Identification of Sociodemographic, Clinical, and Genetic Factors to Aid Alaska Native and American Indian People to Successfully Quit Smoking

Abstract

Introduction

Alaska Native and American Indian (ANAI) people have a smoking prevalence of 23%. Nicotine metabolite ratio (NMR) and genetic testing may enable tailored selection of tobacco cessation medication.

Aims and Methods

The purpose of this study was to evaluate the relative contributions of NMR, cessation medication, demographics, and tobacco use history to cessation. Participants were recruited into an observational cohort study consisting of a baseline visit prior to their quit date and 6-week follow-up. Demographic and tobacco use surveys and blood, urine, and breath samples were collected at each visit. Electronic health records were queried for cessation medications. NMR was categorized into slow or normal nicotine metabolism phenotypes (<0.31 and ≥ 0.31, respectively). The main outcome was cessation at 6 weeks. Analyses consisted of descriptive statistics, medication and phenotype concordance, and estimates of relative risk (RR) of quitting.

Results

We enrolled 151 ANAI adults who smoked cigarettes daily. Two-thirds had normal nicotine metabolism phenotype. Retrospective medication and phenotype concordance was 39%. The overall quit rate was 25%. No demographic factors or tobacco use history were associated with quit success. Varenicline and bupropion increased the likelihood of quitting (RR = 2.93 [1.42, 6.03] and RR = 2.52 [1.12, 5.64], respectively) compared to nicotine replacement therapy. Non-optimal medication and phenotype concordance decreased likelihood of quit success (RR = 0.44 [0.22, 0.91]) compared to optimal concordance.

Conclusions

This exploratory study found associations between quit success and tobacco cessation medication as well as medication and phenotype concordance. Additional research is needed to assess use of NMR for treatment selection among ANAI people.

Implications

These results broadly support additional community-engaged research to improve medication and phenotype concordance in tribal health settings. Such future research on implementing meditcation and phenotype concordance holds promise to improve expectations, quit success, and health outcomes amongst individuals attempting to quit smoking.

Introduction

Tobacco use is the principal cause of preventable disease and death in the United States.¹ Alaska Native and American Indian (ANAI) people have the highest smoking prevalence of any racial and ethnic subgroup in the United States: 23% versus 14% when compared to the U.S. general population.¹ Although 77% percent of ANAI people do not smoke tobacco and those who do on average report smoking fewer cigarettes per day than their peers in the general population, tobacco-related disease burden is higher among ANAI people.^2,3 The ANAI population nationally has a 60% higher rate of smoking-attributable mortality (eg, lung cancer, chronic obstructive pulmonary disease (COPD), heart disease, and stroke) compared to the White population.⁴ In Alaska, two-thirds of people who smoke want to quit, and over half report that they tried to quit in the last year, but quit rates are lower for ANAI adults than others.⁵ With concerted public health efforts, smoking prevalence in the United States has dropped precipitously among non-ANAI populations over the last 30 years (from ~40% to 14%), but ANAI populations have not seen similar reductions.^6–8

Standard of care for tobacco cessation treatment includes both pharmacologic and behavioral counseling approaches. Both strategies are superior to “cold turkey” and a combined strategy produces greater quit rates.⁹ Nicotine replacement therapy (NRT) remains the most used first-line pharmacologic treatment for tobacco cessation and increases the odds of quitting 1.84 times compared to placebo.^9,10 Combination NRT (ie, transdermal patch with gum or lozenge) outperforms single formulations.⁹ Other first-line treatments are varenicline which increases the odds of quitting 2.88 times^9,11 compared to placebo and bupropion which increases odds of quitting 1.82 times compared to placebo.^9,12,13 In clinical practice, the choice of first-line pharmacologic treatment may depend on cost, side effect profile, provider familiarity, pharmacy availability, and patient preference.^14–17

