The Nicotine Metabolite Ratio is Associated With Early Smoking Abstinence Even After Controlling for Factors That Influence the Nicotine Metabolite Ratio

Abstract

Introduction:

The decrease in smoking rates in North America has plateaued, underscoring the need for new approaches to treat nicotine dependence. Inter-individual differences in smoking behavior result, in part, from variation in the rate of CYP2A6-mediated nicotine metabolism. A phenotypic measure of CYP2A6 activity is the nicotine metabolite ratio (NMR), the ratio of 3′hydroxycotinine/cotinine. The NMR is associated with smoking cessation. However, the NMR is also associated with genetic (eg, CYP2A6 genotype) and other (eg, sex and ethnicity) factors. Here we aimed to determine if previously identified non-CYP2A6 sources of variation in the NMR mitigated the association between the NMR and short-term abstinence.

Methods:

The NMR was determined from blood samples collected at intake from daily smokers aged 18–65. Biochemically-verified point prevalence abstinence (exhaled carbon monoxide level ≤ 8 ppm) was measured at 1 week following the target quit date in participants from a smoking cessation clinical trial (NCT01314001). Analyses were restricted to N = 462 blacks and N = 693 whites in the intent-to-treat sample.

Results:

Lower NMR (<0.31) was associated with a higher likelihood of 1-week abstinence (OR = 1.43; 95% CI = 1.12, 1.84). NMR was associated with abstinence even after controlling for treatment arm (nicotine patch or varenicline) and factors previously associated with NMR variation including sex, ethnicity, estrogen-containing hormonal therapy, body mass index, alcohol, and cigarette consumption.

Conclusions:

NMR was associated with 1-week smoking abstinence; NMR may be a useful addition to medication screening approaches evaluating treatments for nicotine dependence.

Introduction

Smoking prevalence has stabilized at ~20% in North America,^1,2 suggesting that the current efforts to reduce smoking initiation and/or increase cessation have plateaued. The Food and Drug Administration has approved three medications (ie, nicotine replacement therapy, bupropion, and varenicline) to treat nicotine dependence; each with modest success (reviewed in ref.³). Bringing a new drug to market typically takes up to 15 years and costs approximately $2 billion.⁴ Despite these extensive costs, over one-third of drugs fail in Phase III trials.⁴ New approaches that incorporate short-term efficacy screening of medications in early phase II may assist in selecting only those compounds with adequate efficacy for further development.⁵ While short-term lab-based studies have been used to assess the ability of new compounds to reduce withdrawal, craving and/or drug reinforcement, they do not always accurately predict cessation outcomes in phase III.⁴

A new screening strategy in smoking cessation drug development has recently been validated using existing smoking cessation drugs,⁶ which assesses short-term abstinence on the medication(s) of interest relative to an existing drug or placebo.^4,5 Successful short-term abstinence increases the likelihood of prolonged cessation,³ suggesting short-term screening approaches may serve as useful predictive models of both medication efficacy and long-term abstinence.

Nicotine, the principal psychoactive compound in cigarette smoke, is inactivated primarily by CYP2A6 to cotinine.⁷ CYP2A6 further converts cotinine to 3′hydroxycotinine,⁸ and the ratio of 3′hydroxycotinine/cotinine is known as the nicotine metabolite ratio (NMR).⁹ NMR, which is stable over time in smokers,¹⁰ is strongly associated with CYP2A6 genotype and nicotine clearance,⁹ and is a genetically-informed biomarker of CYP2A6 activity incorporating environmental influences on nicotine clearance.¹¹ NMR is associated with a variety of smoking behaviors, including cessation^12–14 and early symptoms of withdrawal and craving.¹⁴ Lower NMR, indicative of slower nicotine clearance, is associated with lower cigarette consumption,¹⁵ dependence scores,¹⁶ nicotine-mediated reward,¹⁷ as well as higher quit rates on placebo¹² and nicotine patch.^13,14 In an NMR-stratified smoking cessation clinical trial involving nicotine patch and varenicline, varenicline displayed a higher efficacy relative to patch in those with higher NMR, but quit rates were similar across treatments in those with lower NMR.¹⁸ Treatment with varenicline (vs. placebo) was also associated with greater overall side-effect severity among those with lower NMR.¹⁸ In these clinical trial participants, we previously quantified the effect of sex, ethnicity, estrogen-based hormonal therapy, body mass index, alcohol use, and mentholated cigarette use on baseline NMR.¹⁵ Our current objectives were to determine if NMR is associated with 1-week abstinence following the target quit date, and if controlling for factors associated with baseline NMR¹⁵ mitigate the association between NMR and 1-week abstinence. The incorporation of pharmacogenetic information into early screening approaches, especially factors such as NMR that influence cessation outcomes, may identify new medications for nicotine dependence that work better in some subgroups rather than others.

