Effect of Lifestyle Intervention on Cardiometabolic Risk Factors in Overweight and Obese Women with Polycystic Ovary Syndrome: A Systematic Review and Meta-Analysis

Abstract

The effects of lifestyle interventions (LSIs) on cardiometabolic risk factors in women with polycystic ovary syndrome (PCOS) are controversial. We conducted a systematic review and meta-analysis to assess the effectiveness of LSI on cardiometabolic risk in women with PCOS. We reviewed five databases for randomized controlled trials (RCTs) published between 2012 and December 2016, evaluating the effect of LSI among adult, reproductive age group, overweight, and obese women with PCOS. Outcome measures included weight change, homeostatic model assessment-insulin resistance index (HOMA-IR), total cholesterol (TC), triglyceride (TG), low-density lipoprotein (LDL), and high-density lipoprotein (HDL). A total of 564 studies were screened; 35 full text studies assessed for eligibility, 20 excluded with reasons, 15 included in the systematic review, and 9 studies from 8 RCTs included in a meta-analysis. LSI resulted in significant reduction in weight in sensitivity analyses [study 5 dietary arm as intervention: standardized mean difference (SMD) = −0.81 (95% confidence interval {CI} = −1.24 to −0.39); study 5 combination arm as intervention: SMD = −0.85 (95% CI = −1.26 to −0.45)], HOMA-IR overall [study 5 dietary arm: SMD = −1.46 (95% CI = −2.72 to −0.20); study 5 combination arm SMD = −1.50 (95% CI = −2.76 to −0.23)], and LDL overall [SMD = −1.06 (95% CI = −2.00 to −0.12)]. LSI was more beneficial than Metformin in decreasing LDL [SMD = −2.60 (95% CI = −5.15 to −0.05)] but similar to comparator group in terms of TC, TG, and HDL (P > 0.05). Short-term LSI can effectively improve weight, insulin resistance, and lipid profile and could be recommended as first-line therapy in overweight and obese PCOS women to reduce cardiometabolic risk.

Introduction

Polycystic ovary syndrome (PCOS), a common reproductive endocrinopathy, affects 6%–20% reproductive age women worldwide and 5 million women in the United States.^1,2 Women with PCOS have hyperandrogenism (e.g., hirsutism and acne), menstrual irregularities, reproductive abnormalities (e.g., anovulation and infertility), and metabolic derangements (e.g., insulin resistance and dyslipidemia).² They have higher risk for type II diabetes mellitus, hypertension, obesity, and cardiovascular disease.^3,4

Obesity coexists in up to 80% of women with PCOS.^5,6 Hyperinsulinemia, hyperandrogenism, and anovulatory infertility associated with PCOS could be partly attributed to centripetal obesity.⁷ In fact obesity aggravates the clinical manifestations of PCOS, including cardiometabolic and reproductive features.⁸ It creates a pro-inflammatory environment by increasing cytokine levels and activity, which interfere with insulin action, resulting in compensatory hyperinsulinemia. Obesity also promotes hyperandrogenism through insulin resistance while favoring low-grade chronic inflammation and coagulation disruption, leading to increased cardiovascular risk.^7,9

Primary prevention of cardiovascular disease in women with PCOS is the key. Growing evidence supports cost-effective strategies, including lifestyle interventions (LSIs), in improving cardiometabolic risk in the general population.^10
–12 The American College of Obstetricians and Gynecologists recommends LSI [dietary modifications and/or increased physical activity (PA)] for PCOS women.⁴ These interventions show promise in ameliorating PCOS symptoms.^13,14 However, there are mixed results about the role of LSI in cardiometabolic risk reduction in PCOS populations.¹⁵

A 2011 Cochrane review evaluating the impact of LSI in PCOS women was followed by another meta-analysis of studies until 2013.^15,16 Both report inconsistent LSI effects on insulin resistance—a crucial cardiometabolic risk factor. Haqq et al.,¹⁵ in 2015 reported no change in insulin resistance or insulin levels, while the Cochrane review demonstrated beneficial effect on insulin levels with LSI. None of the included studies in these meta-analyses compared the effect of different kinds of diet alone. Epidemiological evidence suggests beneficial effects of dietary modification alone on PCOS and associated cardiometabolic risk.^5,17,18 Therefore, we performed a systematic review and meta-analysis to summarize and update the evidence regarding LSI effects, including diet, on cardiometabolic risk among overweight and obese women with PCOS, as obesity is so prevalent in PCOS and worsens PCOS phenotype, as well as cardiovascular disease risk.^7,8

Methods

Data sources

A systematic search was conducted from April 2013 to December 2016. A comprehensive search of following databases was performed: PubMed, Scopus, Embase, Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library), and CINAHL Plus. Following search terms were used: “Polycystic Ovary Syndrome,” “Polycystic Ovarian Syndrome,” “Stein Leventhal Syndrome,” “Stein-Leventhal Syndrome,” “lifestyle intervention,” “lifestyle therapy,” “lifestyle modification,” “lifestyle changes,” “behavioral/behavioural intervention,” “behavioral/behavioural therapy,” “behavioral/behavioural modification,” “behavioral/behavioural changes,” “diet,” “nutrition,” “meal,” “food,” “dietary intervention,” “dietary therapy,” “energy restriction,” “energy reduction,” “calorie restriction,” “calorie reduction,” “hypocaloric diet,” “nutrition therapy,” “feeding behavior,” “eating behavior,” “exercise,” “physical exercise,” “aerobic exercise,” “physical activity,” “walk,” “jog,” “swim,” “weight lifting,” “dance,” “aerobics,” “physical training,” “resistance training,” “circuit training,” “weight training,” “endurance training,” “structure training,” “physical education,” and “sports” PA. The bibliographies of selected articles were hand searched for additional related references.

