Effects of Exercise on Insulin Resistance and Body Composition in ...

J C E M

O N L I N E

Hot Topics in Translational Endocrinology—Endocrine Research

Effects of Exercise on Insulin Resistance and Body Composition in Overweight and Obese Women with and without Polycystic Ovary Syndrome Samantha K. Hutchison, Nigel K. Stepto, Cheryce L. Harrison, Lisa J. Moran, Boyd J. Strauss, and Helena J. Teede The Jean Hailes Clinical Research Unit School of Public Health (S.K.H., C.L.H., L.J.M., H.J.T.), Department of Physiology (N.K.S., C.L.H.), and Departments of Medicine, Nutrition, and Dietetics (B.J.S.), Monash University, Clayton, Victoria 3800, Australia; and Diabetes Unit (S.K.H., H.J.T.), Southern Health, Clayton, Victoria 3168, Australia; and Institute of Sport, Exercise, and Active Living (N.K.S.), Victoria University, Footscray, Victoria 3011, Australia

Context: Polycystic ovary syndrome (PCOS) is an insulin-resistant (IR) state. Visceral fat (VF) is independently associated with IR. Objectives: The objectives of the study were to explore mechanisms underpinning IR by assessing the effect of exercise training on IR and body composition in overweight PCOS and non-PCOS women. Design: This was a prospective exercise intervention study. Setting and Participants: The study was conducted at an academic medical center. Participants included 20 overweight PCOS and 14 overweight non-PCOS women. Intervention: The intervention included 12 wk of intensified aerobic exercise (3 h/wk). Main Outcome Measures: IR on euglycemic hyperinsulinemic clamp, body composition including abdominal visceral and sc fat distribution by computer tomography and lipids was measured. Results: PCOS subjects were more IR (P ⫽ 0.02) and had more VF (P ⫽ 0.04 age adjusted) than non-PCOS women. In PCOS women, IR correlated with VF (r ⫽ ⫺0.78, P ⬍ 0.01). With exercise training, both groups maintained weight but within PCOS, VF (⫺12.0 cm2, P ⫽ 0.03) and within non-PCOS abdominal sc fat (⫺40.2 cm2, P ⫽ 0.02) decreased. Despite exercise-induced improvement in IR within PCOS (⫹27.9 mg 䡠 m⫺2 䡠 min⫺1, P ⫽ 0.03), no relationship with decreased VF (r ⫽ ⫺0.08, P ⫽ 0.84) and no differential changes in IR and VF between groups were noted. Triglycerides decreased within PCOS (⫺0.27 mmol/liter, P ⫽ 0.02) and decreased differentially between groups (P ⬍ 0.01). Conclusions: Higher IR was related to increased VF in PCOS, suggesting an etiological role for VF in intrinsic IR in PCOS; however, changes with exercise intervention did not support a causal relationship. Triglycerides were modulated more by exercise training in PCOS than non-PCOS women. Within-group exercise-induced reductions in cardiometabolic risk factors including IR, triglycerides, and VF in PCOS were observed without significant weight loss and if confirmed in future controlled trials, suggest weight loss should not be the sole focus of exercise programs. (J Clin Endocrinol Metab 96: E48 –E56, 2011)

ISSN Print 0021-972X ISSN Online 1945-7197 Printed in U.S.A. Copyright © 2011 by The Endocrine Society doi: 10.1210/jc.2010-0828 Received April 12, 2010. Accepted August 31, 2010. First Published Online October 6, 2010

E48

jcem.endojournals.org

Abbreviations: BMI, Body mass index; CT, computed tomography; DM2, type 2 diabetes; FAI, free androgen index; GIR, glucose infusion rate; HDL, high-density lipoprotein; HOMA, homeostatic model assessment; HRmax, maximal heart rate; IR, insulin resistance; LDL, low-density lipoprotein; PCOS, polycystic ovary syndrome; SCFAT, sc fat; VF, visceral fat; VO2max, maximal oxygen consumption; WC, waist circumference.

