Nicotine Metabolism Predicted by CYP2A6 Genotypes in Relation to Smoking Cessation: A Systematic Review

Abstract

Introduction

Identifying genetic factors associated with smoking cessation could inform precision cessation interventions. Of major interest is genetic variation in nicotine metabolism, largely predicted by CYP2A6 variations.

Aims and Methods

We conducted a systematic literature review to summarize the population-based evidence of the association between CYP2A6 and smoking cessation. In the 12 studies meeting the inclusion criteria, the known functional metabolic effect of CYP2A6 variants was used to classify nicotine metabolism as normal (>75% metabolic activity), intermediate (50.1%–75% activity), slow (25%–50% activity), and poor (<25% activity). Summary odds ratios of smoking cessation were calculated across metabolic groups, stratified by ancestry and whether participants received pharmacotherapy or placebo/no treatment.

Results

Among untreated people of European ancestry (n = 4 studies), those with CYP2A6 reduced metabolism were more likely to quit smoking than those with normal metabolism (Summary OR = 2.05, 95% CI 1.23 to 3.42) and the likelihood of cessation increased as nicotine metabolism decreased. Nicotine replacement therapy attenuated the association at end-of-treatment, while bupropion modified the association such that intermediate/slow metabolizers were less likely to quit than normal metabolizers (Summary OR = 0.86, 95% CI 0.79 to 0.94). Among untreated Asian people (n = 3 studies), results differed compared with those with European ancestry: those with slow metabolism were less likely to have quit smoking than normal metabolizers (Summary OR = 0.52, 95% CI 0.38 to 0.71). Evidence for people of African ancestry (n = 1 study) suggested the CYP2A6 association with cessation may differ compared with those of European ancestry.

Conclusions and Implications

Most studies included in this review were of European ancestry populations; these showed slower nicotine metabolism was associated with increased likelihood of smoking cessation in a dose-related manner. Pharmacotherapy appeared to attenuate or modify this association among people of European ancestry, but it is unclear whether the change in the association remains consistent after treatment ceases. This finding has implications for precision medicine cessation interventions. Based on only a few studies of people of Asian or African ancestry, the association between CYP2A6 variants and cessation may differ from that observed among those of European ancestry, but more evidence is needed.

Introduction

Twin studies suggest nicotine dependence, withdrawal symptoms, and smoking behaviors are between 31% and 60% heritable^1–4 providing evidence of underlying genetic risk factors for cessation failure or relapse. Identifying these risk factors can advance the understanding of precision medicine approaches to smoking cessation. A pathway of major interest for understanding genetic susceptibility to smoking addiction is nicotine metabolism, a key factor influencing nicotine addiction. Most (75%) nicotine is metabolized to cotinine by the cytochrome P450 2A6 (CYP2A6) enzyme⁵ which is encoded by CYP2A6. Cotinine is also metabolized by CYP2A6 to 3′-hydroxycotinine (3-HC), and the ratio of 3-HC to cotinine, referred to as the nicotine metabolite ratio (NMR), is a biomarker of CYP2A6 enzyme activity and of nicotine metabolism rate, with higher ratios indicative of faster nicotine clearance/metabolism. Variations in nicotine metabolism have been attributed in part to CYP2A6 variants⁶ which predict enzyme activity and resulting nicotine metabolite ratio (NMR) levels.⁷ Of note, CYP2A6 variant allele frequencies can vary by ancestral background.⁸ The interplay of nicotine metabolism with the nicotinic acetylcholine receptor (nAChR) pathway plays a large role in the biochemical response to nicotine.

Through coaction with nAChR binding and activation, the rate of nicotine metabolism contributes to the onset of and severity of withdrawal symptoms and cravings.^9,10 Nicotine inhaled through tobacco smoke is absorbed quickly into brain tissue¹¹ where it binds to and activates nAChRs, a family of multisubunit neurotransmitter receptors which typically bind endogenous acetylcholine.¹² Faster nicotine metabolism reduces serum nicotine levels more quickly freeing nAChRs for binding,⁹ and could contribute to craving and withdrawal symptoms^10,13 if homeostasis is not maintained through continued nicotine intake. Thus, smokers with faster nicotine metabolism may need to smoke more cigarettes per day (CPD) to maintain stable serum nicotine levels.¹⁴

Variation in the rate of nicotine metabolism influences smoking behavior.¹⁵ Earlier evidence reviews suggested that slower versus faster nicotine metabolism, as measured by the nicotine metabolite ratio (NMR), is associated with smoking fewer CPD, higher quit rates on nicotine replacement therapy (NRT), and similar quit rates on bupropion.^16,17 A recent (2019) systematic review examined evidence of the association between NMR and smoking cessation from two randomized controlled trials again showing NRT may be more effective for those with slow nicotine metabolism.¹⁸ These reviews demonstrate the importance of understanding the relationship between nicotine metabolism and smoking cessation.

