The Effect of Exercise on Cardiometabolic Risk Factors in Women with Polycystic Ovary Syndrome: A Systematic Review and Meta-Analysis

Background: Polycystic Ovary Syndrome (PCOS), a common endocrine disorder in women of reproductive age, increases the risk for cardiometabolic morbidity. While regular exercise is effective in reducing cardiometabolic risk, women with PCOS may experience condition-specific barriers to exercise thereby limiting its efficacy. Aim: To determine the effect of exercise on cardiometabolic risk factors in women with PCOS. Methods: Five databases (Cochrane, EMBASE, Medline, Scopus and SPORTDiscus) were searched up to December of 2021. Eligible studies included: a randomised controlled design; participants with a diagnosis of PCOS; aerobic and/or resistance exercise intervention lasting ≥4 weeks; cardiometabolic outcomes. Meta-analyses were performed to determine the effect of exercise versus non-exercising control on cardiometabolic outcomes. Results: Of the 4517 studies screened, 18 studies were analysed involving 593 participants. When compared with control, exercise significantly improved cardiorespiratory fitness (weighted mean difference {WMD} = 4.00 mL/kg/min, 95% CI: 2.61 to 5.40, p < 0.001) and waist circumference (WMD = −1.48 cm, 95% CI: −2.35 to −0.62, p = 0.001). Systolic blood pressure, fasting blood glucose, insulin resistance, and lipid profiles remained unchanged. Conclusions: Regular exercise may improve cardiorespiratory fitness and waist circumference in women with PCOS. Further large-scale studies are required to determine whether exercise interventions improve various biochemical and anthropometric parameters in women with PCOS and more severe cardiometabolic abnormalities.


Introduction
Polycystic Ovary Syndrome (PCOS) is a common endocrine disorder which is estimated to affect approximately one in every five women of reproductive age worldwide [1], with variation in prevalence depending on the population and diagnostic criteria used. The Rotterdam 2003 criteria, which is the most widely accepted diagnostic criteria, requires two of the following for a diagnosis of PCOS: oligo/anovulation, clinical and/or biochemical signs of hyperandrogenism, and polycystic ovaries. Furthermore, these characteristics are often associated with a range of symptoms such as hirsutism, infertility, acne, and overweight and obesity.
Women with PCOS are at an increased risk of developing the metabolic syndrome and its individual components, particularly increased waist circumference and elevated fasting glucose [2]. It is estimated that 35% of women with PCOS also present with overweight or se on cardiometabolic health outcomes. Random effects models were used and the weighted mean differences with 95% confidence intervals (CIs) were assessed. The variability of studies was determined using the I 2 measure of consistency, which provides a measure of the amount of variability due to heterogeneity rather than sampling error. I 2 values of ≤25, , and ≥75 were regarded as low, moderate, and high, respectively [16]. Assessment of publication bias was performed via examination of funnel plot asymmetry (precision vs. effect size) and using the Egger's test. Sensitivity analyses were completed where studies revealed publication bias, or where the pooled results were significantly affected by the weighting of individual studies. Where there were three or more studies, sub-analyses were completed to assess the effect of HIIT vs. control and MICT vs. control on cardiometabolic outcomes.

Methodological Quality
Methodological quality and risk of bias of the included studies were assessed by two reviewers (A.B. and A.S.) using a modified Downs and Black Checklist [17], consisting of questions that addressed the ability of each study to clearly state its aims, participants, outcome measurements and interventions, accurately represent its participant groups, perform correct statistical analysis, and report their findings accurately. The scale was modified to include two additional criteria regarding exercise supervision and adherence to reflect their possible impact on the primary outcomes of the study (see Appendix A). A total of 29 questions were scored as no = 0, unable to determine = 0, or yes = 1, hence, the highest possible score for a study was 29. Discrepancies between reviewers were resolved via synchronous review of data presented in the original publications. Studies were classified as being of either low-, moderate-, or high-quality with respective scores of 0-10, 11-20, and 21-29.

Risk of Bias Assessment and Certainty of Evidence
Studies were assessed for bias by one reviewer (A.S.) using the Cochrane Risk of Bias 2 tool, which is structured into a fixed set of domains of bias, including selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias [18]. Other bias was judged by assessing whether studies reported exercise adherence.
The certainty of the evidence was assessed using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) framework [19] by one reviewer (A.S.).