J Clin Endocrinol Metab, January 2011, 96(1):E48 –E56

E49

Eligible on phone screening (n = 117)

Start of Intervention

Excluded (n = 83) = Declined Involvement (n = 51) + Did not meet inclusion criteria (n = 32) Commenced study (n = 34)

PCOS (n = 20)

Non-PCOS women (n =14)

Did not complete (n = 13) = Lost to contact (n = 4 PCOS) + Discontinued intervention (n = 3 PCOS, 5 non-PCOS) + Protocol violation (n = 1 non-PCOS) Completion

olycystic ovary syndrome (PCOS) is the most common endocrinopathy of reproductive-age women affecting 4 –12% (1, 2), depending on the diagnostic criteria used. PCOS women have increased intrinsic insulin resistance (IR) compared with non-PCOS women independent of obesity (3–5). Obesity further exacerbates IR (so-called extrinsic IR). IR in PCOS underpins reproductive and metabolic features (3, 4) including increased metabolic syndrome, impaired glucose tolerance and type 2 diabetes (DM2) (5). The mechanisms underlying intrinsic IR in PCOS remain unclear. Potential mechanisms include increased abdominal visceral fat (VF) (6). In other IR states, VF has the strongest relationship with IR when fat depots such as abdominal sc fat (SCFAT) are considered (reviewed in Ref. 7). Several mechanisms have been postulated (8); however, a causal relationship between VF and IR has not been established (9, 10). Exercise in other IR states reduces VF and IR (11, 12), and one study suggests there may be a differential metabolic response to exercise in IR subjects vs. controls (13). In PCOS, VF correlates with surrogate markers of IR (6), but this has not been confirmed using optimal measures of IR and VF. Studies have inconsistently reported higher VF in PCOS compared with non-PCOS (14, 15). Molecular analyses of VF from obese PCOS and nonPCOS women show differential expression of IR-related genes and proteins, potentially mechanistically linking VF to IR in PCOS (16). Furthermore, metformin, an insulinsensitizer, reduces VF and surrogate IR markers in PCOS women (17). Lifestyle intervention in PCOS is challenging. High dropout rates occur in PCOS dietary intervention studies (18). Potentially exercise may be more sustainable. Although it is reported that exercise improves surrogate markers of IR in PCOS (19), mechanisms by which exercise improves IR and the possible role of VF are unknown. Potential differential metabolic effects of exercise between IR PCOS and non-PCOS women have also not been explored. PCOS presents a useful model in which to study underlying mechanisms of IR before the onset of confounding hyperglycemia. In this mechanistic study, we aimed to explore IR using the gold standard technique and body composition, in particular VF, in overweight IR PCOS and non-PCOS women. By comparing these groups with similar weight, we aimed to investigate mechanisms underpinning intrinsic IR in PCOS at baseline. Subsequently we aimed to explore IR and VF responses to exercise training, focusing on the relationship between these parameters and on whether exercise differentially affects IR and VF between these groups.

P


Recruitment


Completed (n = 21) = PCOS (n = 13) + Non-PCOS women (n = 8) FIG. 1. Recruitment tree.

Participants and Methods Participants Overweight and obese [body mass index (BMI) ⬎27 kg/m2], premenopausal, women aged 20 – 40 yr with (n ⫽ 20) and without (n ⫽ 14) PCOS (Fig. 1) were recruited through community advertisements. PCOS was diagnosed by an endocrinologist (S.K.H.) based on irregular menstrual cycles (⬍21 or ⬎35 d) and clinical (hirsutism, acne) or biochemical (elevation of at least one circulating ovarian androgen) hyperandrogenism [1990 National Institutes of Health criteria (20)]. Hyperprolactinemia, thyroid dysfunction, and specific adrenal disorders were excluded clinically and where indicated biochemically. All nonPCOS women had regular menses and no evidence of clinical or biochemical hyperandrogenism. Exclusion criteria included use of glucocorticoids, antihypertensives, weight loss, lipid-lowering agents, smoking, DM2, participation in regular physical activity, recent weight change, and pregnancy both at screening and during the 3-month run-in. The Southern Health Research Advisory and Ethics Committee approved the study and participants gave written informed consent. The clinical trial registration number is ISRCTN84763265.

