Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Carotid intima-media thickness in polycystic ovary syndrome and its association with hormone and lipid profiles

  • Rhea Jabbour ,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Validation, Visualization, Writing – original draft, Writing – review & editing

    rhea.jab@gmail.com

    Affiliation Division of Gynecological Endocrinology and Reproductive Medicine, Department of Obstetrics and Gynecology, Medical University of Vienna, Vienna, Austria

  • Johannes Ott,

    Roles Visualization, Writing – review & editing

    Affiliation Division of Gynecological Endocrinology and Reproductive Medicine, Department of Obstetrics and Gynecology, Medical University of Vienna, Vienna, Austria

  • Wolfgang Eppel,

    Roles Writing – review & editing

    Affiliation Division of Obstetrics and Feto-Maternal Medicine, Department of Obstetrics and Gynecology, Medical University of Vienna, Vienna, Austria

  • Peter Frigo

    Roles Conceptualization, Data curation, Investigation, Methodology, Resources, Supervision, Validation, Visualization, Writing – review & editing

    Affiliation Division of Gynecological Endocrinology and Reproductive Medicine, Department of Obstetrics and Gynecology, Medical University of Vienna, Vienna, Austria

Abstract

Objective

Polycystic ovary syndrome (PCOS) has been associated with an increased risk of metabolic disturbances and cardiovascular disease. Intima-media thickness of the common carotid artery (CIMT) represents a valid surrogate marker of early systemic atherosclerosis. This study aimed to investigate if CIMT is increased in PCOS patients compared to healthy controls and if there is an association with hormone and metabolic profiles.

Methods

In this prospective cross-sectional study, past medical history, anthropometrical measurements and hormonal, lipidemic and glycemic parameters were obtained in 41 PCOS patients and 43 age-matched healthy controls of similar body mass index (BMI) and frequency of smokers. B-mode ultrasound enabled CIMT measurement at the far wall of the left and right common carotid artery.

Results

Patients with PCOS showed significantly increased CIMT values compared to healthy controls (0.49±0.04mm vs. 0.37±0.04mm respectively, P<0.001). They featured a generally increased cardiovascular risk profile. Correlation analysis showed a positive association between CIMT and the adverse metabolic risk profile. The diagnosis of PCOS was the strongest predictor of CIMT, even after multiple adjustments for BMI, age and smoking status (β = 0.797, P<0.001, R2 = 0.73). A model among oligomenorrhoic patients revealed a relationship between CIMT and the suspected duration of disease (β = 0.373, P = 0.021, R2 = 0.14).

Conclusions

PCOS patients are likely to feature signs of premature systemic atherosclerosis at a young age. Early exposure to adverse cardiovascular risk factors may possibly have long-term consequences on the vascular system. An early vessel screening might thus already be beneficial in these patients at a younger age.

Introduction

Polycystic ovary syndrome (PCOS) represents one of the most common endocrinopathies in women of reproductive age, affecting approximately 4 to 7% of all women [1]. Clinical features include hyperandrogenism, ovulatory dysfunction and polycystic ovarian morphology on ultrasound (PCOM) [2,3]. The fundamental underlying defect in PCOS still remains unclear. However, PCOS seems to be a complex state of multifactorial origin, resulting from genetic components that interact with environmental factors. Intrinsic ovarian dysfunction, leading to dysregulated ovarian steroidogenesis with higher androgen production, neuroendocrine abnormalities and hyperinsulinemia are thought to play significant interactive roles [4,5]. In fact, PCOS seems to be closely associated with a metabolic disorder linked to insulin resistance (IR) [6], as 50–70% of PCOS patients show IR with compensatory hyperinsulinemia [7]. This state is characterized by IR in skeletal muscle, adipose tissue and liver, due to a post-binding defect in insulin receptor signaling caused by increased serine phosphorylation of the insulin receptor and the insulin receptor substrate-1 (IRS-1), but ovarian hypersensitivity to insulin [6,8]. Further clinical manifestations of PCOS include metabolic disturbances, such as abdominal obesity, dyslipidemia, impaired glucose tolerance, type II diabetes mellitus (DM), arterial hypertension and the metabolic syndrome [7,9,10]. IR being a key component of the pathophysiology of PCOS, treatment options with insulin-sensitizers, such as metformin or inositol isoforms [11], are increasingly being used due to their beneficial effects on metabolic and hormonal parameters.

The above-mentioned characteristics represent typical risk factors for early atherosclerosis [12], leading to an increased risk of cardiovascular disease [13]. In fact, a 2-fold increased likelihood of coronary heart disease or stroke [14] and higher prevalence of coronary artery and aortic calcification have been reported among PCOS patients [15]. Measurement of intima-media thickness of the common carotid artery (CIMT) using brightness mode ultrasound (B-mode) represents a valid surrogate marker of early systemic atherosclerosis [16,17,18]. This sensitive and noninvasive method thus enables screening for atherosclerosis and cardiovascular risk assessment. CIMT has been associated with an increased risk of coronary heart disease [19] and permits to predict the likelihood of future cardiovascular events, such as stroke and myocardial infarction [17,18,19].

There is evidence that CIMT is elevated in women suffering from PCOS [20,21], especially when they have passed forty [22,23]. However, CIMT increase can already be seen at a young age, such as in adolescence [24], although discrepant findings have been reported [21,25]. Previous studies have also shown an association between CIMT and lipid profile [26], as well as a relationship with hyperinsulinemia [26] and endogenous androgen levels [20], even though differing results have been reported. In fact, hyperandrogenism is said to lead to an increase in CIMT through its proatherogenic effect [20], but numerous studies have also demonstrated an inverse correlation of CIMT with androgen levels in women [27,28,29,30]. On the other hand, estrogen and sex hormone-binding globulin (SHBG) are found to have beneficial effects on lipid profile, thus leading to a reduced CIMT [29].

Despite increasing knowledge about PCOS in the past years, the findings about CIMT and its predictors in PCOS are still very controversial and little is known about the true impact of the disorder. The aim of the present study was therefore to investigate whether CIMT is increased in PCOS patients compared to healthy control subjects, as well as to find out, if this cardiovascular risk factor is associated with hormone and metabolic profiles.