Pharmacogenetics uses genetic or phenotype testing to proactively identify and characterize interindividual differences in medication response. Pharmacogenetics has shown substantial clinical utility in optimizing pharmacologic treatment in domains including cancer,¹⁸ HIV,¹⁹ and cardiovascular disease.²⁰ It also offers the opportunity to identify sources of interindividual variability in nicotine dose and tobacco cessation treatment disposition and response. Cytochrome enzyme (CYP) 2A6 is an important rate-limiting catalyst for the metabolic inactivation of nicotine to cotinine (COT) and then to trans-3ʹ-hydroxycotinine (3HC).²¹ CYP2A6 is highly genetically polymorphic, resulting in a very large range of activity (phenotype); this activity can be measured in plasma by the ratio of 3HC/COT, referred to as the nicotine metabolite ratio (NMR). Individuals with CYP2A6 variants that result in slow nicotine metabolism and a lower plasma NMR demonstrate higher levels of tobacco cessation treatment success with nicotine replacement patches.^22,23

Pharmacogenetic testing could enable clinicians to individualize tobacco cessation treatment, thereby increasing cessation success, reducing the number of treatment episodes required before a successful quit attempt, and ultimately improving health outcomes.²⁴ For example, data from Lerman et al. (2015), which is prospectively randomized by NMR, suggests that treating individuals with normal NMR with varenicline and those with slow NMR with nicotine patches could increase quit rates and decrease side effects.²⁵ Patterson et al. (2008) suggest that bupropion is more effective than placebo only for individuals with normal NMR.²⁶

Current research findings suggest genetic or phenotypic testing may be a highly beneficial component of the optimal tobacco cessation treatment program, although the mechanisms which influence outcomes may differ by population and gender. Most modifiable and non-modifiable risk factors (eg, genetics, socioeconomic factors) shown to influence response to tobacco cessation treatment are understudied in ANAI populations.^26–30 Important questions remain about the implementation of pharmacogenetics for tobacco cessation treatment in a more heterogenous ANAI community.^31,32 Thus, there is a need to identify and define factors to better evaluate the effectiveness of tobacco cessation pharmacotherapy in ANAI communities. This study aimed to address that gap by evaluating the association between NMR, tobacco cessation treatment, and individual factors with quit success at 6 weeks among a genetically and culturally heterogeneous urban ANAI population in southcentral Alaska. We also assessed optimal therapy (retrospective medication concordance with phenotype) with quit success.

Methods

Setting

The study protocol was reviewed and approved by the Alaska Area Institutional Review Board and Tribal review boards.³³ This study was conducted at Southcentral Foundation (SCF), a nonprofit, tribally owned and operated health system that provides prepaid primary care services to more than 65 000 ANAI people living in Southcentral Alaska.³⁴ SCF offers evidence-based tobacco cessation treatment services to eligible ANAI individuals through its Quit Tobacco Program (QTP). In the program, a health educator provides counseling and follow-up support, and a pharmacist provides medication treatment options following SCF guidelines that include (1) varenicline, (2) NRT (patch, gum, or lozenge), (3) bupropion, or (4) bupropion in combination with NRT. Medication treatment options, preapproved by the primary care provider, are reviewed by the pharmacist for dosing, contraindications, precautions, and side effects, and then selected by the patient. Following a QTP intake appointment, all patients also received 5–15 minute behavioral counseling interventions by phone from a tobacco health educator, on or shortly before the participant-set quit date and at 1, 2, 3, 6, 12, 26, and 52 weeks afterward.³⁵

Recruitment

Participants were recruited from the QTP. The predefined target enrollment goal for this observational study was 150. Our main consideration in determining the sample size of this exploratory study was to recruit enough participants to obtain reasonable estimates of quit rate and to explore associated variables. Eligible participants were those who self-identified as ANAI, were age 18 years or older, were enrolled in the QTP, smoked more than 100 cigarettes in their lifetime, smoked daily, and smoked a cigarette in the past 24 hours. Exclusion criteria included simultaneous use of other tobacco products including chew, iqmik (a form of smokeless tobacco used among ANAI people),³⁶ electronic cigarettes, pipes, and cigars. Additionally, potential participants were excluded from the study if they were being treated for cancer, had hemophilia, or were pregnant. Written informed consent was obtained from all study participants.

