Oral Contraceptives and Cigarette Smoking: A Review of the Literature and Future Directions

Abstract

Introduction

Evidence continues to mount indicating that endogenous sex hormones (eg, progesterone and estradiol) play a significant role in smoking-related outcomes. Although approximately one out of four premenopausal smokers use oral contraceptives (OCs), which significantly alter progesterone and estradiol levels, relatively little is known about how OCs may influence smoking-related outcomes. Thus, the goal of this review article is to describe the state of the literature and offer recommendations for future directions.

Methods

In March 2017, we searched seven databases, with a restriction to articles written in English, using the following keywords: nicotine, smoker(s), smoking, tobacco, cigarettes, abstinence, withdrawal, and craving(s). We did not restrict on the publication date, type, or study design.

Results

A total of 13 studies were identified. Three studies indicated faster nicotine metabolism in OC users compared to nonusers. Five of six laboratory studies that examined physiological stress response noted heightened response in OC users compared to nonusers. Three studies examined cessation-related symptomatology (eg, craving) with mixed results. One cross-sectional study observed greater odds of current smoking among OC users, and no studies have explored the relationship between OC use and cessation outcomes.

Conclusions

Relatively few studies were identified on the role of OCs in smoking-related outcomes. Future work could explore the relationship between OC use and mood, stress, weight gain, and brain function/connectivity, as well as cessation outcomes. Understanding the role of OC use in these areas may lead to the development of novel smoking cessation interventions for premenopausal women.

Implications

This is the first review of the relationship between oral contraceptives (OCs) and smoking-related outcomes. The existing literature suggests that the use of OCs is related to increased nicotine metabolism and physiological stress response. However, the relationship between OC use and smoking-related symptoms (eg, craving) is mixed. Further, no published data were available on OC use and smoking cessation outcomes. Therefore, we recommend additional research be conducted to characterize the relationship between OC use and smoking cessation outcomes, perhaps as a function of the effect of OC use on mood, stress, weight gain, and brain function/connectivity.

Introduction

While the prevalence of smoking in men is greater than that of women in the United States, the gender gap is narrowing.^1,2 A contributing factor to this trend may be that women may have more difficultly quitting smoking, and relatedly, a number of smoking cessation pharmacotherapies may be less effective in women compared to men.^3,4 This difference is of particular concern as women are more susceptible to the harmful effects of smoking.^5,6 Further, smoking in women, as compared to men, also poses a greater overall public health risk given the well-documented adverse effects of smoking during pregnancy and heightened second-hand smoke exposure to offspring.⁵ Fortunately, women quitting smoking before the age of 40 can reduce excess risk by as much as 90%.⁷ The identification of novel modifiable predictors of smoking cessation in premenopausal women may lead to the development of effective smoking cessation interventions tailored to the unique needs of this population.

The existing literature indicates that menstrual phase and/or sex hormones (eg, progesterone and/or estradiol) influence smoking-related symptoms and, possibly cessation outcomes, in women.^8,9 Specifically, a recently conducted meta-analysis of 36 studies found that withdrawal symptoms were significantly greater in the luteal phase as compared to the follicular phase.¹⁰ Further, a trend suggested that craving may also be elevated during the luteal phase.¹⁰ The effect of menstrual phase and/or sex hormones on smoking cessation is less clear though observations suggest that the follicular phase is favorable for smoking cessation when nicotine replacement therapy (NRT) is used^11,12 whereas, the luteal phase is favorable when NRT is not used.^13,14 Further, a recent clinical study observed that an increasing progesterone-to-estradiol ratio, which occurs after ovulation, was favorable for smoking abstinence in women treated with NRT, but not those using varenicline for smoking cessation.¹⁵ Although the specific mechanisms involved in these observations are not yet identified, it is hypothesized that this may be due to progesterone and/or estradiol modifying the neurobiological response to nicotine.¹² Taken together, these observations indicate that menstrual phase and/or sex hormones influence smoking cessation outcomes in women.

Hormonal contraceptives, including oral contraceptives (OCs), are used for birth control, as well as for a variety of off-label uses such as acne, hirsutism, and endometriosis.¹⁶ There are over 90 different types of OCs on the market today, with combination OCs (those that contain both estrogen and progesterone components) as the most common formulation.^17,18 OCs significantly alter progesterone and estradiol levels by preventing ovarian follicle maturation, thus resulting in a low, stable PE ratio throughout the menstrual cycle (Figure 1).¹⁹

Figure 1.

