Using Sleep Timing to Identify Eating Habits in Polycystic Ovary Syndrome

Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders in women of reproductive age reaching a prevalence rate of 5–10% [1]. Distinctive features include chronic anovulation, biochemical and/or clinical hyperandrogenism and polycystic ovarian morphology [2]. Diagnosis of this syndrome is based on the Rotterdam criteria (2003), when 2 out of 3 criteria are satisfied, while other etiologies are excluded [2]. Hyperandrogenism is one of the main features of PCOS and evidence supports that it is one of the most used markers to assess severity of PCOS [3]. In addition, obesity, low-grade chronic inflammatory state and insulin-resistance (IR) often coexist in women with PCOS [4]. Furthermore, we also previously showed that women with PCOS had different body composition compared to women without PCOS, with the lowest values of fat free mass and phase angle (PhA), a marker of the chronic inflammatory state [5]. Beyond genetic predisposition, several pathogenetic factors could be involved in the pathogenesis of PCOS and PCOS-related complications [6,7]. In the management of women with PCOS, the crucial role of lifestyle and in particular of nutrition is well known, and it is currently at the forefront of treatment of this condition [8,9]. However, a tailored nutritional approach is based on an accurate assessment of eating habits at the baseline. The food diary is a useful tool, but it is time-consuming, and it is filled by the patients that may not objectively report eating habits. Thus, in order to drive the nutritional approach, there is the need to find an easy and valid tool to screen eating habits of PCOS in the nutrition outpatient clinic. Chronotype, i.e., the attitude of a subject determining individual circadian preference in behavioral and biological rhythms related to the external light–dark cycle, has been reported to play a role in determining nutrition preferences and the risk of developing chronic diseases [8,9,10,11]. There are three categories of chronotype: morning chronotype (defined “lark”) tends to wake up early and prefers activities earlier in the day, while the evening chronotype (defined “owl”) generally wakes up later and prefers to perform his/her main activities in the late afternoon or evening [10]. Neither chronotype is in an intermediate position between the morning and evening chronotypes [10]. Evening chronotype has been associated with a low adherence to the Mediterranean Diet (MD) [12], obesity [13], and IR [14], all of which are often present in women with PCOS [15].

Based on this background, this observational cross-sectional study aimed to investigate if chronotype categories (morning, evening, and neither) could be used as tool to screen eating habits in women with PCOS in order to drive nutritional approach.

One hundred and twelve women with PCOS (aged 24.21 ± 5.47 years; BMI 30.95 ± 5.66 kg/m²) were enrolled. Specifically, 24 subjects (21.4%) of them were normal weight, 29 subjects (25.9%) were overweight, and 24 subjects (21.4%) and 35 subjects (31.3%) had grade I and II obesity, respectively. The chronotype categories were distributed among women with PCOS as follows: 31 women (27.7%) morning, 33 women (29.5%) neither, and 48 women (42.9%) evening chronotype. BMI categories, lifestyle habits, and HoMA-IR of women with PCOS, according to chronotype categories, are reported in Table 1. Women with PCOS with evening chronotype showed significantly higher percentages of grade I (p = 0.003) and grade II obesity (p = 0.001) and were less prone to follow healthy lifestyles compared to those with neither and morning chronotypes; indeed, they did less regular exercise (p < 0.001) and most of them were smokers (p < 0.001). Evening chronotype also was significantly associated with a higher prevalence of subjects with HoMA-IR cut off > 2.5 than neither and morning chronotypes (p < 0.001).

Differences in age, BMI, WC, HoMA-IR, and adherence to the MD were showed in Table 2. There were no differences in age among the chronotype categories (p = 0.763). Finally, women with PCOS with evening chronotype had the lowest PREDIMED score, i.e., they had the lowest adherence to the MD compared to the other chronotypes (p < 0.001).

Figure 2 reported adherence to the MD in women with PCOS grouped based on chronotype categories. Women with lower adherence to the MD presented the high percentage of evening chronotype (58.3%; p < 0.001). Of note, no women with PCOS with evening chronotype reported high adherence to the MD (0.0%; p < 0.001).

Table 3 reported differences in nutritional parameters of women with PCOS, according to chronotype categories. In detail, women with PCOS with evening chronotype consumed more total energy intake (p < 0.001), total and simple carbohydrates (p < 0.001), total fat, saturated fatty acids (SFA), polyunsaturated fatty acids (PUFA, and n-6 PUFA (p < 0.001 for all), and less fiber (p < 0.001) than women with PCOS with other chronotype categories.