Data Collection

Data were collected at a baseline visit before the participant’s identified quit date and another follow-up visit 6 weeks after the quit date. The follow-up visit occurred within a 2-week window before or after the 6-week date, regardless of quit status. At each visit, participants completed a survey and provided a blood, urine, and breath sample.

Survey

The baseline survey included questions on sociodemographic information, tobacco use history, previous quit attempts, a readiness to quit assessment, the Fagerström Test for Nicotine Dependence,³⁷ and the Smoking Self-Efficacy Questionnaire (SEQ-12).³⁸ After 6 weeks, participants were resurveyed using the initial survey as well as tobacco cessation success (7-day point prevalence of smoking abstinence, if they were currently using tobacco products, and number of cigarettes currently smoking) and tobacco cessation treatment (medication types and frequency of use).

Biospecimens

Blood and urine samples were collected to measure nicotine metabolites to calculate the plasma NMR and quantified urinary recovery of nicotine and eight of its metabolites that together reflect the daily nicotine intake dose. Breath samples were collected to measure exhaled carbon monoxide (CO).

Electronic Health Records

Electronic health records were queried for tobacco cessation medications dispensed to all participants.

Measures

Demographic factors included the following: Gender; age; body mass index³⁹; residence (single-family home, apartment, and other); number of people, children, and tobacco users in residence; education (some high school, high school graduate, some college, or more); and annual household income.

Tobacco use history questions assessed number of cigarettes (normal and menthol) smoked per day; age at which the participant started smoking; number of times participant quit smoking for 24 hours or more in the past year; Fagerström Test for Nicotine Dependence (very low 0–2, low 3–4, moderate 5, high 6–7, very high 8–10)³⁷; Smoking Self-Efficacy Questionnaire³⁸; and Readiness to Quit Smoking Questionnaire for importance, readiness, and confidence.⁴⁰

Clinical factors were plasma NMR, tobacco cessation treatment, and exhaled CO. Plasma NMR was calculated as free 3HC/COT and categorized into slow or normal nicotine metabolism (<0.31 and ≥0.31, respectively) phenotypes.²⁵ Tobacco cessation treatment was categorized in two ways. Current tobacco cessation treatment was (1) NRT (patch, gum, and/or lozenge), (2) varenicline, (3) bupropion alone or in combination with NRT to curb acute cravings, and (4) no tobacco cessation medications dispensed. The second cessation variable a medication and phenotype concordance variable was retrospectively created by combining tobacco cessation treatment and NMR phenotype into three concordance categories: (1) using the tobacco cessation medication optimal for the NMR phenotype, (2) using an alternative tobacco cessation medication (non-optimal for NMR phenotype), and (3) not using a tobacco cessation medication. Concordance included people with slow NMR who used NRT and people with normal NMR who used varenicline or bupropion because they were using optimal medication.^25,26 Individuals with slow NMR who used varenicline or bupropion and people with normal NMR who used NRT were categorized as using non-optimal medication.

The outcome of interest, successfully quitting smoking at 6 weeks, was defined according to three self-reported smoking variables. Self-reported variables included yes–no to currently using tobacco products, 7-day point prevalence of smoking abstinence, and count of current cigarettes smoked. Biomarkers (exhaled CO and urine cotinine) were also used to determine quit status. Participants with a CO ≤8 ppm and or with a urinary cotinine < 10 ng/mL were considered to have quit smoking.⁴¹ If there was disagreement between the three self-reported variables or between one or more of the self-reported variables and biological variables (n = 7), exhaled CO and urine cotinine were used to determine quit status. In two instances when exhaled CO and urine cotinine results disagreed, the urine cotinine level was the final determinant of quit status.

Statistical Analysis

Descriptive statistics on demographics, tobacco use, and clinical factors were calculated as percentages for categorical variables or mean with standard deviation (SD) for continuous variables and stratified by quit status. Chi-square and t-tests were used to compare baseline characteristics in participants who did and did not attend the 6-week visit.