Materials and methods

Study participants

In this prospective cross-sectional study, 41 patients with PCOS between 18 and 34 years of age were consecutively recruited as they attended the endocrine outpatient department of Vienna General Hospital for medical consultation (Department of Gynecology, Division of Gynecological Endocrinology and Reproductive Medicine, Medical University of Vienna) between July 1st and September 30th 2015. Diagnosis of PCOS was made according to the Rotterdam ESHRE/ASRM criteria [3] if two out of the three following features were present: oligoovulation or anovulation, quantified by oligomenorrhea (≤6 menstrual periods in the past year [24]); clinical and/or biochemical signs of hyperandrogenism; PCOM, i.e. at least one ovary with presence of ≥12 follicles measuring 2–9 mm in diameter and/or increased ovarian volume (>10 ml) [3]. In this way, 4 different PCOS phenotypes were defined, i.e. phenotype 1 (hyperandrogenism, oligo-anovulation, PCOM), phenotype 2 (hyperandrogenism with oligo-anovulation), phenotype 3 (hyperandrogenism with PCOM) and phenotype 4 (oligo-anovulation with PCOM). Clinical hyperandrogenism included hirsutism, determined as Ferriman-Gallwey score (FGS) ≥8, and/or acne [24]. Biochemical hyperandrogenism was defined according to the local laboratory’s normal reference ranges as at least one of the following conditions: dehydroepiandrosterone sulfate (DHEAS) >3.7 μg/mL, free testosterone >0.22 ng/mL, total testosterone >0.48 ng/mL, androstenedione >4.1 ng/mL. Hyperandrogenemia was considered mandatory in order to fulfill the criterion of hyperandrogenism in this study. Other etiologies of hyperandrogenism (e.g. congenital adrenal hyperplasia, androgen-secreting tumors, Cushing’s syndrome, non-classic congenital adrenal hyperplasia using 17-hydroxyprogesterone (17-OHP) <2 ng/mL as surrogate parameter [3]) or secondary amenorrhea (anorexia, hyperprolactinemia, i.e. prolactin ≥31 ng/mL after polyethylene glycol immunoprecipitation [31]) were excluded. Thus, 26 newly diagnosed women with PCOS and 15 already diagnosed patients (1 month to 12 years ago) under no treatment for at least 3 months were included in this study.

43 age-matched, healthy and regularly menstruating female volunteers, drawn from the normal community and of similar body mass index (BMI) and frequency of smokers served as control subjects. PCOS was ruled out in all controls according to the Rotterdam ESHRE/ASRM criteria [3], as transvaginal ultrasound was performed between the third and the fifth day of their menstrual cycle in case of hyperandrogenism to exclude PCOS phenotype 3. Metabolic or cardiovascular diseases, e.g. diabetes mellitus, hyperlipidemia, arterial hypertension, coronary heart disease, thyroid dysfunction (i.e. thyroid-stimulating hormone (TSH) <0.44 μU/mL or ≥4.9 μU/mL) or BMI <17 kg/m2 or >50 kg/m2, lactation or pregnancy (beta-human chorionic gonadotropin (beta-hCG) levels >0.1 mU/mL), ongoing medication or hormonal contraception, current psychiatric disorders, heavy consumption of alcohol, caffeine or smoking (>20 cigarettes per day) were further ruled out in all subjects.

The study was approved by the Ethics Committee of the Medical University of Vienna (EK-Nr. 1197/2015). It was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all recruited subjects.

Clinical and biochemical measurements

Initial physical examination included past medical history, age, BMI (BMI = weight[kg]/height[m]2), waist-hip ratio (WHR = waist circumference (WC)[cm]/hip circumference (HC)[cm]), FGS and acne. We further defined the suspected starting point of disease in PCOS patients as the onset of oligomenorrhea. Therefore, when oligomenorrhea started right at menarche, it was noted as starting 0 years after menarche, meaning that 0 years equals menarche. After a 5-minute-rest in a sitting position, resting heart rate, systolic (SBP) and diastolic (DBP) blood pressure were taken with a sphygmomanometer, SBP and DBP being defined by the means of two consecutive measurements on each arm. Pelvic ultrasound was further performed in PCOS patients and controls with hyperandrogenism.

Blood samples were obtained in the morning after overnight fasting in subjects with a normal carbohydrate diet, between the third and the fifth day of their menstrual cycle or randomly in PCOS cases with an irregular cycle for the measurement of hormonal, lipidemic and glycemic parameters, 25-hydroxy-vitamine D and C-reactive protein. Free androgen index (FAI) was calculated as follows: (total testosterone[nmol/L]/SHBG[nmol/L])x100 [32]. Insulin sensitivity was estimated with homeostasis model assessment of insulin resistance (HOMA-IR = (glucose[mmol/L]x insulin[μU/mL])/22.5) [7].

Biochemical and hormonal parameters were acquired by enzyme-linked immuno assays (ELISA). All analyses were performed at the routine hormone laboratory of Vienna General Hospital, the Department of Medical and Chemical Laboratory Diagnostics (http://www.kimcl.at).

CIMT measurement

B-mode ultrasound images of the common carotid artery (CCA) were obtained within the next couple of days, using a 6- to 8-MHz high-resolution linear ultrasound probe for vascular/small parts (GE Medical Systems Ultrasound LOGIQ 9), following a standardized protocol under optimal adjustment of depth, gain and focus and controlled temperature and light conditions. After a resting period of 10 minutes to enable heart rate and blood pressure stabilization, subjects were examined in supine position, with a 35-degree extension and a slight right- or left-turn of the head, depending on the explored side. Left and right CCA were scanned in different transversal and longitudinal planes by the same trained sonographer, who was sitting at the top end of the examination bed behind the patient [33]. Interference of CIMT estimation with a possible plaque or thrombus was thus excluded. After freezing a gray-scale image in a longitudinal plane, CIMT was measured directly during diastole at the far wall of both distal CCAs, as the distance between the lumen-intima and media-adventitia interfaces [16,20,33]. Three consecutive measurements of CIMT were conducted on each side at a distance of 5 millimeters, starting off 1 centimeter prior to the common carotid bulb [20,33] (Fig 1). This resulted in an average value for each left and right CCA. The outcome variable “CIMT” was defined within a single subject as the mean value calculated from left and right CIMT [25], the average from both sides being a more stable representation of CIMT [21]. The intraclass correlation coefficient, measured in 21 subjects, was 0.98 for the right and 0.99 for the left CIMT.

thumbnail
Fig 1. CIMT measurement in PCOS patients and control subjects.