Results
A total of 4517 articles were identified in the database searches. Following the removal of duplicates and elimination of studies deemed ineligible, 19 studies were included in the systematic review of which 18 were pooled for meta-analysis ( Figure 1).

Participant Characteristics
Baseline characteristics for a total of 613 participants from the included studies are summarised in Table 1. Two studies reported different outcomes from the same clinical trial [20,21]. As a result, the number of participants from only one of these studies was used to tally the total number of participants included in this review. Mean age ranged from 24 to 32 years and mean BMI ranged from 21.8 to 41.3 kg/m 2 . All but two studies recruited individuals with overweight or obesity [22,23], and all but three studies classified PCOS using the Rotterdam 2003 diagnostic criteria [20,22,24]. Nine studies recruited inactive individuals [22,23,[25][26][27][28][29][30][31], two recruited individuals with insulin resistance [31,32], and one recruited healthy weight individuals [23].

Participant Characteristics
Baseline characteristics for a total of 613 participants from the included studies summarised in Table 1. Two studies reported different outcomes from the same clin trial [20,21]. As a result, the number of participants from only one of these studies w used to tally the total number of participants included in this review. Mean age ran from 24 to 32 years and mean BMI ranged from 21.8 to 41.3 kg/m 2 . All but two stud recruited individuals with overweight or obesity [22,23], and all but three studies cla fied PCOS using the Rotterdam 2003 diagnostic criteria [20,22,24]. Nine studies recrui inactive individuals [22,23,[25][26][27][28][29][30][31], two recruited individuals with insulin resista [31,32], and one recruited healthy weight individuals [23]. Full-text articles assessed for eligibility (n = 64) Full-text articles excluded (n = 45) • Secondary analysis (13) • Abstract only (11) • No control group (5) • Non-English language (5) • Inadequate detail of results (5) • Inadequate description of exercise intervention (3) • Non-randomised (2) • Subjects without Polycystic Ovary Syndrome (1) Studies included in qualitative synthesis (n = 19) Studies included in quantitative synthesis (meta-analysis) (n = 18)
For HIIT and MICT interventions, both alone or in combination with RT, aerobic exercise intensity was prescribed as a percentage of the participants' maximal heart rate (HR max ), or maximal or peak oxygen uptake (VO 2max or VO 2peak , respectively). HIIT interventions involved high-intensity bouts lasting 20 to 240 s at intensities of 70 to 100% HR max , separated by resting periods of 10 to 180 s. MICT interventions involved intensities of 60 to 85% HR max , 50 to 65% VO 2peak , 60 to 70% VO 2max , and a heart rate of 120 beats per minute or greater. The intensities of RT interventions, while not consistently reported, were prescribed using a percentage of participants' 1RM (one-repetition maximum) or RPE (rating of perceived exertion) scale from 1 to 10. RT interventions involved intensities of 50 to 75% 1RM, an RPE of 5 to 6 out of 10, or aimed to reach muscular fatigue at the end of each set.
The duration of HIIT and MICT sessions ranged from 30 to 50 min and 25 to 65 min for HIIT and MICT, respectively, including warm-up and cool-down if prescribed. RT exercises were performed for one to three sets of 8 to 15 repetitions each. Exercise was fully supervised in nine studies [22,23,[27][28][29]31,33,37,38], weekly support through telephone calls was provided in three studies [20,21,36], and three included both supervised and unsupervised sessions [24,25,33]. The three remaining studies did not report supervision status [26,30,32].

