Comparing the Rate of Nicotine Metabolism among Smokers with Current or Past Major Depressive Disorder

Abstract

Background and Objective:

Persons with current or past major depressive disorder (MDD), vs. those without, have higher smoking rates. The nicotine metabolite ratio (NMR) represents variation in the rate of nicotine metabolism and has been associated with smoking behaviors and response to tobacco treatments. We compared NMR between smokers with current or past MDD (MDD+) vs. smokers without MDD (MDD−). We also assessed correlates of NMR and compared withdrawal and craving between MDD+ and MDD− smokers.

Methods:

Using baseline data from two clinical trials and propensity score weighting based on sex, race, body mass index, and smoking rate, we compared NMR between MDD+ (N=279) and MDD− (N=1,575) smokers. We also compared groups on and nicotine withdrawal and craving.

Results:

Mean NMR (β=−0.02, 95% CI: −0.05 to 0.01, p=0.13) and the distribution of smokers across NMR quartiles (OR=0.76, 95% CI: 0.50 to 1.16, p=0.21) were similar between MDD+ and MDD− samples. This relationship was not affected by antidepressant medication. In the MDD+ sample, African Americans had significantly lower mean NMR, while older smokers and smokers with lower education had higher mean NMR (p’s<0.05). MDD+ smokers had significantly higher withdrawal and craving than MDD− smokers (p’s<0.05).

Discussion and Conclusions:

While variability in NMR may not explain differences in smoking rates between MDD+ and MDD− smokers, MDD+ smokers report increased withdrawal and craving

Scientific Significance:

In this first study to assess NMR among MDD+ smokers, the findings underscore the need to address withdrawal and craving within smoking cessation treatments for those with MDD.

Introduction

Approximately 1 in 10 Americans will receive a diagnosis of major depressive disorder (MDD) during their lifetime,¹ making MDD one of the most prevalent psychiatric conditions. Persons with MDD report a smoking prevalence that is 2–3 times greater than the general population,² which may be associated with a substantial increased risk for cancer and heart disease and a significantly lower life expectancy, vs. the general population.³ Identifying unique characteristics among smokers with MDD may yield recommendations for more effective approaches to treating tobacco use.

Variability in the rate at which nicotine is metabolized may be one factor that contributes to the higher prevalence of smoking and greater smoking intensity among those with MDD compared to the general population. Nicotine is primarily metabolized by the cytochrome P450 (CYP) liver enzyme CYP2A6 into cotinine and then exclusively by CYP2A6 into 3´-hydroxycotinine (3HC).⁴ The ratio of 3HC to cotinine, referred to as the nicotine metabolite ratio (NMR), is a valid biomarker of the rate of nicotine metabolism.⁵ Further, in studies with the general population of smokers, higher NMR (indicating more rapid nicotine clearance) has been associated with smoking more cigarettes per day, greater nicotine dependence, more severe nicotine withdrawal symptoms, and a decreased response to nicotine replacement therapy (NRT) and better response to varenicline.^6,7

While NMR is a marker for genetic variation in the CYP2A6 gene, a range of environmental factors also influence nicotine metabolism, which is captured by the NMR, such as gender, body mass index (BMI), use of mentholated tobacco, and alcohol use.⁸ Certain medications used to treat MDD, such as selective serotonergic reuptake inhibitors, also induce or inhibit CYP2A6 activity and, thus, may alter nicotine metabolism.^9–11 To date, to the best of our knowledge, no study has examined differences in NMR between MDD+ smokers and smokers without past or current MDD. One study with 75 smokers diagnosed with bipolar disorder reported that NMR was significantly higher compared to 86 smokers without bipolar disorder but not when controlling for bipolar subjects taking medications that were hepatic enzyme-inducing including carbamazepine, oxcarbazepine, and topiramate.¹² In addition to antidepressant medication effects, studies have also delineated the shared neurobiology underlying nicotine dependence and depression,¹³ that suggests the need to examine NMR among MDD+ smokers.