Though NMR is influenced by genetic, hormonal, and environmental factors,¹⁹ evidence shows CYP2A6 variants are the strongest predictors of nicotine metabolism.^7,20 Thus, identifying CYP2A6 associations with smoking cessation is a promising line of inquiry with potential relevance to precision medicine strategies in designing and implementing interventions to increase the rates of successful long-term smoking cessation. For these reasons, the objective of this investigation was to perform a systematic review of the epidemiological evidence of CYP2A6 variants in relation to smoking cessation or smoking relapse.

Methods

This systematic review addressed the question “Are CYP2A6 variants associated with smoking cessation in smokers?” The independent variable was thus defined as variants in CYP2A6. The dependent variable was smoking cessation measured as abstinence from smoking either continuously or at a point in time.

Inclusion Criteria

To be eligible for inclusion, a study had to be in humans and present data for at least one CYP2A6 variant in relation to smoking cessation measured as either continuous or point abstinence, number of quit attempts, or time to cessation. Studies with prospective follow-up were included comprising randomized controlled trials, open label trials, cohort studies, nested case-control studies, and studies that analyzed pooled data. Cross-sectional studies relying on a single measurement of smoking status (former vs. current) were also included. Only studies published in peer-reviewed journals and in English were included.

Exclusion Criteria

Articles were excluded if they met one or more of the following criteria: (1) did not have a smoking cessation or relapse outcome, (2) not published in English, (3) did not test a genetic exposure, (4) nonhuman subjects, (5) news, commentary, or letter to the editor, (6) review paper, (7) heritability study such as twin/family studies or linkage studies, and (8) studies evaluating notification of genetic susceptibility to disease on motivation to quit smoking.

Literature Search

A search of both the PubMed and Scopus databases was conducted for evidence of the association between CYP2A6 variants and smoking cessation (Figure 1). These two databases were searched for journal articles published between January 1990 and February 7, 2020. An alert was set to provide notification of articles meeting the search criteria that were published online after February 7, 2020 through October 31, 2020. The PubMed database was searched using MESH search terms “tobacco use cessation” OR “smoking cessation” AND “genetic variation” OR “genetics” OR “precision medicine.” The Scopus database was searched for article titles with the terms “gene*” OR “snp” OR “polymorphism*” OR “telomer*” OR “mutation*” OR “precision medicine” and was then further limited to titles that also included the terms “smoking” OR “tobacco” OR “cessation” OR “quit” OR “nicotine” OR “cigarette” AND the terms “cessation” OR “quit” OR “quitting” OR “stop” OR “stopping” OR “abstin*” OR “relapse” and human and English only studies.

Figure 1.

Summary of search process in systemic review of CYP2A6 variation in relation to smoking cessation in observational studies and clinical trials of smoking cessation interventions.

The PubMed search yielded 625 citations and the Scopus search yielded 353 citations, from which 137 duplicate records present in both databases were removed (Figure 1). The titles and abstracts of the remaining 841 articles were screened using the inclusion and exclusion criteria, leaving 15 articles for full-text review. Upon full-text review, 3 of the 15 remaining articles were excluded based on inclusion/exclusion criteria.

Data Extraction

The data extracted from the 12 included articles constituted effect size measures and confidence intervals, study design, sample size, study inclusion/exclusion criteria (e.g., participant ages, CPD cut-offs excluding or including light smokers), ancestry and sex of participants, and, if relevant, intervention types and protocols and any pharmacogenomic analysis. Some articles had relevant data but the comparisons presented in the published manuscript or supplementary data did not align with the methodology of other studies (e.g., results reported were stratified by genotype or metabolism phenotype versus results reported stratified by pharmacotherapy intervention). In these instances, we contacted the corresponding authors to obtain the relevant results so the studies could be included in the review.