Schematic representation of progesterone to estradiol (PE) ratios for (A) the idealized natural menstrual cycle and (B) during oral contraceptive use.

Although combination OCs are contraindicated in women smokers over the age of 35, especially in those who smoke 15 or more cigarettes per day,²⁰ due to increased risk of cardiovascular complications, the use of OCs in smokers is prevalent. Data from the 2002–2004 Behavioral Risk Factor Surveillance Systems survey suggests that approximately one out of three premenopausal women who smoke use some form of hormonal contraceptive, with 27% using OCs specifically.²¹ More recent estimates, per an online survey conducted with a convenience sample, suggest the prevalence of hormonal contraceptive use in premenopausal smokers is even higher (49%) with a similar prevalence of OC use (29%).²² Despite the known and significant effect OCs have on endogenous levels of progesterone and estradiol (which have been linked to smoking-related outcomes), as well as a high prevalence of use, research on the role of OCs in smoking-related outcomes is lacking. Thus, the goal of this review is to describe the current state of knowledge on the relationship between OCs and smoking-related outcomes. We also discuss the relationship OCs have with other variables (eg, mood, stress, weight gain, and brain structure/function) that may influence smoking behavior and cessation in women, and offer recommendations for future research.

Methods

In March 2017, the following databases were searched: PubMed/MEDLINE (1946–2017), Wiley/Cochrane Library (1888–2017), EBSCO/PsycINFO (1880s–2017), Elsevier/Embase (1947–2017), Elsevier/Scopus (1823–2017), Thomson Reuters/Web of Science (1898–2017), and EBSCO/CINAHL with Full Text (1981–2017). Keywords (and their corresponding controlled vocabulary terms) included hormonal/OCs, nicotine, smoker(s), smoking, tobacco, cigarettes, abstinence, withdrawal, craving(s), and stress. Because of the dearth of articles found in initial screening, heavy use was also made of the “Similar Articles” function in PubMed as well as forward and backward citation analysis of the most relevant articles in Scopus and Web of Science. No limitations were placed on publication dates, publication type, or study design. Articles were chosen if they addressed any direct association between OC use and smoking-related outcomes, and if they were written in English.

Results

A total of 13 publications were identified that reported on nicotine pharmacokinetics (n = 3), physiological stress response (n = 6), smoking reward (n = 1), and/or abstinence-related outcomes (n = 4).

OCs and Nicotine Pharmacokinetics

Three studies examined the relationship between OCs and nicotine pharmacokinetics, including metabolism of nicotine (Supplementary Table 1).^23–25 Berlin et al.²³ examined the ratio of trans-3′-hydroxycotinine (3HC) to cotinine, a measure of nicotine metabolism, in 120 daily smoking adolescents (ages 13–17 years; OC users: n = 30; nonusers: n = 54; boys: n = 36). 3HC-cotinine ratios were significantly higher (ie, nicotine metabolism was faster) in OC users (M = 0.47) than nonusers (M = 0.28) and boys (M = 0.25; p < .001). Similar results were found by Benowitz et al.²⁴ in a sample of 294 adults (OC users: n = 53; nonusers: n = 153; men: n = 88). Specifically, OC users demonstrated faster nicotine metabolism than nonusers and men using multiple measurements, including nicotine clearance, cotinine clearance, and 3HC-cotinine ratio (ps < .05). Faster nicotine metabolism was specific to estrogen-containing OCs and was not found for women using progesterone-only OCs. Additionally, Benowitz et al.²⁴ observed higher levels of total nicotine and cotinine clearance and lower nicotine half-life in OC users versus nonusers. Finally, Chenoweth et al.²⁵ also reported higher 3HC-cotinine ratios for women taking estrogen-containing OCs (n = 17; M = 0.49, SD = 0.24) compared nonusers (n = 739; M = 0.41; SD = 0.22) although the difference did not reach statistical significance (p = .09). Together, these studies suggest faster nicotine metabolism for OC users, especially those containing estrogen, compared to nonusers and men.