CIMT measurement at the far wall of the distal right (A) and left (B) common carotid artery as the distance between the lumen-intima and media-adventitia interfaces. (A) CIMT in a control subject. (B) CIMT in a PCOS patient.

https://doi.org/10.1371/journal.pone.0232299.g001

Statistical analysis

Deviation of continuous data from a normal distribution was assessed through visual inspection of histograms and box and whisker plots. Continuous scaled variables were expressed as means ± standard deviations (SD) if normally distributed, or medians and interquartile ranges (IQR: Q1-Q3). Categorical variables were presented as numbers and percentages (%). Age matching was achieved by ensuring that overall age distribution in terms of mean and standard deviation in both groups is roughly the same.

Continuous parameters were compared between PCOS patients and controls using unpaired two-tailed Student’s t-test when following a Gaussian distribution, or Mann-Whitney U test in case of skewed data. Nominal data were analyzed using χ2 test or Fisher’s exact test. Associations between CIMT and different continuous variables were assessed by Spearman’s rank correlation analysis (rho). The effect of nominal variables on CIMT was determined by Student’s t-test.

Subsequently, multiple linear regression analysis was performed in order to identify independent factors that predict CIMT as the dependent continuous variable. PCOS status, BMI, age and smoking status were therefore entered sequentially as independent factors into the regression model, in order to assess the effect of PCOS and the potential confounding effect of the above-mentioned parameters. They were followed by factors found to be significantly associated with CIMT in individual bivariate correlation. First, the independent variable best correlated with CIMT was included, then the one with the next highest correlation, checking for multicollinearity and normal distribution of the residuals. Nominal variables, i.e. PCOS and smoking status, were defined as dummy variables, coded as 1 for “yes” and 0 for “no”. Models were evaluated with the coefficient of determination R2, which expresses the proportion of variability in the dependent variable, i.e. CIMT, explained by the model. Regression analysis was also done for the oligomenorrhoic PCOS group.

All data analyses were performed using the Statistical Package for the Social Sciences software (IBM SPSS Statistics) version 24.0. Two-sided P-values <0.05 were considered statistically significant. Sample size calculation was performed using “Java applets for power and sample size” (http://homepage.stat.uiowa.edu/~rlenth/Power/) with a significance α of 0.05 and a statistical power of 90% set to find a mean difference in CIMT of 0.08 mm [20,32].

Results

Subject characteristics

Baseline demographic and anthropometrical parameters and clinical characteristics of the study population are exposed in Table 1. Females with PCOS showed increased BMI (P = 0.001) and higher prevalence of smokers (P = 0.007), metabolic syndrome according to the NCEP ATP III criteria [7] (P = 0.011) and parental history of metabolic disorders (i.e. maternal PCOS or DM or paternal DM, P = 0.001) compared to controls, who were slightly older and had higher prevalence of alcohol consumption (P = 0.001). PCOS patients also exhibited preponderance of abdominal obesity, as revealed by significantly increased WC, WHR (P<0.001) and HC (P = 0.019) and elevated prevalence of WC>80 cm (P = 0.001). As expected [3], they further featured significantly higher prevalence of ovulatory dysfunction, hirsutism and hyperandrogenemia (P<0.001) and increased age at menarche (P = 0.010). Accordingly, 29 (70.7%) met the criteria of classic PCOS (phenotype 1), 3 (7.3%) exhibited ovulatory PCOS (phenotype 3) and 9 (22%) had non-hyperandrogenic PCOS (phenotype 4). None showed phenotype 2. In oligomenorrhoic patients (phenotypes 1 and 4), onset of oligomenorrhea had mostly occurred at menarche, resulting in a suspected duration of disease since onset of 9.1 ± 4.8 years.

thumbnail
Table 1. Baseline demographic and anthropometrical parameters and clinical and biochemical diagnostic criteria of the polycystic ovary syndrome among the study population.

https://doi.org/10.1371/journal.pone.0232299.t001

Carotid intima-media thickness

CIMT values were significantly higher in PCOS patients than controls (0.49 ± 0.04 mm vs. 0.37 ± 0.04 mm respectively, P<0.001, Table 2), resulting in a mean difference of 0.12 mm between both study groups. They ranged from 0.40 to 0.63 mm in women with PCOS (n = 41) and from 0.31 to 0.46 mm in control subjects (n = 43, Fig 2). Key exploratory endpoints, i.e. right and left CIMT and maximal and minimal CIMT, were also significantly higher in PCOS patients (P<0.001).

thumbnail
Fig 2. Box and whisker plots of CIMT.

Among each group of interest (i.e. PCOS and controls). Box and whisker plots enable to summarize descriptive analysis of CIMT. They comprise the median (bold horizontal line) and the interquartile range (IQR), represented by a box, the first and third quartile being the borders of this box. Whiskers show values that are within 1.5 IQRs (vertical lines extending from the box). The small circle indicates an outlier (over 1.5 IQRs).

https://doi.org/10.1371/journal.pone.0232299.g002

Cardiovascular, metabolic and hormone profiles

Cardiovascular, metabolic and hormonal parameters are provided in Tables 2 and 3. In comparison to controls, PCOS patients showed higher levels of low-density lipoprotein cholesterol (LDL-C), total cholesterol/high-density lipoprotein cholesterol ratio (TC/HDL-C ratio) and apolipoprotein B (P<0.001), triglycerides (P = 0.001) and TC (P = 0.005), as well as lower levels of HDL-C (P = 0.044), revealing an adverse lipid profile. Increased levels of C-peptide and fasting insulin (P<0.05) also indicated higher endogenous insulin secretion. PCOS patients further presented with significantly higher levels of all measured androgens, FAI, 17-OHP, luteinizing hormone (LH), LH/FSH (follicle-stimulating hormone) ratio and Anti-Müllerian hormone (AMH) (P≤0.001), and lower levels of SHBG (P = 0.006) and FSH (P = 0.008).