Therefore, in this study, we compared NMR between MDD+ smokers and smokers without MDD, using propensity score matching to balance for key variables associated with NMR (e.g., race, gender). We hypothesized that MDD+ smokers would have significantly higher NMR than MDD− smokers, which underlies the increased prevalence rate of smoking among those with current or past MDD. An increased NMR may increase nicotine dependence, the number of cigarettes smoked each day, and a decreased responsiveness to smoking cessation treatment, including NRT, factors which, in turn, lead to the increased smoking prevalence and lower quit rate among those with MDD vs. the general population. We assessed the effect of class of antidepressant medication within MDD+ smokers on NMR as well in order to assess the possible effects of medications that affect CYP2A6 activity. Further, we examined correlates of NMR in the MDD+ sample, and compared MDD+ and MDD− smokers in nicotine withdrawal and craving, since past studies have suggested that such variables are associated with NMR⁷ and may vary between smokers with MDD vs. smokers without MDD.^14,15 Identifying sources of differences between smokers with current or past MDD vs. those without this comorbidity can be used to guide approaches to smoking cessation treatment in this population.

Methods

Samples

MDD+ Participants:

Data for this sample came from a randomized controlled trial testing varenicline plus behavioral activation therapy for smokers with current or past MDD. The study was approved by the Northwestern University and University of Pennsylvania IRBs (ClinicalTrials.gov ID: NCT02378714). Participants were recruited through media advertisements. To be eligible, participants had to be >18 years of age, report daily smoking, have a current or past DSM-5 diagnosis of MDD without psychotic features, able to communicate in English, and able to provide informed consent. Key exclusion criteria included daily use of e-cigarettes or another nicotine/tobacco product, use of mood stabilizers, self-reported suicide attempt in the last 12 months, self-reported current or planned pregnancy, self-reported current use of smoking cessation medication, heavy alcohol consumption defined as more than 28 drinks per week, lifetime bipolar or psychotic disorder as determined by either self-report or the Mini International Neuropsychiatric Interview (MINI-7),¹⁶ and uncontrolled hypertension (systolic > 185 or diastolic > 110). Of the 300 participants enrolled in this trial, 279 provided a sample for NMR and were included in this study.

MDD− Participants:

Data for this sample came from a multi-site NMR-stratified (slow vs. fast) randomized controlled trial testing varenicline, vs. transdermal nicotine, vs. placebo approved by all site IRBs (ClinicalTrials.gov ID: NCT01314001). Participants were age 18–65, reported smoking >10 cigarettes per day for the past 6 months, and provided written informed consent. Exclusion criteria included use of non-cigarette tobacco products, e-cigarettes, or current smoking treatment; history of substance abuse treatment, current use of cocaine or methamphetamine, or >25 alcoholic drinks per week; medical contraindications (e.g., pregnancy, uncontrolled hypertension); history of DSM-IV psychiatric disorder or suicide risk score >1 on the MINI, or current major depression assessed by the MINI; current use of mood stabilizers, antipsychotics, stimulants, metformin, or medications altering CYP2A6 activity (e.g., tricyclic antidepressants). From this trial, we used 1,577 participants screened for the study (before stratification) who provided NMR, self-reported no past or current MDD, and were negative for MDD on MINI assessment.¹⁷

Measures

Demographic and Smoking-related Measures (Both Samples):

Demographic information, including race, age, sex, income, education, and BMI, was collected at baseline. Smoking-related characteristics were also assessed, including current smoking rate (cigarettes per day), the Fagerström Test for Cigarette Dependence (FTCD),¹⁸ and a breath carbon monoxide (CO) sample, measured in parts per million (ppm). Self-reported number of alcoholic drinks per week and oral contraceptive use were recorded.

Nicotine Withdrawal and Craving (Both Samples):

The Minnesota Nicotine Withdrawal Scale (MNWS)¹⁹ measured symptoms of nicotine withdrawal and the brief Questionnaire of Smoking Urges (QSU-B)²⁰ assessed cigarette craving about 2 weeks prior to initiating treatment in both samples.