Analyzing the response frequency of dietary components included in the PREDIMED questionnaire in detail, we found that, when compared to the other two chronotype categories, women with PCOS with evening chronotype consumed less extra virgin olive oil (EVOO) (p = 0.001), legumes (p = 0.038), fish/seafood (p < 0.001), and tree nuts (p = 0.041). In addition, women with PCOS with evening chronotype consumed less red wine (p < 0.001) and more red/processed meat (p < 0.001) than women with PCOS with morning chronotype (Table 4).

Correlation analyses were performed to investigate the association of chronotype scores with anthropometric measurements, metabolic, and nutritional parameters as reported in Table 5. Chronotype score was inversely associated with BMI (r = −0.775; p < 0.001), WC (r = −0.674; p < 0.001), fasting plasma glucose (r = −0.486; p < 0.001), insulin levels (r = −0.568; p < 0.001), HoMA-IR (r = −0.556; p < 0.001), and positively associated with PREDIMED score (r = 0.803; p < 0.001). In addition, the chronotype score was also inversely associated with total energy intake (r = −0.507; p < 0.001), total carbohydrate (r = −0.563; p < 0.001), simple carbohydrates (r = −0.609; p < 0.001), total fat (r = −0.598; p < 0.001), SFA (r = −0.711; p < 0.001), unsaturated fat (r = −0.324; p < 0.001), PUFA (r = −0.465; p < 0.001), and n-6 PUFA (r = −0.456; p < 0.001) and positively associated with fiber intake (r= 0.531; p < 0.001).

Table 6 showed the results of the bivariate proportional OR model performed to assess the association of chronotype score with food items of the PREDIMED questionnaire. EVOO, vegetable, fruit, fish, poultry, nuts, and wine consumption were positively associated with chronotype score, while the highest OR of soda drinks, red meats, butter, cream, margarine commercial sweets and confectionery seemed to have a negative effect on chronotype score.

Women with PCOS and evening chronotype had higher rates of obesity and were less prone to follow healthy lifestyle. In fact, they exercised less regularly and smoked more than women with PCOS with neither and morning chronotypes.

In agreement with our results, subjects with evening chronotype have previously been reported to be more likely to have obesity and an unhealthy lifestyle [12]. In this regard, in a cross-sectional study including 172 middle-aged adults, the lower the chronotype score, the higher the BMI values in the entire population, suggesting that obesity was a common finding in subjects with evening chronotype. Additionally, in this study, subjects with evening chronotype followed an unhealthier lifestyle (less regular activity and more frequently smokers) than other chronotype categories [12].

In our population, evening chronotype was significantly associated with a higher prevalence of IR subjects (HoMA-IR cut off > 2.5) than the other two chronotype categories. In addition, chronotype score was inversely associated with fasting plasma glucose, insulin levels, and HoMA-IR. Evening chronotype involves a misalignment of physiological circadian rhythms [10] and it has already been reported that humans develop impaired glucose tolerance when subjected to circadian misalignment conditions [10,26,27,28]. As PCOS is already a risk factor for IR, evening chronotype may have an additional deleterious effect on the metabolic profile. In this regard, mutations in multiple clock genes in humans have been shown to contribute to genetic susceptibility to obesity, IR, and type 2 diabetes [10,26,27,28]. Indeed, the risk of developing impaired fasting plasma glucose has been associated with polymorphisms in CRY2 [10,26,28] while the polymorphism in ARNTL has been reported to increase the susceptibility to develop type 2 diabetes [27]. Similarly, having polymorphism in NR1D1 has been associated with the risk of developing obesity [29]. Based on these findings, several studies investigated gene-behavior interactions and showed that interactions between diet and clock gene mutations affect fasting plasma glucose [30], IR [31], body weight [32], and type 2 diabetes [27].

For the first time, we report that women with PCOS with evening chronotype had the lowest adherence to the MD and poorer eating habits. In addition, the lower the chronotype score, the lower the PREDIMED score in these women. Considering the nutritional parameters, women with PCOS with evening chronotype consumed more total energy intake, total and simple carbohydrates, total fat, SFA, PUFA, n-6 PUFAs, and less fiber than women with PCOS with other chronotypes. In addition, regarding the dietary components included in the PREDIMED questionnaire, we found that women with PCOS with evening chronotype consumed less EVOO, red wine, legumes, fish/seafood, and tree nuts and more red/processed meat than women with PCOS with neither and morning chronotypes.

In agreement with our results, it has been reported that subjects with obesity and evening chronotype had the lowest adherence to the MD [12]. In fact, in the study previously mentioned, it has been also reported that evening chronotype had the highest percentage of subjects with low adherence to the MD and the lowest percentages of subjects with average and high adherence to the MD [12].

