Background
Polycystic ovary syndrome (PCOS) is a complex endocrine condition traditionally characterized by ovulatory dysfunction and androgen excess [
1]. In addition to reproductive complications, women with PCOS often present with overweight or obesity and associated obesity-related complications, such as insulin resistance, impaired glucoregulatory status, and cardiometabolic dysfunction [
2]. Rates of overweight and obesity in PCOS (up to 88% of patients) well exceed that seen in the general population and suggest that obesity can either contribute to PCOS development and/or is a consequence of PCOS pathogenesis [
2]. Accordingly, expert recommendations support modifications in diet and physical activity primarily targeted at weight loss or management as a means to improve health outcomes in PCOS [
2‐
5]. Current diet and physical activity recommendations for women with PCOS reflect those for the general population or other clinical populations (e.g., diabetes mellitus) that do not necessarily account for the unique biology and/or psychosocial variables associated with PCOS [
6,
7]. An understanding of associations between dietary intake and physical activity levels with PCOS is ultimately needed to determine the relevance of tailored therapies for this patient population.
We previously noted inconsistent findings related to differences in energy intake and dietary composition between women with and without PCOS [
8]. Approximately half of the 10 studies reviewed reported differences in total energy [
9‐
13] and fat intake [
9,
10,
13‐
15] in women with PCOS versus a reference group. Three studies did not detect any association(s) between diet and PCOS [
16‐
18]. With respect to physical activity levels, only two studies reported longer sitting intervals in women with PCOS than the reference [
11,
14]. Most showed no differences in overall activity between groups [
9,
11‐
14,
18]. Collectively, published studies do not provide definitive evidence on associations of diet, physical activity and PCOS [
8].
Comparisons among previous studies of dietary intake and physical activity levels in relation to PCOS are limited by differences between studies in how PCOS phenotypes were defined [
8]. Depending on the PCOS diagnostic criteria used, there are four distinct phenotypes. The National Institutes of Health (NIH) criteria define PCOS by the combined presence of oligomenorrhea (OA) and hyperandrogenism (HA) [
1] to identify a relatively homogenous patient population with the most severe risk for reproductive and metabolic dysfunction [
5]. The Rotterdam criteria define PCOS by the combined presence of two out of three cardinal features: OA, HA, and polycystic ovarian morphology [
19]; the Rotterdam criteria captures a more heterogeneous patient population, including mild PCOS phenotypes (i.e., women with regular menstrual cycles and HA or women with OA and normal androgens) [
20]. Milder phenotype(s) may reflect different etiologies and lower risks for metabolic and reproductive sequelae [
5]. Differences in metabolic and reproductive risk across PCOS phenotypes may be attributed to the presence of isolated OA versus HA and/or the additive effects of these two features in conjunction with adiposity. Higher adiposity (an outcome closely linked with diet and physical activity) has been reported to be more prevalent in severe (i.e., those with HA) versus mild PCOS phenotypes [
20]. There is an emerging consensus that studies should use a more homogenous definition of PCOS [
21] or account for phenotypes in their analyses. Studies of usual dietary intake and physical activity levels that consider isolated versus combined cardinal features of PCOS while accounting for body mass index (BMI) are also needed.
While race influences the likelihood of adverse metabolic outcomes in women with PCOS [
22‐
27], few prior studies investigated race as an effect modifier. To our knowledge, prior studies of the association of dietary intake or physical activity levels with PCOS do not investigate differences by race. Previous studies conducted in the United States (US) reported a higher odds and/or prevalence of obesity, hypertension [
24,
25], and elevated fasting glucose concentrations in Black versus White patients [
23,
26]. Although it is unclear whether race influences the etiology or pathogenesis of PCOS, food choice and medical experiences can vary by race and contribute to differential metabolic manifestations [
28]. New recommendations from the international evidence-based guideline for the assessment and management of PCOS acknowledge this knowledge gap and endorse the need for further research on the impact of race/ethnicity in PCOS and the identification of best approaches for treatment across races and clinical phenotypes [
29]. To that end, the primary objective of the present study was to investigate the cross-sectional associations of usual dietary intake and physical activity levels with (a) PCOS and (b) its isolated features (i.e. OA or HA alone) as defined by NIH criteria to include a more homogenous patient population. We hypothesized that high energy and fat intake, and high sedentary behavior, were associated with PCOS and its isolated features. We also explored whether race had a modifying effect on these associations as our secondary objective.
Discussion
This study investigated the associations of diet and physical activity with PCOS, an important topic given the use of health-related behavioral modifications to treat and prevent PCOS [
38]. Cross-sectional analyses in the CARDIA cohort revealed total daily energy, nutrient intake, diet quality, and physical activity had no association with the odds of PCOS and isolated features. Further analyses also showed a potential impact of race, albeit these findings were from a small sample of women with PCOS.