The main aim of this study was to estimate the association between NMR, tobacco cessation treatment, and individual factors with quit success at 6 weeks. The primary analysis excluded participants who did not attend the 6-week visit and represents a complete case analysis. We used Poisson regression with robust standard error estimation to estimate relative risk (RR) and 95% confidence intervals (CI). We used univariate regression to estimate crude associations between quit success and NMR, tobacco cessation treatment, and individual factors. In multivariable regression, we examined tobacco cessation treatment as the exposure of interest in two ways. First, we examined the association between quit success and treatment type by fitting a model that included variables for each treatment combination used at SCF: NRT, varenicline, bupropion alone or with NRT, or no tobacco cessation medication. Second, we assessed the association between quit success and retrospective medication and phenotype concordance (ie, optimal medication for phenotype, non-optimal medication for phenotype, and no tobacco cessation medication). All models included age; gender; education; income; cigarettes per day; age started smoking; number of tobacco users in residence; Fagerström score; smoking self-efficacy score; score for importance, readiness, and confidence; and plasma NMR phenotype group as covariates. Primary analyses were conducted using R 3.5.3.⁴²

Two sensitivity analyses were conducted due to the loss of follow-up at the 6-week study visit and inability to collect quit status. A limited amount of baseline data were missing (1%–7% range of missing values); however, 43% of participants were lost to follow-up at 6 weeks. First, a sensitivity analysis (intention to treat) was performed assuming that data were not missing at random and assuming all participants lost to follow-up at 6 weeks were assumed to have not quit. This sensitivity analysis may be biased if some participants quit smoking yet were unable to attend the 6-week visit and standard error estimates are too small because variability in the imputation process was not taken into account. A secondary sensitivity analysis (imputation) was conducted due to potential bias in the results if data were not missing completely at random. We used multiple imputation by chained equations to impute missing values for all baseline variables and quit success at 6 weeks using all variables from the multivariable regression analyses and additional variables including number of adults in the residence, number of children in the residence, and baseline CO reading.⁴³ Imputation of 100 datasets were conducted and Rubin’s rule was used to compute combined point estimates and standard errors that account for variability because of the imputation process.⁴⁴ While neither sensitivity analysis likely provides unbiased estimates, they provide a range of possible results under different assumptions. Univariable and multivariable analyses were completed as above for both sensitivity analyses. The sensitivity analyses and subsequent analyses were conducted using Stata 15.1.⁴⁵

Results

Between May 2016 and October 2018, we enrolled 151 ANAI individuals, ≥18 years of age, who smoked cigarettes daily and were enrolled in the SCF QTP. Eighty-six (57%) participants completed the 6-week follow-up visit. Supplementary Table 1 describes the characteristics of study participants who completed the 6-week visit and compares demographic, tobacco use history, and clinical factors by quit status. More than half of the participants were women, most were between 30 and 60 years old, about half completed some college, and over half had an annual household income above $30 000. The mean age of smoking initiation was 16 years of age, and participants smoked a mean of 12 cigarettes per day. Two-thirds of participants had normal NMR. NRT (38%) was the most frequent tobacco cessation treatment used, followed by varenicline (29%), bupropion (21%), and no medication (12%). Compared to those who did not quit, more people who quit were female and had slow nicotine metabolism.

Because of potential bias introduced by the high percentage who were lost to follow-up at 6 weeks, we assessed differences in baseline characteristics between those who attended the 6-week follow-up visit and those lost to follow-up (Supplementary Table 2). The major differences were in type of residence and household income.

Retrospective concordance between chosen tobacco cessation treatment and NMR phenotype was about 40%. Table 1 shows 33 participants chose NRT, but most of them (64%) were people with normal NMR. There was a better match between treatment choice and phenotype for people who chose varenicline and bupropion (56% and 67%, respectively). Everyone who chose to quit without medication had normal NMR.

Table 1.

Retrospective Concordance With TC Medication and NMR Phenotype Among Complete Cases (n = 86)

Plasma NMR phenotypea	Nicotine replace therapyb n = 33	Varenicline n = 25	Bupropion alone or with NRT n = 18	None n = 10
Slow n = 26	9 c	11	6	0
Normal n = 57	21	14 c	12 c	10
Missing n = 3	3	0	0	0

^aSlow (NMR < 0.31); normal (NMR ≥ 0.31).