Study design At screening (3 months before baseline), standard diet and lifestyle advice was delivered [Heart Foundation recommendations (www.heartfoundation.org.au)] and medications affecting end points including insulin sensitizers, antiandrogens, and hor-

E50

Hutchison et al.

Exercise and Insulin Resistance in PCOS

monal contraceptives were ceased. Data were collected at 3 months (baseline) and after 12 wk of exercise (study completion). Data were collected in the follicular phase of the menstrual cycle wherever feasible.

Exercise intervention Participants undertook 12 wk of supervised intensified exercise training on a motorized treadmill (three ⫻ 1 h sessions each week) under supervision of exercise physiologists (C.L.H. and N.K.S.). One session consisted of 60 min of moderate-intensity treadmill walking/jogging that elicited work rates of 75– 85% of maximal heart rate (HRmax) equivalent to 70% of maximal oxygen consumption (VO2max). This alternated with high-intensity intermittent exercise, during which participants walked/ jogged on the treadmill (six ⫻ 5 min work bouts with 2 min of recovery) at an exercise intensity of 95–100% HRmax (equivalent to 90 –100% VO2max). Participants progressed to eight repetitions by wk 4 and reduced recovery time to 1 min by wk 8. Target exercise intensity heart rates were achieved by altering speed and incline on the treadmill according to individual fitness. VO2max tests were repeated at 6 wk to assess changes in fitness and HRmax. Heart rate monitors were used in all sessions (Polar Electro Oy, Kempele, Finland).

Clinical and biochemical measurements Participants were weighed lightly clothed without shoes (TBF310; Tanita, Tokyo, Japan). BMI was calculated [weight (kilograms)/height squared (square meters)], (Stadiometer; Holtain, Wales, UK). Waist circumference (WC) was measured at the umbilicus by an experienced operator. Insulin sensitivity was assessed by the euglycemic hyperinsulinemic clamp described by DeFronzo et al. (21). Clamp timing was standardized to 48 h after exercise in all participants. After a standardized high-carbohydrate diet before an overnight fast, an iv catheter was inserted for blood drawing in the dorsal hand and for glucose and insulin infusion in the contralateral arm. Fasting venous blood samples were collected and stored. Serum total cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, and glucose were measured on a Beckman Coulter LX20PRO analyzer using commercial enzymatic kits (Beckman Coulter Diagnostics Australia, Gladesville, Australia). Low-density lipoprotein (LDL) cholesterol was calculated using the modified Frieldwald equation: LDL (calculation) ⫽ total cholesterol ⫺ HDL ⫺ (triglycerides/2.25) adapted to SI units. Plasma insulin was measured using a commercial human insulin-specific RIA kit (Linco Research, St. Charles, MO). Homeostatic model assessment (HOMA) was calculated as fasting serum insulin (milliunits per liter) ⫻ fasting plasma glucose (millimoles per liter)/22.5] as previously described (22). Serum SHBG was measured by an automated enzyme immunoassay on a Diagnostic Products Corp. Immulite analyzer (Diagnostic Products Corp., Los Angeles, CA). Testosterone was measured on Beckman Coulter Unicel DXI 800 analyzer (Beckman Coulter Diagnostics Australia, Gladesville, Australia) using an automated competitive binding immunoenzymatic assay. Free androgen index (FAI) was calculated as testosterone/SHBG ⫻ 100. The blood collection arm was placed in an electric warming pad for arterialization of venous blood. Insulin (Actrapid; Novo Nordisk, Bagsvaerd, Denmark) was infused at 40 mU/m2 䡠 min for approximately 120 min, with plasma glucose maintained at approximately 5 mmol/liter, using variable infusion rates of 25%