Correlation and regression analysis

Several cardiovascular risk factors and hormonal parameters were found to be significantly positively correlated with CIMT in our subjects (Table 4), including total testosterone, free testosterone, androstenedione, FAI, AMH, LH/FSH ratio, WC, WHR, BMI, FGS and apolipoprotein B (P<0.001), as well as 17-OHP, LDL-C, smoking, TC, triglycerides, TC/HDL-C ratio, DHEAS and SBP (P<0.05). Besides, SHBG showed a significant negative relationship with CIMT (P = 0.023). When carried out separately in PCOS patients, correlation analysis revealed a significant positive association of CIMT with the suspected duration of disease (P = 0.012). Additionally, unpaired two-tailed Student’s t-test showed significantly higher CIMT values among the entire study population in case of hyperandrogenemia (0.47 ± 0.06 (n = 38) vs. 0.40 ± 0.07 (n = 46), P<0.001) and parental history of metabolic disorders (0.48 ± 0.05 (n = 21) vs. 0.41 ± 0.07 (n = 63), P<0.001).

thumbnail
Table 4. Correlation analysis with CIMT and baseline demographic characteristics, metabolic and hormone profiles with Spearman’s correlation coefficient (rho).

https://doi.org/10.1371/journal.pone.0232299.t004

Multiple linear regression analysis was carried out in order to identify independent factors that predict CIMT as the dependent continuous variable (Table 5). When considering PCOS patients and controls as a group, the diagnosis of PCOS was the strongest predictor of CIMT (β = 0.836, P<0.001), explaining 70% of its variability (P<0.001, model 1). In order to assess the potential confounding effect of obesity, age and smoking on CIMT, BMI, age and smoking status were entered sequentially as independent factors into the regression model. Only BMI represented an independent positive predictor of CIMT (β = 0.145, P = 0.025) among these parameters, raising the predicted variability in CIMT to 72% (P<0.001, model 2). However, PCOS status remained the primary predictor of CIMT, even after multiple adjustments for BMI, age and smoking status (P<0.001, R2 = 0.73, models 3 and 4), therefore indicating an independent effect of PCOS on CIMT (β = 0.797, P<0.001). No statistical significance was found for further hormonal and cardiovascular risk factors, probably due to multicollinearity with the PCOS status. When performed separately in oligomenorrhoic PCOS patients (n = 38), regression analysis revealed suspected duration of disease as an independent positive predictor of CIMT (β = 0.373, P = 0.021), accounting for 14% of the variability in CIMT (P = 0.021, model 5).

thumbnail
Table 5. Multiple linear regression models of CIMT as dependent variable.

https://doi.org/10.1371/journal.pone.0232299.t005

Discussion

The present study aimed to assess evidence of early systemic atherosclerosis in PCOS with measurement of CIMT and its possible association with hormone and metabolic profiles. Our results revealed significantly increased CIMT in PCOS patients when compared to healthy controls. This increase was further found to be independent of BMI, age and smoking status, given that PCOS status remained the primary predictor of CIMT after covariate adjustment. These findings suggest that the disorder itself is playing a causative role in CIMT increase.

As described in former studies, measurement of CIMT using B-mode ultrasound represents a valid surrogate marker of subclinical systemic atherosclerosis and enables cardiovascular risk stratification [17,18]. There is evidence that women with PCOS feature an increased CIMT compared to healthy controls [20,21,34], with reported mean differences in CIMT ranging from 0.06 mm [35] to 0.14 mm [24,36]. The mean difference of 0.12 mm found in our study is therefore in accordance with literature, and so are our measured CIMT values among each group of interest, given that a meta-analysis [21] indicates a mean CIMT ranging from 0.41 to 0.75 mm in PCOS and from 0.33 to 0.74 mm in controls in previous studies. Differences in the prevalence of PCOS phenotypes and thus cardiovascular risk profiles between studies can explain discrepant findings regarding elevated CIMT in literature [21,25]. In fact, the prevalence of total and abdominal obesity, metabolic disturbances and cardiovascular risk factors is found to decrease from phenotype 1 to 4, just like the severity of hyperandrogenism and menstrual disturbances [3,10,37,38].

In the present study, several cardiovascular risk factors and hormonal parameters were found to be significantly associated with CIMT, i.e. visceral obesity, dyslipidemia, hyperandrogenemia, AMH, smoking, SBP and parental history of metabolic disorders.

Obesity is a common finding in PCOS and has been linked to increased CIMT in young and middle-aged women [39]. As a matter of fact, women with PCOS are found to exhibit central body fat distribution regardless of BMI, related to hyperandrogenemia and hyperinsulinemia [40]. However, adipose tissue is metabolically active and increased visceral fat thus enhances secretion of free fatty acids (FFA), adipokines, i.e. adiponectin and leptin, growth factors and inflammatory cytokines, e.g. TNF-α [41]. Those substances have a crucial impact on metabolism and cardiovascular system. Visceral adipose tissue is therefore a source of inflammation, the latter representing a key component of atherosclerosis development [42]. Markers of premature atherosclerosis, such as C-reactive protein (CRP), homocysteine and endothelial dysfunction were shown to be involved in CIMT increase [27,32,36]. Visceral obesity is further associated with dyslipidemia and IR, consequently leading to increased risk of DM and cardiovascular disease [40,43,44]. A positive relationship has thus been described with early signs of vascular damage and CIMT increase [32,45]. The higher prevalence of WC >80 cm in our PCOS patients, as defined in European women by the WHO [46], therefore indicates an increased risk of metabolic and cardiovascular complications, WC being a better predictor than WHR or BMI [43]. Metabolic syndrome according to the NCEP ATP III [7] being also more prevalent in our patients and former studies [24] further contributes to dyslipidemia and IR, leading to enhanced atherosclerosis and eventually CIMT increase.

Consistent with literature, PCOS patients featured an adverse lipid profile, including a higher TC/HDL-C ratio, which represents a risk factor for coronary heart disease [24,47]. Dyslipidemia is believed to be the consequence of increased visceral fat [32] and androgen levels [48,49]. Previous studies further confirmed the association found between CIMT and an adverse lipid profile, CIMT being positively associated with the atherogenic LDL-C, TC and triglycerides and negatively correlated with the antiatherogenic HDL-C [26]. As a result, women with PCOS are at increased risk of early systemic atherosclerosis [47].