Methods

Study Participants

Adults (aged 18–65 years) smoking at least 10 cigarettes/d for the previous 6 months were recruited for participation in a smoking cessation clinical trial (NCT01314001). Participants were prospectively randomized to placebo, nicotine patch, or varenicline according to their NMR, assessed at baseline in those interested in participating in the clinical trial after meeting eligibility criteria. Detailed study procedures including trial inclusion/exclusion criteria, as well as participant characteristics, are provided elsewhere.^15,16,18 A flow chart depicting recruitment to each arm of the study and exclusions at each stage is published.¹⁸ The baseline NMR was assessed from blood while participants were smoking as usual, by liquid chromatography-tandem mass spectrometry.¹⁵ A clinical NMR cut-point was used to distinguish faster (≥0.31) from slower (<0.31) metabolizers based on previous clinical trial data as described¹⁸; participants were randomized by treatment site and NMR in a 1:1:1 ratio to treatment arm.¹⁸ Point prevalence abstinence at 1 week following the target quit date was biochemically verified using exhaled carbon monoxide. Those with carbon monoxide ≤ 8 ppm were considered abstinent, while those with carbon monoxide > 8 ppm were classified as smoking.

Statistical Analyses

Mann–Whitney U (two-tailed) and chi-square tests were used to compare continuous and dichotomous characteristics, respectively, between abstainers and non-abstainers at week 1. We used multiple logistic regression analysis to examine the relationship between NMR and 1-week abstinence, after controlling for potential confounding effects of covariates. Covariates included treatment arm and factors previously associated with NMR variation in this study population,¹⁵ as well as treatment site. Dummy-coding was used to control for treatment arm, where the impact of varenicline and nicotine patch on 1-week abstinence was assessed compared with placebo. Statistical analyses were performed using SPSS Version 22 (IBM Corporation).

Results

A total of 1246 comprised the intent-to-treat group for the clinical trial (NCT01314001). We restricted all further analyses to white (N = 693) and black (N = 462) participants in the intent-to-treat group, as the small number of individuals from additional ethnic groups precluded statistical analyses.¹⁵ Of these, 997 (86.3%) completed the week-1 assessment. The remaining 158 participants that did not complete the week-1 assessment were assumed to be smoking and were included in the “non-abstinent” group, as is standard for an intent-to-treat analysis.³ NMR, treatment, and other characteristics according to abstinence status at 1 week are shown in Table 1. Similar results were obtained when these participants were excluded from the analysis (Supplementary Tables 1 and 2).

Table 1.

NMR, Treatment, and Factors Associated With Nicotine Metabolite Ratio (NMR), According to Abstinence Status at 1 Week in Intent-To-Treat (ITT) Group (N = 1155)

	Abstinentb (N = 588)	Not abstinentc (N = 567)	P
NMR
% Slow metabolizers, NMR < 0.31 (ref. normal metabolizers, NMR ≥ 0.31)	56.1	49.2	.019
Treatment arm
% On patch or varenicline (ref. placebo)	75.7	58.9	<.001
Factors associated with NMR
% White (ref. black)	63.6	56.3	.01
% Female (ref. male)	47.3	42.0	.07
% Using estrogen therapyd (ref. no estrogen therapy)	4.3	4.2	.95
Mean BMI (SD)	29.9 (6.8)	29.8 (6.6)	.99a
Mean number of drinks/wk (SD)	3.6 (5.4)	2.8 (4.9)	.005 a
Mean cigarettes per day (SD)	17.5 (6.4)	19.0 (7.6)	<.001 a
% Using mentholated cigarettes (ref. nonmenthol cigarettes)	45.1	50.6	.059
Additional demographic factors
% College or higher	71.1	65.8	.052
% Income > $50 000	41.5	30.7	<.001
% Single	56.8	61.2	.13

BMI = body mass index; CO = carbon monoxide. The bolded P values are those that reached significance at the .05 threshold. P values are derived from chi-square tests unless otherwise indicated.