OCs and Physiological Stress Response

Six laboratory studies examined physiological stress response by OC use in smokers.^26–31 Emmons and Weidner²⁸ compared cardiovascular reactivity to stress among 69 female college students by smoking status (smoker, nonsmoker) and OCs use (OC, no OC). Regardless of smoking status, OC users had a greater cardiovascular response (ie, increase in systolic blood pressure) to a mental arithmetic task compared to nonusers (M = 15.03 vs. M = 12.45) though this difference did not reach statistical significance (p = .09). No differences in cardiovascular reactivity were found by OC use for a cold pressor task. A second study of 35 young adult women smokers²⁷ similarly found OC users had a greater systolic and diastolic blood pressure response to a videotaped speech task than nonusers (ps < .05). Davis²⁹ also used a speech task and a mental arithmetic task to examine stress in 52 women by smoking status and OC use. Among the smokers, OC users demonstrated a greater systolic blood pressure response to both tasks compared to nonusers. Similarly, in a sample of smokers, Masson and Gilbert²⁶ examined nicotine-related changes in cardiovascular activity in 12 women using OCs and 12 women not using OCs. Women using OCs demonstrated a greater increase in heart rate when smoking a cigarette after overnight abstinence compared to women not using OCs (p < .05). In contrast, in a study of 60 smoking and nonsmoking white women (n = 30 OC users), West et al.³⁰ observed that, among smokers who were acutely abstinent (12 h) and either completely deprived of nicotine or were using nicotine gum only, a smaller change in diastolic blood pressure in response to a videotaped speech was observed in OC users as compared to the nonusers. Finally, a recent study³¹ examined cortisol levels over 4 days of smoking abstinence and found that women using OCs (n = 15), compared to women not using OCs (n = 30), had higher levels of morning cortisol the day before quitting and lower levels of morning cortisol on the third day of abstinence during the low progesterone week; lower first morning cortisol on quit day during high progesterone week; and greater nadir change. Together, most of studies have shown increased physiological stress response for OC users compared to nonusers.

OCs and Smoking Reward

One study examined self-reported satisfaction while smoking a cigarette in a sample of adult women using OCs (n = 14) and not using OCs (n = 28).³² Smoking satisfaction was measured during ad lib smoking in both groups twice: once during a low progesterone week (eg, follicular phase in those not using OCs and during the 0.018 norgestimate dose of OC) and once during a high progesterone week (eg, luteal phase in those not using OCs and 0.025 mg dose of OC). OC users reported greater smoking satisfaction during the low progesterone week while non-OC users reported greater smoking satisfaction during the high progesterone week (OC use by week interaction p < .001).

OCs and Abstinence-Related Behaviors

While no study was identified through the literature review that examined cessation outcomes or smoking relapse for women using OCs compared to women not using OCs, one cross-sectional study observed higher odds (2.42, 95% CI: 1.87–3.14) of current smoking among OC users compared to nonusers in the Fourth Korean National Health and Nutrition Examination Survey.³³ Several studies examined behaviors during brief smoking abstinence by OC status including, withdrawal symptoms, cravings, mood, and stress. Adult women taking OCs reported greater physical withdrawal symptoms (eg, heart beat faster) than women not using OCs (mean and SD not reported, p < .05) yet no difference was found for overall withdrawal symptoms.²⁶ No differences in overall withdrawal symptoms were also found for adolescent girls taking OCs compared to those not taking OCs over the first 2 weeks of a quit attempt.³⁴ Interestingly, differences were found for cravings with girls not taking OCs reporting a higher level of cravings (M = 2.38, SD = 0.16) than girls taking OCs (M = 2.11, SD = 0.25) and boys (M = 1.71, SD = 0.16; p < .01).³⁴ A trend toward higher cravings for women not taking OCs compared to women taking OCs was also found by Hinderaker et al.³² (M = 3.02, SD = 1.15 vs. M = 2.73, SD = 1.00, p = .086).

With regard to abstinence-related mood changes, Masson and Gilbert²⁶ found that, following overnight abstinence, women not using OCs reported greater decreases in anxiety after smoking a cigarette than OC users (p < .05). They did not find a difference by group with regard to changes in negative affect after smoking a cigarette. Similarly, Epperson et al.³⁵ found no difference in negative affect by OC status in their clinical trial of the transdermal nicotine patch and naltrexone for smoking cessation. Women using OCs (n = 10; M = 4.5, SD = 3.4) reported a similar average level of negative affect during the first 5 days of the smoking quit attempts as women not using OCs (n = 37; M = 4.7, SD = 3.4; p-value for comparison not reported). Finally, as displayed in Figure 2, Hinderaker et al.³² observed that after 3 days of smoking abstinence, OC users reported higher negative affect during their low progesterone week than their high progesterone week while non-OC users showed the opposite pattern, higher negative affect during the high progesterone week than the low progesterone week (OC use by week interaction p < .001). The authors also found that OC users reported lower positive affect after 4 days of abstinence than non-OC users (p = .032). During smoking abstinence, there is preliminary evidence for differences in positive affect and anxiety relief with OC use and mixed evidence for differences in negative affect with differences appearing when both high and low progesterone time points are examined.