Nicotine Metabolite Ratio (NMR; Both Samples):

MDD+ participants provided a 5mL saliva sample and MDD− participants provided a 10mL blood sample to determine NMR. Samples for both studies were collected at the eligibility visit and frozen until cotinine and 3′-hydroxycotinine (3-HC) were assessed by liquid chromatography-tandem mass spectrometry (LC-MS).²¹ Both saliva and plasma are valid methods for determining NMR²² and are highly correlated.²³ NMR values obtained from saliva are highly correlated with blood NMR measurements (r = 0.9) and can be used interchangeably with blood NMR.^23,24 The continuous measure of the NMR was computed by determining the ratio of 3-HC:cotinine and participants were categorized by NMR quartiles (e.g., lowest quartile equals 25% of people with the slowest nicotine metabolism, vs. highest quartile equals 25% of people with the fastest nicotine metabolism) based on Schnoll et al.²⁵ as done previously.²⁶ The NMR quartile means, medians, upper and lower limits, and sample size were: (1) 0.18, 0.20 (< 0.2591), 142; (2) 0.30, 0.30 (0.2592–0.3519) 142; (3) 0.40, 0.39 (0.352–0.466), 142; and (4) 0.63, 0.56 (> 0.466), 142.

Depression Characteristics (MDD+ Sample Only):

For the MDD+ sample, we collected information related to mental health, including current vs. past MDD, type of current psychiatric medication (those that affect CYP2A6 activity [tricyclic, SSRIs] or not [atypicals, SSNIs]), the presence of a concurrent psychiatric illness, and current depression symptoms measured by the Beck Depression Inventory (BDI-II).²⁷

Data Analysis

We used propensity score weighting to balance covariate values from the MDD+ and MDD− samples. The samples were weighted in terms of sex, race, age, baseline cigarettes per day, and BMI. Using propensity scores to balance covariates in this way has been shown to yield the same results as covariate adjustment within regression models to control for such covariates.²⁸ We used logistic regression to compare the MDD+ and MDD− samples (weighted and unweighted) in terms of demographic and smoking-related variables, and used the predicted probabilities of MDD+ and MDD− to generate the propensity scores and weights. We used descriptive statistics to characterize the sample and examined correlates of NMR within the MDD+ sample and the MDD− sample, separately. A mixed model using the weighted sample of MDD+ and MDD− subjects evaluated the effect of MDD status on NMR. We used logistic regression and predicted odds of assignment to MDD+ as a function of NMR quartile. We examined these models with the total weighted samples and with MDD+ participants who were taking tricyclics or SSRIs removed from the analyses or with only those taking these medications to assess the potential effects of anti-depressant medications on NMR. Lastly, we conducted separate multiple linear regression models using the weighted samples to examine the effects of MDD status and NMR (and the interaction) on nicotine withdrawal and craving; these analyses used a subset of the MDD− sample who had completed these measures (n = 1,308). All analyses were conducted using STATA (StataCorp, College Station, Texas).

Results

Sample Characteristics and Correlates of NMR for MDD+ and MDD− Smokers

Demographic, smoking-related, alcohol use, oral contraceptive use, and mental health-related characteristics are presented in Table 1. About half of the MDD+ sample had current only or current and past MDD, about one-quarter were taking medications for MDD (53 [19.1%] were taking an SSRI or tricyclic), and close to 17% reported one or more psychiatric diagnoses other than MDD. In the unweighted sample, MDD+ smokers, as compared with MDD− smokers, were more likely to be female and African American, were younger, had lower family income, reported less oral contraceptive use, and reported smoking fewer cigarettes per day (p’s < 0.05; see Table 1). In the weighted sample, as compared with MDD− smokers, MDD+ smokers had a higher FTCD (p = 0.0006) and lower oral contraceptive use. Table 2 shows the results of correlational analyses for NMR with the MDD+ sample and the MDD− sample separately. For the MDD+ sample, African American participants had lower mean NMR, while older smokers and smokers with lower educational attainment had higher mean NMR (p’s < 0.05). For the MDD− sample, NMR was higher for females and smokers with higher educational attainment, and lower for African Americans, and NMR was positively correlated with age, alcohol use, and cigarettes per day, and negatively correlated with BMI (p’s < 0.05). For both MDD+ and MDD− samples, women who reported using oral contraceptives had similar NMR compared to women who reported not using oral contraceptives (p’s > 0.05; see Table 2).

