Associations between Plasma 25-Hydroxyvitamin D, Hormonal Contraceptives, and Premenstrual Symptoms

Transcription

1 Associations between Plasma 25-Hydroxyvitamin D, Hormonal Contraceptives, and Premenstrual Symptoms by Alicia Jarosz A thesis submitted in conformity with the requirements for the degree of Master of Science (MSc) Department of Nutritional Sciences University of Toronto Copyright by Alicia Jarosz 2017

2 Associations between Plasma 25-Hydroxyvitamin D, Hormonal Contraceptives, and Premenstrual Symptoms Alicia Jarosz Master of Science (MSc) Department of Nutritional Sciences University of Toronto 2017 Abstract Premenstrual symptoms are experienced by the majority of women and may cause significant personal and professional impairment; however, little is known about their pathophysiology and risk factors. The purpose of this thesis was to determine the prevalence of common premenstrual symptoms in a multiethnic Canadian population and to explore the associations of plasma 25- hydroxyvitamin D and hormonal contraceptive use with these symptoms. Symptom prevalence was found to vary widely between common symptoms, ranging from 11% to 75%. Prevalence of individual symptoms did not differ between ethnic groups, with the exception of cramps. Hormonal contraceptive use was associated with a reduction in the risk of experiencing several symptoms at moderate/severe severity. Plasma 25-hydroxyvitamin D was also inversely associated with the prevalence and severity of several premenstrual symptoms. These findings suggest HC use may be an effective targeted treatment and vitamin D status may be a risk factor for individual premenstrual symptoms. ii

3 Acknowledgments First and foremost, I would like to thank my advisor Dr. Ahmed El-Sohemy for the tremendous support and opportunities he has provided to me in the last two years. I am certain that I could not have had a more encouraging supervisor. His guidance and knowledge were crucial in helping me reach my academic and professional pursuits. Thank you, Ahmed. I had a lot to learn as a young undergraduate student entering your lab and I am incredibly grateful for your support and the countless learning opportunities you have created to encourage my professional development. I would like to thank my advisory committee, Dr. Joanne Kotsopoulos and Dr. Richard Bazinet, for their guidance throughout the progression of this project. Their knowledgeable advice and direction has benefited my thesis and their enthusiasm has encouraged me along the way. Thank you, Dr. Kotsopoulos, for taking the time to help me with my academic writing. My thesis and manuscripts have improved greatly thanks to you. I would also like to thank my professors, Dr. Beatrice Boucher, Dr. Anthony Hanley, and Dr. Paul Corey. The research skills they taught me in epidemiology and statistics helped me immensely with my thesis. Thank you, Dr. Corey, for staying hours late after class to help me understand my data and find the best statistical approach. Finally, thank you to Louisa Kung, who was always there to answer every possible question I could have. My time as a graduate student would not have been so enjoyable without my wonderful El-Sohemy lab mates and colleagues in the Nutritional Sciences department. Thank you to Ohood Alharbi, Neshat Deljoomanesh, Nanci Guest, Riva Sorkin, Sara Mahdavi, Joseph Jamnik, Bryn Dhir, Katie Edmonds, and Daniel Noori. I am grateful to have had the chance to work alongside such kind, helpful, and entertaining lab mates. I will forever look back fondly at the iii

4 time we spent together. Thank you to my honorary advisor, Joseph Jamnik, for teaching me everything I needed to know about SAS and TNH. You were there to help me through every research obstacle I encountered and I could not have done this without you. In many ways, I consider my successful completion of this thesis a team effort with my friends and family who have supported me every step of the way. I would like to thank everyone who has been a source of support and encouragement, and with whom I have celebrated all the small accomplishments along this journey. I would especially like to thank my dad and sister, Jerzy and Isabel, who have always been my biggest supporters in the pursuit of my educational and career dreams. I would also like to thank my boyfriend, Razvan, for his unfaltering support throughout my entire university education. You have all gotten me through the stressful and hectic moments that are an inevitable part of any worthwhile endeavor, and celebrated my triumphs as if they were your own. This accomplishment is as much yours as it is mine, and I am eternally grateful to you all. iv

5 Table of Contents Abstract... ii Acknowledgments... iii Table of Contents... v List of Tables... viii List of Figures... ix List of Abbreviations... x Chapter 1 Introduction Introduction Premenstrual Symptoms Premenstrual Disorders Etiology Treatment Risk Factors Hormonal Contraceptives Introduction Mechanisms of Action HCs and Premenstrual Symptoms Vitamin D Background Vitamin D Metabolism Functions of Vitamin D Determinants of Vitamin D Status Measurement of 25(OH)D Vitamin D and Premenstrual Symptoms Summary and Rationale v

6 1.6 Hypothesis and Objectives Chapter 2 Prevalence of Premenstrual Symptoms and Associations with Use of Hormonal Contraceptives Abstract Introduction Methods Study Population Hormonal Contraceptive Use Anthropometrics and Physical Activity Premenstrual Symptoms Plasma Samples and Vitamin D Measurement Statistical Analysis Results Study Population Prevalence of Premenstrual Symptoms Premenstrual Symptom Associations with HC use Discussion Chapter 3 Association between Plasma 25-Hydroxyvitamin D and Premenstrual Symptoms Abstract Introduction Materials and Methods Study Population Hormonal Contraceptive Use Anthropometrics and Physical Activity Premenstrual Symptoms Plasma Samples and 25-Hydroxyvitamin D Analysis vi

7 3.3.6 Food Frequency Questionnaire Statistical Analysis Results Discussion Chapter 4 Synopsis, Limitations and Future Directions Synopsis Limitations Future Directions References Appendices vii

8 List of Tables Table 2-1 Subject Characteristics Stratified by Hormonal Contraceptive (HC) Use 1, Table 2-2 Subject Characteristics Stratified by Hormonal Contraceptive (HC) Use 1, Table 2-3 Premenstrual Symptom Prevalence by Ethnicity Table 2-4 Associations between HC Use and Premenstrual Symptom Severity Table 2-5 Associations between Duration of HC Use and Premenstrual Symptoms Table 3-6 Subject Characteristics Stratified by Vitamin D Status 1, Table 3-7 Associations between Plasma 25-Hydroxyvitamin D and Premenstrual Symptom Severity Table A-1 GHLQ Premenstrual Symptom Questionnaire viii

9 List of Figures Figure 2-1 Associations between HC Use and Mild Premenstrual Symptoms Figure 2-2 Associations between HC Use and Moderate/Severe Premenstrual Symptoms Figure 2-3 Associations between Duration of HC Use and Mild Premenstrual Symptoms Figure 2-4 Associations between Duration of HC Use and Moderate/Severe Premenstrual Symptoms Figure 3-5 Associations between Plasma 25-Hydroxyvitamin D and Premenstrual Symptom Severity 1, ix

10 List of Abbreviations PMD - Premenstrual Disorder PMS - Premenstrual Syndrome PMDD - Premenstrual Dysphoric Disorder ACOG - American College of Obstetricians and Gynecologists GnRH - Gonadotropin-releasing hormone CNS - Central nervous system CRP - C-reactive protein IL - Interleukin PTH - Parathyroid hormone 1,25(OH)D - 1,25-hydroxyvitamin D 25(OH)D - 25-hydroxyvitamin D SSRI - Selective serotonin uptake inhibitor BMI - Body mass index RCT - Randomized control trial HC - Hormonal contraceptive OC - Oral contraceptive COC - Combined oral contraceptive FSH - Follicle stimulating hormone LH - Luteinizing hormone DMPA - Depot medroxyprogesterone UV - Ultraviolet DBP - Vitamin d binding protein VDR - Vitamin d receptor RAAS - Renin-aldosterone-angiotensin-system VTE - Venous thromboembolism x

11 Chapter 1 Introduction 11

12 1.1 Introduction Premenstrual symptoms are a collection of physiological, behavioral, and psychological symptoms that occur during the late luteal phase of a woman s reproductive cycle. They are characterized by their timing and by their cyclic nature, while the nature, number, and severity of the symptoms varies between women 1. Hundreds of symptoms have been described to date, however, the most commonly experienced somatic symptoms are bloating, headache, fatigue, and muscle cramps. Behavioral and psychological symptoms are also commonly experienced, such as anxiety, mood swings, changes in appetite, and depression 2. The prevalence of experiencing premenstrual symptoms is estimated to be 85-98%, while prevalence of individual symptoms varies widely between studies and populations 2. The social and economic burdens of premenstrual disorders are substantial. It is estimated that premenstrual syndromes result in an increase of $59 and $4333 in American women s individual direct and indirect healthcare costs, respectively 3. Furthermore, women with moderate or severe premenstrual symptoms work fewer days and have reduced productivity compared to those without symptoms 3-5. Although premenstrual symptoms and disorders are common, there are few treatments available for them and little research exists on the topic of dietary influences and risk factors 2. It is generally accepted that prevalence of premenstrual symptoms is influenced by subject characteristics such as BMI, physical activity, and age, however, the effect of race/ethnicity is inconsistent Premenstrual Symptoms Premenstrual Disorders Premenstrual disorders (PMDs) are a group of disorders sharing the commonality of regularly occurring premenstrual symptoms and include Premenstrual Syndrome (PMS) and

13 Premenstrual Dysphoric Disorder (PMDD). Diagnosis of PMDs requires that premenstrual symptoms are linked to the luteal phase. Symptoms must begin during the late luteal phase and resolve within a few days following the onset of menses, with a clear symptom free interval between cycles. The nature or number of the symptoms is not important, so long as the timing and cyclicity criteria are met. PMDs are diagnosed prospectively using symptom diaries where patients record their symptoms daily for at least 2 months. Differential diagnosis must also be excluded, as premenstrual symptoms must be differentiated from exacerbations of underlying disorders 1. PMS diagnosis criteria are defined by the American College of Obstetricians and Gynecologists (ACOG) 7. They require that 1 somatic and 1 affective symptom to be experienced at moderate or severe severity for at least 2 consecutive cycles recorded by prospective recording. Somatic symptoms include: bloating, breast tenderness, headache, joint or muscle pain, swelling of extremities, and weight gain. Affective symptoms include: angry outbursts, anxiety, confusion, depression, irritability, and social withdrawal. Symptoms must begin 5 days prior to menses, subside within 4 days following the onset of menses, and be followed by at least 12 symptom-free days. Symptoms must be recorded in the absence of any pharmacological or hormonal therapy, without use of drugs or alcohol, and be met with identifiable dysfunction in everyday activities such as social or work related activities 7. PMS has been estimated to occur in 20-32% of women 1. The most severe PMD is premenstrual dysphoric disorder (PMDD) which is defined in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM IV) 8. PMDD requires experiencing at least 5 premenstrual symptoms, with at least one affective symptom at moderate/severe severity. Similar to PMS criteria, symptoms must cause identifiable dysfunction in social or work activities, they must not be an exacerbation of other disorders, and 2

