Does ethnicity influence the prevalence of adrenal hyperandrogenism and insulin resistance in polycystic ovary syndrome? Enrico Carmina, MD: Takao Koyama, MD,b Lilly Chang, MD: Frank Z. Stanczyk, PhD: and Rogerio A. Lobo, MD'
Palermo, Italy, Tokyo, Japan, and Los Angeles, California OBJECTIVE: Our purpose was to determine the prevalence of adrenal hyperandrogenism and insulin resistance in patients with hyperandrogenic chronic anovulation, also called polycystiC ovary syndrome, living in the United States, Italy, and Japan. STUDY DESIGN: Seventy-five women with polycystic ovary syndrome, 25 each from the United States, Italy, and Japan, and 10 ovulatory controls were studied. Hirsutism, obesity, and the presence of cystic ovaries were assessed, as were blood levels for estrogen, luteinizing hormone, testosterone, adrenal androgens, and insulin. All patients received an insulin tolerance test to assess insulin resistance. RESULTS: Women from Japan were less obese (p < 0.05) and did not have hirsutism, although the percentage of cystic ovaries (68% to 80%) was comparable. Serum luteinizing hormone, testosterone, and estradiol were similar, but levels of 3a-androstanediol glucuronide, which was elevated in women from the United States and Italy, was normal in women from Japan. The adrenal androgens, dehydroepiandrosterone sulfate and 11 ~-hydroxyandrostenedione were elevated in 48% to 64% of the patients and by a similar percentage in the three groups. Fasting insulin was elevated in all groups, but was significantly higher in women from the United States and Italy compared with women from Japan (p < 0.05). However, insulin resistance as assessed by dissociation constant of insulin tolerance test values was significantly elevated but similar in the three groups and occurred in 68% to 76% of patients. CONCLUSION: In polycystic ovary syndrome, although obesity and hirsutism vary according to dietary, genetiC, and environmental factors, the prevalence of adrenal androgen excess and insulin resistance appear to be fairly uniform. These results suggest that these factors may be involved in the pathophysiologic features of the disorder. (AM J OBSTET GYNECOL 1992;167:1807-12.)
Key words: Ethnic, polycystic ovary syndrome, adrenal androgen, insulin resistance Polycystic ovary syndrome is extremely heterogeneous. Because of this diversity in clinical and biochemical findings, we have argued that the syndrome should be renamed. Thus we have preferred to refer to this entity as the syndrome of hyperandrogenic chronic anovulation on the basis of findings of the two cardinal features: hyperandrogenism and chronic anovulation. 1 In studies of pathogenesis, no unifying hypothesis has been accepted. 1. 2 Also, the presence of previously held characteristic features are no longer necessary for
From the Cattedra di Endocrinologia, Universita di Palermo, a the Division of Reproductive Endocrinology, Department of Obstetrics and Gynecology, Tokyo Medical and Dental University, School of Medicine, b and the Department of Obstetrics and Gynecology, University of Southern California School of Medicine.' Presented in part at the Thirty-ninth Annual Meeting of the Society for Gynecologic Investigation, San Antonio, Texas, March 18-21, 1992.
Reprint requests: Rogerio A. Lobo, MD, Department of Obstetrics and Gynecology, Women's Hospital, Room 1M2, 1240 N. Mission Road, Los Angeles, CA 90033. 6/6/41906
the diagnosis. We 3 and others4 have studied nonhirsute patients with polycystic ovary syndrome-hyperandrogenic chronic anovulation, and others have shown that obesity need not be present. 4 • 5 For some time it has been suggested that adrenal hyperandrogenism and insulin resistance may be involved in the pathogenesis. We wished therefore to study the importance of these factors in patients with hyperandrogenic chronic anovulation and asked the following question: if nonhirsute and nonobese patients diagnosed as having hyperandrogenic chronic anovulation also do not exhibit insulin resistance, then can it be implicated in the pathogenesis of the disorder? Similarly, if adrenal androgen excess is absent, this too may not be a characteristic feature. In this report we have attempted to answer these questions by studying ethnic diversity in hyperandrogenic chronic anovulation among North American, italian, and Japanese women from their respective continents. Specifically, we have investigated the prevalence of adrenal hyperandrogenism and insulin resistance. By so doing, we wished to determine the relative impor1807
1808 Carmina et al.
tance of these features that have been believed to be central to the pathogenesis of the disorder.