Table 1.

Sample Characteristics and Comparisons between Samples

Variable	MDD− (Unweighted)	MDD− (Weighted)	MDD+ (Unweighted)	MDD+ (Weighted)	p-value a
Demographic variables
Female sex (N, %)	670, 42.5%b	383, 44.2%	152, 54.9%b	445, 45.1%	0.85
African American race (N, %)	640, 40.6%b	377, 43.5%	168, 61.1%b	456, 46.5%	0.57
Education HS/GED or less (N, %)	473, 30%	261, 30.1%	87, 31.4%	274, 27.8%	0.57
Income ≤ $20,000 (N, %)	405, 25.7%b	230, 26.5%	300, 30.9%b	101, 36.9%	0.28
Age, years (M, SD)	45.3, 0.29b	45.9, 0.28	49.9, 0.69b	45.8 (1.1)	0.93
Body Mass Index (M, SD)	28.8, 0.18b	28.9, 0.16	30.1, 0.43b	28.4, 0.45	0.30
Alcohol Use (Weekly; M, SD)	3.3, 0.13	3.4, 0.13	3.8, 0.49	3.5, 0.32	0.32
Oral Contraceptive Use (N, % - Female Only)	521, 77.8%b	299, 77.9%	4, 2.7%b	12, 2.8%	0.0001
Smoking-related variables
Cigarettes per day (M, SD)	18.4, 0.18b	18.0, 0.17	15.1, 0.44b	22.1, 2.98	0.17
Nicotine Dependence (FTND; M, SD)	5.2, 0.05	5.2, 0.05	5.2, 0.12	5.8, 0.17	0.0006
Cotinine, ng/ml (M, SD)	247.7, 3.2	248.5, 3.1	243.1, 7.6	269.0, 15.3	0.19
3-HC, ng/ml (M, SD)	89.1, 1.5	88.9, 1.4	85.2, 3.6	93.6, 6.2	0.45
NMR, ng/ml (M, SD)	0.38, 0.01	0.38, 0.01	0.36, 0.01	0.36, 0.01	0.13
MDD characteristics (MDD+ sample only)
MDD Status (N, %)
Current			27, 9.7%
Past			143, 51.3%
Current and Past			109, 39.1%
MDD Medication (N, %)
None			202, 72.9%
SSRI/Tricyclic			53, 19.1%
SSNI/Atypical			22, 7.9%
Comorbid Psychiatric Diagnosis (N, %)
Yes			47, 16.8%
No			232, 83.1%
Depression Symptoms (M, SD)			18.8, 11.5

Note.

^acompares weighted MDD+ to weighted MDD−;

^bcompares unweighted MDD+ vs. unweighted MDD− (p < 0.05).

Table 2.

Associations between NMR and Demographic Characteristics and Smoking-related Variables among MDD+ Participants

Variable	M, SD or Pearson’s r
Sex	MDD+	MDD−
Males	0.34, 0.17	0.36, 0.19
Females	0.39, 0.27	0.41, 0.23**
Race
African-American	0.33, 0.25	0.32, 0.20
Not African-American	0.40, 0.20**	0.41, 0.20**
Education
HS/GED or less	0.32, 0.18	0.35, 0.19
Some college or more	0.38, 0.25*	0.39, 0.21**
Income
≤ $20,000	0.37, 0.27	0.37, 0.22
> $20,000	0.36, 0.20	0.39, 0.20
MDD Status
Current	0.35, 0.17	N/A
Past	0.37, 0.22
Current and Past	0.36, 0.25
MDD Medication
None	0.35, 0.22	N/A
SSRI/Tricyclic	0.40, 0.28
SSNI/Atypical	0.37, 0.20
Comorbid Psychiatric Diagnosis
Yes	0.34, 0.25	N/A
No	0.37, 0.23
Age	0.15*	0.11**
Body Mass Index	0.16	‒0.11**
Cigarettes per day	0.09	0.09**
Nicotine Dependence	‒0.001	‒0.03
Carbon Monoxide	0.02	0.04
Depression Symptoms	‒0.02	N/A
Alcohol Use	0.04	0.11**
Oral Contraceptive Use
Yes	0.29, 0.06	0.41, 0.01
No	0.39, 0.02	0.40, 0.02

Note.