14 must be confirmed using prospective symptom recording for 2 consecutive cycles. Prevalence of PMDD has been found to be between 3-8% Etiology The etiology of premenstrual symptoms is not well understood, however, prevailing theories attribute the occurrence of premenstrual symptoms to women s individual responses to normal fluctuations in gonadal steroid production that occur during the reproductive cycle 9. Premenstrual symptoms are absent during non-ovulatory cycles, menopause, pregnancy, and following oophorectomy 10, when production of the corpus luteum does not occur thus eliminating rises in estrogen and progesterone in the luteal phase. Increased severity of symptoms is likely due to the sensitivity of some women to changes in hormone production, rather than to differences in hormone concentrations 11. Premenstrual symptoms are absent during pregnancy despite high levels of both estrogen and progesterone, and concentrations of these hormones have not been shown to differ between those with PMS compared to controls. However, inducing a chemical menopause using gonadotropin-releasing hormone (GnRH) agonist was shown to abolish symptoms in women with PMS 11. Following this GnRH agonist treatment with introduction of exogenous estrogen or progesterone caused symptoms to return in women with PMS but not in controls 11. This work strongly suggests the occurrence of premenstrual symptoms to be a result of abnormal responses to normal hormonal fluctuations during the luteal phase. Estrogen and progesterone, as well as their metabolites, are involved in the regulation of various physiological processes in the body, some of which have been theorized to be implicated in the pathophysiology of PMDs. Progesterone is metabolized in the brain and ovary to form neuroactive steroids 3-alpha-hydroxy-5-alpha-pregnane-20-one (ALLO) and 3-alpha-hydroxy- 5beta-pregnane-20-one (pregnanolone) 12. ALLO and pregnanolone act as positive allosteric 3

15 modulators of the GABA neurotransmitter system. GABA receptors are widely distributed in the central nervous system (CNS) and are important regulators of anxiety, alertness, stress, and vigilance 12. ALLO binds GABAA receptors and alters their sensitivity to neurosteroids, making them temporarily insensitive to GABA 12. Acute ALLO treatment has been shown to produce anxiolytic and antidepressant effects in the short term, however, long term exposure has been shown to increase anxiety 13. PMS women have been found to have reduced luteal phase ALLO concentrations compared to controls and some studies have shown an association with premenstrual mood symptoms 14, 15. The serotonergic system has also been implicated for its role in PMS. Premenstrual symptoms are very similar to those experienced with reduction of serotonin transmission, such as mood swings, anxiety, depression, irritability, carbohydrate cravings, and difficulty concentrating 16. Ovarian sex steroids are involved in the metabolism of serotonin, as well as its turnover, uptake, binding, and transport 16. Animal studies have demonstrated increased wholebrain serotonin as well as increased serotonin synthesis and decreased re-uptake in the midbrain, hypothalamus, and amygdala in rats following acute and chronic ethynyl estradiol administration Serotonergic function has been found to be altered during the luteal phase in PMS women compared to controls, with PMS women having reducing platelet uptake of serotonin and lower whole-blood serotonin levels 20, 21. The high efficacy of selective serotonin reuptake inhibitors (SSRIs) in the treatment of PMS and PMDD 22 further supports the involvement of the serotonergic system in PMDs. In addition to theories of neurotransmitter involvement in the etiology of PMDs, inflammation has also been linked to the occurrence of premenstrual symptoms. Multiple studies have demonstrated an association between increased inflammatory markers such as c-reactive protein (CRP), and pro-inflammatory cytokines including interleukin (IL)-2, IL-4, IL-10, IL-12, 4

16 and interferon-gamma in PMS women compared to controls The observed increase in CRP found in PMS women was reproduced in another cross-sectional study and associated with symptoms of anxiety and mood changes, muscle aches and cramps, increased appetite and bloating, and breast pain 26. Inflammation has a plausible role in the development of premenstrual symptoms as it is already involved in other aspects of reproductive function such as ovulation, endometrial repair, and follicular recruitment 27. Inflammatory markers, such as CRP, IL-6, IL- 1β, and tumor necrosis factor-α, fluctuate throughout the female reproductive cycle with rises in concentration following ovulation and peaking during menstruation 28. It has been suggested that some premenstrual symptoms may occur as a result of dysregulation in calcium homeostasis and secondary hyperparathyroidism 29. Serum calcium, parathyroid hormone (PTH), and 1,25-hydroxyvitamin D (1,25(OH)D) have been shown to fluctuate across the menstrual cycle 30, 31. Serum calcium levels drop at three stages of the menstrual cycle: during menses, midcycle, and during the late luteal phase 30. Fluctuations in these hormones may differ in women with PMDs, as suggested by two studies, one of which demonstrated that women with PMDD had significantly different fluctuation patterns from controls in 1,25(OH)D, ionized calcium, and urinary calcium 30. Similarly, menstrual cycle fluctuations in PTH, 25(OH)D, and 1,25(OH)D were found to differ in PMS women compared to controls 32. PTH showed midcycle elevations in PMS women but not controls, suggesting that they may be experiencing transient, secondary hyperparathyroidism 32. Higher calcium intake may be protective against these fluctuations, as multiple studies have reported decreased calcium intake in women with PMDs compared to controls 33, 34. Furthermore, clinical studies have demonstrated calcium supplementation reduces the severity of premenstrual symptoms There are similarities between hypocalcaemia symptoms and common premenstrual symptoms, such as those of anxiety, depression, fatigue, impaired 5

17 intellectual capacity, personality disturbances, and muscle cramps 29. Further supporting the hypothesized involvement of calcium in PMS etiology is the observed increased risk of osteoporosis after menopause in women with PMS 40, Treatment The current first line treatment for PMDs is the use of selective serotonin uptake inhibitors (SSRIs) 42. Studies have demonstrated a reduction in premenstrual symptoms with the use of SSRIs with a response rate of 60-90% 43. A 2013 Cochrane review analyzing 31 RCTs found that continuous or luteal-phase SSRIs are effective for premenstrual symptom relief compared to placebo 44. SSRIs have been shown to effectively reduce both premenstrual mood symptoms as well as somatic symptoms such as bloating, breast tenderness, and appetite changes 45. SSRI s are, however, are accompanied by many adverse side effects such as nausea, fatigue, and decreased libido 44 which may make them unsuitable treatment options for some women. PMDs may also be effectively treated by inhibiting ovarian cyclicity through the use of GnRH agonists or through bilateral oophorectomy 43, 46. GnRH agonists act by interrupting the normal pituitary-hypothalamus-gonadal cyclicity which triggers ovulation and premenstrual symptoms. GnRH agonists are considered quite effective in treating somatic and psychological premenstrual symptoms 46 but result in a medically-induced menopause which is accompanied by menopausal symptoms that must also be managed 9. Furthermore, to reduce the risk of cardiovascular disease and hypoestrogenic bone loss resulting from long-term GnRH agonist use, add-back therapy with estrogen and progesterone must often be added which risks reintroducing premenstrual symptoms 2. Surgical bilateral oophorectomy is also effective in abolishing premenstrual symptoms 47 but is considered too invasive of a procedure for a majority of 6

18 patients 9. Although SSRIs and GnRH agonists are effective in treating PMDs, they are either highly invasive or likely to cause severe adverse effects. Consequently, more research into potential therapies and the etiology of premenstrual symptoms is necessary to develop much needed novel therapeutic remedies. Other treatments may be considered when managing premenstrual symptoms such as lifestyle changes to reduce stress and increase exercise, calcium supplementation, hormonal contraceptives, anxiolytics, and herbal preparations 2. Although there is some evidence for the efficacy of these treatments, there is not enough evidence to conclude they are effective in treating premenstrual symptoms Risk Factors Several factors have been identified that put women at risk for experiencing premenstrual symptoms or disorders, and these include age, body mass index (BMI), and physical activity. Regular physical activity is considered a protective factor against premenstrual symptoms and has been inversely associated with the severity of premenstrual symptoms in several epidemiological studies 48, 49, as well as randomized control trials (RCT) For example, a recent RCT demonstrated the efficacy of engaging in regular aerobic exercise three times per week for reducing several individual premenstrual symptoms in young women 50. BMI has been similarly associated with premenstrual symptoms, where those with higher BMI were more likely to be experiencing symptoms at greater severity 53. This was especially pronounced in obese women with BMIs greater than 27.5 kg/m 254, 55. Lastly, increasing age may be a risk factor for premenstrual symptoms, with symptom prevalence peaking at age Ethnic background may also be a contributing risk factor to premenstrual symptomology, however this has not been consistently shown. A couple studies have shown differences in the 7

19 prevalence of premenstrual symptoms between Caucasian and African American women living in America, where Caucasians reported a lower prevalence of some symptoms 56, 57. Similarly, Asians have reported a lesser severity of premenstrual symptoms relative to Caucasians in a previous study 58. However, some cross-sectional studies have not observed any ethnic differences in the prevalence of premenstrual symptoms or PMS. The reason for these ethnic differences in prevalence of symptoms is not known. It is hypothesized that they may be due to underlying genetic or cultural differences 58, Hormonal Contraceptives Introduction Hormonal contraceptives (HC) are formulations of synthetic estrogen and progesterone derivatives which prevent ovulation. They were first introduced to the North American market in the 1960 s as combined oral contraceptives containing first-generation estrogen and progesterone analogues 60. Since that time, their formulations have evolved to include options containing second and third-generation analogues as well as various modes of administration. Rather than taken orally, HCs can now also be administered in skin patches, intra-muscular injections, implants, vaginal rings, or intra-uterine systems. There are numerous HC options available on the Canadian market, whose formulations differ in dose and type of estrogen or progestin used. Oral contraceptives (OC) are used by 16% of Canadian women and are the most common form of HC used in Canada 61. OCs typically contain both an estrogen and progestin (called a combined oral contraceptive (COC)) but may also be composed of only progestin (called a progestin-only pill). Contraceptives were first created to mimic the natural 28-day ovarian cycle by administering COCs for 21 days, followed by 7 days of a dose-free interval during which menstruation took place. However, COCs can also be taken in extended or continuous regimens which delay or eliminate menstruation, respectively. Their formulations may be monophasic, in which the 8

20 estrogen or progestin dose remains constant, or biphasic or triphasic, in which the estrogen or progestin are administered in two or three different doses throughout the cycle, respectively. Biphasic and triphasic regimens have formulations which periodically increase the dose of estrogen of progestin, and were created with the intention of reducing the amount of exogenous hormones administered Mechanisms of Action Although the forms of HCs are varied, they are united in their primary mechanism of action being their ability to inhibit ovulation by exerting an inhibitory effect at the hypothalamic and pituitary levels 62. COCs block the normal production of GnRH and may also act directly on the pituitary gland 63. This exerts an inhibitory effect on the production of follicle-stimulating hormone (FSH) and luteinizing hormone (LH) in the pituitary gland, especially on the midcycle surge of these hormones which would typically induce ovulation 64. These effects inhibit the production of new follicles during the follicular phase and ovulation 62. This prevents the development of the corpus luteum and thus the luteal phase rise in estrogen and progesterone. The progestin component of the COCs is especially effective at preventing this midcycle rise in LH. The estrogen component of the COC amplifies this effect in addition to preventing irregular shedding of the endometrium 62, 65. FSH and LH secretion returns immediately following the discontinuation of COCs. In fact, follicles begin to develop again during the 7-day dose-free interval 66. Other forms on HCs have similar mechanism of action to COCs whereby they suppress follicular development and inhibit ovulation. It is not necessary to administer both estrogen and progestin to suppress ovulation, as either on its own is sufficient to inhibit pre-ovulatory spikes in FSH and LH and thus prevent ovulation 62. The most commonly used progestin-only injectable contraceptive is depot medroxyprogesterone (DMPA). This long-acting progestin interrupts 9