Figure 2.

Mean smoking satisfaction, negative affect, and physical reward by OC Group and testing week from Hinderaker et al.³² (a) Smoking satisfaction during ad libitum smoking. (b) Negative affect on the third day of abstinence. (c) Psychological reward during ad libitum smoking. Originally published in Hinderaker et al.³² Reprinted with permission from Elsevier.

Discussion

The prevalence of OC use in premenopausal smokers is high. However, relatively little is known about OCs and smoking. While the few studies that were identified on this topic demonstrated some mixed findings, these studies have provided some important preliminary information. First, three studies demonstrated that OC use is associated with faster nicotine metabolism.^23–25 These observations, combined with other data demonstrating faster metabolism during pregnancy,^36,37 suggest the role of increased estrogen levels in faster nicotine metabolism. Faster metabolism of nicotine has important clinical relevance as it is related to more intense smoking,^24,38 greater rewarding effects of nicotine,³⁹ greater cravings and withdrawal symptoms,^39,40 and worse smoking cessation outcomes.^41,42 While no studies have yet looked at the relationship between OC-related nicotine metabolism and cessation outcomes, these observations suggest that OC use may increase nicotine metabolism which may make OC users be more dependent on nicotine and, therefore, have more difficulties achieving abstinence. Additional work should also be pursued on potentially important effect modifiers of this relationship such as race.^25,43 Second, five of the six studies that examined physiological stress response contained some evidence of increased response in OC users compared to nonusers.^26–29 It is possible that this increased response may predispose OC users to relapse to smoking after a lapse given the heightened and potentially more reinforcing response to smoking. This hypothesis may also apply to the next observation of this review which is that one study reported greater smoking satisfaction in OC users compared to nonusers but only during the early part of the menstrual cycle.³² Next, the three studies that explored symptoms related to cessation had conflicting results, especially in terms of negative affect and craving. These conflicting observations may be related to methodological study differences such as the study sample and design. For example, the study that observed lower craving in OC users compared to nonusers was in adult women who were temporarily abstaining from smoking³² whereas the study that observed higher craving in OC users than nonusers was in adolescent girls who were in a smoking cessation trial.³⁴ Menstrual cycle timing of assessments may have also contributed to the conflicting observations. For example, the study that observed higher negative affect in OC users compared to nonusers only observed this during the early part of the menstrual cycle³² whereas the study that observed no differences in negative affect by OC use did not time the assessments according to menstrual cycle.³⁵ Lastly, at the time we conducted our literature review, there were no studies on the relationship between OC use and smoking cessation. Since that time, one new study has been published.⁴⁴ This project was a secondary-data analysis that observed that cyclical combination hormonal contraceptive users (eg, OC users, as well as those who used Nuva Ring® and the transdermal patch) had significantly more adverse withdrawal (during ad libitum smoking, on quit day, and at 7 days after quit day), craving (during ad libitum smoking), and negative affect (7 days after quit day) compared to women not using hormonal contraceptives and men. However, women using hormonal contraceptives were at 3.7 higher odds of being abstinent from smoking at end of treatment as compared to men; no difference was observed between the two groups of women. This new publication offers intriguing results and we recommend additional research be conducted on this area, as well as the areas described below.

Other Areas for Future Research

OCs and Mood

One of the most well-studied predictors of smoking behavior and cessation is mood, including depressed mood, positive and negative affect, and anxiety. Individuals with depressed mood have poorer smoking cessation outcomes.⁴⁵ Similar relationships have been observed between negative affect and anxiety with smoking cessation.^46–48

Approximately 4%–10% of women on combination OCs report adverse changes such as depressed mood, irritability, and/or anxiety.⁴⁹ Research on the effect of estrogen and progesterone on mood has a lengthy and robust history. This line of research first began in 1942, well before OCs were developed, when progestins and estrogen were used to modify psychopharmacological effects on mood.⁵⁰ In 1967, Dr Ira Glick concluded that there was not enough evidence to determine the effect of OCs on mood.⁵⁰ The same conclusion remains true today as conflicting evidence continues to mount.^49,51–53 Much of the existing literature to date has been limited by the use of between-subject designs, small study samples, use of imprecise language, and/or a lack of a randomized control approach.^51,54 In addition to these methodological limitations, this large body of literature contains several potential explanations for this conflicting data including, for example, effects that vary by composition of the OC,^54–56 menstrual cycle timing,⁵³ length of time the contraceptive has been used and/or age,^51,57 and family and/or personal history of mood disorders.^49,58,59