^*p < 0.05

^**p < 0.01

Group Differences in NMR by MDD Status

Using the continuous measure of NMR across the weighted samples, NMR was similar between MDD+ and MDD− samples (β = −0.02, 95% CI: −0.05 to 0.01, p = 0.13); the results were the same when using the log of NMR to minimize the possible effect from non-normal distribution of the NMR. Likewise, in a logistic regression model, there were no differences in the distribution of MDD+ vs. MDD− smokers across NMR quartiles (OR = 0.76, 95% CI: 0.50 to 1.16, p = 0.21; see Figure 1). When MDD+ participants who were taking either a SSRI or a tricyclic antidepressant (n = 53) were removed from analyses, the results were similar (OR = 0.81, 95% CI: 0.52 to 1.27, p = 0.36). Likewise, when we included only MDD+ participants taking either a SSRI or a tricyclic antidepressant, the results were similar (OR = 0.49, 95% CI: 0.18 to 1.31, p = 0.16).

Figure 1.

Group Differences in Nicotine Withdrawal and Craving

Table 3 shows the results of the regression models predicting withdrawal and craving. Compared to MDD− smokers (M = 6.8; SD = 4.5), MDD+ smokers (M = 13.2; SD = 7.4) reported significantly greater withdrawal symptoms (OR = 6.16, 95% CI: 4.77 to 7.55, p < 0.001). Likewise, compared to MDD− smokers (M = 30.1; SD = 14.8), MDD+ smokers (M = 38.1; SD = 16.8) reported significantly greater craving (OR = 6.75, 95% CI: 3.42 to 10.09, p < 0.001). NMR and the NMR x MDD status interaction were not significantly correlated with either measure, even when using the log NMR.

Table 3.

Regression Models for Withdrawal and Craving

Model/Variable	Β	95% CI	p-value
Withdrawal
MDD status (Reference: MDD−)	6.16	4.77 to 7.55	<0.001
NMR	‒0.41	‒3.82 to 3.00	0.81
Sex (Reference: Male)	2.22	0.32 to 4.13	0.02
Education (Reference: HS/GED or less)	0.73	‒1.94 to 2.4	0.39
Nicotine Dependence	0.37	0.01 to 0.73	0.64
BMI	0.02	‒0.07 to 0.12	0.64
Craving
MDD status (Reference: MDD−)	6.75	3.41 to 10.09	<0.001
NMR	0.79	‒6.07 to 7.66	0.82
Sex (Reference: Male)	‒0.30	‒4.84 to 4.24	0.90
Education (Reference: HS/GED or less)	‒2.28	‒5.81 to 1.24	0.21
Nicotine Dependence	2.78	2.14 to 3.41	<0.001
BMI	0.26	0.002 to 0.52	0.05

Note. Models use weighted MDD+ and MDD− samples; MDD status remained as a significant predictor of withdrawal in a separate model where only affective items of the withdrawal scale were included (i.e., angry, anxious, depressed, restless, and impatient; β = 2.05, 95% CI: 1.46 to 2.64, p < .001)

Discussion

We utilized baseline data on NMR and common measures (e.g., nicotine withdrawal and craving) from two large clinical trials to explore, for the first time, variability in the rate of nicotine metabolism as one potential explanation for the population-level difference in the prevalence of smoking and quit rate between MDD+ and MDD− individuals. Contrary to our expectation, the rates of nicotine metabolism across the groups were similar, even when considering different antidepressant medications that may influence nicotine metabolism. However, while there was no association with NMR, consistent with the literature, smokers with MDD exhibited significantly higher levels of nicotine withdrawal and craving, which are widely recognized as important determinants of the ability to quit smoking.²⁹

