Clinical Endocrinology (1990), 32,213-220
DIFFERENCES IN CLINICAL BETWEEN OBESE AND WITH POLYCYSTIC AN ANALYSIS OF 263
AND ENDOCRINE FEATURES NON-OBESE SUBJECTS OVARY SYNDROME: CONSECUTIVE CASES
D. S . K I D D Y , P. S . SHARP, D. M. WHITE, M. F. SCANLON, H. D . MASON, C . S . BRAY, D. W. POLSON, M. J . R E E D AND S . FRANKS Departments of Obstetrics and Gynaecology, Clinical Endocrinology and Chemical Pathology, St Mary’s Hospital Medical School, London, UK (Received I 7 May 1989, returned for revision 3 July 1989;finally revised 13 July 1989; accepted 8 August 1989)
SUMM ARY
Two hundred and sixty-three women with ultrasound-diagnosed polycystic ovary syndrome were studied of whom 91 (35%) were obese (BMI > 25 kg/m’). Obese women with PCOS had a greater prevalence of hirsutism (73% compared with 56%) and menstrual disorders than non-obese subjects. Total testosterone and androstenedione concentrations in serum were similar in the two subgroups but SHBG concentrations were significantly lower, and free testosterone levels higher, in obese compared with lean subjects. In addition, concentrations of androsterone glucuronide, a marker of peripheral 5wreductase activity, were higher in obese than in non-obese women with PCOS. There were no significant correlations of either SHBG or free testosterone with androsterone glucuronide suggesting that obesity has independent effects on transport and on metabolism of androgen. There were no significant differences between the subgroups in either baseline gonadotrophin concentrations or the pulsatile pattern of LH and FSH secretion studied over an 8-h period. There was, however, an inverse correlation of FSH with BMI, but only in the obese subgroup. In conclusion, the increased frequency of hirsutism in obese compared with lean women with PCOS is associated with increased bio-availability of androgens to peripheral tissues and enhanced activity of 5a-reductase in obese subjects. The mechanism underlying the higher prevalence of anovulation in obese women remains unexplained. Obesity is a well recognized feature of polycystic ovary syndrome (PCOS) (Stein & Leventhal, 1935; Goldzieher & Green, 1962; Yen, 1980) but the prevalence of obesity in women with PCOS and the differences in clinical and biochemical presentation between Correspondence: Professor S. Franks, Department of Obstetrics and Gynaecology, St Mary’s Hospital Medical School, London W2 IPG, UK.
213
214
D . S. Kiddy et al.
lean and obese women are not clear. Obesity by itself is associated with menstrual disturbances (Hartz et al., 1979; Kopelman et al., 1981)which may be reversed by weight loss (Mitchell & Rogers, 1953; Kopelman et al., 1981; Harlass et al., 1984) but this does not seem to be specific for PCOS. As far as endocrine function tests are concerned, obesity is characterized by low serum concentrations of sex hormone binding globulin (SHBG) (Kopelman et al., 1980; Plymate et al., 1981) and a high androgen production rate (Kirschner et al., 1983; Kurtz et al., 1987). However, Dunaif et al. (1988) found that gonadotrophin and total androgen concentrations were remarkably similar in obese and non-obese subjects with PCO. In this study we assessed the prevalence of obesity in a large series of patients with polycystic ovaries, whose diagnosis had been made primarily by pelvic ultrasound, and compared the clinical and biochemical features in obese and non-obese patients. Our investigations included an analysis of possible effects of body weight on transport and metabolism of androgens. PATIENTS AND METHODS The study group comprised 263 consecutive patients who presented to the gynaecological endocrine clinic at the Samaritan Hospital for Women and were found to have polycystic ovaries on ultrasound