^bPatches, gum, or lozenge.

^cOverall 35/86 (41%) concordant with optimal TC treatment for predicted phenotype from NMR. NRT = nicotine replacement therapy.

In univariate regression (Table 2) using complete case data, no demographic or tobacco use factors were associated with quit success. The quit rate for people with normal NMR was not different from those with slow NMR (RR = 0.67, 95% CI [0.42, 1.06]). Compared to using NRT, using varenicline increased the likelihood of quitting (RR = 2.24,95% CI [1.25, 4.02]), whereas the effect of bupropion was not significant (RR = 1.65, 95% CI [0.82, 3.30]). In addition, using no tobacco cessation was not associated with the likelihood of quitting (RR = 0.33, 95% CI [0.05, 2.27]).

Table 2.

Univariate Poisson Regression for the RR of Quit Success

Variables	Complete case N = 86			Intention to treat N = 151			Imputation N = 151
	RR	95% CI	p	RR	95% CI	p	RR	95% CI	p
Demographics
Gender
Female	Ref		0.52	Ref		0.52	Ref		0.72
Male	0.85	0.51 to 1.40		0.83	0.46 to 1.48		0.90	0.50 to 1.62
Age
18–29	Ref		0.58	Ref		0.56	Ref		0.91
30–39	0.89	0.31 to 2.56		1.04	0.31 to 3.45		0.75	0.27 to 2.14
40–49	1.07	0.38 to 3.03		1.12	0.34 to 3.71		0.87	0.31 to 2.44
50–59	1.52	0.58 to 4.01		1.64	0.53 to 5.07		1.11	0.40 to 3.03
≥60	1.23	0.42 to 3.59		1.88	0.56 to 6.31		0.93	0.29 to 2.95
No. of tobacco users in residence
1	Ref		0.08	Ref		0.02	Ref		0.21
2	0.47	0.22 to 1.02		0.45	0.19 to 1.04		0.59	0.27 to 1.28
3+	1.19	0.72 to 1.95		1.59	0.88 to 2.85		1.26	0.64 to 2.49
Education
Some high school	0.83	0.31 to 2.24		0.67	0.22 to 2.04		0.66	0.22 to 1.99
High school graduate or GED	1.25	0.76 to 2.06		1.21	0.68 to 2.15		1.30	0.70 to 2.40
Some College or more	Ref		0.58	Ref		0.55	Ref		0.43
Annual Household Income
<$9999	Ref		0.68	Ref		0.16	Ref		0.83
$10 000–29 999	2.10	0.82 to 5.37		2.75	0.95 to 7.94		1.63	0.62 to 4.29
$30 000–49 999	1.89	0.72 to 5.01		3.60	1.23 to 10.50		1.49	0.54 to 4.11
$50,000–69 999	1.67	0.57 to 4.91		2.17	0.65 to 7.32		1.26	0.41 to 3.90
≥$70 000	1.78	0.66 to 4.80		3.48	1.17 to 10.33		1.58	0.58 to 1.34
Tobacco use history
No. cigarettes per day	1.00	0.97 to 1.04	0.86	1.00	0.97 to 1.04	0.94	1.01	0.98 to 1.05	0.42
Age started smoking	1.01	0.96 to 1.06	0.67	1.01	0.96 to 1.05	0.77	1.01	0.95 to 1.07	0.68
Fagerström test for nicotine dependence
Very low	Ref		0.93	Ref		0.68	Ref		0.97
Low	0.86	0.47 to 1.58		0.86	0.43 to 1.72		0.87	0.41 to 1.84
Moderate	0.96	0.48 to 1.92		1.03	0.46 to 2.30		1.12	0.48 to 2.61
High	0.69	0.31 to 1.55		0.52	0.21 to 1.31		0.82	0.34 to 1.95
Very high	0.83	0.27 to 2.58		0.69	0.18 to 2.57		0.85	0.22 to 3.26
Smoking self-efficacy	1.02	1.00 to 1.04	0.05	1.02	1.00 to 1.04	0.13	0.05	0.99 to 1.04	0.22
Readiness score
Importance	1.29	0.98 to 1.70	0.07	1.29	0.96 to 1.73	0.09	1.23	0.93 to 1.63	0.14
Ready	1.09	0.92 to 1.30	0.32	1.11	0.92 to 1.33	0.30	1.06	0.90 to 1.25	0.50
Confidence	1.14	0.97 to 1.33	0.11	1.10	0.94 to 1.29	0.22	1.11	0.95 to 1.30	0.18
Clinical
Plasma NMR phenotype
Slow	Ref		0.09	Ref		0.08	Ref		0.31
Normal	0.67	0.42 to 1.06		0.62	0.36 to 1.07		0.73	0.40 to 1.34
TC treatment
NRT	Ref		0.02	Ref		0.002	Ref		0.33
Varenicline	2.24	1.25 to 4.02		3.08	1.57 to 6.03		1.65	0.84 to 3.23
Bupropion	1.65	0.82 to 3.30		2.32	1.06 to 5.06		1.33	0.60 to 2.98
None	0.33	0.05 to 2.27		0.32	0.04 to 2.36		0.63	0.16 to 2.46
Medication phenotype concordance
Using optimal a	Ref		0.19	Ref		0.10	Ref		0.46
Using non-opt b	0.82	0.52 to 1.31		0.73	0.42 to 1.27		0.84	0.46 to 1.52
No TC meds c	0.18	0.03 to 1.21		0.14	0.02 to 0.97		0.46	0.12 to 1.70