The association of chronotype categories with the adherence to the MD might be mediated by the food cluster enclosed in the MD. In fact, a high adherence to the MD could also contribute to normal sleeping pattern [33,34] that in turn plays an important role in determining the chronotype [10]. In this regard, in a cross-sectional study of 172 middle-aged adults with obesity, good sleepers (assessed by Pittsburgh Sleep Quality Index < 5) when compared to poor sleepers had significantly higher adherence to the MD [34]. In another cross-sectional study of 1936 individuals, for every point by which the MD score (MEDI-LITE score) increased, the probability of having adequate sleep quality increased by 10% [33].

It appears that eating patterns change in chronotype categories and this has already been reported previously [35,36]. In agreement with this, it has already been reported that individuals with evening chronotype tend to have a less healthy diet, with higher intakes of sugary drinks and chocolate and lower intakes of vegetables, fish, and fruit than individuals with morning chronotype [35,36]. Mota et al., in a cross-sectional study in medical residents (n = 72) evaluating chronotype and dietary patterns, found that chronotype score was negatively associated with cholesterol and sweet intake [35]. The same finding was then also confirmed in another cross-sectional study carried out in pregnant women (n= 100) [36]. In these women, the morning chronotype was associated with better diet quality [36]. Furthermore, another cross-sectional study reported that a low intake of micronutrients (vitamins and minerals) was also associated with the evening chronotype [37]. In particular, a significant association of the evening chronotype with a lower intake of protein, micronutrients (such as calcium, magnesium, zinc, and vitamins), and vegetables, and a higher intake of noodles has been detected in 112 young women [37]. A deficit of micronutrients such as vitamin D has also been often detected in women with PCOS [6,38] and it has been associated with sleep disturbances [39], a common finding in evening chronotype.

Preference for consumption of specific foods and more generally different dietary patterns may also play a role in worsening normal circadian rhythms [40]. In this regard, it has been reported that the macronutrient composition of the diet can alter the function of central and peripheral circadian clocks in humans [36]. A well-known characteristic of the MD is high fiber content and consequently a low glycemic index (GI) [41]. Our data showed that women with PCOS with evening chronotype consumed more carbohydrates and less fiber than the same women with other two chronotype categories thus suggesting that the diet of these women is characterized by a high GI. Similarly, a recent study by Gangwish et al. carried out in a population of postmenopausal women (n = 53,069) found that high GI diets may be a risk factor for insomnia, whereas the intake of unprocessed, fiber-rich, low GI whole-grain carbohydrates such as whole grains reduced this risk [42]. Hyperinsulinemia, resulting from high glycemic load dietary intake, could result in reactive hypoglycemia, thus triggering the secretion of autonomous counter-regulatory hormones such as adrenaline and cortisol, which may cause symptoms such as heart palpitations, tremors, anxiety, and irritability causing waking from sleep [42]. Indeed, St-Onge et al. observed that a low fiber intake and a high intake of SFA and sugars in normal weight adults (n = 26) was associated with less restorative sleep resulting in reduced overall sleep quality [43]. In addition, high GI food predisposes to obesity, which is a well-known risk factor for sleep disturbances [44]. Therefore, it is possible that a high-fiber diet such as the MD may be a useful tool for improving sleep in individuals with sleep disturbances [43]. Furthermore, in a cross-sectional study of non-shift workers (n = 4435) conducted by Tanaka et al., the association among carbohydrate, lipid, and protein intake and insomnia symptoms, including difficulty initiating sleep, difficulty maintaining sleep, and poor quality of sleep was assessed [45]. This study showed that protein and carbohydrate intake were associated with insomnia; in particular, a low protein intake (19% of energy from protein) and low carbohydrate intake (50% of energy from carbohydrate) were associated with difficulty maintaining sleep. This can probably be explained by the fact that a low protein intake could lead to a low availability of tryptophan, a precursor of serotonin and melatonin, which are well known hormones promoting sleep, while a high protein intake can reduce tryptophan concentrations in the brain due to competition for the blood–brain barrier transporter with other neutral amino acids, causing in both cases symptoms of insomnia [45]. The MD contains various dietary sources of tryptophan such as milk, turkey, chicken, fish, eggs, beans, cheese, and green leafy vegetables. Thus, low adherence to the MD, as in the case of women with PCOS with evening chronotype, results in poor intake of tryptophan-rich foods. The intake of tryptophan rich food has been associated with an improvement of primary insomnia in healthy subjects (n = 49) [46]. In this study, administration of food bars containing small dose (250 mg) of tryptophan achieved significant results in reducing waking time during the night, increasing sleep efficiency, and increasing subjective sleep quality [46]. Furthermore, the depletion of tryptophan resulted in sleep fragmentation, rapid eye movement sleep latency and rapid eye movement density compared to baseline and placebo [47,48]. Consumption of tryptophan-rich proteins (e.g., milk protein) may also improve sleep quality by influencing changes in core temperature [49].