scanning (Adams et al., 1985, 1986). The patients had presented with either hirsutism or anovulation or both. The group was subdivided according to body weight. Those subjects with a Body Mass Index (BMI) greater than 25 kg/m2were considered obese and those with a BMI equal to or less than 25, non-obese. Women with polycystic ovaries but who had other specific endocrine abnormalities including oestrogen-deficient hyperprolactinaemia or weight-loss related amenorrhoea, were excluded from this analysis. Women who had hyperprolactinaemia but who had menstrual cycles or progestagen positive amenorrhoea were included since such subjects have clinical and endocrine characteristics which are indistinguishable from those in normo-prolactinaemic patients with PCOS (Franks, 1989a). Serum LH, FSH, prolactin and testosterone were measured by radioimmunoassay as previously described (Adams et al., 1985). Analysis of LH and FSH pulses in serial samples taken at 15-min intervals for 8 h in 20 lean and 13 obese women with PCOS were performed by a modification of the method of Santen and Bardin (1973) as described by Mason et al. (1988). Only anovulatory subjects were included so as to avoid any influence of cyclicity on gonadotrophin concentrations. Serum free testosterone fraction was measured using the Amicon Centrifree micropartition system (Danvers, Mass, USA) according to the method of Vlahos et al. (1982). The free testosterone concentration was derived from the product of this fraction and the total testosterone measured in the same sample of serum. Sex hormone binding globulin was measured by an immunoradiometric assay (Farmos Diagnostics, Turku, Finland) as previously reported (Polson et al., 1987). Androsterone glucuronide (AG) was measured by the method of Scanlon et al., (1988) as described for measurement of androstanediol glucuronide (Adiol G), the only significant differencebeing the use of antiserum against androsterone, kindly donated by Dr Stephen Hillier of the Department of Obstetrics and Gynaecology, University of Edinburgh (Uilenbroek er al., 1983). Androsterone glucuronide was chosen as a marker of peripheral 5a-reductase activity in preference to Adiol G because preliminary data indicated a better correlation of AG with the presence of hirsutism (Scanlon et al., 1987; Franks, 1989b).
215
PCO in obese and non-obese subjects Table I . Distribution of hirsutism and menstrual symptoms between non-obese and obese subjects with PCOS Non-obese
Obese
n (YO)
n (W,)
Hirsutism Hirsute Non-hirsute
96 (56) 76 (44)
66 (73) 25(27)
Menstrual cycle Regular? Irregular Amenorrhoea
49 (28) 103 (60) 20(12)
11 (12) 73 (80) 7(8)
x2 = 6.3* P 700 mU/1) but there was no difference in the distribution of serum prolactin concentrations between the two groups. In the obese group only, there was a weak but statistically significant inverse correlation of FSH with BMI (r = -0.23, n = 84, P = 0.028; Spearman) and of prolactin with BMI (r = -0.35, n = 62, P= 0.006).
D . S. Kiddy et al.
216
Table 2. Ultrasound and endocrine results in 263 women with PCOS, divided according to body mass index Non-obese mean BMI (kg/m2) 21.0 Uterine area (cm2) 25.5 Ovarian volume (ml) 11.5 LH (IU/I) 13.9 FSH (IU/l) 4.2 Testosterone (nmolil) 2.9 Androstenedione (nmol/l) 9.2 225 Prolactin (mU/l)*
(SD)
Obese
n mean -
172 I30 103 (5.1) 162 (9.8) 161 (1.9) 149 (1.4) 60 (3.7) (31-3350) I 24 (1.8) (9.3)
30 3 27.7 12.1 13.4 4.4 3.0 9.6 225
(SD)
n
P
(4.2)
91 c 0.001 71 NS (8.5) 49 NS NS (7.1) 84 (1.9) 84 NS NS (1.1) 85 NS (4.3) 49 (63-1620) 65 NS (10.1)
Values are mean (SD) or * median (range). Groups were compared using Student’s ttest (after logarithmic transformation of the data in the case of ovarian volume and LH) or * Mann-Whitney U-test.