21 ovulation by mechanisms similar to those of COCs, where it prevents the midcycle LH surge by acting at hypothalamic and pituitary levels 67. HCs also prevent pregnancy through secondary mechanisms such as thickening of the cervical mucus, affecting peristalsis and secretion within the fallopian tube, and by affecting the uterine lining making it unsuitable for implantation HCs also exert several other biologic effects in the body that are unrelated to their contraceptive action. HCs have been shown to impact bone mass 73, vitamin D status 74, the cardiovascular system 75-77, cognitive outcomes 78, inflammatory markers 79, as well as many other physiological processes. Their physiological effects can be dependant on the type of HC used. For example, while COCs have been associated with increased 1,25-hydroxyvitamin D concentrations due to the estrogenic effect on vitamin D metabolism, formulations containing medroxyprogesterone acetate did not exert the same effects on 1,25-hydroxyvitamin D concentrations. HC formulations using newer formulations also have a more net-estrogenic effect than early generation progestins since the new progestins have very little androgenic activity 80, 81, and also exert an anti-aldosterone effect 82. COCs have also been associated with increased risk of adverse cardiovascular outcomes, such as increased risk of venous thromboembolism (VTE) and altered lipid profiles 75, 83, 84. This effect is also dependent on the COC formulation used, as rates of VTE differ between formulations 75, 76, 85. There is also evidence that these cardiovascular outcomes may be dependent on the duration of use of hormonal contraceptives. Risk of VTE is highest in the first few months of use and then declines 75, 84. Similarly, total serum lipids are observed to increase with the start of HC use but return to baseline following 24 months of use HCs and Premenstrual Symptoms Hormonal contraceptive use may be effective in treating premenstrual symptoms by preventing ovulation and stabilizing fluctuations in estrogen and progesterone during the luteal 10

22 phase. Since premenstrual symptoms have been observed to be absent during non-ovulatory cycles 9, preventing ovulation may be an effective therapeutic strategy for premenstrual syndromes. This relationship, however, is complicated by the addition of exogenous estrogens and progestins with HCs as well as by the minor fluctuations in hormones that occur during the pill-free days 9. This is supported by evidence showing the comparatively improved efficacy of extended HC regimens in treating premenstrual symptoms 86. Due to their widespread use and the inconsistency of the available research it is important to determine whether HCs are useful in treating premenstrual symptoms and, if so, for which symptoms they are effective. The few placebo-controlled randomized trials that have been conducted on this topic have obtained mixed results. Some randomized control trials (RCTs) did not find HCs to be effective in treating premenstrual symptoms 87, 88, while others found them to be effective for some premenstrual symptoms but not others 89, 90. Numerous open-label studies found hormonal contraceptives to be effective in decreasing the severity of some premenstrual symptoms, but not all There is also evidence that more recent HC formulations may be more effective in treating premenstrual symptoms than older formulations, such as those containing drospirenone 94. Observational studies conducted on the topic and have also obtained conflicting results. Three large observational studies have found a substantially lesser prevalence of premenstrual symptoms in women using HCs compared to non-users 58, 95. An analysis of a health maintenance organization found an inverse relationship between hormonal contraceptive use and both the number and severity of emotional premenstrual symptoms (p<0.01), but not physical symptoms 58. Similarly, investigations of a large multiethnic US cohort of premenopausal women as well as a large French cohort showed significantly lower prevalence of PMS in HC users 96, 97. Conversely, a small nested case-control study within a US cohort found no difference in total 11

23 symptom prevalence between HC users and non-users 98. Furthermore, one study instigating premenstrual symptom associations with HC use among women with PMDD found an increase in the prevalence of some individual symptoms in HC users compared to non-users 99. Symptoms that were more prevalent among HC users included: anxiety, anger, avoided social activities, weight gain, joint/muscle pain, and difficulty concentrating 99. Novel studies assessing the effects of newer HC formulations on individual premenstrual symptoms may elucidate the conflicting findings. To date, no studies have investigated whether duration of HC use plays a role in their effect on premenstrual symptoms. The effects of HCs on risk of VTE, lipid profiles and cholesterol metabolism have been shown to be related to be dependent on duration of HC use 75, 83, 84, 100. The decline in VTE risk occurs after a few months 75, 84 and lipid profiles return to baseline after two years 84, although the mechanisms for this are unknown. Current trials investigating the effect of HCs on premenstrual symptoms have been conducted no longer than 6-8 months. It is possible that HCs may become more or less effective in the treatment of premenstrual symptoms with time. 1.4 Vitamin D Background Vitamin D status is determined by a combination of dietary Vitamin D consumption and cutaneous production 101. Since vitamin D can be synthesized in the skin with sun exposure and it is not necessary to obtain it in the diet, it is technically a prohormone rather than a vitamin 102. It is produced in the skin phytochemically from 7-dehydrocholesterol and has a structure similar to that of classic steroid hormones such as estradiol, aldosterone, and cortisol 102. Vitamin D is found in two forms- ergocalciferol (vitamin D2) and cholecalciferol (vitamin D3). Ergocalciferol and cholecalciferol are very similar in structure, the only difference being the existence of a 12

24 double bond at carbons and a methyl group on carbon 24 in ergocalciferol 103. Ergocalciferol is obtained from plant and fungal sources while cholecalciferol can be obtained from animal sources or endogenous production, and both are found in supplements 102. Although vitamin D may be obtained from the diet, most circulating vitamin D is derived from cutaneous production 104. Those living in Canada are at risk of vitamin D deficiency resulting from inadequate sun exposure due to its high latitude 101. According to the most recent Canadian Health Measures Survey (CHMS), a third of Canadians had vitamin D levels below the cut-off for sufficiency, and this number increased to about 40% during wintertime months 105. In response to the widespread vitamin D deficiency in Canada, laws now mandate the fortification of margarine, milk and plant-based beverages with vitamin D 101. Nevertheless, dietary vitamin D consumption accounts for about IU of daily vitamin D 104 which in Canada is mostly derived from fortified dairy products and fruit juices 101. Few foods naturally contain vitamin D, the largest sources are eggs, liver, and oily fish which can provide IU/serving Vitamin D Metabolism Vitamin D consumed in the diet is absorbed in the small intestine by enterocytes and subsequently packed into chylomicrons with other lipids for delivery to the liver 102, 107. While ergocalciferol and cholecalciferol can be obtained from foods and supplements, cholecalciferol can also be synthesized in the body dehydrocholesterol is embedded in the plasma membrane and is converted to previtamin D3 in response to ultraviolet (UV) B exposure. Since previtamin D3 is thermodynamically unstable, it is quickly converted to cholecalciferol via rearrangement of its double bonds 103. As a consequence of this reaction, cholecalciferol is ejected from the plasma membrane and diffuses into the dermal capillary. Here it becomes bound to its transport protein, vitamin D binding protein (DBP), which has a very strong affinity for cholecalciferol, and is transported to the liver 103. In the liver, cholecalciferol is hydroxylated by 13

25 25-hydroxylase (CYP27A1) to form the main circulating vitamin D metabolite, 25- hydroxyvitamin D (25(OH)D) (OH)D is thought to be biologically inert and its half-life is about 3 to 4 weeks 102, (OH)D can be converted into its active form of 1,25-dihydroxyvitamin D (1,25(OH)D) in the kidneys and other target tissues such as the brain 102. This conversion is done by the mitochondrial enzyme 1-hydroxylase (CYP27B1) 102, ,25(OH)D is a steroid hormone which regulates gene transcription and affects various signal transduction pathways through vitamin D receptor (VDR) binding 111. Its half-life is about 24 hours 112. Unlike 25(OH)D, the production of 1,25(OH)D is very tightly regulated, with 1,25(OH)D and high calcium levels decreasing its production and PTH and calcitonin increasing its production 113. Low calcium levels cause an increase in PTH production which leads to both and increase in 1,25(OH)D production as well as increased calcium absorption in the intestine 113, Functions of Vitamin D Vitamin D is a fat-soluble vitamin which behaves like a steroid hormone. 1,25(OH)D exerts a majority of its biological actions by binding to the vitamin D receptor (VDR) and regulating gene transcription 115. VDRs and CYP27B1 are expressed widely throughout the body, suggesting that vitamin D has widespread roles in many biologic systems such as the immune system, brain, bone, and skin 116. It plays an important in bone health and regulating calcium homeostasis. 1,25(OH)D stimulates calcium and phosphorous absorption in the intestine in response to low calcium levels ,25(OH)D stimulates the proliferation and differentiation of osteoblasts, as well as increases bone mineralization Vitamin D deficiency is associated with low bone mineral density and increased fractures 101. It is also associated with decreased falls 121, 122, which is attributed to improvements in muscle strength and lower-extremity function 14

26 with vitamin D supplementation 123. VDRs are located on fast-twitch muscle fibers 124 and vitamin D deficiency is associated with muscle weakness and myopathy 125, 126. In recent years, vitamin D research has extended to areas beyond the skeletal system which has given some insight into its widespread effects in the body. Vitamin D appears to exert beneficial effects on the cardiovascular system, where 25(OH)D has been inversely associated with risk of heart disease and hypertension in several large observational studies These cardiovascular effects of vitamin D are thought to be mediated by vasculoprotection, decreased inflammation, beneficial effects on calcium homeostasis, and suppression of the RAAS 131. Vitamin D has shown anti-inflammatory properties in vitro 132, 133 and has been associated with lower levels of inflammation in some , but not all 137, 138, observational studies. Lastly, vitamin D may be involved in disorders affecting the neurological system. 1,25(OH)D is a neurosteroid which can cross the blood-brain barrier and act on most areas of the brain through VDRs 117, and vitamin D deficiency has been associated with depression and cognitive impairment 139, Determinants of Vitamin D Status Vitamin D status can be influenced by several lifestyle, demographic, and biologic factors. The main factors known to influence 25(OH)D levels include UV exposure, body composition, dietary consumption, medications, ethnocultural status, and genetics. Since cholecalciferol in synthesized in the body in response to UVB exposure, any factor affecting the degree of UVB exposure can influence an individual s vitamin D status. This includes time spent outdoors, skin pigmentation, clothing, use of sunscreen, and environmental factors influencing the strength of UVB radiation 141. UVB radiation is diminished in environments with latitudes farther from the equator, and is also affected by weather, time of day, altitude, and season 141. Generally, vitamin D can not be synthesized by those living in regions above 33 north latitude 15