RR = relative risk; CI = confidence interval, NRT = nicotine replacement therapy.

^aUsing therapy that would be optimal based on nicotine metabolic ratio: Slow metabolizers using NRT or normal metabolizers using varenicline or bupropion.

^bUsing therapy that would NOT be optimal based on nicotine metabolic ratio: slow metabolizers using varenicline or bupropion or normal metabolizers using NRT.

^cThose not using medication may have received behavioral counseling by telephone.

Compared to the complete case analysis, RR estimates for the intention to treat sensitivity analysis were not sufficiently precise, but did have stronger estimates than the complete case analysis. Again, compared to the complete case analysis, RR estimates from the imputation analysis were often attenuated toward the null and CI were narrower due to the increased sample size. The most notable attenuation between imputation and complete case analysis was for tobacco cessation treatment, particularly varenicline compared to NRT (RR = 1.65 95% CI [0.84, 3.23] vs. 2.24 95% CI [1.25, 4.02], respectively).

We tested different exposure variables for tobacco cessation treatment and quit success (Table 3) after controlling for other variables. For tobacco cessation treatment, varenicline and bupropion increased the likelihood of quitting (RR = 2.93 95% CI [1.42, 6.03] and RR = 2.52 95% CI [1.12, 5.64], respectively). Again, using no tobacco cessation medication did not affect the likelihood of quitting (RR = 0.40 95% CI [0.05, 3.00]) compared to NRT. Tobacco cessation medication to NMR phenotype concordance was associated with quit success. However, those not on the optimal medication for their NMR phenotype were less likely to quit (RR = 0.44 95% CI [0.22, 0.91]).

Table 3.

Multivariable Regression for the Relative Risk (RR) of Quit Success by Tobacco Cessation Treatment Used, Controlled for Covariates^a

Exposure		Complete case N = 73		Intention to treat N = 128		Imputation N = 151
		RR	95% CI	RR	95% CI	RR	95% CI
Model 1	TC treatment	Overall p = .80			Overall p = .01		Overall p = .39
	NRT	Ref		Ref		Ref
	Varenicline	2.93	1.42 to 6.03	3.68	1.42 to 9.57	1.76	0.78 to 3.97
	Bupropion	2.52	1.12 to 5.64	2.94	1.23 to 7.00	1.17	0.48 to 2.81
	None	0.40	0.05 to 3.00	0.49	0.07 to 3.29	0.64	0.15 to 2.62
Model 2	TC medication concordance Overall p = .77				Overall p = .05		Overall p = .44
	Using optimal	Ref		Ref		Ref
	Using non-opt	0.44	0.22 to 0.91	0.60	0.31 to 1.14	0.77	0.40 to 1.47
	No TC meds	0.22	0.05 to 1.04	0.20	0.04 to 0.97	0.46	0.12 to 1.77