glucose. Glucose was assessed every 5 min using a glucose analyzer (YSI 2300 STAT glucose/L-lactate analyzer; Yellow Springs Instruments, Yellow Springs, OH). Glucose infusion rates (GIRs) were calculated during steady state, defined as the last 30 min of the insulin-stimulated period and expressed as glucose (milligrams) per body surface area (square meter) per minute. VO2max was assessed using MOXUS modular VO2 system (AEI Technologies, Pittsburgh, PA) while participants exercised on a treadmill [Biodex RTM 500 (model no. 945-295) New York, NY] until volitional fatigue using a modified Bruce protocol (23). VO2max was defined as the highest oxygen uptake during a 1-min sampling period and HRmax defined as the highest heart rate during a 15-sec sampling period. Fat mass, abdominal fat mass, and fat-free mass were measured by dual-energy x-ray absorptiometry [GE Lunar Prodigy (GE Lunar Corp., Madison, WI) using operating system version 9] and interpreted by a body composition physician (B.J.S.). Abdominal VF and SCFAT were assessed with participants placed supine with arms extended above their head. Single-slice computer tomography (CT) axial images of the abdomen were acquired at L4 –L5 intervertebral disc space level without angulation, using a lateral pilot for location. All scans were performed using a General Electric Lightspeed CT (GE Medical Systems, Milwaukee, WI) scanner and saved as DICOM images for analysis. Standard CT procedures of 120 kV, 5 mm thickness, and a 512 ⫻ 512 matrix were used. The measurement boundary for VF was defined as the innermost aspect of the abdominal and oblique muscle walls and the posterior aspect of the vertebral body. SCFAT area at the L4 –L5 intervertebral disc space was obtained. CT scans were analyzed using Slice-O-Matic version 4.3 software (TomoVision, Magog, Canada). Fat cross-sectional areas were calculated using standard Hounsfield unit ranges by delineating regions of interest with a mouse computer interface. The thresholding function was initially used to set the adipose tissue Hounsfield unit ranges. Compartmental segmentation was computed using standard Hounsfield unit ranges (adipose tissue: ⫺190 to ⫺30 and skeletal muscle: ⫺29 to 150). Adipose tissue cross-sectional area (centimeters squared) was calculated from the pixel areas associated with each region of interest (24). The intrareader variability (coefficient of variation) in VF and SCFAT was less than 1%.

Statistics All data are presented as mean ⫾ SEM. Results are presented for 34 participants (20 PCOS and 14 non-PCOS women) at baseline except for GIR (n ⫽ 29; PCOS, n ⫽ 17; non-PCOS, n ⫽ 12) and CT data (n ⫽ 33; PCOS, n ⫽ 19; non-PCOS, n ⫽ 14). At completion, results are presented for n ⫽ 21 (PCOS, n ⫽ 13; non-PCOS, n ⫽ 8) except for GIR (n ⫽ 16; PCOS, n ⫽ 9; nonPCOS, n ⫽ 7) and CT data (n ⫽ 20; PCOS, n ⫽ 13; non-PCOS, n ⫽ 7). Two-tailed statistical analysis was performed using SPSS for Windows 17.0 software (SPSS Inc., Chicago, IL) with statistical significance set at ␣-level of P ⬍ 0.05. Data were log transformed if not normally distributed (insulin, HOMA) and assessed using Student’s t test with general linear modeling to correct for age. The effect of exercise was assessed using repeated-measures ANOVA with PCOS status as between-subject factor and exercise as within-subject factor. Relationships between variables were examined using bivariate (Pearson) correlations and the impact of covariates assessed using linear regression. Change in variable was defined as the percentage difference between pre- and posttraining values.