27 during wintertime months of November to March or between the hours of 3pm and 10am 142. Similarly, increased skin pigmentation can act as a natural sunscreen by absorbing UVB rays and prevent the synthesis of cholecalciferol 143. For this reason, ethnic groups with increased skin pigmentation produce less vitamin D in response to sun exposure and are at risk of vitamin D deficiency 144, 145. Dietary vitamin D intake can also contribute to circulating levels of 25(OH)D, although, their correlations range from r=0.2 to r=0.7 suggesting that other factors are contributing significantly to vitamin D status 101, 146. Other factors, such as body mass, genetics, and medication use, can also affect circulating levels of 25(OH)D by affecting its bioavailability, sequestration, and rate of breakdown. Obesity has been reported to be inversely associated with 25(OH)D in several cross-sectional studies and this is thought to be the result of sequestration of vitamin D by adipose tissue 149, 150. Similarly, some common medications are known to affect vitamin D metabolism, such as those containing sex-hormones 151, 152. HC use is positively associated with 25(OH)D levels and it is thought that this effect is mediated by the estrogen component 74, 152, 153 Estrogen has been shown in animal and in vitro studies to upregulate CYP27B1 and downregulate CYP24A1, enzymes responsible for the conversion of 25(OH)D to 1,25(OH)D and the catabolism of 25(OH)D as well as 1,25(OH)D, respectively 154, 155. Estrogen is also associated with elevated DBP concentrations and may upregulate VDR expression 153, 156. A recent cross-sectional study found HC users to have significantly higher 25(OH)D concentrations than non-users, who had similar levels to men 74. Finally, several common genetic variants along the vitamin D metabolic pathway have been associated with vitamin D status including those in CYP27B1 157, CYP24A1 158, and VDR 159. Heritability estimates for plasma 25(OH)D concentrations based on twin studies range between 45% and 75% , suggesting genetic variation may be an important factor in determining an individual s vitamin D status. 16

28 1.4.5 Measurement of 25(OH)D Vitamin D status is measured by serum concentrations of the main circulating vitamin D metabolite, 25(OH)D, which is the best indicator of functional vitamin D status as it takes into account contributions from dietary intake and cutaneous production 163. In comparison to 1,25(OH)D, 25(OH)D has a much longer half-life, is present in much higher concentrations, and its formation is not regulated, thus making it a better estimate of vitamin D status 112. Approximately 85% of 25(OH)D is bound to DBP, while 15% is bound to albumin and 0.03% is free 164. There are several methods available for measuring 25(OH)D including by radioimmunoassay (RIA), high-performance liquid chromatography (HPLC), and LC MS/MS 112. The strengths and limitations of each of these methods differ, but HPLC with UV detection and LC MS/MS may be considered the gold standard methods because of their abilities to differentiate between 25(OH)D2 and 25(OH)D3, their precision, and their accuracy (OH)D concentration cut points determining vitamin D status were suggested by the Institute of Medicine (IOM), and according to these criteria, deficient, inadequate, and sufficient vitamin D statuses correspond to plasma 25(OH)D concentrations of <30 nmol/l, nmol/l, and >50 nmol/l 163. These criteria were largely based on maintaining optimal bone health 163. The IOM criteria for determining vitamin D status have been controversial and widely criticized as overly conservative by not considering the important role of vitamin D in health outcomes beyond bone metabolism Many investigators as well as the Canadian Osteoporosis Society (COS) and the Endocrine Society (ES) have proposed that optimal vitamin D status should be considered at 25(OH)D plasma concentrations >75 nmol/l 101,

29 1.4.6 Vitamin D and Premenstrual Symptoms The evidence of calcium homeostasis dysregulation and luteal-phase calcium deficiency playing a role in the etiology of premenstrual symptoms has prompted the investigation of associations between vitamin D and premenstrual symptoms. It is proposed that vitamin D may be protective again premenstrual symptoms through its involvement in the regulation of calcium homeostasis 29. Plasma calcium and 1,25(OH)D concentrations have been shown to fluctuate during the menstrual cycle, with levels of calcium decreasing and 1,25(OH)D increasing during the luteal phase 30, (OH)D concentrations were not shown to fluctuate throughout the menstrual cycle 30. Alternative theories for the involvement of vitamin D in premenstrual symptoms include its role in reducing inflammation as well as its direct effect on the brain. 1,25(OH)D can cross the blood-brain-barrier and is capable of binding to VDRs located in the brain VDRs are distributed throughout areas of the brain known to be involved in mood and psychologic disorders, such as depression, which share common symptoms with PMS including depression, loss of appetite, and insomnia Furthermore, vitamin D plays a role in immune regulation and is associated with a reduction in inflammation 176, 177. Since premenstrual symptoms have been associated with elevated inflammatory markers 26, vitamin D may reduce symptoms through this pathway. Investigations into the role of dietary vitamin D have shown relatively consistent inverse associations between vitamin D intake and premenstrual symptoms. Analysis of the Nurse s Health Study II data has revealed an association between high dietary intake of vitamin D and a decreased risk of PMS 33. High vitamin D intakes have also been associated with decreased severity of premenstrual symptoms in the general population 178. Few studies have examined the association between vitamin D status and premenstrual symptoms, but those that have report 18

30 conflicting results. Most found no association between 25(OH)D and risk of premenstrual symptoms 34, , while one larger study found a negative association between 25(OH)D and symptoms of breast tenderness, fatigue, diarrhea and/or constipation, and depression 181. This negative association between 25(OH)D and premenstrual symptoms was observed in a large prospective cohort which examined the Nurses Health Study II data and evaluated premenstrual symptoms individually 181. Conversely, all other observational studies examined the association between plasma 25(OH)D and PMS, without assessing symptoms or symptom severities individually 34, Discrepancies in the analysis of individual and grouped symptoms may help explain these conflicting findings, as it has been previously shown that treatment response may be specific to the premenstrual symptom 182. One clinical trial has been conducted in which adolescent patients with severe vitamin D deficiency were supplemented with 25,000 IU biweekly vitamin D for four months and this was found to be effective in reducing their mean premenstrual symptom severity scores as well as the severity of all individual premenstrual symptoms studied, which included anxiety, irritability, crying easily, and sadness 183. One RCT has examined the effect of 200 mg daily vitamin D supplementation on premenstrual symptoms in an Iranian population. Their findings showed that after two months of supplementation symptom scores significantly decreased in the intervention group compared to placebo for all symptoms studied, which included depression, cravings, water retention, anxiety, and somatic changes Summary and Rationale Premenstrual symptoms are common in the North American population however their prevalence has not been determined previously in a Canadian population. Few suitable treatments are available for symptoms, particularly for those at mild or moderate symptom severity. Hormonal contraceptives may be effective in treating many common symptoms, but 19

31 this research is inconsistent. Furthermore, it is not clear for which symptoms and severities they are most effective. Little is also known about dietary associations with premenstrual symptoms. Women are advised to improve diet and lifestyle, but not enough evidence exists for specific claims. Vitamin D has been shown to be associated with the prevalence of premenstrual symptoms in some studies, but this has not been researched in relation to the severity of individual premenstrual symptoms. 1.6 Hypothesis and Objectives The objectives of this dissertation were to characterize the prevalence of common premenstrual symptoms in a Canadian population and to determine their associations with use of hormonal contraceptives and vitamin D status. It was hypothesized that HC use and plasma 25(OH)D concentrations are inversely associated premenstrual symptom prevalence and severity. Chapter-specific objectives are as follows: Objective 1: To determine the prevalence of premenstrual symptoms in a multiethnic population and to investigate their associations with use of hormonal contraceptives. Objective 2: To determine the associations between plasma 25-hydroxyvitamin D concentrations and the prevalence and severity of premenstrual symptoms. 20

32 Chapter 2 Prevalence of Premenstrual Symptoms and Associations with Use of Hormonal Contraceptives 21

33 2.1 Abstract Background: Hormonal contraceptive (HC) use may be associated with a reduction in some premenstrual symptoms, however, the evidence remains equivocal. Objective: To determine the prevalence of premenstrual symptoms in a multiethnic population of women and to investigate the association between hormonal contraceptive use and premenstrual symptoms. Methods: 1,048 women participating in the Toronto Nutrigenomics and Health Study provided data on their premenstrual symptoms and HC use. Severity of symptoms was classified as none, mild, moderate, or severe. Logistic regressions were used to calculate the relative risk (RR) and 95% confidence interval (CI) to determine the associations between HC use and duration of HC use with premenstrual symptoms, adjusting for ethnicity and other covariates. Results: Prevalence of individual symptoms varied, and the most commonly reported were cramps (75%), bloating (75%), mood swings (73%), increased appetite (64%), and acne (62%). Prevalence of cramps differed between ethnic groups (p<0.05). Use of HCs was associated with a lower RR (95% CI) of experiencing moderate/severe: cramps (0.82, ), clumsiness (0.22, ), confusion (0.22, ) and desire to be alone (0.45, ). HC use was not associated with the risk of premenstrual symptoms at mild severity. HC use was not associated with symptoms of anxiety, bloating, mood swings, increased appetite, acne, fatigue, sexual desire, depression, nausea, headache and insomnia. Premenstrual symptoms of acne, mood swings, bloating, increased appetite, headache, insomnia, nausea, clumsiness, and sexual desire were not associated with HC use. Each year of HC use was associated with a decreased RR of experience mild confusion (0.85, ), and insomnia (0.77, ), as well as moderate/severe fatigue (0.89, ), and mood swings (0.91, ), although these did 22

34 not meet Benjamini-Yekutieli criteria for multiple comparisons. Other premenstrual symptoms were not associated with duration of HC use. Conclusion: This study demonstrates that the prevalence of some premenstrual symptoms differs between ethnic groups and that HC use is associated with a lower risk of experiencing many, but not all premenstrual symptoms, only at moderate/severe severity. It also suggests that duration of HC use is not associated with the severity of premenstrual symptoms. 23

35 2.2 Introduction Premenstrual symptoms include a wide range of physical, psychological and behavioral symptoms, which occur in the late luteal phase of a woman s reproductive cycle and subside a few days following the onset of menses 1. Many symptoms have been described to date, and a few most commonly experienced somatic symptoms are bloating, headache, fatigue, and muscle cramps. Behavioural and psychological symptoms are also commonly experienced, such as anxiety, mood swings, changes in appetite, and depression 2, 185. It is estimated that more than 80% of women regularly experience premenstrual symptoms, however, prevalence varies between studies and populations It is generally accepted that the prevalence is influenced by factors such as body weight and age, however, the association with ethnicity has been inconsistent 58, 193. Little is known about the pathophysiology of premenstrual symptoms, and consequently, few effective therapies have been developed for them 1. Due to the timing of the symptoms, changes in plasma levels of progesterone and estradiol are thought to be involved in their etiology 1. Stabilizing fluctuations of these hormones during the luteal phase with the use of hormonal contraceptives may be effective in treating premenstrual symptoms 1, but the studies have been inconsistent 1, 2. Furthermore, the physiological effects of HCs have previously been shown to vary with their duration of use, such as in their effects on plasma lipid levels and VTE risk 75, 83, 84. This suggests that the effects of HCs on premenstrual symptoms may also differ with time, although this has not been previously investigated. Due to the large variations in frequencies of reported symptoms and their possible associations with ethnicity and hormonal contraceptive use, the objectives of this study were to 24

36 determine the prevalence of various premenstrual symptoms in a multiethnic Canadian population and to assess their associations with hormonal contraceptive use. 2.3 Methods Study Population Subjects included 1,636 men and women aged years who participated in the Toronto Nutrigenomics and Health (TNH) study, which is a cross-sectional examination of young adults investigating genetics, lifestyle, and biomarkers of health 194, 195. Recruitment occurred between 2004 and Participants completed a general health and lifestyle questionnaire (GHLQ), a physical activity questionnaire, and a food frequency questionnaire (FFQ). Overnight fasting blood samples were also collected for genotyping and biomarker analysis. Exclusion criteria included current pregnancy or breastfeeding. The study protocol was approved by the Ethics Review Board of the University of Toronto and participants provided written informed consent. From the initial 1,636 subjects, 520 men were excluded, 10 subjects were excluded due to missing GHLQ information, and 4 were excluded for lack of blood samples. The remaining 1,102 subjects were categorized into four ethnic groups based on self-reported ethnic status: Caucasian (n=514), East Asian (n=401), South Asian (n=105), or Other (n=82), as described previously 196. Caucasians included those self-reported as European, Middle Eastern, or Hispanic. East Asians consisted of Chinese, Japanese, Korean, Filipino, Vietnamese, Thai, and Cambodian. South Asians included Bangladeshi, Indian, Pakistani, and Sri Lankan. The Other category included self-reported ethnicities of Aboriginal Canadians, Afro-Caribbeans, and those who self-reported belonging to 2 ethnic groups not included in the same category. 25