The positive association of androgen levels with CIMT and higher CIMT values found in case of hyperandrogenemia are in accordance with previous findings [20], suggesting that CIMT increase in PCOS is being caused by hyperandrogenism. High levels of androgens are believed to lead to an increased CIMT through a direct effect on the vascular system. In fact, testosterone has proatherogenic effects on macrophage function by facilitating the uptake of modified lipoproteins [50] and enhancing monocyte adhesion to vascular endothelial cells [51]. Furthermore, androgens regulate lipolysis and lipogenesis in women, having an impact on lipid metabolism and adipocyte production [52]. Testosterone has been found to inhibit catecholamine-stimulated lipolysis in subcutaneous fat cells through decreased expression of hormone-sensitive lipase (HSL) and β2-adrenoceptors [53]. However, it seems to exert the opposite effect on visceral fat, leading to an increased sensitivity to catecholamines through upregulation of lipolytic β3-adrenergic receptors and HSL [49,54] and a reduced sensitivity to the antilipolytic effect of insulin [40,55]. As a consequence, the elevated release of FFA affects liver functioning and insulin signaling, leading to dyslipidemia, hyperinsulinemia, glucose intolerance and IR. It also reduces hepatic SHBG production, amplifying hyperandrogenemia [56]. Androgens further lead to an adverse lipid profile through inhibition of lipoprotein lipase (LPL) in adipose tissue [48,49]. Moreover, serum testosterone has been associated with advanced glycation end products (AGEs) [57]. These proinflammatory and oxidant mediators result from non-enzymatic glycation of proteins or lipids and are believed to play a contributive role in PCOS development [58,59], given that elevated levels have been reported [10,38,57,58]. AGEs further exert a proatherogenic effect on the vascular system [60]. They have been related to tissue damaging in the framework of atherosclerosis and may therefore contribute to the elevated cardiovascular risk in PCOS [38,57]. Besides, the role of testosterone as crucial determinant of CIMT increase is further emphasized by the positive relationship of CIMT seen with FAI, which represents a good reflection of testosterone action. Furthermore, the observed association between CIMT and hirsutism demonstrates the impact played by dihydrotestosterone (DHT), which leads to similar regional differences concerning lipolysis as testosterone [48,53]. The positive relationship of CIMT found with AMH further underlines the impact of hyperandrogenemia. In fact, granulosa cells of growing ovarian follicles in PCOS secrete elevated levels of AMH, which inhibit aromatase activity, thus contributing to androgen excess [61].

Smoking and SBP were also positively associated with CIMT. Indeed, smoking enhances the progression of extracoronary atherosclerosis [12] and hypertension is involved in the pathogenesis of vascular damage and thus atherosclerosis [62]. Moreover, higher CIMT values were seen in case of parental history of metabolic disorders, putting an emphasis on genetic predisposition in PCOS. Despite showing no correlation with CIMT in our study, higher endogenous insulin secretion, as found in our patients, is thought to contribute to CIMT increase. In fact, insulin exerts an atherogenic effect by increasing cholesterol transport into arteriolar smooth muscle cells. In this way, it stimulates their proliferation and endogenous cholesterol and collagen synthesis, eventually leading to elevated formation of lipid plaques [63].

Finally, our results from multiple regression analysis are in accordance with previous studies, in which the diagnosis of PCOS and BMI were found to predict CIMT [24,27], with PCOS status being the strongest predictor, independent of BMI [27]. Furthermore, our study revealed suspected duration of disease as predictor of CIMT, which is a new finding that has not been reported so far to our knowledge. Therefore, oligomenorrhoic patients seem to feature higher CIMT the earlier onset of disease occurs. Although suspected duration of disease is related to the subjects’ age, it also depends on the age of menarche that varies in each patient. Besides, PCOS patients have later menarche than healthy controls [30,64,65], therefore it does not totally equate to the age. The observed relationship between suspected duration of disease and CIMT in oligomenorrhoic patients thus suggests even more that PCOS itself contributes to the enhancement of atherosclerosis, probably due to hyperandrogenism, an adverse lipid profile and hyperinsulinemia.

However, although women with PCOS feature increased CIMT and metabolic and cardiovascular risk at a younger age, there is evidence suggesting that they do not show higher prevalence of cardiovascular events than controls after menopause [30,64,65]. Possible explanations could be a protective effect of delayed menopause with a consequently prolonged estrogen exposure in PCOS [30,64,65] or even hyperandrogenism itself [64,65] in peri- and postmenopausal PCOS patients, as indicated by previous studies [27,28,29], probably mainly due to enzymatic conversion to estrogen. Moreover, differences in cardiovascular risk factors (e.g. diabetes, abdominal obesity) between PCOS and controls seem to be less preponderant in aging women, explaining the similar cardiovascular morbidity and mortality later in life [65].

Strengths of this study include the comparable sample size and age in both groups of interest. Controls were slightly older than PCOS patients, enabling to exclude even more a potential confounding effect of age on CIMT. All subjects further benefited from a thorough hormone, metabolic and cardiovascular assessment. Regarding limitations of the current study, even though most patients featured the classic PCOS phenotype, 29.3% belonged to phenotypes 3 or 4, with lesser severity of clinical and metabolic manifestations [3]. This could explain the only mild metabolic disturbances seen in our patients. Given that 9 PCOS patients without hyperandrogenism (i.e. phenotype 4) and 6 controls with hyperandrogenemia were included in the study, the impact of androgen excess on CIMT may have been underestimated and thus no significant association was found in multiple regression analysis. Besides, the presence of acne was quantified subjectively, thus no major focus should be put on its prevalence. Moreover, estrogen and progesterone effect on CIMT could not be examined in our study, given that blood tests were not always performed during follicular phase in oligomenorrhoic PCOS patients. Analyses should therefore be conducted in these women after progestin-induced withdrawal bleeding. Further studies with a larger sample size, especially among each PCOS phenotype, are required in order to unravel the exact underlying pathogenetic mechanisms in PCOS and to identify which components of this disorder have the greatest effect on CIMT.

Conclusions

The present study indicates that patients with PCOS are likely to feature signs of premature systemic atherosclerosis already at a young age, as shown by their elevated CIMT values. Our findings further suggest that early exposure to adverse cardiovascular risk factors in the framework of this disorder may possibly have long-term consequences on the vascular system, given that the duration of disease seems to have a predictive impact on the extent of CIMT increase. In fact, metabolic disturbances, including visceral obesity, IR with compensatory hyperinsulinemia and dyslipidemia were found to be involved. Moreover, our results revealed that hyperandrogenism, a central feature of PCOS, represents a crucial determinant of CIMT elevation in women with this disorder and is thus implicated in the enhancement of atherosclerosis. However, there seems to exist a still unexplained independent effect of PCOS. A possible contributing factor to the elevated cardiovascular risk could be the proatherogenic role of AGEs, leading to tissue damage in the framework of atherosclerosis. PCOS patients thus seem to be at greater risk for early systemic atherosclerosis. This emphasizes the importance of thorough metabolic and cardiovascular evaluation in all women with PCOS. An early vessel screening might therefore already be beneficial in these patients at a younger age.