Material and methods Subjects. Seventy-five patients with hyperandrogenic chronic anovulation were studied. The diagnosis was based on the findings of hyperandrogenism (elevated testosterone or dehydroepiandrosterone [DHEA] sulfate) and chronic anovulation of perimenarchial onset. No patient had virilization, tumors, or congenital adrenal hyperplasia (on the basis of normal levels of 17hydroxyprogesterone). In three separate clinics, 25 consecutive patients were recruited for study between 1990 and 1991. Thus 25 women (mean age 30 ± 2 years) from East Los Angeles, California, of Hispanic origin (Mexican and Central American); 25 from Tokyo, Japan (mean age 24 ± 1 years); and 25 from Palermo, Italy (mean age 24 ± 1 years) were recruited. None of the Japanese women had DHEA sulfate values determined as a criterion for their diagnosis. From each country 10 ovulatory controls of the same ages were selected for comparisons. Protocol. Between 8:00 and 9:00 AM patients and controls had fasting blood obtained between days 5 and 8 after a progestin-induced or spontaneous menses. In addition, all patients received an insulin tolerance test. Regular insulin, 0.1 U/kg, was administered intravenously as previously described. 6 Hirsutism was evaluated by a modification of the scoring system of Ferriman-Gallwey.7 Only scores > 8 were considered to constitute hirsutism. Body mass was evaluated in all patients and expressed as percent of ideal body weight. In all patients ovarian morphologic condition was assessed by ultrasonography by means of a 5 mHz abdominal or vaginal probe. Classic ultrasonographic criteria of pericortical cysts and increased stroma was sought. 8 Measurements. In all samples serum luteinizing hormone (LH), estradiol, testosterone, DHEA sulfate, 1113hydroxyandrostenedione, 3a-androstanediol glucuronide, and insulin were measured. During the insulin tolerance test glucose was measured. Assays. All hormones were assayed by established radioimmunoassay techniques. 9 - 12 Hormonal determinations for all patients were made in the same laboratory (Los Angeles, Calif.). Estradiol, testosterone, and 1113-hydroxyandrostenedione assays included extraction and Celite chromatography. In the LH assay the standard used was LER 907, which was referenced to the Second International Reference Preparation. The sensitivity of this assay was 1.5 mIU/m!' The 3a-androstanediol glucuronide assay used specific hydrolysis, followed by extraction, Celite column chromatography, and radioimmunoassay. 13 In all assays, intraassay and interassay coefficients of variation did not exceed 6% and 13%, respectively. Statistical analyses. To assess insulin resistance, a
December 1992 Am J Obstet Gynecol
Kitt value, or dissociation constant of the insulin tolerance test expressed as percent per minute was calculated by regression analysis. 6 The Student t test with and without log transformation was used, as was analysis of variance for group comparisons.
Results Comparisons of ideal body weight in North American, Japanese, and Italian women may be found in Fig. 1. Compared with Japanese women, North American women had significantly higher ideal body weight (122% ± 3% vs III % ± 4%, P < 0.05). Ideal body weight in Italian women was intermediate and not significantly different from either group. The presence of cystic ovaries was similar in the three groups: 80%, 68%, and 76%, respectively (Fig. 1). Ferriman-Gallwey scores were significantly elevated and similar in United States and Italian women (12 ± 1 and 12.5 ± 1). In controls, by definition, all Ferriman-Gallwey scores were < 8. The Japanese women had a mean score of 3.5 ± 0.2, which was not statistically different from the scores of Japanese controls. The three ethnic control groups had similar hormone levels, and thus for comparisons with the hyperandrogenic chronic anovulation groups the following composite mean values were used: LH, 9.1 ± 0.5 mU/ml; estradiol, 56 ± 5 pg/ml; testosterone, 32.3 ± 3 ng/dl; DHEA sulfate, 1.8 ± 1 j.Lg/ml; 1113-hydroxyandrostenedione, 1.2 ± 0.1 ng/ml; 3a-androstanediol glucuronide, 3 ± 0.2 ng/ml; and insulin, 8 ± 0.2 j.LU/m!. Serum LH was elevated in all three hyperandrogenic chronic anovulation groups compared with controls (P < 0.01) but were similar to one another (Fig. 2). Serum estradiol was also similar between groups and comparable to control values. Serum testosterone was elevated in all groups compared with controls (P < 0.05), but the values were similar in the three hyperandrogenic chronic anovulation groups (Fig. 2). Serum 3a-androstanediol glucuronide was elevated in United States and Italian women (p < 0.01) when compared with both Japanese women and controls. Japanese women and control values were similar, and United States and Italian women had similar levels (Fig. 2). Fig. 3 depicts levels of the adrenal androgens DHEA sulfate and 1113-hydroxyandrostenedione and the percentage of patients having elevated levels. Levels of both DHEA sulfate and 1113-hydroxyandrostenedione were significantly elevated in the three groups compared with controls (p < 0.01). However, values were similar among the three groups for each androgen. Serum DHEA sulfate was elevated in 52%, 48%, and 48% of North American, Japanese, and Italian women, respectively. Serum 1113-hydroxyandrostenedione was elevated in 64%, 64%, and 56% of patients in each group. Fasting insulin levels were significantly elevated in
Ethnic variations in polycystic ovary syndrome
Volume 167 "umber 6
.........