37 2.3.2 Hormonal Contraceptive Use Hormonal contraceptive use was self-reported in the GHLQ. Subjects were categorized as HC users (n=320) and non-users (n=782). HC users included subjects indicating current use of HCs, regardless of HC type or delivery method (transdermal, oral, vaginal, injection, etc.). HC users also indicated how long they have been using HCs. Subjects also reported use of any medications in the past month. Use of anti-depressants, analgesics, or anxiolytics was considered in the present study as PMS medication use, due to their effects on premenstrual symptoms Anthropometrics and Physical Activity Subjects height and weight were measured using the protocol previously described by Garcia-Bailo et al. (2012) 197. Subjects wore light clothing and removed their shoes during the measurements. Body mass index (BMI) was subsequently calculated in kg/m 2. Subjects selfreported their physical activity in the GHLQ by estimating the amount of time they spent sleeping and engaging in light, moderate, and vigorous activity. Values were then converted into metabolic equivalent (MET) levels Premenstrual Symptoms Premenstrual symptoms and severities were self-reported in a questionnaire included in the GHLQ. The questionnaire included the following symptoms: cramps/skin blemish; bloating/swelling/breast tenderness; mood swings/crying easily/irritability/angry outbursts; increased appetite/food cravings; acne; sexual desire/activity change; fatigue; anxiety/tension/nervousness; depression; desire to be alone; confusion/difficulty concentrating/forgetfulness; nausea; insomnia; headache; and clumsiness. Symptom severities were classified as none, mild, moderate, or severe. Subjects indicated the severity at which they experienced each symptom, within the 5 days before the onset of their period and ending by the 26

38 4th day of their period. Subjects could also list other premenstrual symptoms experienced, however, due to the scarcity of other symptoms they were not included in the analyses Plasma Samples and Vitamin D Measurement Participants provided blood samples following a minimum 12-hour overnight fast. Participants experiencing a temporary inflammatory condition (including a recent piercing or tattoo, acupuncture, a medical or dental procedure, a vaccination or immunization, flu, an infection, or a fever) underwent a two-week recovery period prior to providing blood samples. Samples were collected at LifeLabs Medical Laboratory Services (Toronto, Ontario, Canada), and 25(OH)D levels were measured at the University Health Network Specialty Lab at Toronto General Hospital (Toronto, Ont., Canada). Plasma 25(OH)D was measured by high-performance liquid chromatography tandem mass spectrometry Statistical Analysis All statistical analyses were conducted using SAS (version 9.4; SAS Institute Inc, Cary, NC, USA). The α was set at 0.05 and all reported p-values are 2-sided. Subject characteristics were compared between HC users and non-users by chi-square analysis for categorical variables and t-tests for continuous variables. Distribution of continuous variables was assessed prior to analysis and log-transformed BMI was used in all subsequent analyses. Crude mean BMI values were reported for ease of interpretation. The prevalence of premenstrual symptoms was defined as the frequency of subjects experiencing the symptoms at any severity (mild, moderate, or severe). Prevalence was calculated for each symptom in the total population, and separately for the major ethnic groups: Caucasians (n=514), East Asians (n=401), South Asians (n=104), and Other (n=82). Logistic regressions were used to determine differences in the prevalence of symptoms between the four 27

39 ethnic groups. P-values were calculated in both unadjusted models as well as adjusted models which included the following covariates: age, BMI, HC use, physical activity, PMS medication use and plasma 25(OH)D concentrations. Benjamini-Yekutieli adjustments for multiple comparisons were applied (15 tests, α = 0.05: p<0.015). Differences in the prevalence of each symptom between each pair of ethnic groups were also examined (Caucasians vs East Asians; Caucasians vs South Asians; Caucasians vs Other; East Asians vs South Asians; East Asians vs Other; South Asians vs Other) using logistic regressions. Logistic regressions were used to examine the associations between HC use and premenstrual symptom severities. The proc genmod procedure was conducted with a binomial distribution and a log link function. Moderate and severe symptom severities were combined due to the small number of subjects reporting severe symptoms. Relative risks (RR) and 95% confidence intervals (CI) were reported for associations between HC use and premenstrual symptoms. Univariate models were first used in Model 1, followed by multivariate models in Model 2 which adjusted for ethnicity, BMI, physical activity, PMS medication use and age. Covariates were selected based on their associations with HC use or premenstrual symptoms in the TNH study population and previous studies. Benjamini-Yekutieli adjustments for multiple comparisons were applied (30 tests, α = 0.05: p<0.013). 2.4 Results Study Population Subject characteristics are reported in Table 2-1 Subject Characteristics Stratified by Hormonal Contraceptive (HC) Use1,2. The mean age of participants was 22.6 years. HC users were on average older (23 years) than non-users (22.4 years) (p=0.0006). HC use differed between ethnic groups (p<0.0001), with Caucasian women reporting the greatest use of HCs 28

40 (43%), followed by Other (34%), South Asians (17%) and East Asians (13.0%). Reported physical activity was greater for HC users (8.1 met-hours/week) than non-users (7.5 methours/week) (p=0.003). Log-transformed BMI did not differ between HC users and non-users. The distribution of HC types in the study population is reported in Table 2-2 Subject Characteristics Stratified by Hormonal Contraceptive (HC) Use1,2. The most commonly used HC brand was Tri-Cyclen (25%) followed by Alesse (22%). Yasmin, Diane35, and Marvelon were each used by 9% of participants. Evra was used by 3% of participants, while Demulen, Triphasel, and Triquillar were used by 1% of participants. All the aforementioned HC s are combinations of ethinyl estradiol and various progestins in varying doses which are administered orally. Depo-provera, which was used by 1% of participants, is an injectable progestin-only HC containing 150 mg medroxyprogesterone. 29

41 Table 2-1 Subject Characteristics Stratified by Hormonal Contraceptive (HC) Use 1,2 HC Non-Users (%) HC Users (%) p-value N Age (years) ± ± Ethnicity (%) < Caucasian 293 (57) 221 (43) East Asian 348 (87) 53 (13) South Asian 87 (83) 18 (17) Other 54 (66) 28 (34) Body mass index (kg/m 2 )* 22.4± ± Physical activity (met-h/wk) 7.5± ± Medication Use 205 (26) 68 (21) 0.08 Physical activity (met-h/wk) 7.5± ± Shown are crude means and standard errors of continuous variables, and n (%) of categorical variables 2 P-values were obtained using chi-square tests for categorical variables and t-tests for continuous variables. * Indicates log-transformed variable was used for obtaining p-value 30

42 Table 2-2 Subject Characteristics Stratified by Hormonal Contraceptive (HC) Use 1,2 HC Brand N (%) Estrogen (mg) Progestin (mg) Tri-Cyclen 1 71 (25) Ethinyl Estradiol (0.035) Norgestimate (0.18,0.215, 0.25) Alesse 63 (22) Ethinyl Estradiol (0.02) Levonorgestrel (0.1) Other 3 41 (14) - - Yasmin 27 (9) Diane (9) Marvelon 25 (9) Cyclen 13 (5) Evra 9 (3) Demulen 4 (1) Triphasel 1 4 (1) Ethinyl Estradiol (0.03) Ethinyl Estradiol (0.035) Ethinyl Estradiol (0.03) Ethinyl Estradiol (0.035) Ethinyl Estradiol (0.02) Ethinyl Estradiol (0.03) Ethinyl Estradiol (0.03, 0.04, 0.03) Drospirenone (3) Cyproterone Acetate (2) Desogestrel (0.15) Norgestimate (0.25) Norgestimate (0.15) Ethynodiol diacetate (2) Levonorgestrel (0.05, 0.075, 0.125) Depo-provera 2 2 (1) - Medroxyprogesterone (150) Triquillar 1 2 (1) Ethinyl Estradiol (0.03, 0.04, 0.03) Levonorgestrel (0.05, 0.075, 0.125) 1 Indicates triphasic HCs, with dose of hormone reported in the order of administration 2 HC administered in the form of an intra-muscular injection every 3 months 3 Category composed of various HC formulations each used by <1% of subjects * 3% of subjects did not report type of HC used 31

43 2.4.2 Prevalence of Premenstrual Symptoms Prevalence of experiencing any premenstrual symptoms in the total population was 99%, and did not differ significantly between ethnic groups (p=0.11). Prevalence of each premenstrual symptom in the total population as well as stratified by ethnicity is shown in Table 2-3 Premenstrual Symptom Prevalence by Ethnicity. The most common symptoms experienced were cramps (75%), bloating (75%), mood swings (73%), increased appetite (64%), and acne (62%). Other premenstrual symptoms experienced were fatigue (55%), sexual desire (50%), anxiety (37%), desire to be alone (33%), depression (29%), headache (27%), confusion (21%), clumsiness (15%), nausea (15%), and insomnia (11%). In the unadjusted model symptom prevalence differed between ethnic groups for symptoms of cramps, bloating, sexual desire, headache, and confusion (p<0.05). However, after adjustments for age, BMI, HC use, physical activity, medication use, and plasma 25(OH)D concentrations, only the prevalence of cramps differed between ethnic groups (p<0.05). This association met adjustments for multiple comparisons (p<0.015), where East Asians reported a lower prevalence of cramps than Caucasians and South Asians. Prevalence of bloating, mood swings, increased appetite, acne, fatigue, sexual desire, anxiety, desire to be alone, depression, headache, confusion, clumsiness, nausea, and insomnia did not differ between ethnic groups in the adjusted model. 32

44 Table 2-3 Premenstrual Symptom Prevalence by Ethnicity Symptom Total (%) N=1048 Caucasian (%) N=46 East Asian (%) N=395 South Asian (%) N=100 Other (%) N=78 Model 1 p-value Model 2 p-value Cramps a 67 b 84 a 78 ab < < Bloating/Swelling/Breast Tenderness Mood Swings/Irritability Increased Appetite/Food Cravings Acne Fatigue Sexual Desire/Activity Change Anxiety/Tension/Nervousness Desire to be alone Depression Headache Confusion/Difficulty Concentrating/Forgetfulness Clumsiness Nausea Insomnia

45 Sorted by total premenstrual symptom prevalence. Letters indicate prevalence values which differed significantly from each other in the adjusted model (p<0.05). 34