Study selection

Eligibility criteria were initially applied by two independent reviewers; any disagreement was resolved through a consensus among all study authors. Based on the population, intervention, comparison, and outcome framework, we established a priori inclusion/exclusion criteria for study identification.¹⁸ We included randomized controlled trials (RCTs) evaluating the role of LSIs on a range of health outcomes of adult (≥18 years) reproductive age group, overweight, and obese women, with PCOS, regardless of their pregnancy status and criteria used to diagnose PCOS. LSI was operationally defined as: dietary intervention alone; combination of diet and PA; or a behavioral intervention to modify either healthy eating alone or combination of healthy eating and PA, irrespective of the duration of the intervention. Studies were considered eligible if the comparison arm received any form of PA alone, any dietary intervention other than the one received by the intervention arm, usual diet, Food and Drug Administration-approved antiobesity medicines (Orlistat, Lorcaserin, Liraglutide, Phentermine/Topiramate, and Naltrexone/Bupropion), and drugs used to manage PCOS symptoms, such as Metformin and Oral Contraceptives. Outcomes of interest included change in weight, insulin resistance marker [homeostatic model assessment-insulin resistance index (HOMA-IR)], and lipid profile, such as total cholesterol (TC), triglycerides (TGs), low-density lipoprotein (LDL), and high-density lipoprotein (HDL). Animal studies, studies comprising of healthy/normal weight/underweight women, with participants receiving supplements/herbal medicines/alternate medicine (e.g., acupuncture)/bariatric surgery, and those without a full text in English language were excluded. We also excluded studies where dietary intervention involved reducing calorie intake to <1200 kcal/day.¹⁹ In addition, conference articles or abstracts and unpublished data were also excluded.

Using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram, two independent reviewers initially identified 850 potential articles by their titles and abstracts²⁰ (Fig. 1). After removing duplicates, we screened titles or abstracts of 564 articles. Thirty-five full text articles were retrieved for assessment of eligibility where titles and abstracts could not provide relevant information; of these 20 were excluded with reasons. Finally 16 articles (15 RCTs) were included for qualitative synthesis, and 9 studies (8 RCTs) were included in the quantitative synthesis.^{21

–29}

FIG. 1.

PRISMA flow diagram of study selection. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

The selected studies provided data on 290 women with PCOS. Four RCTs compared two different diets^{22,23,27
–29}; one intervened with diet versus metformin,²⁵ while two others used a combination of diet and PA versus metformin/orlistat.^24,26 The duration of interventions ranged from 8 to 24 weeks. Six out of eight RCTs were conducted in low–middle-income countries (Brazil, India, and Iran), one in Denmark, while only one was carried out in the United States. The study participants' age ranged from 18 to 40 years, with body mass index (BMI) from 25 to 55 kg/m². One study adjusted BMI according to Asian standards and included participants with BMI of 23 kg/m² and above.²⁶ Another study did not specify BMI in its selection criteria and reported the average BMI at baseline as 30.6 ± 7.8 and 30.5 ± 8.5 in the intervention and comparator arms, respectively.²⁹ A comprehensive description of included studies' characteristics is given in Table 1.

Table 1.