.........
c
I-
:r: c.9
W
1809
~
'-'"
*
C/l
W ...... 0:: «
>
:3:
0
>0
u
f=
0
C/l
CD
>-
U
**
** C/l W 0::
o
U
(/)
c.9
I.J...
Fig. 1. Percent ideal body weight. hirsutism (Ferriman-Gallway [FG) score) and prevalence of cystic ovaries in three ethnic groups. D. United States; ~. Japan; IllS. Italy. Differences between United States and Italy compared with Japan: One astnisk. p < 0.05; two asterisks. p < 0.01.
r--.
--.J
E
...........
::J E
I --.J
r--.
--.J
E
........... CJ)
a..
'-'
N W
r--.
~
--.J -0
...........
CJ)5
c
'--'"
l-
--.J E ...........
**
**
CJ)
c
'-' ~
I
0 M
Fig. 2. Serum levels of LH. testosterone (T). estradiol (E 2 ). and 3a-androstanediol glucuronide (3a-G) in three ethnic groups of patients with polycystic ovary syndrome. D. United States; ~. Japan; Italy. Significance of 3a-androstanediol glucuronide values between the U.S. and Italy compared to Japan: Two asterisks. p < 0.01.
~.
the United States and Italian groups compared with control values (p < 0.01). In Japanese women this increase was smaller but still of significance (p < 0.05). Both United States and Italian group values for insulin
were significantly higher than values in Japanese women (Fig. 4). Fasting insulin showed a highly significant correlation with percent of ideal body weight in all three groups (p < 0.01).
1810 Carmina at al.
December 1992 Am J Obstet Gynecol
,-..
,-.. .....J E
.....J E
.........
.........
Ol
Ol
::j.
C
'-"
'-"
C/)
Palermo, Italy, Tokyo, Japan, and Los Angeles, California OBJECTIVE: Our purpose was to determine the prevalence of adrenal hyperandrogenism and insulin resistance in patients with hyperandrogenic chronic anovulation, also called polycystiC ovary syndrome, living in the United States, Italy, and Japan. STUDY DESIGN: Seventy-five women with polycystic ovary syndrome, 25 each from the United States, Italy, and Japan, and 10 ovulatory controls were studied. Hirsutism, obesity, and the presence of cystic ovaries were assessed, as were blood levels for estrogen, luteinizing hormone, testosterone, adrenal androgens, and insulin. All patients received an insulin tolerance test to assess insulin resistance. RESULTS: Women from Japan were less obese (p < 0.05) and did not have hirsutism, although the percentage of cystic ovaries (68% to 80%) was comparable. Serum luteinizing hormone, testosterone, and estradiol were similar, but levels of 3a-androstanediol glucuronide, which was elevated in women from the United States and Italy, was normal in women from Japan. The adrenal androgens, dehydroepiandrosterone sulfate and 11 ~-hydroxyandrostenedione were elevated in 48% to 64% of the patients and by a similar percentage in the three groups. Fasting insulin was elevated in all groups, but was significantly higher in women from the United States and Italy compared with women from Japan (p < 0.05). However, insulin resistance as assessed by dissociation constant of insulin tolerance test values was significantly elevated but similar in the three groups and occurred in 68% to 76% of patients. CONCLUSION: In polycystic ovary syndrome, although obesity and hirsutism vary according to dietary, genetiC, and environmental factors, the prevalence of adrenal androgen excess and insulin resistance appear to be fairly uniform. These results suggest that these factors may be involved in the pathophysiologic features of the disorder. (AM J OBSTET GYNECOL 1992;167:1807-12.)