E51

TABLE 1. Baseline characteristics Characteristic Waist (cm) Weight (kg) BMI (kg/m2) Age (yr) VO2max (ml 䡠 kg⫺1 䡠 min⫺1) Androgens Testosterone (nmol/liter) SHBG (nmol/liter) FAI Lipids Cholesterol (mmol/liter) Triglycerides (mmol/liter) HDL (mmol/liter) LDL (mmol/liter) IR and glucose metabolism Fasting glucose (mmol/liter) Fasting insulin (pmol/liter) GIR (mg 䡠 m⫺2 䡠 min⫺1) HOMA Body composition and CT Lean tissue mass (kg) Total fat mass (kg) Abdominal fat mass (kg) VF (cm2) SCFAT (cm2)

PCOS (n ⴝ 20) 106.8 ⫾ 3.4 100.5 ⫾ 4.5 37.4 ⫾ 1.5 29.5 ⫾ 1.4 25.2 ⫾ 1.4

Non-PCOS (n ⴝ 14) 102.8 ⫾ 2.6 96.2 ⫾ 3.5 35.7 ⫾ 1.3 35.0 ⫾ 1.1 25.2 ⫾ 0.9

P 0.39 0.42 0.43 0.01 0.98

P adjusteda 0.29 0.74 0.54

2.9 ⫾ 0.2 29.0 ⫾ 1.8 10.7 ⫾ 1.1

1.6 ⫾ 0.2 43.6 ⫾ 7.8 4.6 ⫾ 0.9

⬍0.01 0.04 ⬍0.01

⬍0.01 0.07 ⬍0.01

5.0 ⫾ 0.3 1.4 ⫾ 0.2 1.0 ⫾ 0.1 3.3 ⫾ 0.2

4.7 ⫾ 0.2 1.2 ⫾ 0.2 1.2 ⫾ 0.1 3.0 ⫾ 0.2

0.81 0.46 0.04 0.48

0.85 0.30 0.02 0.54

5.0 ⫾ 0.1 141.6 (100.8 –181.2) 175.6 ⫾ 23.4 5.0 (3.5–7.1)

4.8 ⫾ 0.1 72.6 (58.8 –115.8) 257.2 ⫾ 18.6 2.5 (2.1– 4.0)

0.57 0.02 0.02 0.02

0.21 0.02 ⬍0.01 0.01

48.5 ⫾ 2.0 48.6 ⫾ 2.4 4.7 ⫾ 0.3 129.2 ⫾ 12.8 590.2 ⫾ 35.2

44.7 ⫾ 1.2 47.6 ⫾ 2.8 4.3 ⫾ 0.3 121.5 ⫾ 9.4 550.3 ⫾ 45.2

0.25 0.94 0.40 0.65 0.49

0.24 0.81 0.46 0.04 0.77

0.60

Data are means ⫾ SEM except for insulin and HOMA 关median (interquartile range) P values from log transformed data兴. a

P value adjusted for age.

Power calculations, based on a previous non-PCOS exercise study demonstrating a decrease in IR and a decrease in VF (11), suggested that the current study has a power of 80% and an ␣-level 0.05 with a required sample size of 7.

Results Twenty PCOS and 14 non-PCOS women completed the 3-month run-in with a stable diet and the withdrawal of relevant medications. Thirteen PCOS and eight non-PCOS women completed 12 wk of exercise. The recruitment tree with dropouts is provided (Fig. 1). One participant, eligible at baseline, commenced significant sustained physical activity during the study in violation of the protocol and was excluded from final analysis. PCOS vs. non-PCOS women: baseline characteristics (Table 1) PCOS women were younger than non-PCOS women (29.5 ⫾ 1.4 vs. 35.0 ⫾ 1.1 yr, P ⫽ 0.01) and had higher IR (P ⫽ 0.02) and androgens and lower SHBG and HDL. Despite similar fitness and body composition parameters, age-adjusted VF was higher in PCOS than non-PCOS women (129.2 ⫾ 12.8 vs. 121.5 ⫾ 9.4 cm2, age adjusted P ⫽ 0.04).