Characteristics of Studies Included in the Meta-Analysis

Study	Population and setting	Sample size	Intervention description	Outcomes	Findings
Curi et al.²⁴	Population: Women with PCOS, BMI >25 kg/m², and age 18–34 years Setting: Gynecology Division of the Obstetrics and Gynecology Department, Sao Paulo, Brazil	I: 12; C: 15	Design: Open-label RCT Intervention: Combination of diet and PA Comparator: Metformin 250 mg capsule PO × 2/day Duration: 6 months	BMI, HOMA-IR	BMI = (I: 31.8 ± 1.6 at baseline vs. 30.1 ± 1.5 at 6 months; P = 0.004) and (C: 31.1 ± 1.4 at baseline vs. 30.2 ± 0.8 at 6 months; P = 0.041) HOMA-IR = (I: 3.373 ± 0.847 at baseline vs. 2.765 ± 0.444 at 6 months; P = 0.440) and (C: 3.901 ± 0.566 at baseline vs. 3.309 ± 0.417 at 6 months; P = 0.378)
Mehrabani et al.²⁸	Population: Volunteer women with PCOS, 20–40 years old, BMI >25 and <38 kg/m² Setting: Tehran, Iran	I: 23, C: 26	Design: RCT Intervention: Modified hypocaloric diet Comparator: Conventional hypocaloric diet Duration: 12 weeks	Weight loss %, HOMA-IR, TC, TG, HDL, and LDL	Δ weight % = I: 4.1% ± 0.58%, C: 3.3% ± 0.62% (P < 0.05 for both) Δ HOMA-IR = I: −0.8 ± 0.2 (P < 0.05) vs. C: −0.3 ± 0.2 (NS) Δ TC/TG/HDL/LDL = NS in both groups
Sorensen et al.²⁹	Population: Women with PCOS Setting: Copenhagen, Denmark	I: 14, C: 13	Design: Parallel group, controlled trial (assignment through rank minimization) Intervention: High protein diet Comparator: Standard protein diet Duration: 6 months	Body weight, BMI, TC, TG, HDL, LDL	Weight = I: 81.8 (68.5–95.1) at baseline vs. 74.1 (63.3–84.9) at 6 months and C: 78.7 (69.7–87.7) at baseline vs. 75.4 (68.1–82.8) at 6 months; P = 0.002 BMI = I: 29.7 (24.8–34.6) at baseline vs. 26.9 (22.9–30.9); P < 0.001 Δ TC/TG/HDL/LDL = NS both groups
Esfahanian et al.²⁵	Population: Women aged 20–30 years, BMI ≥27 kg/m² with PCOS Setting: Outpatient endocrine clinic, Tehran, Iran	I: 13; C: 17	Design: RCT Intervention: hypocaloric diet Comparator: metformin only Duration: 12 weeks	BMI, HOMA-IR, TC, TG, HDL, LDL	BMI = I: pre 34.1 ± 5.4 vs. post 30.1 ± 5.5 and C: pre 31.1 ± 3.3 vs. post 30.3 ± 3.5; P < 0.001 for both HOMA = I: pre 3.81 ± 1.83 vs. post 2.03 ± 1.09; P = 0.001 and C: pre 3.28 ± 1.9 vs. post 2.64 ± 1.7; P = 0.012 TC = I: pre 217.5 ± 66 vs. post 175.6 ± 42; P = 0.007 and C: pre 220.5 ± 54.1 vs. post 212.8 ± 58.2; NS LDL = pre I: 110.4 ± 32.4 vs. post 86.4 ± 23.4; P = 0.003 and C: pre 114.5 ± 29.2 vs. post 107.6 ± 25.3; NS TG/HDL: NS in both groups
Nybacka et al.²¹	Population: women with PCOS, BMI >27 kg/m², and age 18–40 years Setting: Women's Health Research Unit, Karolinska University Hospital, Stockholm, Sweden	I: 14 and 12; C: 17	Design: 3 arm RCT Interventions: 1. Diet 2. Combination of diet and PA Comparator: PA alone Duration: 4 months	BMI, HOMA-IR	BMI = diet: pre 35.4 ± 4.9 vs. post 32.9 ± 5.5; P < 0.001; Combined: pre 38.1 ± 7.0 vs. post 36.9 ± 8.0; P < 0.001; PA only: pre 34.8 ± 5.2 vs. post 34.1 ± 5.7; P < 0.05. HOMA-IR = diet: pre 3.6 (3.0–6.9) vs. post 2.9 (2.2–3.6); P < 0.01 and NS in combined and PA only groups
Kumar and Arora²⁶	Population: Overweight and obese infertile women (BMI >23 kg/m², adapted version for Asians), euthyroid, with PCOS, age <40 years Setting: Assisted reproduction center, Manipal, Karnataka, India	I: 30; C: 30 each	Design: 3 arm, RCT Intervention: Combination of diet and PA Comparator: 1. Metformin with diet and PA combined 2. Orlistat with diet and PA combined Duration: 12 weeks	Weight, BMI, HOMA-IR, TC, TG, LDL, HDL	% Δ weight = I: −0.47 ± 0.26 vs. M: −7.78 ± 0.57; P < 0.001 % Δ BMI = I: −4.60 ± 0.27 vs. M: −8.40 ± 0.65; P < 0.001 % Δ TC = I: 0.3 ± 1.9 vs. M: −4.33 ± 2.90; NS % Δ TG = I: 5.89 ± 4.53 vs. M: 0.98 ± 3.34; NS % Δ LDL = I: −4.47 ± 2.66 vs. M: −1.32 ± 1.82; NS % Δ HDL = I: 2.23 ± 1.27 vs. M: 2.37 ± 2.68; NS
Turner-McGrievy et al.²⁷	Population: Overweight and obese women (BMI 25–49.9 kg/m²), age 18–35 years, with PCOS Setting: South-eastern United States	I: 9; C: 9	Design: RCT Intervention: Vegan diet, low-fat, low-glycemic index Comparator: Low calorie diet Duration: 6 months	% Δ weight	% weight loss at 6 months = I: 0.0 (−5.1 to 0.0) vs. C: 0.0 (−0.3 to 0.0); NS
Asemi et al.²³	Population: Overweight and obese women (BMI ≥25 kg/m²), with PCOS, aged 18–40 years Setting: Gynecology clinics, Kashan, Iran	I: 24; C: 24	Design: Parallel, RCT Intervention: DASH diet Comparator: Iranian traditional diet Duration: 8 weeks	Weight, BMI, TC, TG, HDL, LDL	Δ weight = I: −4.4 ± 2.7 vs. C: −1.5 ± 2.6; P < 0.001 Δ BMI = I: −1.7 ± 1.1 vs. C: −0.6 ± 0.9; P < 0.001 Δ TG = I: −10 vs. C: 19; P < 0.005 Δ TC/LDL/HDL: NS
Asemi et al.²²	Population: Overweight and obese women (BMI ≥25 kg/m²), with PCOS, aged 18–40 years. Setting: Gynecology clinics, Kashan, Iran	I: 24; C: 24	Design: Parallel, RCT Intervention: DASH diet Comparator: Iranian traditional diet Duration: 8 weeks	HOMA-IR	Δ HOMA-IR = I: −0.45 vs. C: 0.8; P = 0.01

BMI, body mass index; C, control; DASH, dietary approaches to stop hypertension; HDL, high-density lipoprotein; HOMA-IR, homeostatic model assessment-insulin resistance index; I, intervention; LDL, low-density lipoprotein; NS, not significant; PA, physical activity; PCOS, polycystic ovary syndrome; PO, per oral; RCT, randomized controlled trial; TC, total cholesterol; TG, triglyceride; Δ, change.