Table 3. Serum concentrations of SHBG, free testosterone and androgen metabolites in non-obese and obese women with PCOS Non-obese mean (SD) ___
~~
~
~
Obese
n
64.1 (29.4) 30 SHBG (nmol/l) Free testosterone (pmol/l) 38.4 (21.5) 15 Androsterone glucuronide (nmol/l) 15.3 (7.2) 23 1Ig-Hydroxy androstenedione (nmol/l) (1.8) 13 5.5 ~
~~
~
~~
~
mean (SD)
~
~
n
P*
~
30.5 50.5 23.0 5.6
(19.8) 25
DIFFERENCES IN CLINICAL BETWEEN OBESE AND WITH POLYCYSTIC AN ANALYSIS OF 263
AND ENDOCRINE FEATURES NON-OBESE SUBJECTS OVARY SYNDROME: CONSECUTIVE CASES
D. S . K I D D Y , P. S . SHARP, D. M. WHITE, M. F. SCANLON, H. D . MASON, C . S . BRAY, D. W. POLSON, M. J . R E E D AND S . FRANKS Departments of Obstetrics and Gynaecology, Clinical Endocrinology and Chemical Pathology, St Mary’s Hospital Medical School, London, UK (Received I 7 May 1989, returned for revision 3 July 1989;finally revised 13 July 1989; accepted 8 August 1989)
SUMM ARY
Two hundred and sixty-three women with ultrasound-diagnosed polycystic ovary syndrome were studied of whom 91 (35%) were obese (BMI > 25 kg/m’). Obese women with PCOS had a greater prevalence of hirsutism (73% compared with 56%) and menstrual disorders than non-obese subjects. Total testosterone and androstenedione concentrations in serum were similar in the two subgroups but SHBG concentrations were significantly lower, and free testosterone levels higher, in obese compared with lean subjects. In addition, concentrations of androsterone glucuronide, a marker of peripheral 5wreductase activity, were higher in obese than in non-obese women with PCOS. There were no significant correlations of either SHBG or free testosterone with androsterone glucuronide suggesting that obesity has independent effects on transport and on metabolism of androgen. There were no significant differences between the subgroups in either baseline gonadotrophin concentrations or the pulsatile pattern of LH and FSH secretion studied over an 8-h period. There was, however, an inverse correlation of FSH with BMI, but only in the obese subgroup. In conclusion, the increased frequency of hirsutism in obese compared with lean women with PCOS is associated with increased bio-availability of androgens to peripheral tissues and enhanced activity of 5a-reductase in obese subjects. The mechanism underlying the higher prevalence of anovulation in obese women remains unexplained. Obesity is a well recognized feature of polycystic ovary syndrome (PCOS) (Stein & Leventhal, 1935; Goldzieher & Green, 1962; Yen, 1980) but the prevalence of obesity in women with PCOS and the differences in clinical and biochemical presentation between Correspondence: Professor S. Franks, Department of Obstetrics and Gynaecology, St Mary’s Hospital Medical School, London W2 IPG, UK.
213
214
D . S. Kiddy et al.
lean and obese women are not clear. Obesity by itself is associated with menstrual disturbances (Hartz et al., 1979; Kopelman et al., 1981)which may be reversed by weight loss (Mitchell & Rogers, 1953; Kopelman et al., 1981; Harlass et al., 1984) but this does not seem to be specific for PCOS. As far as endocrine function tests are concerned, obesity is characterized by low serum concentrations of sex hormone binding globulin (SHBG) (Kopelman et al., 1980; Plymate et al., 1981) and a high androgen production rate (Kirschner et al., 1983; Kurtz et al., 1987). However, Dunaif et al. (1988) found that gonadotrophin and total androgen concentrations were remarkably similar in obese and non-obese subjects with PCO. In this study we assessed the prevalence of obesity in a large series of patients with polycystic ovaries, whose diagnosis had been made primarily by pelvic ultrasound, and compared the clinical and biochemical features in obese and non-obese patients. Our investigations included an analysis of possible effects of body weight on transport and metabolism of androgens. PATIENTS AND METHODS The study group comprised 263 consecutive patients who presented to the gynaecological endocrine clinic at the Samaritan Hospital for Women and were found to have polycystic ovaries on ultrasound scanning (Adams