PCOS vs. non-PCOS women: effect of exercise training (Table 2) Exercise attendance was similar for both groups [97% PCOS, 92% non-PCOS (P ⫽ 0.19)]. Fitness (VO2max) improved with exercise training (P ⬍ 0.01) and improved significantly within each group with no significant between-group differences. Whole-group weight, BMI, and WC decreased after exercise training. Within groups, weight did not decrease significantly (PCOS, ⫺1.6 ⫾ 0.7 kg, P ⫽ 0.05; non-PCOS, ⫺2.5 ⫾ 1.2 kg, P ⫽ 0.08). BMI was significantly reduced in PCOS (⫺0.6 ⫾ 0.3 kg/m2, P ⫽ 0.03), and WC was significantly reduced in non-PCOS (P ⫽ 0.02). There were no between-group changes with training in weight, BMI, or WC. Total and abdominal fat mass were reduced after training (P ⬍ 0.01) in both groups with no between-group difference. With exercise training, VF decreased in PCOS (P ⫽ 0.03) but not in non-PCOS women (P ⫽ 0.75) (Fig. 2). Conversely, SCFAT decreased in non-PCOS (P ⫽ 0.02) but not in PCOS women (P ⫽ 0.08). There were no between-group differences in change in VF or SCFAT. IR (as measured by GIR) improved in PCOS after training by 16% (P ⫽ 0.03) with no change in the non-PCOS women (P ⫽ 0.07) (Fig. 3A) and no between-group dif-

E52

Hutchison et al.

Exercise and Insulin Resistance in PCOS


TABLE 2. Effect of exercise training PCOS (n ⴝ 13)

Non-PCOS (n ⴝ 8)

Characteristic

Pre

Post

Pre

Post

P value for change with exercise training

P value for change over study PCOS vs. non-PCOS

Waist (cm) Weight (kg) BMI (kg/m2) VO2max (ml 䡠 kg⫺1 䡠 min⫺1) Androgens Testosterone (nmol/liter) SHBG (nmol/liter) FAI Lipids Cholesterol (mmol/liter) Triglycerides (mmol/liter) HDL (mmol/liter) LDL (mmol/liter) IR and glucose metabolism Fasting glucose (mmol/liter) Fasting insulin (pmol/liter) GIR (mg 䡠 m⫺2 䡠 min⫺1) HOMA Body composition and CT Lean tissue mass (kg) Fat mass (kg) Abdominal fat mass (kg) VF (cm2) SCFAT (cm2)

103.6 ⫾ 4.0 96.9 ⫾ 4.8 35.6 ⫾ 1.6 25.9 ⫾ 1.8

103.1 ⫾ 4.0 95.3 ⫾ 4.8 35.0 ⫾ 1.6a 31.4 ⫾ 1.9a

104.8 ⫾ 3.3 99.4 ⫾ 5.4 36.9 ⫾ 2.1 26.1 ⫾ 2.2

99.9 ⫾ 4.1a 96.9 ⫾ 4.5 35.9 ⫾ 1.8 30.7 ⫾ 2.3a

0.02 0.01 ⬍0.01 ⬍0.01

0.05 0.51 0.49 0.54

2.8 ⫾ 0.2 28.3 ⫾ 2.4 10.7 ⫾ 1.4

2.8 ⫾ 0.3 30.7 ⫾ 2.8 10.1 ⫾ 1.6

1.4 ⫾ 0.2 51.7 ⫾ 12.6 3.5 ⫾ 0.7

1.8 ⫾ 0.3 54.3 ⫾ 10.6 4.1 ⫾ 1.1

0.12 0.30 0.99

0.13 0.97 0.18

4.5 ⫾ 0.3 1.1 ⫾ 0.2 1.0 ⫾ 0.1 3.0 ⫾ 0.3

4.4 ⫾ 0.2 0.9 ⫾ 0.1a 1.0 ⫾ 0.1 3.0 ⫾ 0.2

4.6 ⫾ 0.4 1.1 ⫾ 0.1 1.2 ⫾ 0.1 2.9 ⫾ 0.3

4.8 ⫾ 0.4 1.3 ⫾ 0.1 1.2 ⫾ 0.1 3.1 ⫾ 0.4

0.61 0.39 0.21 0.63

0.23 0.01 0.61 0.59

5.0 ⫾ 0.1 139.8 (104.4 –207.0) 171.3 ⫾ 40.2 5.0 (3.7– 8.2)