Data extraction

We used a prior developed data extraction form, based on our selection criteria, for data extraction. Information was retrieved by one reviewer after consultation with all study authors on: (1) study (study design, setting, eligibility criteria, sample size, withdrawal/dropout rate, and risk of bias); (2) participants (baseline characteristics of intervention and comparator groups; (3) intervention versus comparator arm (type, content, frequency, and duration of delivery); and (4) treatment effect on outcome measures. In instances where a trial had multiple publications, we inspected all publications to get comprehensive and accurate information on all aspects of the trial. However, for the sake of outcomes' analysis, we included only those articles that were published after Haqq et al.'s work.¹⁵ Missing data were obtained by contacting study authors of the included studies.

Risk of bias assessment

An assessment of the validity of included studies was carried out in accordance with the Cochrane risk of bias tool (RoB).³⁰ Bias evaluation was performed using the risk of bias table (Supplementary Table S1). The domains examined for each study included the following: sequence generation (selection bias), allocation concealment (selection bias), blinding of patients, investigators (performance bias) or outcome assessors (detection bias), incomplete outcome data (attrition bias), selective outcome reporting (reporting bias), and other sources of bias. A decision was made for the severity of bias for each of the aforementioned domains and ranked as “low risk” (all or most of the criteria fulfilled, the likelihood of bias affecting effect estimates was small), “high risk” (all or most of the criteria were unfulfilled and bias likely to affect the effect estimates), and “uncertain risk” (insufficient information about any of the criteria to make a clear-cut decision).^18,30 Disagreements about risk of bias were resolved by consensus among the study authors.

Statistical analyses

Statistical analyses were conducted using RevMan Software 5.3.³¹ The following data were entered into the RevMan software for both the intervention and the comparator arm of each study: total number of participants and mean changes in outcome measures along with standard deviations (SDs). Where studies reported pre/postintervention outcome measurements, mean changes in outcome measures were first computed for both arms by subtracting the pre- from the postmeasurements. For missing SDs, we either contacted the individual study authors to obtain missing information or calculated SDs using the P values given for the postintervention change in mean for each group. One of the studies included in the meta-analysis—referred to as study 5 or Nyback et al.'s study in our article-had three arms: receiving diet only, combination of diet and PA, and PA only (Table 1). in our article—had three arms: receiving diet only, combination of diet and PA, and PA only.²¹ Therefore, we performed two separate meta-analyses for outcomes reported by study 5, that is, weight loss and HOMA-IR—one using study 5 diet-only arm as intervention and the other with combination arm as intervention, while using exercise arm as comparator on both occasions.

The pooled effect sizes were computed using random effect models and reported as standardized mean differences (SMDs) along with 95% confidence intervals (CIs). The effect size was considered significant if the CI of any SMD did not include zero. The SMD is a useful summary statistic in instances where studies use different measurement units or outcome measures, for example, weight loss measured as change in BMI or body weight. It thus makes the results less heterogeneous and more generalizable compared to mean difference.³² The significance level was set at 5% (P < 0.05). The I ² test statistic was used to quantify percentage of inconsistency across studies' results not attributed to chance alone.³³

To account for the high variance across studies and given the use of Metformin to treat insulin resistance in PCOS, we performed two different sensitivity analyses³⁴: (1) excluding RCTs that used Metformin in the comparison arm and (2) including only those RCTs that used Metformin in the comparison arm.

Results

In this meta-analysis, LSI resulted in significant reduction in weight in sensitivity analysis, insulin resistance index in both main and sensitivity analyses, as well as LDL in both main and sensitivity analyses.

Weight change

Overall LSI, including both dietary and combination interventions, did not have significant impact on weight in overweight and obese women with PCOS. This relationship remained unchanged whether we included study 5 dietary arm as intervention or combination arm (Fig. 2A, B). The change in weight with inclusion of study 5 dietary intervention was at SMD = 0.12 (95% CI = −1.04 to 1.28) and with study 5 combination arm at SMD = 0.10 (95% CI = −1.07 to 1.27).

FIG. 2.

Overall effect of LSI on weight loss. (A) Comparison of LSI versus comparator using Nybacka A's dietary arm as intervention. (B) Comparison of LSI versus comparator using Nybacka A's combined arm as intervention. CI, confidence interval; LSI, lifestyle intervention; SD, standard deviation.

Sensitivity analyses

After excluding RCTs that used Metformin in the comparison arm (Fig. 3A–C), LSI versus comparator resulted in significant weight loss [using study 5 dietary arm as intervention: SMD = −0.81 (95% CI = −1.24 to −0.39); using study 5 combination arm as intervention: SMD = −0.85 (95% CI = −1.26 to −0.45)]. In contrast, the effect of LSI was not significantly different than Metformin in terms of change in weight with SMD = 1.76 (95% CI = −2.09 to 5.61).

FIG. 3.

Weight loss sensitivity analyses. (A) Excluding RCTs using Metformin in comparison arm and using Nybacka A's dietary arm as intervention. (B) Excluding RCTs using Metformin in comparison arm and using Nybacka A's combined arm as intervention. (C) Comparison of LSI versus Metformin. RCTs, randomized controlled trials.

Insulin resistance

Overall LSI, compared with comparator groups, resulted in significant improvement in insulin resistance. The insulin resistance index, HOMA-IR, decreased significantly, with SMD = −1.46 (95% CI = −2.72 to −0.20) after inclusion of study 5 dietary arm as intervention and SMD = −1.50 (95% CI = −2.76 to −0.23) after inclusion of study 5 combination arm as intervention (Fig. 4A, B).