Statistical Analysis

The methods for categorizing metabolism varied across studies.^6,21 Nine of the twelve studies categorized participants’ nicotine-to-cotinine metabolism function by the presence of variants predominantly known to impair CYP2A6 enzyme activity and tested the association of these nicotine metabolism categories with smoking cessation. The three remaining studies tested the association of individual CYP2A6 variant genotypes with cessation outcomes. To harmonize these findings that were presented differently across studies, we separately determined nicotine metabolism categories based on the known loss of function and decrease of function effects of CYP2A6 variants on metabolic activity described in Table 1. We designated the following nicotine metabolism categories: normal metabolism (NM) (>75% metabolic activity, e.g. wild type), intermediate metabolism (IM) (50.1%–75% activity), slow metabolism (SM) (25–50% activity), and poor metabolism (PM) (<25% activity). The specific CYP2A6 variants from Table 1 that were available in each study to designate these metabolic categories are listed in Table 2. When data on multiple variants were available, nicotine metabolism was categorized based on the estimated combined effect of the variants (e.g., one decrease of function variant = IM, whereas two decrease of function variants = SM). It was possible to categorize the comparison for all studies except for the study by Stevens et al. that tested a CYP2A6 variant with unknown metabolic effects.²² Metabolic categories were evaluated on their own if the study data available made it possible to distinguish each category. Some studies did not have data available to enable separating participants with slow metabolism from those with poor or intermediate metabolism; in this case, the categories were grouped together as having at least slow metabolism but possibly poor (PM/SM) or having at least intermediate metabolism but possibly slow metabolism (SM/IM).

Table 1.

CYP2A6 Alleles and Functional Effect on Nicotine Metabolism

	CYP2A6 allele	Identifying variant(s)	Genotype	Reference
Decreased function alleles reducing metabolic activity to <75% of normal (wild type)levela	CYP2A6*9	rs28399433 (possibly rs4803381)	Heterozygotes	6,20,23,24
	CYP2A6*12	Partial gene deletion	Heterozygotes	6,20,25
	CYP2A6*23	rs56256500	Heterozygotes	26
Decreased or Loss of function alleles reducing metabolic activity to <50% of normal (wild type) levela	CYP2A6*2	rs1801272	Heterozygotes	20,27,28
	CYP2A6*4	Gene deletion	Heterozygotes	20,27,28
	CYP2A6*5	rs5031017	Heterozygotes	27,28
	CYP2A6*7	rs5031016	Heterozygotes	27,29
	CYP2A6*9	rs28399433 (possibly rs4803381)	Homozygotes	6,20,23,24
	CYP2A6*10	rs5031016, rs28399468	Heterozygotes	27
	CYP2A6*12	Partial gene deletion	Homozygotes	6,20,25
	CYP2A6*17	rs28399454	Heterozygotes	26,30
	CYP2A6*20	rs28399444	Heterozygotes	31
Loss of function alleles reducing metabolic activity to <25% of normal (wild type) levela	CYP2A6*2	rs1801272	Homozygotes	20,27,28
	CYP2A6*4	Gene deletion	Homozygotes	20,27,28
	CYP2A6*5	rs5031017	Homozygotes	27,28
	CYP2A6*7	rs5031016	Homozygotes	27,29
	CYP2A6*10	rs5031016, rs28399468	Homozygotes	27
	CYP2A6*17	rs28399454	Homozygotes	26,30
	CYP2A6*20	rs28399444	Homozygotes	31
	CYP2A6*23	rs56256500	Homozygotes	26

^aThe CYP2A6 *1 allele is considered normal (wild type) level metabolism. The *1 variant includes both *1A and *1B when differentiated.

Table 2.