Despite the conflicts in these data, according to a review conducted by Oinonen and Mazmanian,⁵¹ two relatively consistent findings have emerged. First, OC users have less menstrual cycle variability in mood as compared to nonusers. This may be relevant to smoking cessation as previous work indicates that less variability in smoking-related symptomatology over the course of the menstrual cycle in naturally cycling women is associated with improved cessation outcomes.⁶⁰ The second consistent finding is that OC users tend to have less negative affect than nonusers. As an extension of this finding, Böttcher et al.⁵⁴ concluded that the estrogen contained in combination OCs may have a positive influence on mood, whereas progestin-only OCs may have a negative influence. This view was further supported by the most recent review conducted by Schaffir et al.⁵⁸ that concluded that OCs containing less androgenic progestins have a lower likelihood of adversely influencing mood. While Schaffir et al.⁵⁸ also concluded that OC use is not related to adverse mood changes in most women, a recent study of over 1 million Danish women found a 20%–30% increase in risk of subsequent use of antidepressants and a 10%–20% increase risk of subsequent first depression diagnosis in women using OCs versus nonusers.⁵⁶ While one of the two studies included in this review did not observe a significant difference in negative affect by OC use,³⁵ the other observed higher negative affect in OC users than nonusers but this varied by menstrual phase.³² Specifically, Hinderaker et al. observed higher negative affect during the follicular phase compared to the luteal phase in OC users whereas the nonusers had lower negative affect in the follicular phase compared to the luteal phase. A similar pattern in “irritability” was observed in a recent publication by Lundin et al.,⁵³ though it was not statistically significant. If menstrual phase variability of negative affect varies by OC use, this may have implications for smoking cessation (eg, the ideal menstrual cycle timing of quit date may differ by OC use). Clearly, additional work is needed to disentangle the relationship between OC use and mood both in the general population, as well as within the context of smoking.

OCs and Stress

Similar to mood, there is a clear relationship between smoking and stress (see review for additional information⁶¹) and stress may play an especially important role in the smoking behavior of women.⁶² A number of studies have examined the association of ovarian hormones and OCs in response to stress. OCs have been implicated in hypothalamic–pituitary–adrenal (HPA) axis responsivity with progesterone appearing to have an inhibitory effect on HPA axis reactivity.⁶³ There has been a general consistency in study results with OC use related to reduced responsivity (eg, blunted cortisol response) to physical and mental stress (eg, refs.^64,65). This decreased reactivity may be related to increased levels of corticosteroid binding globulin among women taking OCs.^66,67 Further, a study of 28 women found that reduced HPA axis reactivity during a physical stress task was found for women using OC but was not associated with menstrual cycle phase in naturally cycling women.⁶⁵ As reviewed above, women taking OCs demonstrate increased physiological stress response. More research is needed to further understand the relationship between smoking, OCs, and stress including the impact of OCs and smoking on measures of stress beyond cardiovascular reactivity (eg, cortisol) and stress-related variables such as memory and learning which have shown differences by OC use (eg, refs.^68,69). In addition, more research is needed to better understand the relationships of OC use and stress to important smoking-related outcomes such as smoking reward, cessation, and relapse.

OCs and Weight Gain

The relationship between weight and smoking is well established. Smokers, on average, weigh approximately 10 pounds less than nonsmokers.⁷⁰ Nicotine increases the resting metabolic rate and energy expenditure, each by approximately 10%.^71,72 Following smoking cessation, a recent meta-analysis identified that quitters will gain an average of 9 pounds over a 5-year period.⁷³ Given this well-known relationship, smokers, particularly female smokers, report that they smoke to manage their weight,⁷⁴ and that weight gain is a risk associated with smoking cessation that reduces their motivation to quit.⁷⁵ Further, women smokers are more likely to relapse back to smoking following weight gain during a quit attempt.⁷⁶ The impact of OCs on the relationship between weight and smoking is unknown. We could not identify any literature which has addressed this topic. While there is much literature examining whether OCs have a negative impact on weight, these investigations have not encompassed smoking status. Cochrane reviews examining both progestin-only and combination contraceptives demonstrate that OCs do not negatively influence weight, although they acknowledge that there is a wide-spread belief to the counter.^77,78 Given the wide-held and erroneous belief that OCs cause weight gain, women taking OCs may be more inclined to continue smoking for weight management and see smoking cessation as riskier with regards to weight gain, although this is currently unknown. It is also unknown whether OCs moderate weight gain following cessation, either through expectations or some other biological process.