FIG. 4.

Overall effect of LSI on HOMA-IR. (A) Comparison of LSI versus comparator using Nybacka A's dietary arm as intervention. (B) Comparison of LSI versus comparator using Nybacka A's combined arm as intervention. HOMA-IR, homeostatic model assessment-insulin resistance index.

Sensitivity analyses

When we excluded RCTs that used Metformin in the comparison arm (Fig. 5A–C), the HOMA-IR reduced further with SMD = −2.18 (95% CI = −3.75 to −0.60) and SMD = −2.24 (95% CI = −3.68 to −0.80) using study 5 dietary arm as intervention and study 5 combination arm as intervention, respectively. The effect of LSI on HOMA-IR was not significantly different compared with Metformin with SMD = −0.74 (95% CI = −2.28 to 0.81).

FIG. 5.

HOMA-IR sensitivity analyses. (A) Excluding RCTs using Metformin in comparison arm and using Nybacka A's dietary arm as intervention. (B) Excluding RCTs using Metformin in comparison arm and using Nybacka A's combined arm as intervention. (C) Comparison of LSI versus Metformin.

Lipid profile

In terms of lipid control, overall LSI versus comparator groups had a beneficial impact. LDL [SMD = −1.06 (95% CI = −2.00 to −0.12)] decreased significantly (Fig. 6A), while there was a downward although insignificant trend in TC [SMD = −1.03 (95% CI = −2.58 to 0.53)], TGs [SMD = −0.81 (95% CI = −2.12 to 0.51)], and HDL [SMD = −0.57 (95% CI = −1.74 to 0.60)] (Fig. 7A–C).

FIG. 6.

Lipids-LDL: (A) Overall effect of LSI on LDL. Sensitivity analyses: (B) Comparison of LSI versus comparator after excluding RCTs using Metformin in comparison arm. (C) Comparison of LSI versus Metformin. LDL, low-density lipoprotein.

FIG. 7.

Overall effect of LSI on other lipids. (A) Total cholesterol. (B) Total triglyceride. (C) High-density lipoprotein.

Sensitivity analyses

Even after excluding RCTs with Metformin as comparator, LSI versus comparator had a nonsignificant effect on TC [SMD = −0.36 (95% CI = −0.85 to 0.12)], TG [SMD = −0.27 (95% CI = −1.04 to 0.51)], and HDL [SMD = 0.55 (95% CI = −0.06 to 1.15)]. Compared with Metformin, LSI had significant improvement in LDL only [SMD = −2.60 (95% CI = −5.15 to −0.05)] (Fig. 6B, C) and nonsignificant effect on TC [SMD = −3.17 (95% CI = −13.18 to 6.83)], TG [SMD = −2.07 (95% CI = −8.61 to 4.47)], and HDL [SMD = −3.17 (95% CI = −9.39 to 3.05)].

Risk of bias analysis

Risk of bias assessment is presented in Fig. 8 and Supplementary Fig. S1. All RCTs described adequate methods of random sequence generation except for one.²⁴ However, only 2 of 8 studies concealed the allocation assignment.^21,28 Lack of adequate allocation concealment despite random assignment raises concern for selection bias.³⁰ Except for Asemi et al.²³ and Mehrabani et al.,²⁸ majority of the studies did not blind the participants or the study personnel as it is difficult to perform blinding in most behavioral interventions. This creates potential source of performance bias. However, Sørensen et al.'s²⁹ and Turner-McGrievy et al.'s²⁷ studies were regarded at low risk of performance bias as both the arms in both studies received dietary interventions. Only Asemi et al.²³ blinded outcome assessment (by blinding all study personnel, except for the study dietician). Nevertheless, we regarded the objectively assessed outcomes at low risk of detection bias (insulin resistance index and lipid profile) compared to the subjectively assessed weight change.³⁰ As with behavioral interventions, most of the studies encountered high dropout rates. Three of the seven trials reporting dropouts accounted for the attrition bias by conducting intention-to-treat analysis.^23,27,29 In terms of reporting bias, all studies reported all prespecified outcomes that are of interest in this meta-analysis except for Curi et al.,²⁴ which did not report data on lipid profile. We did not identify any other serious potential sources of bias in the included studies.

FIG. 8.

Risk of Bias Summary. Review authors' judgments about each risk of bias item for each included study. Green circles with positive symbols represent low risk of bias, red circles with negative symbols represent high risk of bias, and yellow circles with question marks represent uncertain risk of bias.

Discussion

This study presents a systematic review and meta-analysis on the effectiveness of LSI, both dietary and combination of diet and PA, versus comparator groups on cardiometabolic risk factors of overweight and obese women with PCOS. Our work is an update of a prior meta-analysis by Haqq et al.¹⁵ In addition, we evaluated the effect of different forms of diet, independent of PA, on a range of cardiometabolic risk factors, namely weight, insulin resistance, and lipid profile. Our analyses reveal that short-term LSI (both dietary and combination) results in beneficial effect on weight insulin resistance and lipid control. In addition, LSI was not significantly different than Metformin in terms of its effect on weight change and insulin resistance and more effective than Metformin in reducing LDL.