Methodological Quality, Risk of Bias, and Certainty of Evidence
The results of the methodological quality and risk of bias assessment are presented in Table 3. The scores ranged from 16 to 25, with an average of 20.8 ± 2.5. Six studies were classified as being of moderate quality [20,22,27,31,32,34], while the remainder were considered to be of high-quality.
The results of the risk of bias assessment are summarised in Figure 2. Twelve studies scored an unclear or high risk of bias on five or more domains [20,21,24,27,28,[30][31][32][33]35,36,38]. Two studies scored an unclear or high risk of bias on four domains [22,37]. Four studies scored an unclear or high risk of bias on three or less domains [23,25,26,29]. The level of certainty of the results produced are detailed in Table 4. There was a low certainty of evidence showing that exercise may result in higher cardiorespiratory fitness when compared to control. There was a very low certainty of evidence showing that exercise may reduce waist circumference when compared to control. There was a very low certainty of evidence showing that exercise is unlikely to induce any meaningful improvement in systolic blood pressure in normotensive women with PCOS when compared to control. There was a low level of certainty showing that exercise is unlikely to induce any meaningful improvement in HOMA-IR when compared to control. There was a low level of certainty that exercise is unlikely to result in any meaningful improvement in fasting blood glucose, blood triglycerides, or HDL-C in women with PCOS (and normal scores for these parameters) when compared to control. The level of certainty of the results produced are detailed in Table 4. There was a low certainty of evidence showing that exercise may result in higher cardiorespiratory fitness when compared to control. There was a very low certainty of evidence showing that exercise may reduce waist circumference when compared to control. There was a very low certainty of evidence showing that exercise is unlikely to induce any meaningful improvement in systolic blood pressure in normotensive women with PCOS when compared to control. There was a low level of certainty showing that exercise is unlikely to induce any meaningful improvement in HOMA-IR when compared to control. There was a low level of certainty that exercise is unlikely to result in any meaningful improvement in fasting blood glucose, blood triglycerides, or HDL-C in women with PCOS (and normal scores for these parameters) when compared to control.  [22] 0

GRADE Working Group grades of evidence
High certainty: We are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.