Key words: Ethnic, polycystic ovary syndrome, adrenal androgen, insulin resistance Polycystic ovary syndrome is extremely heterogeneous. Because of this diversity in clinical and biochemical findings, we have argued that the syndrome should be renamed. Thus we have preferred to refer to this entity as the syndrome of hyperandrogenic chronic anovulation on the basis of findings of the two cardinal features: hyperandrogenism and chronic anovulation. 1 In studies of pathogenesis, no unifying hypothesis has been accepted. 1. 2 Also, the presence of previously held characteristic features are no longer necessary for
From the Cattedra di Endocrinologia, Universita di Palermo, a the Division of Reproductive Endocrinology, Department of Obstetrics and Gynecology, Tokyo Medical and Dental University, School of Medicine, b and the Department of Obstetrics and Gynecology, University of Southern California School of Medicine.' Presented in part at the Thirty-ninth Annual Meeting of the Society for Gynecologic Investigation, San Antonio, Texas, March 18-21, 1992.
Reprint requests: Rogerio A. Lobo, MD, Department of Obstetrics and Gynecology, Women's Hospital, Room 1M2, 1240 N. Mission Road, Los Angeles, CA 90033. 6/6/41906
the diagnosis. We 3 and others4 have studied nonhirsute patients with polycystic ovary syndrome-hyperandrogenic chronic anovulation, and others have shown that obesity need not be present. 4 • 5 For some time it has been suggested that adrenal hyperandrogenism and insulin resistance may be involved in the pathogenesis. We wished therefore to study the importance of these factors in patients with hyperandrogenic chronic anovulation and asked the following question: if nonhirsute and nonobese patients diagnosed as having hyperandrogenic chronic anovulation also do not exhibit insulin resistance, then can it be implicated in the pathogenesis of the disorder? Similarly, if adrenal androgen excess is absent, this too may not be a characteristic feature. In this report we have attempted to answer these questions by studying ethnic diversity in hyperandrogenic chronic anovulation among North American, italian, and Japanese women from their respective continents. Specifically, we have investigated the prevalence of adrenal hyperandrogenism and insulin resistance. By so doing, we wished to determine the relative impor1807
1808 Carmina et al.
tance of these features that have been believed to be central to the pathogenesis of the disorder.
Material and methods Subjects. Seventy-five patients with hyperandrogenic chronic anovulation were studied. The diagnosis was based on the findings of hyperandrogenism (elevated testosterone or dehydroepiandrosterone [DHEA] sulfate) and chronic anovulation of perimenarchial onset. No patient had virilization, tumors, or congenital adrenal hyperplasia (on the basis of normal levels of 17hydroxyprogesterone). In three separate clinics, 25 consecutive patients were recruited for study between 1990 and 1991. Thus 25 women (mean age 30 ± 2 years) from East Los Angeles, California, of Hispanic origin (Mexican and Central American); 25 from Tokyo, Japan (mean age 24 ± 1 years); and 25 from Palermo, Italy (mean age 24 ± 1 years) were recruited. None of the Japanese women had DHEA sulfate values determined as a criterion for their diagnosis. From each country 10 ovulatory controls of the same ages were selected for comparisons. Protocol. Between 8:00 and 9:00 AM patients and controls had fasting blood obtained between days 5 and 8 after a progestin-induced or spontaneous menses. In addition, all patients received an insulin tolerance test. Regular insulin, 0.1 U/kg, was administered intravenously as previously described. 6 Hirsutism was evaluated by a modification of the scoring system of Ferriman-Gallwey.7 Only scores > 8 were considered to constitute hirsutism. Body mass was evaluated in all patients and expressed as percent of ideal body weight. In all patients ovarian morphologic condition was assessed by ultrasonography by means of a 5 mHz abdominal or vaginal probe. Classic ultrasonographic criteria of pericortical cysts and increased stroma was sought. 8 Measurements. In all samples serum luteinizing hormone (LH), estradiol, testosterone, DHEA sulfate, 1113hydroxyandrostenedione, 3a-androstanediol glucuronide, and insulin were measured. During the insulin tolerance test glucose was measured. Assays. All hormones were assayed by established radioimmunoassay techniques. 9 - 12 Hormonal determinations for all patients were made in the same laboratory (Los Angeles, Calif.). Estradiol, testosterone, and 1113-hydroxyandrostenedione assays included extraction and Celite chromatography. In the LH assay the standard used was LER 907, which was referenced to the Second International Reference Preparation. The sensitivity of this assay was 1.5 mIU/m!' The 3a-androstanediol glucuronide assay used specific hydrolysis, followed by extraction, Celite column chromatography, and radioimmunoassay. 13 In all assays, intraassay and interassay coefficients of variation did not exceed 6% and 13%, respectively. Statistical analyses. To assess insulin resistance, a
December 1992 Am J Obstet Gynecol
Kitt value, or dissociation constant of the insulin tolerance test expressed as percent per minute was calculated by regression analysis. 6 The Student t test with and without log transformation was used, as was analysis of variance for group comparisons.