The MD also includes a substantial consumption of fruits and clinical studies suggested sleep-promoting effects of some fruits [45,46,47,48]. In humans, consumption of two kiwis one hour before bedtime improved sleep onset, duration, and efficiency in healthy adults (n = 24) during a 4-week open clinical trial [50]. In addition, a double-blind pilot study showed that fresh tartar cherry juice reduced insomnia in elderly subjects (n = 15) [51]. The effects of cherries have also been reported to increase sleep duration and to reduce the number of awakenings, as measured by actigraphy in Spanish participants (n = 12) [52]. These benefits may be facilitated by dietary polyphenols, which have been shown to modulate circadian rhythms [53] and sleep–wake cycles in rodents [54]. According to another hypothesis, these results (on improvement of sleep quality) could probably be due to the high content of antioxidants in fruit. In fact, antioxidants may have beneficial effects in subjects with sleep disorders as they may develop oxidative stress due to the accumulation of free radicals [55]. During sleep, the antioxidant activity increases in the whole body, resulting in the elimination of oxygen free radicals accumulated during wakefulness [55]. In addition to fruit, both polyphenols and antioxidants are also present in other foods that are typical of the MD such as EVOO and red wine that were less consumed in our population of women with PCOS with evening chronotype. In particular, it has been identified that several polyphenols, such as resveratrol, a polyphenol found in red wine, act as dietary activators of Sirtuin 1 (SIRT1) [56]. In turn, SIRT1 modulates transcription factors including period circadian clock 2 (PER2), which are circadian clock genes that in turn regulate the daily rhythms of locomotor activity, metabolism, and behavior [57]. Additionally, SIRT1 modulates the ventromedial hypothalamic clock, a brain region that contains neuronal food-synchronized clocks contributing to regulation of the circadian rhythm in feeding behavior [57]. Thus, it is possible to speculate that a high adherence to the MD and more specifically the consumption of certain foods with well-known positive effects on sleep may contribute to the maintenance of normal circadian rhythms and therefore have an influence on chronotype [10].

The main strength of our study was the novelty of highlighting the role of chronotype assessment in women with PCOS. Identifying women with PCOS and evening chronotype becomes of paramount importance because they might be at risk of following unhealthy diet thus being at high risk of developing metabolic diseases. In this context, using a tool as easy as the MEQ questionnaire, women with PCOS with a worse clinical and metabolic picture could be identified early and thus could be tackled with more tailored diet plans. Indeed, an intervention that seeks to align the attitude of the evening chronotype with the circadian time could be an early attempt to improve the metabolic status of women with PCOS. The limitation of the study is represented above all by the cross-sectional experimental design, which, although showing the association of the evening chronotype with a worsening of IR in women with PCOS, fails to provide any explanation of the causality of this association. It is recommended that future longitudinal studies investigate the causality among these variables with particular emphasis on the implementation of lifestyle interventions. Although a CLOCK-gene assessment of chronotype should be more accurate but more difficult to carry out, there are several studies that validated the association of chronotype, and thus the assessment of chronotype categories through a questionnaire, with CLOCK-gene based assessed of chronotype [58]. This allows us to use this tool for the assessment of chronotype in our study. Finally, both genetic and environmental factors, including nutrition and gut microbiota, influence the distribution of chronotypes. However, we did not include the gut microbiota or gut-derived metabolites, including trimethylamine N-oxide, in this study [17].

In summary, the current study presented the first evidence that women with PCOS with evening chronotype have worse anthropometric measurements, more severe IR, and in general follow an unhealthier lifestyle than women with PCOS with other chronotype categories. These same women have also the lowest adherence to the MD and poorest eating habits. Our results support the evidence that chronotype assessment could be a new and effective tool to screen the eating habits, and more generally the lifestyle, of women with PCOS that could be used by multidisciplinary team that usually take care of women with PCOS such as gynecologists, endocrinologists, and dietitians. Thus, chronotype assessment could be a preliminary examination to be performed in women with PCOS in order to drive the most appropriate nutritional approach.