Acknowledgments

Thanks to the Departments of Radiology and Neurology of the Medical University of Vienna for their support. We also thank Georg Dorffner, Ph.D., M.S. (Center for Medical Statistics, Informatics and Intelligent Systems, Institute of Artificial Intelligence) and Christian Göbl, M.D. (Department of Obstetrics and Gynecology, Medical University of Vienna) for their statistical advice.

References

  1. 1. Azziz R, Woods KS, Reyna R, Key TJ, Knochenhauer ES, Yildiz BO. The prevalence and features of the polycystic ovary syndrome in an unselected population. J Clin Endocrinol Metab. 2004 Jun;89(6):2745–9. pmid:15181052
  2. 2. Stein I, Leventhal M. Amenorrhea associated with bilateral polycystic ovaries. Am J Obs Gynecol. 1935;29(2):181–91.
  3. 3. Rotterdam. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Hum Reprod. 2004 Jan;19(1):41–7. pmid:14688154
  4. 4. Rosenfield RL. Ovarian and adrenal function in polycystic ovary syndrome. Endocrinol Metab Clin North Am. 1999 Jun;28(2):265–93. pmid:10352919
  5. 5. Jakimiuk AJ, Weitsman SR, Navab A, Magoffin DA. Luteinizing Hormone Receptor, Steroidogenesis Acute Regulatory Protein, and Steroidogenic Enzyme Messenger Ribonucleic Acids Are Overexpressed in Thecal and Granulosa Cells from Polycystic Ovaries 1. J Clin Endocrinol Metab. 2001 Mar;86(3):1318–23. pmid:11238527
  6. 6. Goodarzi MO, Dumesic DA, Chazenbalk G, Azziz R. Polycystic ovary syndrome: etiology, pathogenesis and diagnosis. Nat Rev Endocrinol. 2011 Apr 25;7(4):219–31. pmid:21263450
  7. 7. Azziz R. Evaluation for Insulin Resistance and Comorbidities Related to Insulin Resistance In Polycystic Ovary Syndrome. In: Diamanti-Kandarakis E, Nestler JE, Panidis D, Pasquali R, editors. Insulin Resistance and Polycystic Ovarian Syndrome—Pathogenesis, Evaluation and Treatment. Totowa, New Jersey: Humana Press Inc.; 2007. p. 1–13.
  8. 8. Diamanti-Kandarakis E, Dunaif A. Insulin Resistance and the Polycystic Ovary Syndrome Revisited: An Update on Mechanisms and Implications. Endocr Rev. 2012 Dec;33(6):981–1030. pmid:23065822
  9. 9. Moran LJ, Misso ML, Wild RA, Norman RJ. Impaired glucose tolerance, type 2 diabetes and metabolic syndrome in polycystic ovary syndrome: a systematic review and meta-analysis. Hum Reprod Update. 2010 Jan;16(4):347–63. pmid:20159883
  10. 10. Bajuk Studen K, Pfeifer M. Cardiometabolic risk in polycystic ovary syndrome. Endocr Connect. 2018 Jul;7(7):R238–51. pmid:29844207
  11. 11. Laganà AS, Garzon S, Casarin J, Franchi M, Ghezzi F. Inositol in Polycystic Ovary Syndrome: Restoring Fertility through a Pathophysiology-Based Approach. Vol. 29, Trends in Endocrinology and Metabolism. Elsevier Inc.; 2018. p. 768–80. https://doi.org/10.1016/j.tem.2018.09.001 pmid:30270194
  12. 12. van der Meer IM, Iglesias del Sol A, Hak AE, Bots ML, Hofman A, Witteman JCM. Risk factors for progression of atherosclerosis measured at multiple sites in the arterial tree: the Rotterdam Study. Stroke. 2003 Oct;34(10):2374–9. pmid:12947155
  13. 13. Legro RS. Polycystic Ovary Syndrome and Cardiovascular Disease: A Premature Association? Endocr Rev. 2003 Jun;24(3):302–12. pmid:12788801
  14. 14. de Groot PCM, Dekkers OM, Romijn JA, Dieben SWM, Helmerhorst FM. PCOS, coronary heart disease, stroke and the influence of obesity: a systematic review and meta-analysis. Hum Reprod Update. 2011 Jul 1;17(4):495–500. pmid:21335359
  15. 15. Talbott EO, Zborowski J V., Rager JR, Boudreaux MY, Edmundowicz DA, Guzick DS. Evidence for an Association between Metabolic Cardiovascular Syndrome and Coronary and Aortic Calcification among Women with Polycystic Ovary Syndrome. J Clin Endocrinol Metab. 2004 Nov;89(11):5454–61. pmid:15531497
  16. 16. Pignoli P, Tremoli E, Poli A, Oreste P, Paoletti R. Intimal plus medial thickness of the arterial wall: a direct measurement with ultrasound imaging. Circulation. 1986 Dec;74(6):1399–406. pmid:3536154
  17. 17. Poredos P. Intima-media thickness: indicator of cardiovascular risk and measure of the extent of atherosclerosis. Vasc Med. 2004 Feb;9(1):46–54. pmid:15230488
  18. 18. Lorenz MW, Markus HS, Bots ML, Rosvall M, Sitzer M. Prediction of clinical cardiovascular events with carotid intima-media thickness: a systematic review and meta-analysis. Circulation. 2007 Jan 30;115(4):459–67. pmid:17242284
  19. 19. Chambless LE, Heiss G, Folsom AR, Rosamond W, Szklo M, Sharrett AR, et al. Association of coronary heart disease incidence with carotid arterial wall thickness and major risk factors: the Atherosclerosis Risk in Communities (ARIC) Study, 1987–1993. Am J Epidemiol. 1997 Sep 15;146(6):483–94. pmid:9290509
  20. 20. Luque-Ramírez M, Mendieta-Azcona C, Alvarez-Blasco F, Escobar-Morreale HF. Androgen excess is associated with the increased carotid intima-media thickness observed in young women with polycystic ovary syndrome. Hum Reprod. 2007 Dec;22(12):3197–203. pmid:17933750
  21. 21. Meyer ML, Malek AM, Wild RA, Korytkowski MT, Talbott EO. Carotid artery intima-media thickness in polycystic ovary syndrome: a systematic review and meta-analysis. Hum Reprod Update. 18(2):112–26. pmid:22108382