4.9 ⫾ 0.1 97.8 (66.6 –231.0)a 199.2 ⫾ 35.1a 3.4 (2.3–9.2)a

4.8 ⫾ 0.1 100.2 (65.4 –129.0) 240.4 ⫾ 20.0 3.7 (2.2– 4.2)

4.9 ⫾ 0.1 115.2 (76.2–177.6) 297.5 ⫾ 34.7 4.1 (2.5– 6.3)

0.93 0.73 0.05 0.43

0.56 0.06 0.28 0.05

47.5 ⫾ 2.1 45.6 ⫾ 2.8 4.4 ⫾ 0.3 119.5 ⫾ 16.1 561.4 ⫾ 40.5

47.6 ⫾ 2.3 44.4 ⫾ 2.9a 4.2 ⫾ 0.4a 107.6 ⫾ 15.1a 538.4 ⫾ 40.2

45.3 ⫾ 1.3 49.9 ⫾ 4.4 4.5 ⫾ 0.5 135.1 ⫾ 15.7 598.7 ⫾ 81.7

45.9 ⫾ 9.0 47.0 ⫾ 3.9a 4.2 ⫾ 0.5a 132.7 ⫾ 18.1 558.5 ⫾ 74.5a

0.42 ⬍0.01 ⬍0.01 0.11 ⬍0.01

0.55 0.16 0.71 0.28 0.37

Data are means ⫾ SEM except for insulin and HOMA 关median (interquartile range) P values from log transformed data兴. P ⬍ 0.05 for change within group with exercise training.

CT scan cross-sectional area (cm2)

Correlations At baseline, GIR inversely correlated with VF in PCOS (r ⫽ ⫺0.78, P ⬍ 0.01) and the whole group (r ⫽ ⫺0.69, P ⬍ 0.01) but not in non-PCOS women (Fig. 4). HDL (r ⫽ 0.66, P ⬍ 0.01) and SHBG (r ⫽ 0.68, P ⬍ 0.01) correlated with GIR across the whole group. VF was the only variable to remain independently correlated with GIR when these factors were entered into a linear regression model. VF did not correlate with androgens but did correlate with age (r ⫽ 0.46, P ⫽ 0.01). *

50 40 30 20

Change in GIR with exercise training in PCOS was not correlated with change in VF despite both significantly decreasing from baseline (r ⫽ ⫺0.08, P ⫽ 0.84). The only variable to correlate with the change in GIR in PCOS was change in SHBG (r ⫽ 0.70, P ⫽ 0.04).

A 350 GIR (mg.m-2.min-1)

ference. Androgens did not change. There was a significant between-group difference in the change in triglycerides (P ⫽ 0.01), with PCOS women demonstrating a reduction in triglycerides (⫺0.27 mmol/liter, P ⫽ 0.02) and no change in non-PCOS women (P ⫽ 0.09) (Fig. 3B). No other significant between-group changes were noted.

300 250 200 150

*

100 50 0

B Triglycerides (mmol/L)

a

1.6 1.4

}

1.2

**

1 0.8

*

0.6 0.4 0.2

*

0 Baseline

10 0 Visceral Fat

Subcutaneous Fat

FIG. 2. Decrease in abdominal fat on CT after training. PCOS (gray bars, n ⫽ 13) and non-PCOS (black bars, n ⫽ 7). Data are mean ⫾ SEM. *, P ⬍ 0.05 within-group difference.

Post-training

FIG. 3. A, Change in GIR with training in the PCOS (E, n ⫽ 9) and nonPCOS women (F, n ⫽ 7). *, P ⫽ 0.03 for change in GIR (paired t test) in PCOS with training. B, Change in triglycerides with training in PCOS (E, n ⫽ 13) and non-PCOS (F, n ⫽ 8) subjects. *, P ⫽ 0.02 for change in triglycerides (paired t test) with training in the PCOS women; **, P ⫽ 0.01 for change with training PCOS vs. non-PCOS. Data are mean ⫾ SEM.

GIR (mg.m-2.min-1)


400 350 300 250 200 150 100 50 0


R2 = 0.60, p