Description of Studies of CYP2A6 and Smoking Cessation Included in the Review

Study	Metabolism comparison	Variants used to predict metabolism	Study design	Sample size	Ancestry	Key eligibility criteria	Outcome/follow-up timepoint	Treatment
Bergen (2015)	IM vs. NM	rs4803381 and rs1137115 (reduced activity, may be associated with *9 haplotype, <70% activity)	RCT (pooled data from 8 RCT)	N = 2499 (56.1% female)	European	≥18 years old, ≥10 CPD	Point prevalent abstinence, at EOT (8–12 weeks) and 26 weeks	Placebo, bupropion, NRT, bupropion + NRT, varenicline
Chen (2014)	PM vs. IM/NM	PM: bottom 25% of CYP2A6 activity based on *2, *9, *12, and *4 variants	RCT	N = 709 (58.7% female)	European	≥18 years old, ≥10 CPD	Relapse 90 days post-cessation	Placebo, Bup, NRT, Bup + NRT
Chenoweth (2013)	SM vs. NM and IM vs. NM	IM: 1 copy of decreased function variant (*9 or *12).	Prospective cohort	N = 308 (63.7% female)	European	12–17 years old, ever a puff of a cigarette, no CPD requirement	1-year abstinence	None
		SM: 2 copies of decreased function variant, or 1 or 2 copies of loss of function variant (*2 or *4)
Ho (2009)	PM/SM vs. IM/NM	IM: 1 copy of decreased function variant (*9 or *12).	RCT	N = 588 (67.5% female)	African	≥18 years old, <10 CPD, interested in quitting within 2 weeks	Point prevalent abstinence, at EOT (8 weeks)	NRT gum/Placebo
		PM/SM: 2 copies of loss of function (*2, *4, *17, *20, *23-27, *35) or 1 copy of loss of function and 1 copy of decreased function (*9 or *12)						NRT (females only, n = 200)
Lerman (2010)	SM/IM vs NM	*2, *9, *12, *4 CYP2A6 haplotypes	RCT	N = 471 (58% male)	European	18–65 years old, ≥10 CPD	Point prevalent abstinence, at EOT (8 weeks)	Transdermal NRT
Liu (2011)	PM vs. NM, SM vs. NM, IM vs. NM	IM: 1 copy of decreased function variant (*9).	Cross-sectional	N = 1328 (92.4% male)	Asian (Han Chinese)	≥20 years old, ever smoker—100 lifetime cigarettes	Current vs. Former smoker (self-reported quit and abstinent at time of interview)	None
		SM: 2 copies of decreased function variant (*9) or 1 copy of loss of function variant (*4, *5, *7, or *10)
		PM: 2 loss of function variants
Minematsu (2003)	SM vs. NM	SM: 1 or 2 copies of *4 loss of function variant	Cross-sectional	N = 203 (93.1% male)	Asian (Japanese)	≥50 years old, ≥10 pack-years	Current vs. Former Smokers (quit for at least 6 months)	None
Ohmoto (2014)	SM vs. NM	SM: 1 or 2 copies of *4 loss of function variant	Cross-sectional	N = 96 (91.7% male)	Asian (Japanese)	>18 years old, ever smoker*	Current vs Former smoker (quit for at least 1 year)	None
Stevens (2017)	N/A	rs8192729 (unknown effect on metabolic activity)	Prospective cohort	N = 5277 (75.6% female)	European	29–65 years old, ever smoker—≥1 CPD for 1 year	9+ failed quit attempts vs. 1 quit attempt and abstinent >1 year)	None
Styn (2013)	SM/IM vs. NM	SM: 1 or more copies of *2, *4, or *9 variants	Nested case control	N = 878 (52% female)	92% European	50–79 years old, ≥10 CPD for ≥25 years	30-day abstinence at 1-year follow-up	None
Tomaz (2019)	SM vs. NM	*2 allele (rs1801272) (<50% activity)	Prospective cohort	N = 167 (57.2% female)	83% European	≥18 years old	6-month continuous abstinence	Varenicline
	IM vs. NM	*9 allele (rs28399433) (<70% activity)
Verde (2014)	SM/IM vs. NM	SM: 1 or 2 copies of *2 variant, or IM: 1 copy of *9 variant	Open label trial	N = 70 (77% women)	European (Spanish)	≥18 years old, >10 CPD, >10 pack-years	Point abstinence at 1-year post-treatment	Transdermal NRT or bupropion

NM = normal metabolism (>75%), IM = intermediate metabolism (50.1%–75%), SM = slow metabolism (25%–50%), PM = poor Metabolism (<25%). The CYP2A6 *1 allele is considered normal (wild type) level metabolism. The *1 variant includes both *1A and *1B when differentiated. EOT = end of treatment. *“Ever smoker” was not defined by Ohmoto et al.

