamilton MA (1960): A rating scale for depression. J Neural Neurosurg Psychiatry 23:46-62. Kaplan AS, Garfinkel PE, Brown GM (1989): The DST and TRH test in bulimia nervosa. Br J Psychiatry 154:86-92. Kennedy SH, Heslegrave ILl (1989): Heart regulation in bulimia nervosa. J Psychiatr Res 23:267273. Kennedy SH, Garfinkel PE, Parienti V, Costa D, Brown GM (1989a): Changes in melatomn levels but not cortisol levels are associated with depression in patients with eating disorders. Arch Gen Psychiatry 46:73-78. Kennedy SH, Tighe S, MeVey G, Brown GM (1989b): Melatonin and cortisol "switches" d ~ g mania, depression, and euthynfia in a drug-free bipolar patient. J Nerv Merit Dis 177:300-303. Kennedy SH, Brown GM, Garfinkel PE, McVey G, Costa D, Parienti V (1990): Sulphatoxy melatonin: An index of depression in anorexia nervosa and b u ~ a nervosa. Psychiatry Res 32:221-227. Kerr A, Kennedy SH, Stem D (1987): Inpatient treatment for anorexia nervosa and bulimia Br Rev Anorexia Bulimia. 2:5-16. Lewy AJ (1984): Human melatonin secretion It: A marker for the c i r c ~ system and the affects of light. In Post RM, Ballenger JC (eds), Neurobiology of Mood Disorders. Balt~ore: Williams & Wilkins, pp 2 ! 5-226. Metropolitan Life Insurance Company (1983): 1983 height and weight tables. Stat Bull 64:3-9. Miles A, Philbrick DRS (1988): Melatonin and psychiatry: A review. Biol Psychiatry 23:405--425. Spitzer RL, Williams JBW, Gibbon M, First MB (1988): Structured Clinical Interview for DSMIll-R-Patient Version. Biometrics Research Department, New York State P s y c ~ c Institute, New York, NY. Tamarkin L, Baird CJ, Alraeida OFS (1985): Melatonin: A coordinating signal for reproduction? Science 227:714-720. Treasure J, Wheeler MJ, Salieh B, Russell GFM (1985): Anorexia nervosa and the adrenal: The effect of weight gain. J Psychiatr Res 19:221-225.
224
BIOL PSYCHIATRY 1991:30:216-224
S.H. Kennedy et al
Waldhauser F, Steger H (1986): Changes in melatonin secretion with age and pubescence. J Neural Transm 21(Suppl): 183-197. Walsh TT, Katz JL, Levin J, et al (1978): Adrenal activity in anorexia nervosa. Psychosom Med 40:499-506. Wetterberg L, Arendt J, Paunier L. Sizonenko PC, Donselaar W, Heyden T (1976): Human serum melatonin changes during the menstrual cycle. J Clin Endocrinol Metab 42:185-188, Wetterberg L, Beck-Friis J, Aperia B, Petterson V (1979): Melatonin: Cortisol ratio in depression. Lancet 2:1361. Wong PY, Mee AV, Ho FFK (1979): A direct radioimmunoassay of serum cortisol with in-house 125 l-tracer and preconjugated double antibody. Clin Chem 25:914-917.