In our study, LSI resulted in significant weight loss only after excluding RCTs comparing LSI versus metformin (sensitivity analysis). However, in the main analysis, weight loss was not statistically significant. The pooled effect estimate in the main analysis was driven predominantly by Kumar and Arora study²⁶ (Fig. 2). In this study, compared to LSI only arm, the Metformin arm received both Metformin and LSI and subsequently lost significantly more weight probably because the effect of Metformin was potentiated by LSI or vice versa. Nevertheless, the finding of significant weight loss due to LSI in the sensitivity analyses (Fig. 3A, B) corroborates with earlier research and is encouraging, given the high prevalence of obesity among women with PCOS.^6,15,16,35 Moreover, weight loss helps in achieving optimal glycemic and lipid control and is, thus, advantageous to overall cardiovascular health.³⁶ It is noteworthy here that assessing weight change in terms of BMI could be misleading due to a variety of reasons. BMI doesn't differentiate between loss of lean body mass and body fat—the latter being the most beneficial aspect of weight loss.^37,38 Nonessential body fat poses health risks, such as diabetes mellitus, hypertension, cardiovascular disease, and cerebrovascular accident, and is increasingly being targeted by weight loss/LSIs.³⁸ Furthermore, BMI has limited sensitivity and specificity and it does not account for other weight influencing factors, namely age, race, sex, and muscle mass.³⁷ However, both researchers and clinicians continue using BMI and weight to assess an individual's health for ease of measurement and interpretation, as well as inexpensiveness.³⁸

In terms of insulin resistance in women with PCOS, prior meta-analyses present mixed results.^15,16 Of note, 5 of 8 studies included in our meta-analysis evaluated the effect of dietary modification only versus usual care/Metformin, whereas studies included in prior meta-analyses were assessing PA or combination of PA. We found a beneficial effect of LSI on insulin resistance and LDL, regardless of the dietary composition, in overweight and obese women with PCOS. Visceral adiposity—a common occurrence in women with PCOS—is strongly associated with insulin resistance and hyperlipidemia, both of which are harbingers of cardiovascular disease.⁷ However, the relationship between insulin resistance and weight loss is not directly linear.³⁹ In fact, LSI can positively influence insulin resistance and lipid metabolism, even in the absence of significant weight loss.⁴⁰ Consistent with prior studies, we observed a significant decline in LDL and a trend toward improvement in TC and TG.^15,16 Similar to the results of the Diabetes Prevention Program (DPP) research, LSI did not favorably affect HDL in our study.³⁶ Most of the studies assessing HDL in our meta-analysis focused on the dietary aspect of LSI. It is postulated that reducing total dietary fat intake inevitably decreases HDL level, while exercise tends to increase it.^41
–43 Given that dietary changes tend to cause a decrease in TC, TG, and LDL and increased PA favors increase in HDL, we recommend combining the two approaches as they are likely to have a synergistic effect.

Of particular note is the finding that compared to Metformin, LSI was not significantly different in improving insulin resistance and exhibited superiority in reducing LDL levels in overweight and obese women with PCOS. Perhaps, if the follow-up in the included studies was longer, we could observe a similar decline in insulin resistance also.^35,39 Interestingly, LSI versus Metformin was found to be more effective in preventing diabetes among older participants (>45 years) in the DPP study,³⁵ whereas our analysis was composed of a relatively younger population. Nevertheless, the above findings and the adverse effects of Metformin justify preferential treatment with LSI while managing women with PCOS having cardiometabolic risk factors.³⁴

Limitations

Our analyses have several limitations, which warrant further investigation by future studies. Lack of information on blood pressure in the included studies precluded evaluation of this important cardiometabolic risk factor. Based on Cochrane RoB, the evidence presented in the included studies is of low-to-moderate quality. Our study was limited by small sample size. Moreover, the I ² test statistic, which quantifies the degree of heterogeneity due to variability between studies, was high. Therefore, the results need a cautious interpretation. Possible explanation of high heterogeneity could be variability in terms of duration of intervention, content of intervention, including prescription of diet and PA, study population, as well as the comparator group treatment. In addition, the included studies were conducted in different countries. Therefore, a proportion of heterogeneity in the results could be attributed to different sociocultural trends of the study populations, which shape dietary habits and PA, thereby potentially affecting the outcomes.⁴⁴ The comparator group treatment was variable; some followed usual dietary pattern, some received a specific dietary prescription different from that given to the intervention arm, some received an exercise prescription, while others were prescribed metformin. We addressed the variance by performing sensitivity analyses, which decreased the I ² value for certain outcomes, such as weight loss and LDL. The attrition rate in most of the studies was high as is the case in studies targeting behavioral or lifestyle modification. Selective attrition rate in either group can cause selection bias and compromise external validity of results.⁴⁵ This can be efficiently tackled through intention-to-treat analysis, which only few of the included studies performed. Furthermore, the included studies looked at short-term effect of the LSI (≤24 months). It is unclear whether LSI would similarly affect the cardiometabolic profile of overweight and obese women with PCOS in the long term.

Conclusion

Short-term LSI is more effective compared with a comparator group and somewhat comparable with pharmacologic approach, Metformin, in improving weight, insulin resistance, and dyslipidemia. Compared with Metformin, LSI was more effective in reducing LDL in women with PCOS. However, larger studies in diverse populations are warranted to further establish evidence. LSI is a low cost, noninvasive, noninjurious approach that could be used as the first step in the management of overweight and obese women with PCOS, especially those at high risk of developing cardiovascular disease. Our findings could help formulate guidelines for clinicians in the management of overweight and obese women with PCOS. This is crucial given the high prevalence of cardiometabolic risk factors among women with PCOS.⁴⁶ Individualized prescription of LSI may result in sustainable benefits and possibly prevent cardiovascular disease, as well as associated complications, in this patient population.^46,47

Footnotes

Acknowledgment

The authors acknowledge Kira Pavlik at Yale School of Medicine who served as the second reviewer and assessed potential articles for study selection.