Studies were stratified by ancestry of the study population to address differences in allelic frequencies between ancestral groups. Results were further stratified according to whether participants received a smoking cessation pharmacotherapy (e.g., NRT, varenicline, bupropion) or received placebo or no intervention. The Laird and DerSimonian estimators for the fixed effects, the log odds of abstinence, within each strata of studies were obtained using the %METAANAL SAS macro, which also tested for between-study heterogeneity.^32,33 Random effects pooling was assumed, weighting the studies proportionate to the inverse of the study variance plus the between-studies variance.

Results

Of the twelve studies identified examining CYP2A6 genetic variation and smoking cessation (Table 2), the participants were comprised of people of European or predominantly European ancestry in eight studies, Asian ancestry in three studies, and African ancestry in one study. Among the studies of European populations, four were trials of pharmaceutical interventions of bupropion, nicotine replacement therapy (NRT), or varenicline; two of the trials included a placebo arm. Inclusion criteria for these four trials were similar: adult smokers consuming 10 or more CPD. Three of the remaining studies of people of European ancestry were prospective cohort studies, two that included no pharmacotherapy, and one that included a subset of participants from a smoking assistance program cohort who reported using varenicline. These cohorts did not report using a CPD cutoff for inclusion/exclusion. The cohort using varenicline recruited adult smokers (≥18 years of age), and one other cohort also included adult smokers 29–65 years old. The third cohort was comprised of adolescents (12–17 years old). The remaining study of people of European ancestry was a nested case-control study of participants 50–79 years old who had smoked 10 or more CPD for at least 25 years. The sample sizes of these eight studies ranged from 70 to 5277. All three studies of people of Asian ancestry were predominantly male (91.7–93.1% male) and were cross-sectional studies relying on smoking status at the time of recruitment; in two of these studies the participants were of Japanese ancestry and in the other the participants were of Han Chinese ancestry. Among the studies of people of Japanese ancestry, one included only participants that were at least 50 years old who had smoked at least 10 pack-years, whereas the other study included adult ever smokers, but did not define cut-offs for determining ever smoker status. The study of people of Han Chinese ancestry included participants at least 20 years old who were ever-smokers, defined as smoking at least 100 lifetime cigarettes. The sample sizes of these three studies ranged from 96 to 1328. The remaining study of adult (≥18 years old) people of African ancestry was a randomized controlled trial of 588 smokers which tested the efficacy of NRT compared with placebo and limited inclusion to only light smokers (<10 CPD).

European Ancestry—PLACEBO/No Pharmaceutical Intervention (n = 4 Studies)

The results of studies of populations of European ancestry with no pharmacotherapy intervention showed that slower nicotine metabolism was associated with increased likelihood of smoking cessation. The results of a cohort study,³⁴ from the placebo arm of two trials,^23,35 and a nested case-control study,³⁶ showed those with at least some reduction in nicotine metabolism activity were significantly more likely to quit smoking compared with those with normal metabolism (Summary OR = 2.05, 95% CI 1.23 to 3.42) (Figure 2a). Significant heterogeneity was present across these studies, likely due to the differences in metabolism phenotype comparisons. Assessing the dose-related relationship of decreasing nicotine metabolism, the summary ORs compared with normal metabolism were 1.25 for intermediate metabolism (95% CI 0.93 to 1.68), 2.26 for slow or poor metabolism (95% CI 1.28 to 3.99), and 5.92 for poor metabolism (95% CI 2.40 to 4.63). These associations were consistent with a dose-related relationship showing increased odds of smoking cessation with slow nicotine metabolism. The dose-related comparisons eliminated the significant heterogeneity between studies. (Q statistics, I², and p-values resulting from tests for heterogeneity between studies are available in Supplementary Figure 1.)

Figure 2.

CYP2A6 results. (a) European ancestry with no pharmaceutical treatment or in a placebo arm, (b) European ancestry in pharmaceutical treatments groups, (c) Asian ancestry, (d) African ancestry (results are for women). NM = normal metabolism, IM = intermediate metabolism, SM = slow metabolism, PM = poor metabolism, EOT = end of treatment, NRT = nicotine replacement therapy. *Significant heterogeneity.

A study in the American Cancer Society’s Cancer Prevention Study 3 (CPS 3) cohort found those carrying the rs8192729 G>A minor (A) allele were more prone to relapse than non-carriers.²² This study could not be included in the meta-analysis because neither this variant’s impact on nicotine metabolism nor its association with CYP2A6 functional variants are known.