OCs and Brain Structure and Function

Over the past two decades, neuroimaging has provided insight into the structural and functional changes associated with cigarette smoking.^79,80 According to a recent meta-analysis of structural magnetic resonance imaging (MRI) studies,⁸¹ cigarette smoking is associated with gray matter volume decreases in the left insula, right cerebellum, parahippocampus, prefrontal cortex (PFC) regions, and the thalamus. Further, cigarette smoking is associated with functional brain changes in brain regions involved in reward processing and cognitive control.^82,83 In fact, recent research suggests that sex differences in the effects of cigarette smoking on the brain exist^82,84 which may be related to the effect of ovarian hormones on the brain.^85,86 This suggests that OCs could alter brain structure and function in different ways.⁴⁹ Unfortunately, there are no known studies examining the effects of OCs on brain structure and function in cigarette smokers.

OCs and Brain Structure and Function: Emotion

In the general population, OC users exhibit altered emotion reactivity and regulation, which refers to the extent to which an individual responds to emotional stimuli (ie, sensitivity), as well as the intensity and persistence of that response.⁸⁷ In general, adolescents using OCs show a blunted cortisol response upon awakening.⁸⁸ This contrasts with work done in adult OC users who smoke compared to their nonuser smoking peers which demonstrated higher cortisol at waking during ad libitum smoking but lower levels during acute smoking abstinence.³¹ Adolescent OC users also differed in response to certain stressors (eg, social stress test,⁸⁹ viewing an emotional story⁶⁹); however performance on emotion-related tasks and neural responses during emotion-related tasks varies across studies. For example, on tasks of emotional memory, OC users exhibit attenuated bilateral amygdala response when exposed to negatively valenced, emotionally arousing stimuli compared to nonusers,⁹⁰ whereas in tasks examining aspects of fear (ie, conditioning, extinction, and recall), women taking OCs show lower activations in the insular and cingulate cortices, amygdala, hippocampus, and hypothalamus than naturally cycling (NC) women with high estrogen levels.⁹¹ In a study exploring brain responses to faces, women using OCs (both adults and adolescents) showed stronger responses to faces than NC women in the fusiform face area (FFA), and this neural response to faces in the FFA correlated to duration of OC use.⁹² However, in a double-blind, placebo-controlled study examining responses during an emotional face task in women who were randomized to either active OC or placebo, women taking OCs exhibited lower emotion-induced reactivity in the left insula, left middle frontal gyrus, and bilateral inferior frontal cortex compared to those randomized to placebo.⁹³ Further, compared to pre-OC treatment, women taking OCs showed decreased emotional reactivity in the bilateral inferior frontal cortex.⁹³ Interestingly, women taking OCs reported increased depressed mood, mood swings, and fatigue compared to those in the placebo group, which is consistent with recent findings showing that women taking OCs have increased incidence rate ratios of depression diagnosis and use of antidepressants. Further, the pattern of hypoactivation in the insula, cingulate, and frontal regions during emotional processing in women taking OCs parallels findings from a recent meta-analysis of neuroimaging studies of emotion and cognition in major depression.⁹⁴ Together, the findings from these studies suggest that OC use may dampen stress response and emotional reactivity during emotional processing tasks, perhaps in a pattern similar to that of individuals with depression; however, brain responses vary across studies depending on the methodological approach.

OCs and Brain Structure and Function: Reward

To date, there are few neuroimaging studies examining the effects of OC use on brain responses during tasks associated with reward. Indeed, we found only one neuroimaging study examining reward, specifically monetary reward, by OC use. Bonenberger et al.⁹⁵ examined differences in neural activation during a monetary incentive task and found that women taking OCs showed greater neural response during expectation of monetary reward in the anterior insula and inferior lateral prefrontal cortex. In a recent double-blind placebo-controlled study, 58 naturally cycling women were randomized to receive either placebo or the gonadotropin-releasing hormone agonist (GnRHa) goserelin, which causes a decrease in sex-steroid levels (ie, estradiol, testosterone).⁹⁶ Participants completed a monetary reward task during neuroimaging conducted in the mid-follicular phase and following placebo or GnRHa administration. While GnRHa is not an OC, it changes sex-hormone levels and could provide additional information on the effect of changes in sex-hormone levels on reward processing. Findings revealed that GnRHa caused a net reduction in estradiol and testosterone levels, an increase in depressive symptoms, and, compared to placebo, reduced amygdala reactivity to high monetary rewards. Together, these findings suggest that hormone levels clearly influence reward processing; however, future research should explore the effects of OCs on reward-related processing particularly as it pertains to cigarette smoking.