46 2.4.3 Premenstrual Symptom Associations with HC use Associations between premenstrual symptoms and HC use is shown in Table 2-4 Associations between HC Use and Premenstrual Symptom Severity and displayed graphically in Figure 2-1 Associations between HC Use and Mild Premenstrual Symptoms and Figure 2-2 Associations between HC Use and Moderate/Severe Premenstrual Symptoms. In the unadjusted model, HC use was associated with a lower risk of experiencing mild acne and the following symptoms at moderate/severe severity: cramps, fatigue, anxiety, clumsiness, confusion, nausea, depression, and desire to be alone. In Model 2, which adjusted for ethnicity, BMI, physical activity, PMS medication use and age, HC use was not associated with any symptoms at mild severity. HC use was associated with a lower RR (95% CI) of experiencing moderate/severe: cramps (0.81, ), anxiety (0.62, ), clumsiness (0.13, ), confusion (0.22, ), depression (0.55, ), and desire to be alone (0.45, ). All symptoms with the exception of depression met Benjamini-Yekutieli criteria for multiple comparisons (30 tests, α = 0.05: p<0.013). Premenstrual symptoms of bloating, mood swings, increased appetite, acne, fatigue, sexual desire, headache, and insomnia were not associated with HC use in Model 2. Low sample size precluded calculations of adjusted relative risks for moderate/severe nausea, where unadjusted RRs were: 0.53 (0.25, 1.13). Associations between duration of HC use and premenstrual symptoms severities are reported in Table 2-5 Associations between Duration of HC Use and Premenstrual Symptoms and displayed graphically in Figure 2-3 Associations between Duration of HC Use and Mild Premenstrual Symptomsand Figure 2-4 Associations between Duration of HC Use and Moderate/Severe Premenstrual Symptoms. In Model 1, duration of HC use was associated with a decreased risk of experiencing mild insomnia and cramps. In Model 2, after adjustments for 35

47 ethnicity, BMI, physical activity, PMS medication use, and age, each year of HC use was associated with a lower risk of mild cramps (0.94, ), confusion (0.89, ) and insomnia (0.79, ), as well as moderate/severe fatigue (0.89, ). None of these associations met Benjamini-Yekutieli criteria for multiple comparisons. Low sample size precluded calculations of adjusted RRs for moderate/severe symptoms of confusion, insomnia, and nausea, where unadjusted RRs were: 0.63 (0.33,1.18), 0.53 (0.16,1.82), and 0.69 (0.44,1.08), respectively. 36

48 Table 2-4 Associations between HC Use and Premenstrual Symptom Severity Symptom Severity HC Non-Users (%) N = 782 Acne / Skin Blemish HC Users (%) N = 320 Model 1 Relative Risk Model 1 p-value Model 2 Relative Risk None 310 (40) 111 (35) REF REF Model 2 p-value Mild 318 (41) 161 (50) 1.17 (1.03,1.33) (0.98,1.28) 0.10 Moderate/Severe 154 (20) 48 (15) 0.91 (0.69,1.19) (0.63,1.12) 0.23 Bloating / Swelling / Breast Tenderness None 207 (26) 71 (22) REF REF Mild 309 (40) 143 (45) 1.12 (0.99,1.26) (0.95,1.22) 0.27 Moderate/Severe 266 (34) 106 (33) 1.06 (0.92,1.23) (0.86,1.16) 0.99 Cramps None 185 (24) 89 (28) REF REF Mild 256 (33) 128 (40) 1.02 (0.89,1.16) (0.81,1.07) 0.29 Moderate/Severe 341 (44) 103 (32) 0.83 (0.72,0.96) (0.71,0.93) Mood Swings / Crying Easily / Irritability /Angry Outbursts None 212 (27) 92 (29) REF REF Mild 287 (37) 128 (40) 1.01 (0.88,1.16) (0.88,1.18) 0.80 Moderate/Severe 283 (36) 100 (31) 0.91 (0.78,1.06) (0.76,1.06) 0.19 Increased Appetite / Food Cravings None 282 (36) 112 (35) REF REF Mild 232 (30) 107 (33) 1.08 (0.91,1.28) (0.89,1.27) 0.49 Moderate/Severe 268 (34) 101 (32) 0.97 (0.82,1.15) (0.82,1.16) 0.76 Fatigue None 342 (44) 153 (48) REF REF Mild 245 (31) 107 (33) 0.99 (0.83,1.17) (0.83,1.20)

49 Symptom Severity HC Non-Users (%) N = 782 HC Users (%) N = 320 Model 1 Relative Risk Model 1 p-value Model 2 Relative Risk Model 2 p-value Moderate/Severe 195 (25) 60 (19) 0.78 (0.61,0.99) (0.64,1.05) 0.11 Headache None 575 (74) 231 (72) REF REF Mild 131 (17) 58 (18) 1.08 (0.82,1.43) (0.84,1.49) 0.46 Moderate/Severe 76 (10) 31 (10) 1.01 (0.68,1.50) (0.67,1.51) 0.97 Anxiety / Tension / Nervousness None 480 (61) 220 (69) REF REF Mild 201 (26) 73 (23) 0.84 (0.67,1.06) (0.69,1.13) 0.31 Moderate/Severe 101 (13) 27 (8) 0.63 (0.42,0.94) (0.42,0.95) 0.03 Clumsiness None 655 (84) 278 (87) REF REF Mild 90 (12) 39 (12) 1.02 (0.72,1.45) (0.74,1.56) 0.71 Moderate/Severe 37 (5) 3 (1) 0.20 (0.06,0.64) (0.07,0.73) 0.01 Confusion / Difficulty Concentrating / Forgetfulness None 599 (77) 267 (83) REF REF Mild 121 (15) 48 (15) 0.91 (0.67,1.23) (0.72,1.38) 1.00 Moderate/Severe 62 (8) 5 (2) 0.20 (0.08,0.48) (0.09,0.54) Sexual Desire / Activity Change None 407 (52) 147 (46) REF REF Mild 233 (30) 112 (35) 1.19 (1.00,1.41) (0.95,1.37) 0.16 Moderate/Severe 142 (18) 61 (19) 1.13 (0.88,1.46) (0.74,1.23) 0.74 Insomnia None 688 (88) 293 (92) REF REF Mild 75 (10) 25 (8) 0.80 (0.52,1.23) (0.57,1.43)

50 Symptom Severity HC Non-Users (%) N = 782 HC Users (%) N = 320 Model 1 Relative Risk Model 1 p-value Model 2 Relative Risk Model 2 p-value Moderate/Severe 19 (2) 2 (1) 0.25 (0.06,1.08) (0.05,0.99) 0.05 Nausea* None 664 (85) 277 (87) REF REF Mild 81 (10) 35 (10) 1.03 (0.71,1.50) (0.64,1.41) 0.80 Moderate/Severe 37 (5) 8 (3) 0.53 (0.25,1.13) 0.10 N/A N/A Depression None 543 (69) 236 (74) REF REF Mild 150 (19) 65 (20) 1.00 (0.77,1.29) (0.73,1.26) 0.77 Moderate/Severe 89 (11) 19 (6) 0.53 (0.33,0.85) (0.34,0.90) 0.02 Desire to be alone None 499 (64) 238 (75) REF REF Mild 180 (23) 62 (19) 0.78 (0.60,1.01) (0.62,1.04) 0.10 Moderate/Severe 103 (13) 19 (6) 0.43 (0.27,0.69) (0.28,0.73) Model 1 contains unadjusted relative risks and p-values Model 2 contains relative risks and p-values adjusted for ethnicity, log-transformed BMI, physical activity, age, and medication use * indicates no values obtained in Model 2 due to low sample size 39

51 Figure 2-1 Associations between HC Use and Mild Premenstrual Symptoms Contains relative risks and confidence intervals of experiencing each mild premenstrual symptom in HC users, adjusted for ethnicity, log-transformed BMI, physical activity, age, and medication use 40

52 Figure 2-2 Associations between HC Use and Moderate/Severe Premenstrual Symptoms Contains relative risks and confidence intervals of experiencing each moderate/severe premenstrual symptom in HC users, adjusted for ethnicity, log-transformed BMI, physical activity, age, and medication use 41

53 Table 2-5 Associations between Duration of HC Use and Premenstrual Symptoms Symptom Severity Mean Duration of Use (Years) 1 Acne / Skin Blemish Model 1 Per Year RR Model 1 p-value Model 2 Per Year RR Model 2 p-value N = 1,102 None 2.9 ± 0.2 REF REF Mild 3.3 ± (0.99,1.06) (0.98,1.06) 0.36 Moderate/Severe 2.6 ± (0.85,1.07) (0.87,1.10) 0.70 Bloating / Swelling / Breast Tenderness Cramps Mood Swings / Crying Easily / Irritability /Angry Outbursts Increased Appetite / Food Cravings Fatigue None 3.0 ± 0.3 REF REF Mild 3.4 ± (0.98,1.05) (0.96,1.03) 0.91 Moderate/Severe 2.7 ± (0.96,1.03) (0.94,1.01) 0.22 None 3.6 ± 0.3 REF REF Mild 3.1 ± (0.88,0.99) (0.88,0.99) 0.03 Moderate/Severe 2.6 ± (0.93,1.01) (0.83,1.03) 0.45 None 3.4 ± 0.3 REF REF Mild 3.1 ± (0.94,1.03) (0.91,1.01) 0.11 Moderate/Severe 2.7 ± (0.93,1.01) (0.88,1.01) 0.09 None 3.1 ± 0.2 REF REF Mild 3.3 ± (0.97,1.06) (0.95,1.06) 0.82 Moderate/Severe 2.8 ± (0.92,1.03) (0.92,1.05) 0.62 None 3.4 ± 0.2 REF REF Mild 2.9 ± (0.90,1.02) (0.88,1.01) 0.08 Moderate/Severe 2.7 ± (0.85,1.02) (0.80,0.98)

54 Symptom Severity Mean Duration of Use (Years) 1 N = 1,102 Model 1 Per Year RR Model 1 p-value Model 2 Per Year RR Model 2 p-value Headache Anxiety / Tension / Nervousness Clumsiness Confusion / Difficulty Concentrating / Forgetfulness* Sexual Desire / Activity Change Insomnia* None 3.1 ± 0.2 REF REF Mild 2.8 ± (0.90,1.08) (0.89,1.09) 0.80 Moderate/Severe 3.0 ± (0.83,1.09) (0.80,1.06) 0.23 None 3.2 ± 0.2 REF REF Mild 2.8 ± (0.88,1.03) (0.86,1.02) 0.13 Moderate/Severe 3.0 ± (0.85,1.11) (0.84,1.14) 0.82 None 3.1 ± 0.2 REF REF Mild 2.7 ± (0.84,1.06) (0.82,1.06) 0.27 Moderate/Severe 3.5 ± (0.74,1.47) (0.66,1.33) 0.73 None 3.2 ± 0.2 REF REF Mild 2.6 ± (0.82,1.03) (0.79,1.00) 0.05 Moderate/Severe 1.4 ± (0.33,1.18) 0.15 N/A N/A None 3.3 ± 0.2 REF REF Mild 3.0 ± (0.93,1.03) (0.90,1.01) 0.13 Moderate/Severe 2.8 ± (0.87,1.04) (0.85,1.04) 0.22 None 3.2 ± 0.2 REF REF Mild 1.9 ± (0.64,0.96) (0.64,0.98) 0.03 Moderate/Severe 1.1 ± (0.16,1.82) 0.31 N/A N/A 43