Ethics Approval and Consent to Participate

Not applicable. The research was conducted in accordance with the Helsinki Declaration as revised in 2013.

Authors' Contributions

K.K. conceived the article; K.K. and V.N. performed the statistical analysis; K.K., V.N., and V.C. interpreted the results; K.K. drafted the initial article; K.K., V.N., and V.C. provided feedback on the content of the article and editorial support; and K.K. modified all subsequent drafts. All authors read and approved the final draft of the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This research project is supported by the Health Resources and Services Administration (HRSA) of the U.S. Department of Health and Human Services (HHS)—HRSA Primary Care Training and Enhancement Grant, Grant Number TOBHP285720100-01-00—as part of an award totaling $1,686,490.00 with 0 percent financed with nongovernmental sources. The contents are those of the author(s) and do not necessarily represent the official views of, nor an endorsement, by HRSA, HHS, or the U.S. Government. HRSA has no role in the design of the study, collection, analysis, and interpretation of data, and in writing the article. For more information, please visit HRSA.gov

Supplementary Material

Supplementary Figure S1

Supplementary Table S1

References

CDC PCOS and Diabetes, Heart Disease and Stroke. 2018. Accessed at https://www.cdc.gov/diabetes/library/spotlights/pcos.html on 17 July, 2017.

Sirmans

, Pate

. Epidemiology, diagnosis, and management of polycystic ovary syndrome. Clin Epidemiol, 2013; 6:1–13.

Rotterdam ESHRE/ASRM-Sponsored PCOS consensus workshop group. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Hum Reprod, 2004; 19:41–47.

ACOG Committee on Practice Bulletins—Gynecology. ACOG Practice Bulletin No. 108: Polycystic ovary syndrome. Obstet Gynecol, 2009; 114:936–949.

Lin

, Lujan

. Comparison of dietary intake and physical activity between women with and without polycystic ovary syndrome: A review. Adv Nutr, 2014; 5:486–496.

Glueck

, Dharashivkar

, Wang

, et al. Obesity and extreme obesity, manifest by ages 20–24 years, continuing through 32–41 years in women, should alert physicians to the diagnostic likelihood of polycystic ovary syndrome as a reversible underlying endocrinopathy. Eur J Obstet Gynecol Reprod Biol, 2005; 122:206–212.

Zheng

, Li

. Visceral adiposity index as a predictor of clinical severity and therapeutic outcome of PCOS. Gynecol Endocrinol, 2016; 32:177–183.

Legro

. Obesity and PCOS: Implications for diagnosis and treatment. Semin Reprod Med, 2012; 30:496–506.

Orio

, Muscogiuri

, Nese

, et al. Obesity, type 2 diabetes mellitus and cardiovascular disease risk: An uptodate in the management of polycystic ovary syndrome. Eur J Obstet Gynecol Reprod Biol, 2016; 207:214–219.

10.

Mudaliar

, Zabetian

, Goodman

, et al. Cardiometabolic risk factor changes observed in Diabetes Prevention Programs in US settings: A systematic review and meta-analysis. PLoS Med, 2016; 13:e1002095.

11.

Siebenhofer

, Jeitler

, Berghold

, et al. Long-term effects of weight-reducing diets in hypertensive patients. Cochrane Database Syst Rev, 2011; 9:CD008274.

12.

Vetter

, Wadden

, Chittams

, et al. Effect of lifestyle intervention on cardiometabolic risk factors: Results of the POWER-UP trial. Int J Obes, 2013; 37(Suppl. 1):S19–S24.

13.

Orozco

, Buchleitner

, Gimenez-Perez

, et al. Exercise or exercise and diet for preventing type 2 diabetes mellitus. Cochrane Database Syst Rev, 2008; 3:CD003054.

14.

Bates

, Legro

. Longterm management of polycystic ovarian syndrome (PCOS). Mol Cell Endocrinol, 2013; 373:91–97.

15.

Haqq

, McFarlane

, Dieberg

, et al. The effect of lifestyle intervention on body composition, glycemic control, and cardiorespiratory fitness in polycystic ovarian syndrome: A systematic review and meta-analysis. Int J Sport Nutr Exerc Metab, 2015; 25:533–540.

16.

Moran

, Hutchison

, Norman

, et al. Lifestyle changes in women with polycystic ovary syndrome. Cochrane Database Syst Rev, 2011; 7:CD007506.

17.

Panidis

, Tziomalos

, Papadakis

, et al. Lifestyle intervention and anti-obesity therapies in the polycystic ovary syndrome: Impact on metabolism and fertility. Endocrine, 2013; 44:583–590.

18.

Moran

, Ko

, Misso

, et al. Dietary composition in the treatment of polycystic ovary syndrome: A systematic review to inform evidence-based guidelines. J Acad Nutr Diet, 2013; 113:520–545.

19.

Wadden

, Foster

, Letizia

. One-year behavioral treatment of obesity: Comparison of moderate and severe caloric restriction and the effects of weight maintenance therapy. J Consult Clin Psychol, 1994; 62:165–171.

20.

Liberati

, Altman

, Tetzlaff

, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: Explanation and elaboration. PLoS Med, 2009; 6:e1000100.

21.

Nybacka

, Carlstrom

, Fabri

, et al. Serum antimullerian hormone in response to dietary management and/or physical exercise in overweight/obese women with polycystic ovary syndrome: Secondary analysis of a randomized controlled trial. Fertil Steril, 2013; 100:1096–1102.