There were no differences in macro- and micronutrient intakes between women with and without PCOS in the CARDIA cohort. Previous U.S. [
14,
17] and international studies [
15,
16,
18] mainly reported no differences in total energy and/or macro- or micronutrient intake between women with and without PCOS, although three prior studies found that women with PCOS consumed more carbohydrates (ranged from 31 to 43 g) than comparison groups [
9,
10,
14]. The conflicting evidence may be explained by the use of unexplained or unique criteria to define PCOS (i.e., luteinizing hormone: follicle stimulating hormone) in these earlier studies, and/or their comparison of women with PCOS to infertile women with various etiologies (rather than a comparison to reproductively healthy women). Our study of relatively younger women, taken together with another US study of older women [
14], suggests that there are no differences in total daily energy intake between women with and without PCOS during the reproductive and peri-menopausal years. Additionally, when examining diet quality of women with PCOS in the US, our finding that there were no differences in diet quality between women with and without PCOS confirmed our observations from a recent study with a different US participant sample [
39].
Our findings of no associations between nutrients with OA contrast with two longitudinal cohort studies that reported a greater intake of carbohydrate and folic acid was associated with a reduced risk of anovulatory infertility [
40,
41] but agree with findings of a longitudinal cohort study of B vitamins and self-reported anovulatory infertility, which reported no significant associations [
41]. The inconsistencies between study findings may be attributed to the heterogeneity in the causes of HA and OA [
42]. For example, anovulatory infertility can manifest secondary to stress and nutrient deficiency, and not all women with OA are at risk of progressing to PCOS due to differences in etiology [
43]. We did not observe any associations of diet and physical activity with HA.
There was little to no association of physical activity behaviors with PCOS, which agrees with previous studies that noted no differences in self-reported moderate and vigorous intensity physical activity between women with and without PCOS [
8]. Similar to previous studies that investigated physical activity in adult women with PCOS [
11‐
14,
18], our findings do not support the hypothesis that women with PCOS engage in less moderate and vigorous physical activity [
38]. Other researchers reported differences in sedentary and physical activity patterns in women with PCOS (i.e. longer sitting intervals), which have been associated with increased risk of all-cause mortality [
44]. The CARDIA measurement of physical activity did not capture patterns of exercise types and sedentary behaviors across time, and thus, objective physical activity data are needed to assess the full spectrum of movement intensity, from sedentary to vigorous activities.
Given that racial differences in the risk of glucoregulatory and cardiometabolic conditions have been identified in PCOS patients [
23‐
26], a modifying effect of race on the associations between diet and physical activity with PCOS warranted consideration. We found that a higher whole grain intake, and a lower sugar-sweetened beverage and fruit juice intake, were associated with higher odds of PCOS and/or OA in Black women. These findings are paradoxical given that these two dietary patterns are typically associated with better health outcomes [
45]. Our conclusions are limited by the small number of Black women with PCOS in this study and the multiple tests for interaction that may lead to type I errors.
Strengths of this study include the lower likelihood of reverse causality as an explanation for these findings because diet and physical activity data were collected prior to widespread recognition of PCOS, which mainly occurred after the establishment of formal NIH (1990) criteria for PCOS. The cohort also contained an equal proportion of Black and White participants, which provided the opportunity to explore differential diet, physical activity and PCOS associations by race. Notably, the CARDIA Women’s study was developed to identify those with PCOS features within the CARDIA cohort, thus allowing us and others to evaluate the isolated and combined features of PCOS using a well-established dataset [
31,
46‐
48]. A limitation of this study includes the potential residual confounding by AHEI-2010 groups that are highly correlated, although there will be minimal issue with groups that are not closely associated with one another. Additionally, there is greater potential for type I error due to multiple comparisons, therefore our study results should be considered exploratory. Further, self-reported and retrospective accounts of menstrual cycle history and clinical signs of androgen excess were not confirmed with clinical assessments. Our findings should also be viewed as being relevant for the most severe clinical phenotype of PCOS (defined as hyperandrogenism and irregular menses) because the absence of data on ovarian morphology meant we could not consider the association of diet and/or physical activity with the other phenotypes of PCOS. Future studies of how diet and physical activity associate with outcomes across the various PCOS definitions are needed particularly in light of the recent release of international evidence-based guidelines for PCOS, which support ovarian morphology as a PCOS criterion [
29]. Previous studies have identified nutrient predictors of anovulatory infertility [
40,
41], but associations between food groups and physical activity with PCOS remained largely unexplored. Though nutrients are important components of the diet, evaluation of diet quality - particularly by race - provides further knowledge about the potential targets for intervention since alterations in foods and/or dietary patterns, rather than nutrients, are critical information for patients.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.