^a P values derived from Mann–Whitney U tests.

^bIncludes those who completed the week-1 assessment and were abstinent (ie, CO ≤ 8 ppm).

^cIncludes those who were in the ITT group and completed the week-1 assessment and were not abstinent (ie, CO > 8 ppm), or did not complete the week-1 assessment (and were therefore assumed to be smoking).

^dIncludes the use of estrogen-containing birth control pills and hormone replacement therapy; assessed in women only.

We first examined the association between NMR group and 1-week abstinence; compared to normal metabolizers, slow metabolizers were significantly more likely to achieve abstinence (odds ratio [OR] = 1.32, 95% confidence interval [CI] = 1.05, 1.67; P = .019). Nicotine patch and varenicline, relative to placebo, significantly influenced 1-week abstinence with ORs (95% CI) of 1.80 (1.35, 2.41) and 2.60 (1.94, 3.48), respectively (P < .001 vs. placebo). The association between slow metabolism and increased abstinence likelihood did not change (OR = 1.32; P = .022) when we controlled for treatment.

In a final logistic regression model, we included treatment, NMR, and all variables previously associated with NMR in this population, as independent variables.¹⁵ After controlling for these factors, NMR was significantly associated with 1-week abstinence (OR for abstinence in slow vs. normal metabolizers = 1.43; Table 2). In addition, white (vs. black) ethnicity, higher baseline alcohol consumption, and lower baseline cigarette consumption were all significantly associated with abstinence (Table 2), even while controlling for treatment and NMR. Including treatment site as a covariate in the logistic regression models did not alter any of these relationships (data not shown). In 2×2 models, there was no interaction between NMR and ethnicity (OR = 0.84, 95% CI = 0.52, 1.38; P = .50), or between NMR and sex (OR = 0.65, 95% CI = 0.40, 1.03; P = .067), on 1-week abstinence, suggesting similar associations between NMR and abstinence in whites and blacks, and in men and women.

Table 2.

Association Between Nicotine Metabolite Ratio (NMR) and 1-Week Abstinence (CO ≤ 8 ppm) After Controlling for Treatment and Factors Associated With NMR Variation in the Intent-To-Treat (ITT) Group (N = 1155)

Predictor	OR	95% CI	P
Treatment arm
Nicotine patch (ref. placebo)	1.80	1.34, 2.41	<.001
Varenicline (ref. placebo)	2.58	1.92, 3.47	<.001
Factors associated with NMR
Female sex (ref. males)	1.28	1.00, 1.65	.054
White ethnicity (ref. blacks)	1.56	1.13, 2.16	.007
Estrogen therapya (ref. no estrogen therapy)	1.00	0.41, 2.43	.99
BMI (continuous)	1.01	1.00, 1.03	.16
Standard alcoholic drinks per week (continuous)	1.03	1.01, 1.06	.009
Cigarettes per day (continuous)	0.97	0.95, 0.99	<.001
Menthol cigarettes (ref. non-menthol cigarettes)	0.89	0.66, 1.21	.46
NMR
Slow metabolism, NMR < 0.31 (ref. NMR ≥ 0.31)	1.43	1.12, 1.84	.004

BMI = body mass index; CI = confidence interval; CO = carbon monoxide; OR = odds ratio. The bolded P values are those that reached significance at the .05 threshold.

^aIncludes the use of estrogen-containing birth control pills and hormone replacement therapy; women not taking estrogen therapy, as well as men, were coded as “0”, whereas women taking estrogen therapy were coded as “1”.

Discussion

One-week abstinence was strongly associated with end-of-treatment abstinence (OR = 9.6, 95% CI = 6.7, 13.8; P < .001) consistent with it being a good predictor of long-term abstinence.³ At 1 week, lower NMR was associated with an increased likelihood of abstinence, as had been observed at later time-points (ie, end-of-treatment and 6 months),^12–14,19 even after controlling for factors that influence NMR.¹⁵ The relationship between NMR and early abstinence also remained significant when a median split on body mass index, alcohol consumption, and cigarette consumption was performed (Supplementary Table 3). Thus, it may be worthwhile to incorporate NMR, a genetically-informed biomarker of cessation,¹¹ into short-term efficacy screening approaches during the development of novel smoking cessation medications. These studies together provide support for the development of a test kit for assessing NMR for future use in clinical settings.