Although our propensity weighting procedure balanced factors across the samples that can influence NMR, this weighting adjusts for differences in a non-parametric manner; there can still be a parametric effect of a covariate. The MDD+ sample was significantly more likely to be African American, who in this sample and in past studies,³⁰ are more likely to be slow metabolizers of nicotine. Likewise, MDD+ smokers in the present study reported smoking fewer cigarettes per day, compared to MDD− smokers, which has also been associated with slower nicotine metabolism,⁷ although this may have also been associated with more liberal inclusion criteria regarding smoking rate in the MDD+ sample. Lastly, MDD+ smokers were significantly older than MDD− smokers, and increased age has been associated with slower nicotine clearance.³¹ Taken together, differences in race, cigarettes per day, and age between the samples, which affect NMR, should be recognized.

In contrast, we found that MDD+ smokers report significantly greater levels of nicotine withdrawal and craving prior to treatment initiation. While few studies have compared smokers with or without MDD on key variables like withdrawal and craving, this result is consistent with an early epidemiologic study¹⁴ and a later twin study¹⁵ that observed MDD status to be associated with greater withdrawal severity. Nicotine withdrawal and craving are widely recognized as core features of nicotine dependence²⁹ and are strongly associated with the ability to quit smoking and avoid relapse.³² Affective features of nicotine dependence may assume particular salience for MDD+ smokers given that nicotine can exert antidepressant like effects, with periods of abstinence exacerbating the absence of a critical source of negative reinforcement.^33,34 Since measures of nicotine withdrawal and craving include items that overlap with depression symptoms (e.g., depressed mood, smoking would make me less depressed), these results may reflect proxies for greater use of tobacco among those with MDD to alleviate depression symptoms compared to those without MDD; in fact MDD+ smokers reported increased withdrawal symptoms when withdrawal was defined as only affective items from the withdrawal scale; see Table 3). As such, ensuring that MDD+ smokers attempting to quit receive treatments that can effectively mitigate withdrawal and craving, including FDA-approved medications, is critical.

These results should be considered in the context of study limitations. First, the data from both samples were from two separate randomized controlled trials, which each had inclusion and exclusion criteria that limit the generalizability of the results. Second, we were limited to examining variables that were measured in the same way across both trial samples. In particular, we were unable to control for current depressive symptoms across the samples or evaluate current depressive symptoms as a potential factor that differed across samples as we did with withdrawal and craving. Likewise, we were limited to assessing baseline data because there were differences in the timing and nature of the treatments provided across the two studies. This limitation is particularly relevant for interpreting the results concerning withdrawal and craving since baseline assessments of these variables do not reflect a state of nicotine abstinence, and the observed group differences may reflect broader affective differences (e.g., negative affectivity). Third, although NMR assessed by saliva and blood are highly correlated and can be used to compare NMRs^23,24, and the same laboratory analyzed NMR samples for both studies, there may be some variability across these biosamples in measuring individual nicotine metabolism and this approach was not optimal. Fourth, given that the MDD+ community sample was comprised of individuals with current and/or past MDD and that almost three-quarters were not currently taking anti-depressant medications, the generalizability of the MDD+ sample to the broader population of those with MDD may be limited. Lastly, the data from this study are non-randomized so no causal interpretations of the findings should be made. Overall, the present results should be considered preliminary and in need of further study using prospective data to control for potential confounding variables.

Nevertheless, this is the first study to compare NMR values across smokers that differ in terms of current or a past MDD and has done so with relatively large samples with well-characterized mental health histories. Variability in NMR may not account for the significant differences between MDD+ and MDD− smokers in terms of rates of smoking, but our study highlights the elevated rates of nicotine withdrawal and craving in this MDD+ under-served sub-population of smokers. Indeed, elevated nicotine withdrawal and craving among smokers with MDD has previously been highlighted as key factors that need to be addressed when designing smoking cessation treatments for this sub-group of smokers.³⁵ Doing so, may help reduce important disparities among smokers with past and current MDD.