European Ancestry—Pharmaceutical Interventions (n = 5 Studies)

In studies of populations of European ancestry who were administered an NRT intervention (n = 4 studies), those with reduced metabolism (IM, SM, or PM) had greater odds of smoking cessation but the association was not statistically significant (Summary OR = 1.38, 95% CI 0.91 to 2.11) (Figure 2b).^35,37 There was significant heterogeneity among these studies due to one outlying result from Verde et al. which observed that reduced metabolizers (IM or SM) had 3.86 (95% CI 1.94 to 7.72) the odds of abstinence compared with normal metabolizers.³⁸ This outlier could be due to several factors such as the extremely small sample size (n = 36), the open label nature of the trial, or the 12 months post-treatment abstinence outcome. In sensitivity analysis with this study removed, the association was substantially attenuated (Summary OR = 1.13, 95% CI 0.94 to 1.37) and there was no longer significant heterogeneity across studies. Null findings comparing reduced metabolizers (IM or SM) to normal metabolizers were also observed among study participants receiving varenicline (n = 2 studies) (Summary OR = 0.94, 95% CI 0.73 to 1.22).^23,39 Among people of European ancestry administered bupropion (n = 2 studies), reduced metabolizers (IM or SM) were significantly less likely to quit smoking compared with normal metabolizers at end-of-treatment (Summary OR = 0.86, 95% CI 0.79 to 0.94).^23,38

Asian Ancestry—Cross-Sectional Studies (n = 3 Studies)

In contrast to findings in those of European ancestry, three studies of Asian people^40–42 suggest slow nicotine metabolism may be associated with reduced odds of abstinence in untreated populations (Summary OR = 0.52, 95% CI 0.38–0.71) (Figure 2c).

African Ancestry (n = 1 Study)

In the sole study of people of African ancestry, among participants randomized to placebo, no significant difference in odds of abstinence was found between slow and normal metabolizers (OR = 0.78, 95% CI 0.38–1.61) (Figure 2d). However, among women in the NRT arm, those with slow metabolism were more likely to be abstinent at the end of eight weeks than those with normal metabolism (OR = 2.54, 95% CI 1.28–5.02).⁴³ This association was not found in men; however, the odds ratio and specific p-value were not reported.

Discussion

Given the important role of CYP2A6 in nicotine metabolism, this systematic review aimed to synthesize the evidence of the potential role of CYP2A6 variants in smoking cessation and relapse. The bulk of the evidence stems from studies of populations of European ancestry, all of which used prospective follow-up to evaluate changes in smoking status. Among untreated people of European ancestry, a monotonic dose-related relationship provided convincing evidence that the likelihood of smoking cessation increases as the rate of CYP2A6 predicted nicotine metabolism decreases. The dose-related relationship, the strength of the association, and prospective follow-up provide robust evidence that among smokers of European ancestry, slow nicotine metabolism increases the likelihood of smoking cessation.

The results in those of European ancestry also suggest the use of pharmacotherapy smoking cessation interventions may attenuate or modify the CYP2A6 genetic association with smoking cessation but that the change in the association may depend upon which pharmacotherapy is administered. The administration of NRT or varenicline attenuated the association such that the difference in likelihood of smoking cessation between intermediate or slow metabolizers and normal metabolizers in these treatment groups was not statistically significant. However, the use of bupropion appeared to significantly reverse the direction of the association. Due to the differences in CYP2A6 variants genotyped and the corresponding nicotine metabolism classifications across the limited number of studies, it was not feasible to look for the dose-related relationship within each treatment group.

In addition to insufficient data to test for the dose-related relationship within pharmacotherapy groups, it was also unclear if the mitigation of the associations observed in these groups remains consistent long-term after pharmacotherapy use has ceased. The studies suggest the difference in cessation outcomes between slow and normal metabolizers is attenuated during NRT administration because NRT may more effectively promote end of treatment abstinence in fast metabolizers than in slower metabolizers.^35,37 For example, Chen et al. found that end of treatment abstinence among those randomized to placebo was significantly lower among fast metabolizers compared with poor metabolizers; among poor metabolizers, abstinence was similar in both the placebo and NRT arms. However, among participants randomized to NRT, abstinence rates were higher among fast metabolizers so that the rate was similar to that observed among the poor metabolizers. We would expect to observe the largest pharmacotherapy effect on smoking cessation during and at the end-of-treatment and some attenuation in the effect long-term, but the studies with a 12-month follow-up found contradictory results. These contradictory results make it difficult to infer the degree to which associations observed at end of treatment would be observable long-term. Verde et al. observed that 12 months after NRT use ended, slow/intermediate metabolizers had increased odds of abstinence compared with normal metabolizers,³⁸ whereas Lerman et al. observed no statistically significant difference in abstinence 12-month post-cessation between slow/intermediate and normal metabolizers who had received either standard (8 weeks) or extended (6 months) NRT.³⁷ The fact that Verde et al. evaluated only 36 smokers receiving NRT may help explain the disparate findings. More research is needed to fully understand these associations.