Discussion
The results from this systematic review and meta-analysis provide novel and upto-date data regarding the effect of exercise interventions on key cardiometabolic risk factors in women with PCOS. Eighteen studies involving a total of 593 participants were combined in the meta-analyses, which showed that regular exercise significantly improved cardiorespiratory fitness and central obesity. The results also showed that exercise did not improve other cardiometabolic outcomes including systolic blood pressure, insulin resistance, blood glucose levels, and lipid profiles, albeit the baseline values were all within normal ranges. The results of the sub-analyses suggest that both HIIT and MICT interventions significantly improve cardiorespiratory fitness, and while both interventions also reduced waist circumference, only MICT led to statistically significant benefit.
Low cardiorespiratory fitness is a strong predictor of cardiometabolic risk [39] and all-cause mortality [40]. The results of this study showed that exercise significantly improved cardiorespiratory fitness by more than one metabolic equivalent (MET), equal to 3.5 mL/kg/min, when compared to a non-exercise control. These results indicate that exercise may play a critical part in cardiovascular health management for women with PCOS, who experience increased cardiovascular risk [7], as previous studies have shown that improvements in cardiorespiratory fitness of one MET were associated with a 13% risk reduction in all-cause mortality and a 15% reduction in the incidence of cardiovascular disease [41]. This finding is in accordance with an earlier meta-analysis which reported similar improvements in cardiorespiratory fitness of +3.84 mL/kg/min [42]. Furthermore, the results of the sub-analysis expand on previous findings by suggesting that both HIIT and MICT improve cardiorespiratory fitness, with MICT achieving greater improvements (HIIT: WMD = 2.87 mL/kg/min, p < 0.001; MICT: WMD = 5.33 mL/kg/min, p < 0.001).
Waist circumference is included in the diagnostic criteria for the metabolic syndrome [43] and more accurately predicts cardiovascular risk than BMI [44], with a higher value also associated with an increased risk of developing type 2 diabetes and cardiovascular disease [45]. In fact, it is estimated that for every 1 cm increase in waist circumference, the relative risk of a cardiovascular event is estimated to increases by 2% [46]. The results of this study showed that exercise decreased waist circumference by −1.48 cm. Taken together, these results highlight the potential benefit of exercise for reducing the risk of cardiovascular disease-related events by approximately 3% through the reduction of waist circumference alone in women with PCOS. Furthermore, the results of the sub-analyses revealed that while both HIIT and MICT improved waist circumference, only MICT achieved statistical significance (HIIT: WMD = −2.41 cm, p = 2.05; MICT: WMD = −1.69 cm, p = 0.027). Importantly, while the magnitude of effect seen with exercise may not be large, the results support the utility of exercise as a therapeutic option for the management of central obesity, i.e., preventing central adiposity.
Previous reports have shown that young women with PCOS often present with elevated blood pressure independent of BMI [47]. In clinical practice, maintaining systolic blood pressure below 130 mmHg is the primary goal for individuals with, or at risk of, hypertension. To this end, lifestyle modification involving diet modulation and increased physical activity is the first line of therapy [48]. The results of this study show that exercise was associated with only modest non-significant reductions in systolic blood pressure when compared with control. Although the findings were non-significant, mean baseline systolic blood pressure across studies was 116 mmHg, which is considered normotensive (i.e., below 120 mmHg) [47].
Insulin resistance, which contributes to elevated fasting blood glucose levels, is highly prevalent in women with PCOS and significantly elevates the risk for ensuing type 2 diabetes [49]. Yet, fasting blood glucose, and not insulin resistance measured by HOMA-IR, is included in the diagnostic criteria for the metabolic syndrome [43]. The results of this study support the results of existing systematic reviews, which demonstrate that exercise does not significantly improve insulin resistance or fasting blood glucose levels beyond control in women with PCOS [42]. Furthermore, the results of the sub-analysis add to previous findings by showing that neither HIIT nor MICT improve fasting blood glucose levels. Importantly, these results may be explained by the normal baseline fasting blood glucose levels of participants included in the meta-analysis which averaged 4.93 mmol/L and were <5.44 mmol/L across all studies. As a result of these relatively normal values, exercise is not expected to further improve fasting blood glucose levels. Similarly, although exercise did not significantly improve insulin resistance, 10 of the 12 groups included in the meta-analysis favoured exercise for the reduction of insulin resistance when measured as HOMA-IR. Again, this non-significant result may be explained by the HOMA-IR baseline values averaging 2.67 across studies, which is on the cusp of normal. Furthermore, existing evidence supports the utility of exercise for improving glucose metabolism in individuals with more severe metabolic abnormalities, such as those with type 2 diabetes [50], and as a result, exercise may help prevent the onset of insulin resistance in women with PCOS.
Dyslipidaemia, involving abnormal circulating blood lipid levels, is associated with an array of cardiometabolic conditions such as type 2 diabetes [51], non-alcoholic fatty liver disease [52], and PCOS [53]. While the analyses undertaken in this review did not show a significant effect of exercise on HDL-C or triglyceride levels, all studies involved women with normal values for both HDL-C and triglycerides at baseline. As it is known that regular exercise improves lipid metabolism [54], it is reasonable to expect that dyslipidaemia in PCOS may be ameliorated following increased physical activity, yet further research is required to clarify this.
Previous systematic reviews have shown the efficacy of exercise for improving an array of cardiometabolic outcomes [55,56]. The results of this study go beyond previous reviews by providing up-to-date evidence surrounding the effects of exercise on cardiometabolic health in women with PCOS. Furthermore, the sub-analyses assessing the efficacy of HIIT and MICT on the aforementioned outcomes may be used to provide clinicians with more evidence-based prescription options when designing and delivering tailored exercise interventions. Importantly, the vast majority of studies included in this review had significant bias (Figure 2), and as a result, the certainty of evidence for the results generated was low at best (Table 4). Further high-quality studies are required to strengthen these findings and elucidate the role of exercise in the management of PCOS.

Implications of the Research
Although PCOS is associated with increased cardiometabolic risk factors, these may become exacerbated by comorbid and age-related factors such as reduced cardiorespiratory fitness and obesity. The results of this study show that regular exercise is an effective therapy for the management of multiple cardiometabolic risk factors including low cardiorespiratory fitness and abdominal adiposity in women with PCOS. While the results did not show a significant effect of exercise on systolic blood pressure, insulin resistance, fasting blood glucose levels, or lipid profiles, this result may be partly due to baseline values being within a relatively normal range. As cardiorespiratory fitness has been shown to be inversely associated with a range of cardiometabolic risk factors such as HOMA-IR, waist circumference, systolic blood pressure, and fasting glucose levels in individuals with obesity and type 2 diabetes [57], the results of the analyses indicate that exercise may be an effective strategy for the management of cardiometabolic risk in individuals with more severe metabolic abnormalities. Consequently, regular exercise may be useful for ameliorating the deterioration of cardiometabolic health in women with PCOS. Furthermore, while both HIIT and MICT improved cardiorespiratory fitness, MICT resulted in an almost two-fold greater increase. As a result, women with PCOS who are aiming to improve their cardiorespiratory fitness may see greater improvements by undertaking MICT rather than HIIT. While not as effective, HIIT may significantly improve cardiorespiratory fitness also, and this strategy may be particularly useful for time-poor individuals. Therefore, on the basis of the findings reported herein, as well as current available evidence, exercise appears to be an effective first-line therapy for the management of cardiometabolic health in women with PCOS.