OCs and Brain Structure and Function: Cognitive Function

Recent studies and reviews exploring the effect of OC use on cognition suggest that women taking OCs show better cognitive performance on tasks of verbal memory,^97,98 visual spatial ability, speed and flexibility,⁹⁹ associative learning, and spatial attention¹⁰⁰ than nonusers. In the neuroimaging literature, however, findings are inconsistent. In an fMRI study exploring differences in neural responses and performance on a verb generation task among women taking OCs, men, and nonusers scanned once during menses and once during the luteal phase of their menstrual cycle, there were no differences in task performance, but differential activation across groups. Women taking OCs had similar neural responses to men, whereas women taking OCs showed greater response in the right superior temporal cortex compared to NC women during menses and greater response in the inferior frontal cortex than nonusers during the luteal phase.¹⁰¹ Pletzer et al.¹⁰² compared number processing performance and neural responses during the task in women taking OCs, men, and nonusers scanned once during the follicular phase and once during the luteal phase and found that women taking OCs performed similarly to NC women in their follicular phase, yet had similar neural activation patterns to men. Both women taking OCs and men had less neural response during the task than nonusers in their follicular phase in the medial prefrontal cortex. Thus, in these two studies that compared men, nonusers at different times of their menstrual cycle, and women taking OCs, performance was similar across groups, but neural responses of women taking OCs was similar to men and differed to NC women depending on the assessment time during the menstrual cycle.

Using a randomized placebo-controlled design, Gingnell et al.¹⁰³ investigated neural response and performance during a Go/NoGo response inhibition task before and during treatment (OC vs. placebo). Before treatment, groups did not differ on performance, but only women in the OC group showed improvement in performance during treatment. Further, women in the OC group displayed decreased activity in the right middle frontal gyrus during treatment compared to placebo. In this study, women in the OC group showed improved response inhibition performance with less neural activity, suggesting more efficient cognitive control among OC users.

OCs and Brain Structure and Function: Connectivity

Given that resting-state functional connectivity (rsFC) is a relatively recent advance in the neuroimaging field, very few studies have explored the effects of OC on rsFC. Briefly, rsFC approaches allow for an examination of functional interactions between brain regions and represents inherent brain organization underlying brain function¹⁰⁴ and behavior.¹⁰⁵ Of the multiple neural networks that exist, three networks are particularly relevant to cognition and behavior, and include the default mode network (DMN), the executive control network (ECN), and the salience network (SN).¹⁰⁶ Given the importance of these networks, Petersen et al.¹⁰⁷ analyzed resting-state data from 45 nonusers (20 in early follicular phase, 25 in luteal phase) and 46 women taking OCs (24 active pill, 22 inactive pill) and compared rsFC of the anterior DMN (aDMN) and the ECN. Women in the early follicular phase had increased connectivity of the aDMN with the left angular gyrus compared to women taking OCs and women in the early follicular phase had greater connectivity between the ECN and the left middle frontal cortex than women taking OCs. The authors suggest that these findings indicate that women using OCs could experience deficits in abstract and self-referential processing as well as difficulties integrating cognition and emotion; however, this is purely speculative as behavior and/or cognitive performance was not examined or associated with these rsFC differences.

There are relatively few studies exploring how OC use influences brain structure. There are cross-sectional studies that compare brain structure of OC users compared to nonusers,^108–111 but findings across these studies are inconsistent. One study examined the effects of OC use on cortical thickness and found that women taking OCs had decreased cortical thickness in the lateral orbitofrontal cortex and posterior cingulate cortex compared to NC women.¹¹⁰ Findings become more complicated in studies comparing gray matter volume (GMV) by OC use. In fact, there were no clear, consistent findings across studies^108,109,111 as each study used different comparison groups and different analytic approaches.