The results also suggested bupropion pharmacotherapy may be less effective or not effective for slow nicotine metabolizers. These findings were observed during either 6- or 12-month follow-up implicating that the effect may be present long-term. Though based on only two studies, these results are supported by a previous study that based nicotine metabolism on NMR quartiles, with results showing that bupropion was effective over placebo for smoking cessation among those with faster nicotine metabolism but had little to no impact on cessation over placebo for those with slower metabolism.⁴⁴ The association between CYP2A6 genotype and smoking cessation among smokers using a bupropion intervention could be due to gene-gene interactions between CYP2A6 and CYP2B6 variants.^45,46 The CYP2B6 gene encodes for an enzyme that metabolizes bupropion. The role of CYP2B6 on nicotine metabolism is likely minor, but some CYP2B6 variants may increase nicotine metabolism and the degree of increase may be larger in those with reduced function CYP2A6 variants.^45,46

Though studies of people of non-European ancestry were few, and often used cross-sectional study designs, the evidence suggested variation in the association of CYP2A6 with smoking cessation by ancestry. In the single study of people of African ancestry, CYP2A6 genotype was not associated with abstinence in the placebo arm; however, the contrasting results could be due to the study population being light smokers,⁴³ the difference in allelic frequencies between ancestral groups,²⁷ or a combination of these, or other, factors. Among untreated people of Asian ancestry, a strong association in the opposite direction as that observed in people of European ancestry was present such that, slow metabolizers were significantly less likely to have quit smoking than normal metabolizers. All three studies of Asian people used cross-sectional measurement of smoking status (former vs. current smoker) at the time of recruitment into the study whereas the studies of populations of European ancestry evaluated prospective changes in smoking status. These differences in study design could contribute to the discordant findings between people of Asian ancestry compared with people of European ancestry. The disparate findings could also be due to significantly greater frequencies of loss or reduced function CYP2A6 alleles among populations in China and Japan.²⁷ Further complicating comparisons, China and Japan have a higher prevalence of heavy smokers, those consuming at least 20 CPD, than the United States or Europe^27,47 and East Asian smokers express lower interest in quitting⁴⁸ compared with smokers in the United States.⁴⁹ This phenomenon may be partially explained by social or cultural factors within these East Asian populations such as the importance of smoking as part of social and familial networking in mainland China^48,50 or lower perceived health risks of cigarette smoking in East Asia.^51,52 The social acceptance of cigarette smoking and decreased motivation to quit among East Asian populations may confound detection of genetic associations with smoking cessation.

None of the studies of people of Asian ancestry examined pharmacotherapy interventions, thus it remains unknown if the effects of pharmacotherapy would differ in these populations compared with those observed among people of European ancestry. The findings among people of African ancestry who were light smokers receiving NRT is the opposite of those observed among populations of European ancestry: NRT gum appeared to be more effective in promoting end-of-treatment abstinence in slow metabolizer genotypes,⁴³ but this association was found only among the females. The disparate findings between ancestral groups may be due to differences in CYP2A6 allelic frequencies between populations of African ancestry compared with European ancestry; ²⁷ however, the results among women of African ancestry are consistent with the study of people of European ancestry by Verde et al.³⁸ in which the study population was predominantly female. Though the endpoint in Verde et al. was not at end-of-treatment with NRT but rather 12-month post-treatment, it is plausible that there may be sex differences in the effectiveness of NRT by CYP2A6 predicted nicotine metabolism. Support for potential sex differences in the effectiveness of NRT is found in a meta-analysis of 14 randomized controlled trials of transdermal nicotine patch interventions for smoking cessation which found that the nicotine patch increased abstinence 6-months post-cessation in both men and women, but the patch was less effective among women compared with men.⁵³CYP2A6 is a strong determinant of nicotine metabolism, but additional factors such as the endogenous hormone estrogen also contribute to nicotine metabolism and may partially explain the differences between men and women.¹⁹