55 Nausea* Symptom Severity Mean Duration of Use (Years) 1 Model 1 Per Year RR Model 1 p-value Model 2 Per Year RR Model 2 p-value N = 1,102 None 3.1 ± 0.2 REF REF Mild 3.3 ± (0.91,1.14) (0.91,1.15) 0.72 Moderate/Severe 1.5 ± (0.44,1.08) 0.11 N/A N/A Depression None 3.2 ± 0.2 REF REF Mild 3.0 ± (0.91,1.06) (0.90,1.06) 0.55 Moderate/Severe 2.6 ± (0.77,1.11) (0.77,1.12) 0.42 Desire to be alone None 3.2 ± 0.2 REF REF Mild 2.9 ± (0.88,1.05) (0.86,1.04) 0.28 Moderate/Severe 2.5 ± (0.75,1.09) (0.70,1.09) Shown are crude means ± standard errors Model 1 contains unadjusted per-year relative risks and p-values Model 2 contains adjusted per-year relative risks and p-values. Adjusted for ethnicity, log-transformed BMI, physical activity, age, and medication use * indicates no values obtained in Model 2 due to low sample size 44

56 Figure 2-3 Associations between Duration of HC Use and Mild Premenstrual Symptoms Contains per-year relative risks and confidence intervals of experiencing each mild premenstrual symptom with each year of HC use, adjusted for ethnicity, logtransformed BMI, physical activity, age, and medication use 45

57 Figure 2-4 Associations between Duration of HC Use and Moderate/Severe Premenstrual Symptoms Contains per-year relative risks and confidence intervals of experiencing each moderate/severe premenstrual symptom with each year of HC use, adjusted for ethnicity, log-transformed BMI, physical activity, age, and medication use 46

58 2.5 Discussion In this study, we investigated the prevalence of 15 common premenstrual symptoms and their associations with hormonal contraceptive use in a multiethnic population of young adults living in Canada. Our findings show that the prevalence of individual premenstrual symptoms varies widely between the symptoms, and we observed ethnic differences in the prevalence of several symptoms. We also found that HC use was associated with a lower risk of experiencing several, but not all, premenstrual symptoms at moderate/severe severity. No associations were observed between HC use and the risk of experiencing mild premenstrual symptoms. Duration of HC use was also inversely associated with experiencing some, but not all, premenstrual symptoms. In our population 99% of the subjects reported experiencing premenstrual symptoms. The same prevalence estimates were found in female university students in Thailand and Iran 190, 191. Prevalence reported in other studies have been slightly lower and have ranged from 80% to 95% , 198. These variations in prevalence estimates may be explained by differences in symptom assessment, subject population, and subject characteristics such as age 199. For example, the lowest prevalence of 80% was reported in a German community survey which included adolescent subjects aged years 188. The inclusion of adolescents could explain the lower prevalence, as was shown in a previous study which found that subjects under 20 or over 45 years of age had the lowest symptom prevalence, with prevalence peaking at age Alternatively, a survey of only married Iranian women from health clinics aged reported a prevalence of 86% 198. Two previous studies that included women of similar age as in the present study reported similar prevalence for the various premenstrual symptoms 190,

59 The most commonly experienced symptoms in the present study were cramps (75%), bloating (75%), irritability (73%), increased appetite (64%), and acne (62%). These differed from those reported in other studies, and as expected, investigations into the nature of the most commonly experienced symptoms have yielded varying results depending on the population studied 198, 200, 201. In a recent survey of Iranian women, the most common symptoms reported were tiredness (70%), backache (68%), headache (59%), and insomnia (50%) 198. The most common premenstrual symptoms reported in a population of Turkish medical students were bloating (90%), irritability (88%), breast tenderness (83%), and anxiety (74%) 200. However, a study involving a Mexican population demonstrated abdominal cramping to be the most prevalent symptom (54%), while only 8% of women reported irritability 201. Discrepancies in the prevalence of symptoms may be explained by several factors including variations in premenstrual symptom questionnaires, BMI, age, cultural factors, and environmental exposures. The questionnaire used in the present study differed from those used by others 198, 200, 201, which could account for some of the variation in symptom reporting. For example, the questionnaire used by Goker et al did not include symptoms of acne, appetite changes, or cramps which were among the five most commonly experienced symptoms in the present population 200. The effect of ethnicity in relation to premenstrual symptoms remains controversial. Sternfeld et al. showed that relative to Whites, Hispanics reported a greater severity of premenstrual symptoms whereas Asians reported a lesser severity 58. Several studies involving US populations have shown significant differences in symptom prevalence between White and Black women, but these racial differences were not present for all symptoms 56, 57, 202. This is in line with the results of the present study which revealed ethnic differences in the prevalence of some, but not all, symptoms and no ethnic differences in the total prevalence. In the present study, many symptoms were observed to differ by ethnicity in our unadjusted models but after 48

60 adjustments for potential confounding variables these differences were no longer significant. Following adjustments, ethnic differences in prevalence were observed only for cramps, which remained significant after adjustments for multiple comparisons. East Asian participants reported a lower prevalence of cramps compared to all other ethnic groups. Although this may reflect differences in genetics or cultural factors that may put East Asians at lesser risk of some premenstrual symptoms, it could also be explained by cultural differences in the interpretation and reporting of symptoms 59. Ethnic differences in premenstrual symptom reporting have been previously observed and it was suggested that differences in the clustering of symptoms in Chinese women compared to Western women may be a result of differences in the conceptualization of the integration of organ systems and their relation to health and disease influenced by Traditional Chinese Medicine 59. Nonetheless, these findings may guide researchers and healthcare practitioners in determining high-risk populations for premenstrual symptoms, and should be supported by future large-scale studies on Canadian populations. In the present study, hormonal contraceptive use was associated with a lower risk of experiencing moderate/severe cramps, depression, desire to be alone, confusion, and anxiety. Use of hormonal contraceptives was not associated with mild premenstrual symptoms. These findings are in agreement with three previous studies that found a decrease in the overall prevalence of symptoms as well as a decrease in the number and severity of emotional symptoms in women using oral contraceptives 58, 95, 97. Two studies found no association between HC use and premenstrual symptoms 98, 192. One study sampling 400 Iranian women observed a greater prevalence of several premenstrual symptoms in HC users versus non-users 198. These studies, however, assessed the effects of HC use on grouped symptom prevalence and severity, while the present study identified specific premenstrual symptoms and severities which are associated with HC use. Grouping of symptoms likely accounted for these discrepancies in findings of 49

61 associations between HC use and premenstrual symptoms. As shown in the present study, not all symptoms are associated with HC use and including their prevalence likely impacted previous findings. The present findings emphasize the importance of examining individual premenstrual symptoms in research investigating the efficacy of treatments for PMDs. Furthermore, future research into PMD treatment considering individual premenstrual symptom and severity will help guide clinicians in making individualized treatment decisions for patients. Duration of HC use was associated with a lower risk of moderate/severe fatigue and mild cramps, confusion, and insomnia, but not after adjustments for multiple comparisons. This suggests that our findings may have been due to chance and do not represent a true effect of duration of use. Results from the present study should be confirmed by future analysis, as to our knowledge, this is the first study that has examined the relationship between duration of HC use and premenstrual symptoms. Other health outcomes have previously been associated with duration of HC use, such as risk of VTE, lipid profiles, and cognitive outcomes 75, 78, 83. The mechanisms of these effects are not known and may be outcome-specific. The associations observed between duration of HC use and some premenstrual symptoms in the present study do not support any clinical recommendations as our findings did not meet the threshold for multiple comparisons and should be confirmed in future studies. The observed improvement of premenstrual symptoms with HC use has largely been attributed to stabilizing ovarian sex steroid fluctuations during the reproductive cycle 43. Treatments preventing ovulation, such as long-acting GnRH agonists and bilateral oophorectomy, have been highly effective in diminishing premenstrual symptoms 43. HCs may present a more favorable option for the management of premenstrual symptoms as they are accompanied by far fewer and less severe side effects 1. Some HCs also possess anti-aldosterone and anti-androgenic properties that likely enhance their effects on premenstrual symptoms 1,

62 There is some evidence that the effect of HC use on premenstrual symptoms is dependent on the HC formulation and regimen 86, 204. In the present study, sample size limitations precluded the ability to study the effects of different HC formulations on premenstrual symptoms. Interestingly, HC use is associated with a higher concentration of pro-inflammatory proteins 79 which have also been linked to an increase in the severity of some premenstrual symptoms 23, 26. This cross-sectional examination of a young multiethnic population of Canadian women found that 99% of women experienced some type of premenstrual symptom and that prevalence of individual symptoms differed across ethnic groups. The present study identifies the most common premenstrual symptoms in a Canadian population, and reports those symptoms that are more common in some ethnic groups than in others. The findings also show that HC use was associated with a lower risk of experiencing several premenstrual symptoms. 51

63 Chapter 3 Association between Plasma 25-Hydroxyvitamin D and Premenstrual Symptoms 52

64 3.1 Abstract Background: Premenstrual symptoms are experienced by up to 95% of women and little is known about dietary risk factors. Previous studies suggest that 25(OH)D may be inversely associated with the severity of premenstrual symptoms, but the findings have been inconclusive. Objective: The objective of this study was to determine whether plasma 25(OH)D is associated with premenstrual symptoms. Methods: 1,051 women aged years participating in the cross-sectional Toronto Nutrigenomics and Health Study provided data on their premenstrual symptoms and fasting blood samples were collected for plasma 25(OH)D analysis. Multinomial logistic regressions were used to determine the association between vitamin D and the severity of individual premenstrual symptoms. Adjustments were made for age, BMI, ethnicity, physical activity, hormonal contraceptive use, season of blood draw, use of anxiolytics, antidepressants, or analgesics, and calcium intake. Results: Significant inverse associations were found between 25-hydroxyvitamin D and the severity of premenstrual cramps, depression, and confusion (p<0.05). Only confusion met Benjamini-Yekutieli adjustment for multiple comparisons (p<0.015). Plasma 25-hydroxyvitamin D was not associated with any of the other premenstrual symptoms (acne, bloating, mood swings, increased appetite, fatigue, headache, anxiety, sexual desire, insomnia, nausea, clumsiness, or desire to be alone). Conclusion: Our findings indicate that plasma 25-hydroxyvitamin D status is inversely associated with some, but not all, premenstrual symptoms. 53

65 3.2 Introduction Few treatments are available for premenstrual symptoms and little is known about their risk factors, particularly when considering dietary recommendations 43. Calcium has been the most well-studied nutrient in relation to premenstrual symptoms and there is strong evidence that calcium supplementation may improve premenstrual symptoms 9, 37, 205, 206. Similarly, low dietary calcium intake has been shown to put women at risk of experiencing premenstrual symptoms 33. Vitamin D, which aids the absorption of calcium, has also recently been investigated for its role in premenstrual symptoms. Evidence has suggested an association between vitamin D and premenstrual symptoms, where those with lower vitamin D intakes experienced more severe premenstrual symptoms 33, 178. Examinations of Nurse s Health Study II data indicated that consumption of an average of 400 IU of vitamin D per day was associated with a 40% decreased risk of developing premenstrual syndrome (PMS) compared to those consuming an average of 100 IU per day. 33. This was supported by the findings of a small intervention trial showing that administration of 25,000 IU of vitamin D for two weeks reduced the severity of premenstrual symptoms in adolescents with severe hypovitaminosis 183. Vitamin D status is determined by a combination of dietary vitamin D consumption and cutaneous production, and is measured by the main circulating vitamin D metabolite 25- hydroxyvitamin D (25(OH)D). Insufficient vitamin D status is particularly prevalent in highlatitude countries such as Canada due to low wintertime sun exposure 207 and this may put Canadian women at risk for increased premenstrual symptoms. Research evaluating the association between 25(OH)D and premenstrual symptoms has been inconsistent 34, Few of these studies have measured the association between 25(OH)D and individual premenstrual symptoms, and no studies have characterised these associations by symptom severity. Treatment 54