Results Comparisons of ideal body weight in North American, Japanese, and Italian women may be found in Fig. 1. Compared with Japanese women, North American women had significantly higher ideal body weight (122% ± 3% vs III % ± 4%, P < 0.05). Ideal body weight in Italian women was intermediate and not significantly different from either group. The presence of cystic ovaries was similar in the three groups: 80%, 68%, and 76%, respectively (Fig. 1). Ferriman-Gallwey scores were significantly elevated and similar in United States and Italian women (12 ± 1 and 12.5 ± 1). In controls, by definition, all Ferriman-Gallwey scores were < 8. The Japanese women had a mean score of 3.5 ± 0.2, which was not statistically different from the scores of Japanese controls. The three ethnic control groups had similar hormone levels, and thus for comparisons with the hyperandrogenic chronic anovulation groups the following composite mean values were used: LH, 9.1 ± 0.5 mU/ml; estradiol, 56 ± 5 pg/ml; testosterone, 32.3 ± 3 ng/dl; DHEA sulfate, 1.8 ± 1 j.Lg/ml; 1113-hydroxyandrostenedione, 1.2 ± 0.1 ng/ml; 3a-androstanediol glucuronide, 3 ± 0.2 ng/ml; and insulin, 8 ± 0.2 j.LU/m!. Serum LH was elevated in all three hyperandrogenic chronic anovulation groups compared with controls (P < 0.01) but were similar to one another (Fig. 2). Serum estradiol was also similar between groups and comparable to control values. Serum testosterone was elevated in all groups compared with controls (P < 0.05), but the values were similar in the three hyperandrogenic chronic anovulation groups (Fig. 2). Serum 3a-androstanediol glucuronide was elevated in United States and Italian women (p < 0.01) when compared with both Japanese women and controls. Japanese women and control values were similar, and United States and Italian women had similar levels (Fig. 2). Fig. 3 depicts levels of the adrenal androgens DHEA sulfate and 1113-hydroxyandrostenedione and the percentage of patients having elevated levels. Levels of both DHEA sulfate and 1113-hydroxyandrostenedione were significantly elevated in the three groups compared with controls (p < 0.01). However, values were similar among the three groups for each androgen. Serum DHEA sulfate was elevated in 52%, 48%, and 48% of North American, Japanese, and Italian women, respectively. Serum 1113-hydroxyandrostenedione was elevated in 64%, 64%, and 56% of patients in each group. Fasting insulin levels were significantly elevated in
Ethnic variations in polycystic ovary syndrome
Volume 167 "umber 6
.........
.........
c
I-
:r: c.9
W
1809
~
'-'"
*
C/l
W ...... 0:: «
>
:3:
0
>0
u
f=
0
C/l
CD
>-
U
**
** C/l W 0::
o
U
(/)
c.9
I.J...
Fig. 1. Percent ideal body weight. hirsutism (Ferriman-Gallway [FG) score) and prevalence of cystic ovaries in three ethnic groups. D. United States; ~. Japan; IllS. Italy. Differences between United States and Italy compared with Japan: One astnisk. p < 0.05; two asterisks. p < 0.01.
r--.
--.J
E
...........
::J E
I --.J
r--.
--.J
E
........... CJ)
a..
'-'
N W
r--.
~
--.J -0
...........
CJ)5
c
'--'"
l-
--.J E ...........
**
**
CJ)
c
'-' ~
I
0 M
Fig. 2. Serum levels of LH. testosterone (T). estradiol (E 2 ). and 3a-androstanediol glucuronide (3a-G) in three ethnic groups of patients with polycystic ovary syndrome. D. United States; ~. Japan; Italy. Significance of 3a-androstanediol glucuronide values between the U.S. and Italy compared to Japan: Two asterisks. p < 0.01.
~.
the United States and Italian groups compared with control values (p < 0.01). In Japanese women this increase was smaller but still of significance (p < 0.05). Both United States and Italian group values for insulin
were significantly higher than values in Japanese women (Fig. 4). Fasting insulin showed a highly significant correlation with percent of ideal body weight in all three groups (p < 0.01).
1810 Carmina at al.
December 1992 Am J Obstet Gynecol
,-..
,-.. .....J E
.....J E
.........
.........
Ol
Ol
::j.
C
'-"
'-"
C/)