  22. 22. Talbott EO, Guzick DS, Sutton-Tyrrell K, McHugh-Pemu KP, Zborowski J V, Remsberg KE, et al. Evidence for association between polycystic ovary syndrome and premature carotid atherosclerosis in middle-aged women. Arterioscler Thromb Vasc Biol. 2000 Nov;20(11):2414–21. pmid:11073846
  23. 23. Guzick DS, Talbott EO, Sutton-Tyrrell K, Herzog HC, Kuller LH, Wolfson SK. Carotid atherosclerosis in women with polycystic ovary syndrome: initial results from a case-control study. Am J Obstet Gynecol. 1996 Apr;174(4):1224–32. pmid:8623850
  24. 24. Vural B, Caliskan E, Turkoz E, Kilic T, Demirci A. Evaluation of metabolic syndrome frequency and premature carotid atherosclerosis in young women with polycystic ovary syndrome. Hum Reprod. 2005 Sep;20(9):2409–13. pmid:15919774
  25. 25. Kim JJ, Choi YM, Kang JH, Hwang KR, Chae SJ, Kim SM, et al. Carotid intima-media thickness in mainly non-obese women with polycystic ovary syndrome and age-matched controls. Obstet Gynecol Sci. 2013 Jul;56(4):249–55. pmid:24328010
  26. 26. Saha S, Sarkar C, Biswas SC, Karim R. Correlation between serum lipid profile and carotid intima-media thickness in Polycystic Ovarian Syndrome. Indian J Clin Biochem. 2008 Jul;23(3):262–6. pmid:23105767
  27. 27. Vryonidou A, Papatheodorou A, Tavridou A, Terzi T, Loi V, Vatalas I-A, et al. Association of hyperandrogenemic and metabolic phenotype with carotid intima-media thickness in young women with polycystic ovary syndrome. J Clin Endocrinol Metab. 2005 May;90(5):2740–6. pmid:15741256
  28. 28. Bernini GP, Sgro’ M, Moretti A, Argenio GF, Barlascini CO, Cristofani R, et al. Endogenous androgens and carotid intimal-medial thickness in women. J Clin Endocrinol Metab. 1999 Jun;84(6):2008–12. pmid:10372702
  29. 29. Karim R, Hodis HN, Stanczyk FZ, Lobo RA, Mack WJ. Relationship between serum levels of sex hormones and progression of subclinical atherosclerosis in postmenopausal women. J Clin Endocrinol Metab. 2008 Jan;93(1):131–8. pmid:17925335
  30. 30. Meun C, Franco OH, Dhana K, Jaspers L, Muka T, Louwers Y, et al. High Androgens in Postmenopausal Women and the Risk for Atherosclerosis and Cardiovascular Disease: The Rotterdam Study. J Clin Endocrinol Metab. 2018;103(4):1622–30. pmid:29408955
  31. 31. Suliman AM, Smith TP, Gibney J, McKenna TJ. Frequent misdiagnosis and mismanagement of hyperprolactinemic patients before the introduction of macroprolactin screening: application of a new strict laboratory definition of macroprolactinemia. Clin Chem. 2003 Sep;49(9):1504–9. pmid:12928232
  32. 32. Cascella T, Palomba S, De Sio I, Manguso F, Giallauria F, De Simone B, et al. Visceral fat is associated with cardiovascular risk in women with polycystic ovary syndrome. Hum Reprod. 2008 Jan;23(1):153–9. pmid:18024952
  33. 33. Nagele W, Nagele J. Sonographie der Karotis-Intima-Dicke. Zeitschrift für Gefäßmedizin. 2015;12(1):5–9.
  34. 34. Garg N, Dharmalingam M, Prabhu V, Murthy NS. Carotid intimo-medial thickness: A predictor for cardiovascular disorder in patients with polycystic ovarian syndrome in the South Indian population. Indian J Endocrinol Metab. 20(5):662–6. pmid:27730077
  35. 35. Heutling D, Schulz H, Nickel I, Kleinstein J, Kaltwasser P, Westphal S, et al. Asymmetrical dimethylarginine, inflammatory and metabolic parameters in women with polycystic ovary syndrome before and after metformin treatment. J Clin Endocrinol Metab. 2008 Jan;93(1):82–90. pmid:17986642
  36. 36. Orio F, Palomba S, Cascella T, De Simone B, Di Biase S, Russo T, et al. Early impairment of endothelial structure and function in young normal-weight women with polycystic ovary syndrome. J Clin Endocrinol Metab. 2004 Sep;89(9):4588–93. pmid:15356067
  37. 37. Kim JJ, Hwang KR, Choi YM, Moon SY, Chae SJ, Park CW, et al. Complete phenotypic and metabolic profiles of a large consecutive cohort of untreated Korean women with polycystic ovary syndrome. Fertil Steril. 2014 May;101(5):1424–30. pmid:24613538
  38. 38. Sathyapalan T, Atkin SL. Recent advances in cardiovascular aspects of polycystic ovary syndrome. Eur J Endocrinol. 2012 Apr;166(4):575–83. pmid:22096112
  39. 39. Lo J, Dolan SE, Kanter JR, Hemphill LC, Connelly JM, Lees RS, et al. Effects of obesity, body composition, and adiponectin on carotid intima-media thickness in healthy women. J Clin Endocrinol Metab. 2006 May;91(5):1677–82. pmid:16522690
  40. 40. Carmina E. Fat Distribution and Adipose Products in Polycystic Ovary Syndrome. In: Diamanti-Kandarakis E, Nestler JE, Panidis D, Pasquali R, editors. Insulin Resistance and Polycystic Ovarian Syndrome—Pathogenesis, Evaluation and Treatment. Totowa, New Jersey: Humana Press Inc.; 2007. p. 15–24.
  41. 41. Peppa M, Koliaki C, Nikolopoulos P, Raptis SA. Skeletal muscle insulin resistance in endocrine disease. J Biomed Biotechnol. 2010;2010:13.
  42. 42. Alexopoulos N, Katritsis D, Raggi P. Visceral adipose tissue as a source of inflammation and promoter of atherosclerosis. Atherosclerosis. 2014 Mar;233(1):104–12. pmid:24529130