Limitations
This review has limitations that should be considered when interpreting the results. First, the majority of studies demonstrated unclear or high risk of bias on multiple domains as per the ROB2 assessment. As a result, the certainty of evidence was very low or low across all outcomes. Second, as the majority of studies included in the analysis incorporated sample sizes as low as five participants per group, there is a high risk for type II errors. Third, although compliance was high when reported, the impact of unreported and possibly poor adherence rates on the analyses undertaken could not be determined. As supervised exercise demonstrates greater improvements in adherence and overall fitness [58], studies with unsupervised sessions may differ in their adherence and intensity due to reliance on participant-reported measures. Fourth, because the research question was limited to women with PCOS and most studies were conducted in Western countries, the results may not be generalisable to other population groups. Fifth, given that women with PCOS can exhibit multiple phenotypes [59], each with possible variances in cardiometabolic risk, the paucity of data precluded our ability to determine the effect of exercise on specific PCOS phenotypes. Sixth, non-English language articles and grey literature were excluded which may result in publication bias, as was evident for some outcomes assessed in this review. Finally, although our meta-analysis suggests that exercise may improve surrogate markers of cardiometabolic health such as cardiorespiratory fitness over the short-to mediumterm, the observed effect was modest. As such, further studies are required to determine whether exercise-induced improvements in health outcomes in women with PCOS can be maintained in the long-term. Despite these limitations, our review included only RCTsthe gold-standard design for determining causality. We incorporated a comprehensive search strategy conforming to international reporting guidelines, and performed sensitivity analyses where possible to account for high heterogeneity and publication bias. In doing so, we were able to provide an in-depth and up-to-date synthesis of current evidence regarding the effects of exercise interventions on cardiometabolic risk in women with PCOS.

Conclusions
The results of this study show that regular exercise is an effective therapy for the management of multiple cardiometabolic risk factors including low cardiorespiratory fitness and abdominal adiposity in women with PCOS. While HIIT and MICT interventions both improve cardiorespiratory fitness, the efficacy of HIIT for the management of other cardiometabolic risk factors requires further investigation. Additionally, because the majority of participants in this review did not have impaired fasting blood glucose levels or insulin resistance, further studies involving women with PCOS and more severe metabolic abnormalities should be undertaken to determine the efficacy of exercise in this subset of women with PCOS.
Author Contributions: A.B.-S. was involved in the search strategy, study selection, data extraction, risk of bias scoring, and drafting of the manuscript. A.M. was involved in the study design, data interpretation and drafting of the manuscript. H.J.T. was involved in the study design and drafting the manuscript. N.A.J. was involved in the study design, search strategy, study selection, and drafting of the manuscript. A.S. was involved in the study design, the search strategy, study selection, data extraction, statistical analysis, data interpretation, risk of bias and study quality assessment, and drafting of the manuscript. All authors have read and agreed to the published version of the manuscript. Data Availability Statement: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.

8.
Have all important adverse events that may be a consequence of the intervention been reported? 0-1

9.
Have the characteristics of patients lost to follow-up been described? 0-1 10. Have actual probability values been reported (e.g., 0.035 rather than <0.05) for the main outcomes except where the probability value is less than 0.001? 0-1 External validity 11. Were the subjects asked to participate in the study representative of the entire population from which they were recruited? 0-1 12. Were those subjects who were prepared to participate representative of the entire population from which they were recruited? 0-1 13. Were the staff, places, and facilities where the patients were treated representative of the treatment the majority of patients received?