66 response rates have been shown to vary significantly between individual premenstrual symptoms 182, and findings from objective 1 demonstrated that associations between hormonal contraceptives and premenstrual symptoms differed by symptom severity, suggesting that these may be important parameters to consider. Therefore, the objective of this study was to determine whether plasma 25(OH)D is associated with the prevalence or severity of individual premenstrual symptoms. 3.3 Materials and Methods Study Population Refer to Chapter 2, section Additional exclusion criteria were applied in Study 2. From the 1,102 subjects remaining following Study 1 exclusion criteria, 51 were excluded due to current smoking. The remaining 1,051 subjects in Study 2 were categorized into four ethnic groups based on self-reported ethnicity: Caucasian (n=481), East Asian (n=391), South Asian (n=101), or Other (n=78) Hormonal Contraceptive Use Refer to Chapter 2, section Anthropometrics and Physical Activity Refer to Chapter 2, section Premenstrual Symptoms Refer to Chapter 2, section

67 3.3.5 Plasma Samples and 25-Hydroxyvitamin D Analysis Participants provided blood samples following a minimum 12-hour overnight fast. Participants experiencing a temporary inflammatory condition (including a recent piercing or tattoo, acupuncture, a medical or dental procedure, a vaccination or immunization, flu, an infection, or a fever) underwent a two-week recovery period prior to providing blood samples. Samples were collected at LifeLabs Medical Laboratory Services (Toronto, Ontario, Canada), and 25(OH)D levels were measured at the University Health Network Specialty Lab at Toronto General Hospital (Toronto, Ont., Canada). Plasma 25(OH)D was measured by high-performance liquid chromatography tandem mass spectrometry, as described previously 208. Reported plasma 25(OH)D concentrations are the sum of measured 25(OH)D3 and 25(OH)D2. The season of blood draw was classified as follows based on month of blood draw: spring (March, April, May), summer (June, July, August), fall (September, October, November), and winter (December, January, February). Vitamin D status categories were created based on recommendations from the Canadian Osteoporosis Society, the Endocrine Society, and the Institute of Medicine 163, 171, 172. Deficient vitamin D status was defined as 25(OH)D <30 nmol/l, insufficiency was nmol/l, sufficiency was nmol/l, and optimal status was 75 nmol/l Food Frequency Questionnaire Participants completed a 196-item Toronto-Modified Willett Food Frequency Questionnaire (FFQ), which was used to estimate their dietary intakes of various foods, beverages, and supplements, including calcium-containing foods and supplements. Subjects estimated their consumption of a preassigned portion of each item over the past month by choosing from several frequency options. Responses were then converted to estimate daily 56

68 averages of total calcium intake from foods and supplements. 329 participants reported currently consuming calcium or vitamin D containing supplements Statistical Analysis All statistical analyses were conducted using SAS (version 9.4; SAS Institute Inc, Cary, NC, USA). The α was set at 0.05 and all reported p-values are 2-sided. Distribution of continuous variables was assessed prior to analysis and non-normally distributed variables were log-transformed (BMI) or square root-transformed (25(OH)D). P-values are reported from analyses using transformed variables, while untransformed means and standard errors are reported for ease of interpretation. Subject characteristics were compared between subjects in four vitamin D status categories (deficient, insufficient, sufficient, optimal) using chi-square tests for categorical variables and ANOVA for continuous variables. Multinomial logistic regressions were used to determine the associations between 25(OH)D and the severity of premenstrual symptoms. Moderate and severe symptom severities were combined due to the small number of subjects reporting severe symptoms. Univariate associations between 25(OH)D and premenstrual symptom severities were calculated in Model 1. Multivariate models were conducted in Model 2, which included adjustments for age, BMI, ethnicity, physical activity, season of blood draw, total calcium intake, and use of hormonal contraceptives, anxiolytics, anti-depressants and analgesics. Benjamini-Yekutieli adjustments for multiple comparisons were applied (15 tests, α = 0.05: p<0.015). 3.4 Results Subject characteristics are reported in Table 3-6 Subject Characteristics Stratified by Vitamin D Status1,2. The distribution of vitamin D status differed between ethnic groups. The 57

69 majority of Caucasian (75%) and Other (51%) participants had a sufficient or optimal vitamin D status, while only 37% of East Asian and 25% of South Asian participants had sufficient or optimal vitamin D status. HC use was highest among those with optimal vitamin D status and lowest in those with deficient status. Age, physical activity, and calcium intake were also highest in those with optimal vitamin D status. BMI and use of anti-depressants, anxiolytics, or analgesics did not significantly differ by vitamin D status. Mean plasma 25(OH)D of subjects was 58.7 nmol/l, which represents a sufficient vitamin D status according to criteria by the Institute of Medicine and the Canadian Osteoporosis Society 101, 163. According to these criteria, 14% of participants had a deficient vitamin D status (<30 nmol/l), 31% had an insufficient status ( nmol/l), 31% had a sufficient status ( nmol/l), and 24% had optimal status (>75 nmol/l). Mean total calcium intake was 1,019 mg/day, which is above the Recommended Dietary Allowance (RDA) of 1000 mg/day determined by the Canadian Osteoporosis Society for premenopausal women aged

70 Table 3-6 Subject Characteristics Stratified by Vitamin D Status 1,2 Deficient n (%) Insufficient n (%) Sufficient n (%) Optimal n (%) N 147 (14) 323 (31) 331 (31) 250 (24) p-value Age (years) 22.3 ± ± ± ± 0.2 <.0001 Ethnicity <0.001 Caucasian 22 (5) 96 (20) 160 (33) 203 (42) East Asian 69 (18) 177 (45) 117 (30) 28 (7) South Asian 43 (43) 33 (33) 21 (21) 4 (4) Other 13 (17) 17 (22) 33 (42) 15 (19) HC Users 18 (12) 47 (15) 92 (28) 147 (59) <0.001 Medication Users 3 37 (25) 79 (24) 68 (21) 69 (28) 0.25 BMI, (kg/m 2 ) 22.6 ± ± ± ± Calcium, (mg/d) 788 ± ± ± ± 31 < Shown are crude means ± standard errors of continuous variables, and n (%) of categorical variables. P- values are from tests using log- transformed BMI to improve fit 2 Differences between groups were compared using chi-square tests for categorical variables and ANOVA for continuous variables HC: hormonal contraceptive; BMI: body mass index 59

71 Associations between 25(OH)D and premenstrual symptom severities are shown in Table 3-7 Associations between Plasma 25-Hydroxyvitamin D and Premenstrual Symptom Severity and graphically in Figure 3-5 Associations between Plasma 25-Hydroxyvitamin D and Premenstrual Symptom Severity1,2. Model 1 includes unadjusted associations and adjusted associations are shown in Model 2. In Model 1, significant inverse associations were observed between 25(OH)D and premenstrual symptoms of acne, cramps, clumsiness, confusion, and desire to be alone (p<0.05). However, following adjustments for age, BMI, ethnicity, physical activity, season of blood draw, calcium intake, and use of hormonal contraceptives, analgesics, anxiolytics and antidepressants, only symptoms of cramps, confusion, and depression remained significantly inversely associated with 25(OH)D concentrations (p<0.05). 25(OH)D concentrations were significantly associated with the prevalence of confusion, where experiencing confusion at any severity was associated with decreased 25(OH)D concentrations. Decreased 25(OH)D concentrations were also observed in those experiencing depression at mild severity, and moderate/severe cramps. No other symptoms were associated with 25(OH)D. Following Benjamini-Yekutieli adjustments for multiple comparisons, only confusion remained significantly associated with 25(OH)D (p<0.015). Results were similar within all major ethnic groups (data not shown). 60

72 Table 3-7 Associations between Plasma 25-Hydroxyvitamin D and Premenstrual Symptom Severity Symptom Severity Plasma 25(OH)D nmol/l ± SE N = 999 Acne / Skin Blemish Bloating / Swelling / Breast Tenderness Cramps Mood Swings / Crying Easily / Irritability /Angry Outbursts Increased Appetite / Food Cravings Fatigue Model 1 p-value Model 2 p-value None 55.8 ± Mild 60.8 ± 1.5 Moderate/Severe 59.8 ± None 56.8 ± 1.8 Mild 58.6 ± 1.4 Moderate/Severe 60.2 ± None Mild Moderate/Severe 60.3 ± 1.9 a 61.2 ± 1.6 a 55.5 ± 1.4 b None 57.9 ± 1.7 Mild 60.2 ± 1.6 Moderate/Severe 57.6 ± 1.5 None 59.0 ± 1.5 Mild 59.6 ± 1.7 Moderate/Severe 57.5 ± 1.6 None 59.9 ± 1.4 Mild 59.0 ± 1.7 Moderate/Severe 55.8 ±

73 Symptom Severity Plasma 25(OH)D nmol/l ± SE N = 999 Model 1 p-value Model 2 p-value Headache Anxiety / Tension / Nervousness Clumsiness Confusion / Difficulty Concentrating / Forgetfulness Sexual Desire / Activity Change Insomnia None 58.5 ± 1.0 Mild 59.9 ± 2.5 Moderate/Severe 57.8 ± 3.0 None 59.4 ± 1.1 Mild 58.5 ± 1.9 Moderate/Severe 55.2 ± 2.7 None 59.4 ± 1.1 Mild 56.0 ± 2.9 Moderate/Severe 49.5 ± 4.7 None Mild Moderate/Severe 60.5 ± 1.0 a 54.4 ± 2.3 b 46.6 ± 3.4 b None 58.0 ± 1.3 Mild 59.9 ± 1.7 Moderate/Severe 58.3 ± 2.0 None 59.4 ± 1.0 Mild 53.2 ± 2.4 Moderate/Severe 51.0 ± <

74 Symptom Severity Plasma 25(OH)D nmol/l ± SE N = 999 None 59.0 ± 1.4 Nausea Mild 58.0 ± 2.7 Moderate/Severe 53.9 ± 3.6 Depression Desire to be alone None Mild Moderate/Severe 59.9 ± 1.1 a 56.0 ± 1.8 b 55.0 ± 3.1 ab None 60.3 ± 1.2 Mild 57.1 ± 1.8 Moderate/Severe 51.5 ± 2.5 Model 1 p-value Model 2 p-value Model 1 contains unadjusted p-values Model 2 contains p-values adjusted for ethnicity, log-transformed BMI, physical activity, age, season of blood draw, use of anxiolytics, anti-depressants, or analgesics, and total calcium intake. Letters indicate means which significantly differed in Model 2 63

75 Figure 3-5 Associations between Plasma 25-Hydroxyvitamin D and Premenstrual Symptom Severity 1,2 1 Shown are mean plasma 25(OH)D concentrations and standard errors 2 P-values are adjusted for ethnicity, log-transformed BMI, physical activity, age, season of blood draw, use of anxiolytics, anti-depressants, or analgesics, and total calcium intake. Letters indicate means which significantly differed 64