22.

Asemi

, Esmaillzadeh

. DASH diet, insulin resistance, and serum hs-CRP in polycystic ovary syndrome: A randomized controlled clinical trial. Horm Metab Res, 2015; 47:232–238.

23.

Asemi

, Samimi

, Tabassi

, et al. Effects of DASH diet on lipid profiles and biomarkers of oxidative stress in overweight and obese women with polycystic ovary syndrome: A randomized clinical trial. Nutrition, 2014; 30:1287–1293.

24.

Curi

DDG

, Fonseca

, Marcondes

JAM

, et al. Metformin versus lifestyle changes in treating women with polycystic ovary syndrome. Gynecol Endocrinol, 2012; 28:182–185.

25.

Esfahanian

, Zamani

, Heshmat

, et al. Effect of Metformin compared with hypocaloric diet on serum C-reactive protein level and insulin resistance in obese and overweight women with polycystic ovary syndrome. J Obstetr Gynaecol Res, 2013; 39:806–813.

26.

Kumar

, Arora

. Orlistat in polycystic ovarian syndrome reduces weight with improvement in lipid profile and pregnancy rates. J Hum Reprod Sci, 2014; 7:255–261.

27.

Turner-McGrievy

, Davidson

, Wingard

, et al. Low glycemic index vegan or low-calorie weight loss diets for women with polycystic ovary syndrome: A randomized controlled feasibility study. Nutr Res, 2014; 34:552–558.

28.

Mehrabani

, Salehpour

, Amiri

, et al. Beneficial effects of a high-protein, low-glycemic-load hypocaloric diet in overweight and obese women with polycystic ovary syndrome: A randomized controlled intervention study. J Am Coll Nutr, 2012; 31:117–125.

29.

Sørensen

, Søe

, Halkier

, et al. Effects of increased dietary protein-to-carbohydrate ratios in women with polycystic ovary syndrome. Am J Clin Nutr, 2012; 95:39–48.

30.

Higgins

JPT

, Altman

, Sterne

JAC

. Chapter 8: Assessing risk of bias in included studies. In: Higgins

JPT

, Green

(eds). Cochrane Handbook for Systematic Reviews of Interventions. Chicester, UK: John Wiley & Sons; 2008:187–235.

31.

Windows. Review Manager (RevMan) [Computer program]. Version 5.3. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration; 2014.

32.

Takeshima

, Sozu

, Tajika

, et al. Which is more generalizable, powerful and interpretable in meta-analyses, mean difference or standardized mean difference?. BMC Med Res Methodol, 2014; 14:30.

33.

Higgins

, Thompson

, Deeks

, et al. Measuring inconsistency in meta-analyses. BMJ, 2003; 327:557–560.

34.

Faure

, Bertoldo

, Khoueiry

, et al. Metformin in reproductive biology. Front Endocrinol (Lausanne), 2018; 9:675.

35.

Knowler

, Barrett-Connor

, Fowler

, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med, 2002; 346:393–403.

36.

Orchard

, Temprosa

, Goldberg

, et al. The effect of metformin and intensive lifestyle intervention on the metabolic syndrome: The Diabetes Prevention Program randomized trial. Ann Intern Med, 2005; 142:611–619.

37.

Rothman

. BMI-related errors in the measurement of obesity. Int J Obes, 2008; 32(Suppl. 3):S56–S59.

38.

Millstein

. Measuring outcomes in adult weight loss studies that include diet and physical activity: A systematic review. J Nutr Metab, 2014; 2014:421423.

39.

Lindstrom

, Louheranta

, Mannelin

, et al.; The Finnish Diabetes Prevention Study (DPS): Lifestyle intervention and 3-year results on diet and physical activity. Diabetes Care, 2003; 26:3230–3236.

40.

Duncan

, Perri

, Theriaque

, et al. Exercise training, without weight loss, increases insulin sensitivity and postheparin plasma lipase activity in previously sedentary adults. Diabetes Care, 2003; 26:557–562.

41.

Kodama

, Tanaka

, Saito

, et al. Effect of aerobic exercise training on serum levels of high-density lipoprotein cholesterol: A meta-analysis. Arch Intern Med, 2007; 167:999–1008.

42.

Brown

, Setji

, Sanders

, et al. Effects of exercise on lipoprotein particles in women with polycystic ovary syndrome. Med Sci Sports Exerc, 2009; 41:497–504.

43.

Stefanick

, Mackey

, Sheehan

, et al. Effects of diet and exercise in men and postmenopausal women with low levels of HDL cholesterol and high levels of LDL cholesterol. N Engl J Med, 1998; 339:12–20.

44.

Delormier

, Frohlich

, Potvin

. Food and eating as social practice—understanding eating patterns as social phenomena and implications for public health. Sociol Health Illn, 2009; 31:215–228.

45.

Groeneveld

, Proper

, van der Beek

, et al. Factors associated with non-participation and drop-out in a lifestyle intervention for workers with an elevated risk of cardiovascular disease. Int J Behav Nutr Phys Act, 2009; 6:80.

46.

Dokras

. Cardiovascular disease risk in women with PCOS. Steroids, 2013; 78:773–776.

47.

Wild

, Carmina

, Diamanti-Kandarakis

, et al. Assessment of cardiovascular risk and prevention of cardiovascular disease in women with the polycystic ovary syndrome: A consensus statement by the Androgen Excess and Polycystic Ovary Syndrome (AE-PCOS) Society. J Clin Endocrinol Metab, 2010; 95:2038–2049.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB

0.85 MB