Findings from a recent longitudinal neuroimaging study that investigated structural brain changes following 3 months of OC use and associated changes in rsFC among women using OCs compared to age-matched NC women indicate that the left amygdala/anterior parahippocampal gyrus volume decreased in OC users versus nonusers.⁵⁷ Further, functional connectivity of the amygdala/parahippocampal region showing reduced GMV changed pre- to post-OC use such that the connectivity of the region to the dorsolateral prefrontal cortex (dlPFC) was positive prior to OC use, then negative after 3 months of OC use. The nonusers showed an opposite pattern of rsFC between the amygdala/parahippocampus and the dlPFC, despite being scanned during the early follicular phase at both time points. Overall, the ways in which OCs affect brain structure remain unclear. However, the findings of reduced GMV in the amygdala/parahippocampus following 3 months of OC use and associated changes from positive to negative rsFC between this area and the dlPFC align with functional and behavioral findings of reduced emotional reactivity and difficulties integrating cognition and emotion. Given the inconsistencies within the neuroimaging literature on the effect of OC use on the brain, and the lack of research examining the effects of OCs on brain structure and function in cigarette smokers, additional research in these areas are clearly warranted.

Limitations

Given this review is exploring a novel potential predictor of smoking behavior, this work contains a number of limitations. First, the formulations of OCs have evolved over time. Older OCs contained higher doses of estrogen and progesterone than modern OCs. Thus, the effect of modern OCs may differ from the effects observed in older OCs. Despite this concern, we did not restrict our review by publication date, including one article from 1980s and three from 1990s. Second, and similarly, most studies (9 out of 13) did not include a description of the OC dosage, nor did they standardize the type of OC investigated. Therefore, we cannot assess the direct effect of the dose of OC on smoking-related outcomes. Third, and relatedly, many formulations of OCs have variable doses of estrogen and progesterone. However, most of these studies (10 of 13) did not take into consideration these varying hormone levels. Fourth, methods varied from study to study. For example, one study²³ included injectable progesterone-only hormonal contraceptive users within the sample of combination OC users; one study did not clarify if the progesterone-only contraceptive studied was an oral formulation or another type of hormonal contraceptive³⁴; and one study included both hormone replacement therapy and OC users.²⁴ Study samples thus included heterogeneous groups of hormone users which may have blunted the association between OC use and smoking-related outcomes. Fourth, the study samples were, in general, small and homogenous with respect to demographic variables such as race. Next, and relatedly, combination OC use is contraindicated in some smokers including those over the age of 35²⁰ and none of the studies included here randomly assigned participants to OC condition. Thus, the observed differences between OC users and nonusers may be influenced by selection bias. Lastly, we restricted studies to those published in English and, therefore, may have not have included all available work on this topic.

Conclusions and Recommendations

In summary, OC use is high in smokers and the use of OCs appears to affect a variety of smoking-related outcomes including nicotine metabolism, cardiovascular reactivity, mood, craving, and withdrawal. Further, there is substantial existing literature suggesting that OC use may influence several factors related to smoking behavior such as cognitive function and reward, as well as weight gain. Future work on OC users, either alone or in comparison with naturally cycling women, would be strengthened by the inclusion of five additional methodological considerations: (1) standardizing the timing of data collection periods to occur during specific points in the menstrual cycle to limit the variability of hormones, (2) examining the effect of exogenous hormone values by measuring, analyzing, and reporting the type (eg, combination vs. progestin-only, or mono-, bi-, vs. tri-phasic) and dose of OC, (3) including refined measurement of the relationship between endogenous hormones and smoking-related outcomes by including absolute and change of endogenous hormone values given smoking behavior has been associated with endogenous hormone changes in naturally cycling women,^15,112 (4) comparing OC use by “long-time users” versus “first-time users” as recommended by Oinonen and Mazmanian⁵¹ to avoid a survivor effect (ie, those who have been using OCs for a lengthy period of time may be effected by OCs differently than those who are using OCs for the first time), and (5) including examination of potential effect modifiers such as race and personal/family history of depression. We also recommend that OC use be assessed in important epidemiological surveillance studies focused on tobacco use or drug use more generally (eg, National Survey and Drug Use and Health, Population Assessment of Tobacco and Health). Understanding the impact that OC use has on smoking and cessation behaviors will help to optimize treatment outcomes for premenopausal women.

Funding

Support for this project was provided by National Institutes of Health and National Institute of Alcohol Abuse and Alcoholism (K23AA023894 to RRW) and Office of Women’s Health Research and National Institute of Drug Abuse (P50DA033945 to SAM).

Declaration of Interests

None declared.

Acknowledgments

We extend our thanks to Evelyn Rens, Family and Community Medicine, for her editorial assistance and Steve Tkachyk, Biocommunications, for his graphic design. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.