et al., 1985, 1986). The patients had presented with either hirsutism or anovulation or both. The group was subdivided according to body weight. Those subjects with a Body Mass Index (BMI) greater than 25 kg/m2were considered obese and those with a BMI equal to or less than 25, non-obese. Women with polycystic ovaries but who had other specific endocrine abnormalities including oestrogen-deficient hyperprolactinaemia or weight-loss related amenorrhoea, were excluded from this analysis. Women who had hyperprolactinaemia but who had menstrual cycles or progestagen positive amenorrhoea were included since such subjects have clinical and endocrine characteristics which are indistinguishable from those in normo-prolactinaemic patients with PCOS (Franks, 1989a). Serum LH, FSH, prolactin and testosterone were measured by radioimmunoassay as previously described (Adams et al., 1985). Analysis of LH and FSH pulses in serial samples taken at 15-min intervals for 8 h in 20 lean and 13 obese women with PCOS were performed by a modification of the method of Santen and Bardin (1973) as described by Mason et al. (1988). Only anovulatory subjects were included so as to avoid any influence of cyclicity on gonadotrophin concentrations. Serum free testosterone fraction was measured using the Amicon Centrifree micropartition system (Danvers, Mass, USA) according to the method of Vlahos et al. (1982). The free testosterone concentration was derived from the product of this fraction and the total testosterone measured in the same sample of serum. Sex hormone binding globulin was measured by an immunoradiometric assay (Farmos Diagnostics, Turku, Finland) as previously reported (Polson et al., 1987). Androsterone glucuronide (AG) was measured by the method of Scanlon et al., (1988) as described for measurement of androstanediol glucuronide (Adiol G), the only significant differencebeing the use of antiserum against androsterone, kindly donated by Dr Stephen Hillier of the Department of Obstetrics and Gynaecology, University of Edinburgh (Uilenbroek er al., 1983). Androsterone glucuronide was chosen as a marker of peripheral 5a-reductase activity in preference to Adiol G because preliminary data indicated a better correlation of AG with the presence of hirsutism (Scanlon et al., 1987; Franks, 1989b).
215
PCO in obese and non-obese subjects Table I . Distribution of hirsutism and menstrual symptoms between non-obese and obese subjects with PCOS Non-obese
Obese
n (YO)
n (W,)
Hirsutism Hirsute Non-hirsute
96 (56) 76 (44)
66 (73) 25(27)
Menstrual cycle Regular? Irregular Amenorrhoea
49 (28) 103 (60) 20(12)
11 (12) 73 (80) 7(8)
x2 = 6.3* P 700 mU/1) but there was no difference in the distribution of serum prolactin concentrations between the two groups. In the obese group only, there was a weak but statistically significant inverse correlation of FSH with BMI (r = -0.23, n = 84, P = 0.028; Spearman) and of prolactin with BMI (r = -0.35, n = 62, P= 0.006).
D . S. Kiddy et al.
216
Table 2. Ultrasound and endocrine results in 263 women with PCOS, divided according to body mass index Non-obese mean BMI (kg/m2) 21.0 Uterine area (cm2) 25.5 Ovarian volume (ml) 11.5 LH (IU/I) 13.9 FSH (IU/l) 4.2 Testosterone (nmolil) 2.9 Androstenedione (nmol/l) 9.2 225 Prolactin (mU/l)*
(SD)
Obese
n mean -
172 I30 103 (5.1) 162 (9.8) 161 (1.9) 149 (1.4) 60 (3.7) (31-3350) I 24 (1.8) (9.3)
30 3 27.7 12.1 13.4 4.4 3.0 9.6 225
(SD)
n
P
(4.2)
91 c 0.001 71 NS (8.5) 49 NS NS (7.1) 84 (1.9) 84 NS NS (1.1) 85 NS (4.3) 49 (63-1620) 65 NS (10.1)
Values are mean (SD) or * median (range). Groups were compared using Student’s ttest (after logarithmic transformation of the data in the case of ovarian volume and LH) or * Mann-Whitney U-test.
Table 3. Serum concentrations of SHBG, free testosterone and androgen metabolites in non-obese and obese women with PCOS Non-obese mean (SD) ___
~~
~
~
Obese
n
64.1 (29.4) 30 SHBG (nmol/l) Free testosterone (pmol/l) 38.4 (21.5) 15 Androsterone glucuronide (nmol/l) 15.3 (7.2) 23 1Ig-Hydroxy androstenedione (nmol/l) (1.8) 13 5.5 ~
~~
~
~~
~
mean (SD)
~
~
n
P*
~
30.5 50.5 23.0 5.6
(19.8) 25