In addition to NMR and treatment arm, our overall model suggests a direct influence of ethnicity, alcohol consumption, and cigarette consumption on 1-week abstinence. Sex had a modest influence on abstinence, which varied depending on whether body mass index, alcohol consumption, and cigarette consumption were included as continuous measures (OR = 1.28, P = .054, Table 2) or as categorical variables (OR = 1.31, P = .034, Supplementary Table 3). Compared with men, women may display an enhanced ability to maintain short-term abstinence, when nicotine withdrawal symptoms are strongest, as women appear to smoke more for reasons unrelated to nicotine.²⁰ However, women generally have lower cessation rates in studies evaluating longer-term abstinence.^21,22 Taken together, each of these factors would likely need to be considered when tailoring smoking cessation strategies, but as separate entities from the NMR. Consistent with prior findings on smoking cessation, we observed a greater likelihood of abstinence among whites relative to blacks. Relative to whites, black smokers display lower quit ratios (ie, a lower proportion of former smokers among ever smokers²³). In our clinical trial population, more than 85% of black smokers used mentholated cigarettes, compared to less than 25% of white smokers.¹⁵ The use of mentholated cigarettes was not associated with 1-week abstinence in our study or at end-of-treatment.¹⁸ Mentholated cigarette use has not been consistently associated with a reduced likelihood of cessation or with an increased risk for lung cancer,²⁴ however this remains to be investigated further.

Alcohol consumption (number of standard alcoholic drinks/wk) was positively associated with 1-week abstinence, with the caveat that self-reported alcohol consumption was measured at intake and may not accurately reflect the level of alcohol consumption 1 week following the target quit date. Follow-up of alternative drug usage during 1-week abstinence studies might help clarify if there are compensatory increases in consumption levels over the course of a smoking cessation attempt. A previous cohort study in community-based adult smokers showed a negative influence of daily alcohol consumption on self-reported smoking cessation 5 years later,²¹ suggesting that this level of alcohol consumption impedes attempts to quit smoking. The association with the level of alcohol was weak, with abstainers consuming one more standard drink per week relative to non-abstainers (~4 vs. ~3 drinks/wk). The low level of consumption likely reflects clinical trial exclusion criteria (consumption of more than 25 standard drinks/wk).

Consistent with previous findings,²¹ lower baseline cigarette consumption was associated with a higher likelihood of abstinence in our study. In these clinical trial participants, we previously showed that lower NMR was associated with lower cigarette consumption.¹⁵ In our overall model (Table 2), both baseline cigarette consumption and NMR were associated with abstinence, suggesting at least part of the effects were independent of one another.

Overall, we showed that NMR, which is associated with treatment outcome at end-of-treatment and 6-month follow-up,¹⁸ was also associated with 1-week abstinence, and factors which account for variability in NMR did not remove the association between NMR and 1-week abstinence. This early period of abstinence represents a period of heightened vulnerability during the process of smoking cessation, and NMR may be useful in informing early efficacy screening approaches for compounds undergoing development. Screening approaches may include NMR-based randomization to treatment, as we have done in this clinical trial,¹⁸ or simply including NMR in analytic models as a covariate known to alter smoking cessation and treatment response.

Supplementary Material

Supplementary Tables 1–3 can be found online at http://www.ntr.oxfordjournals.org

Funding

This work was supported by the Endowed Chair in Addictions (RFT), Canadian Institutes of Health Research (CIHR)-CGSD and Ontario Graduate Scholarship (MJC), National Institutes of Health (NIH) PGRN grant DA020830 (to RFT and CL), CIHR (grants MOP86471 and TMH-109787 to RFT), the Campbell Family Mental Health Research Institute of CAMH, the CAMH Foundation, the Canada Foundation for Innovation (#20289 and #16014 to RFT and TPG) and the Ontario Ministry of Research and Innovation.

Declaration of Interests

TPG has consulted for Novartis. RFT has consulted for Apotex and McNeil. RAS has consulted for GlaxoSmithKline. LWH Jr has consulted on investigator-initiated smoking cessation studies funded by Pfizer and by the state of Florida. CL and RAS have received medication and placebo from Pfizer. CL, TPG, and PMC have received research funding from Pfizer. The remaining authors declare no conflicts of interest.