  43. 43. Tchernof A, Despres J-P. Pathophysiology of Human Visceral Obesity: An Update. Physiol Rev. 2013 Jan 1;93(1):359–404. pmid:23303913
  44. 44. Carmina E, Bucchieri S, Esposito A, Del Puente A, Mansueto P, Orio F, et al. Abdominal fat quantity and distribution in women with polycystic ovary syndrome and extent of its relation to insulin resistance. J Clin Endocrinol Metab. 2007 Jul;92(7):2500–5. pmid:17405838
  45. 45. Karabulut A, Yaylali GF, Demirlenk S, Sevket O, Acun A. Evaluation of body fat distribution in PCOS and its association with carotid atherosclerosis and insulin resistance. Gynecol Endocrinol. 2012 Feb;28(2):111–4. pmid:21770828
  46. 46. Waist Circumference and Waist-Hip Ratio—Report of a WHO Expert Consultation, Geneva, December 8-11th, 2008. In.
  47. 47. Macut D, Bačević M, Božić-Antić I, Bjekić-Macut J, Čivčić M, Erceg S, et al. Predictors of subclinical cardiovascular disease in women with polycystic ovary syndrome: interrelationship of dyslipidemia and arterial blood pressure. Int J Endocrinol. 2015;2015:9.
  48. 48. Blouin K, Nadeau M, Perreault M, Veilleux A, Drolet R, Marceau P, et al. Effects of androgens on adipocyte differentiation and adipose tissue explant metabolism in men and women. Clin Endocrinol (Oxf). 2010 Feb;72(2):176–88.
  49. 49. De Pergola G. The adipose tissue metabolism: role of testosterone and dehydroepiandrosterone. Int J Obes Relat Metab Disord. 2000 Jun;24 Suppl 2:59–63.
  50. 50. Wu FCW, von Eckardstein A. Androgens and Coronary Artery Disease. Endocr Rev. 2003 Apr;24(2):183–217. pmid:12700179
  51. 51. McCrohon JA, Jessup W, Handelsman DJ, Celermajer DS. Androgen exposure increases human monocyte adhesion to vascular endothelium and endothelial cell expression of vascular cell adhesion molecule-1. Circulation. 1999 May 4;99(17):2317–22. pmid:10226099
  52. 52. Corbould A. Chronic testosterone treatment induces selective insulin resistance in subcutaneous adipocytes of women. J Endocrinol. 2007 Mar 1;192(3):585–94. pmid:17332526
  53. 53. Dicker A, Rydén M, Näslund E, Muehlen IE, Wirén M, Lafontan M, et al. Effect of testosterone on lipolysis in human pre-adipocytes from different fat depots. Diabetologia. 2004 Mar 1;47(3):420–8. pmid:14752604
  54. 54. Rojas J, Chávez M, Olivar L, Rojas M, Morillo J, Mejías J, et al. Polycystic ovary syndrome, insulin resistance, and obesity: navigating the pathophysiologic labyrinth. Int J Reprod Med. 2014;2014:719050. pmid:25763405
  55. 55. Wajchenberg BL. Subcutaneous and Visceral Adipose Tissue: Their Relation to the Metabolic Syndrome. Endocr Rev. 2000 Dec;21(6):697–738. pmid:11133069
  56. 56. Ek I, Arner P, Rydén M, Holm C, Thörne A, Hoffstedt J, et al. A unique defect in the regulation of visceral fat cell lipolysis in the polycystic ovary syndrome as an early link to insulin resistance. Diabetes. 2002 Feb;51(2):484–92. pmid:11812759
  57. 57. Diamanti-Kandarakis E, Piperi C, Kalofoutis A, Creatsas G. Increased levels of serum advanced glycation end-products in women with polycystic ovary syndrome. Clin Endocrinol (Oxf). 2005 Jan;62(1):37–43.
  58. 58. Diamanti-Kandarakis E, Katsikis I, Piperi C, Kandaraki E, Piouka A, Papavassiliou AG, et al. Increased serum advanced glycation end-products is a distinct finding in lean women with polycystic ovary syndrome (PCOS). Clin Endocrinol (Oxf). 2008 Oct;69(4):634–41.
  59. 59. Diamanti-Kandarakis E, Piperi C, Patsouris E, Korkolopoulou P, Panidis D, Pawelczyk L, et al. Immunohistochemical localization of advanced glycation end-products (AGEs) and their receptor (RAGE) in polycystic and normal ovaries. Histochem Cell Biol. 2007 Jun 5;127(6):581–9. pmid:17205306
  60. 60. VLASSARA H. Advanced Glycation in Health and Disease: Role of the Modern Environment. Ann N Y Acad Sci. 2005 Jun;1043(1):452–60.
  61. 61. Jonard S, Dewailly D. The follicular excess in polycystic ovaries, due to intra-ovarian hyperandrogenism, may be the main culprit for the follicular arrest. Hum Reprod Update. 2004 Mar 1;10(2):107–17. pmid:15073141
  62. 62. Pauletto P, Scannapieco G, Pessina AC. Sympathetic drive and vascular damage in hypertension and atherosclerosis. Hypertens (Dallas, Tex 1979). 1991 Apr;17(4 Suppl):III75–81.
  63. 63. DeFronzo RA, Ferrannini E. Insulin resistance. A multifaceted syndrome responsible for NIDDM, obesity, hypertension, dyslipidemia, and atherosclerotic cardiovascular disease. Diabetes Care. 1991 Mar;14(3):173–94. pmid:2044434
  64. 64. Fauser BCJM, Bouchard P. Uncertainty remains in women with PCOS regarding the increased incidence of cardiovascular disease later in life, despite the indisputable presence of multiple cardiovascular risk factors at a young age. Vol. 96, Journal of Clinical Endocrinology and Metabolism. 2011. p. 3675–7. pmid:22143833
  65. 65. Gunning MN, Fauser BCJM. Are women with polycystic ovary syndrome at increased cardiovascular disease risk later in life? Vol. 20, Climacteric. Taylor and Francis Ltd; 2017. p. 222–7. https://doi.org/10.1080/13697137.2017.1316256 pmid:28457146