In addition to allelic frequencies and sex differences, the different results could also be an artifact of the inclusion of only light smokers in the study of people of African ancestry and the exclusion of light smokers in studies of populations of European ancestry. Further complicating the comparison is that the study of people of African ancestry used NRT gum, whereas the trials of people of European ancestry used transdermal patches or combined NRT methods such as patches and lozenges. Smokers with faster rate of nicotine metabolism not only tend to smoke more CPD,¹⁴ but likely experience more severe and earlier onset of withdrawal symptoms^9,10 and thus require more robust NRT delivery to combat the greater severity of the symptoms and cravings. For light smokers with less severe withdrawal, NRT gum may be effective. Consequently, the use of different NRT products may complicate comparisons across studies.

In summary, most genetic studies of CYP2A6 and smoking cessation are of populations of European ancestry and these studies show a dose-related relationship that the likelihood of smoking cessation increases as the rate of CYP2A6 predicted nicotine metabolism decreases. The use of pharmacotherapy appeared to attenuate associations between nicotine metabolism and cessation among people of European ancestry, but this was based on relatively few studies that used different designs and pharmacotherapy interventions. In addition, the few studies of Asian people have all used a cross-sectional design, often with small sample sizes (e.g., n = 96), and at this time only one study of people of African ancestry is available. Thus, there is a need for more large prospective studies and randomized controlled trials, especially of ancestrally diverse populations, to enrich understanding of the relationship between CYP2A6 and smoking cessation. Additionally, because NMR is a biomarker of CYP2A6 enzyme activity, a future systematic review synthesizing the evidence related to NMR and smoking cessation will provide valuable complementary evidence to the present systematic review to provide a more complete understanding of CYP2A6 in relation to smoking cessation.

Conclusion

The evidence to date suggests that among smokers of European ancestry not receiving pharmacotherapy those with CYP2A6 reduced metabolism genotypes are more likely to quit smoking than those with normal metabolism genotypes. Support for this association is strengthened by the finding of a dose-related relationship showing that the likelihood of quitting increases as metabolism decreases. Among people of European ancestry, findings also suggest NRT may reduce the higher likelihood of cessation among those with CYP2A6 slow metabolism; however, it remains unclear if such attenuation has long-term effects after the NRT intervention ends. Further research to understand the long-term effectiveness of NRT interventions by nicotine metabolism genotype will help to inform precision medicine strategies.

Though evidence suggests the association between CYP2A6 and cessation among people of Asian ancestry and African ancestry may differ from the associations observed among those of European ancestry, this inference is based on only three cross-sectional studies of Asian people and one randomized clinical trial of people of African ancestry. Nonetheless, differences in associations by ancestry are plausible and may be due to differences in social acceptance and perceived risk of smoking, variation in allelic frequencies between ancestral groups, and differences in smoking topography, such as depth of inhalation and cigarettes per day. More research is needed to gain an understanding of the association between CYP2A6 and smoking cessation among diverse populations.

Supplementary Material

A Contributorship Form detailing each author’s specific involvement with this content, as well as any supplementary data, are available online at https://academic.oup.com/ntr.

Funding

Stephanie Jones is a Hollings Cancer Center Abney Graduate Fellow at the Medical University of South Carolina. This research was further supported by National Institutes of Health National Institute of Arthritis and Musculoskeletal and Skin Diseases grant number P30 AR072582 (Medical University of South Carolina Core Center for Clinical Research Improving Minority Health in Rheumatic Diseases) and National Institutes of Health National Center for Advancing Translational Sciences grant numbers TL1 TR001451 and UL1 TR0011450 (South Carolina Clinical and Translational Research Institute, with an academic home at the Medical University of South Carolina).

Declaration of Interests

The authors declare that there is no conflict of interest. Dr. Brett Froeliger is a consultant for Promentis Pharmaceuticals, Inc. (BF) for work unrelated to the content of the manuscript. Dr. Matthew Carpenter has received consulting honoraria from both Pfizer and Frutarom for work unrelated to the content of the manuscript.

Data Availability

There are no new data associated with this article.