^aAll models included age; gender; education; income; cigarettes per day; age started smoking; number of tobacco users in residence; Fagerström score; Smoking self-efficacy score; Readiness score for importance, ready, and confidence; and NMR phenotype as covariates.

Compared to the complete case analysis, RR estimates for the intention to treat sensitivity analysis were more strongly associated with quit success among patients using varenicline or bupropion compared to those using NRT, although CI were less precise. Compared to the complete case analysis, RR estimates from the imputation sensitivity analysis were attenuated toward the null, most notably for quit success among participants using varenicline (RR = 1.76, 95% CI [0.78, 3.97]) or bupropion (RR = 1.17, 95% CI [0.48, 2.81]) compared to those using NRT.

Discussion

This observational study found varenicline and bupropion increased the likelihood of quitting at 6 weeks compared to NRT. Use of medication not optimal for NMR phenotype decreased participants’ likelihood of quit success compared to use of optimal medications. This decrease was most pronounced for those with normal NMR who received NRT. These results were supported in complete case and intention to treat analyses. Unlike studies in other populations,^46–48 demographic and tobacco use history were not associated with quit success.

The results of this study support the potential use of NMR phenotype to inform selection of tobacco cessation treatment. We evaluated the use of NMR phenotype, rather than testing for CYP2A6 genetic variation, based on previous studies demonstrating its superiority as a tobacco cessation biomarker²⁶ and a strong CYP2A6 genotype—plasma NMR phenotype relationship observed previously in the study population.³⁰ The potential benefit of NMR phenotype testing is predicated on previous qualitative work that found pharmacogenetic testing to be acceptable in the same ANAI patient population with proper community-level protective measures in place.⁴⁹

Further research is needed to develop and pilot an intervention to perform NMR phenotyping, identify optimal pharmacotherapies, and inform individual patients and their healthcare providers of the resulting therapeutic recommendations in this population. It is critical to determine whether these recommendations and the underlying rationale behind the medication and phenotype concordance may serve to increase patients’ uptake of tobacco cessation treatment identified as optimal per their NMR phenotype; and if not, to characterize other factors that influence patients’ and providers’ choices. Further research is needed to improve the precision of found associations. Pharmacogenetic results may not only improve treatment, at the level of biological effect, but may also shape patients’ expectancies in a helpful direction (eg, “This is the best treatment for me personally.”). Informing patients of the potential for NMR phenotyping to increase the likelihood of successful cessation has the potential to increase treatment rates.

Strengths and Limitations

The study has several strengths. It explored the innovative use of pharmacogenetic testing in a genetically and culturally heterogeneous population of ANAI people that bears a disproportionate health burden due to tobacco use. It took place within a unique tribal healthcare setting based on relationships and which encourage shared decision-making with patients. It also avoided the potential bias of self-reported quit rates by verifying quit status via biomarker measurement. In addition, it included data derived from surveys, electronic health records abstraction, and biospecimens, enabling associations with quit status to be determined for sociodemographic, clinical, and genetic factors.

Like all observational cohort studies, this investigation had several inherent threats to internal validity, including potential self-selection bias at both enrollment and follow-up. Of note, treatment decisions were based on patient characteristics, clinical contraindications, and, ultimately, patient preference. This could have resulted in selection bias, with greater mismatch in patients selecting NRT compared to those selecting varenicline or bupropion (without or with NRT), exaggerating the effectiveness of an optimal tobacco cessation treatment-NMR phenotype match. A relatively high rate of missingness in data, due to loss to follow-up, resulted in some association analyses being underpowered. Additionally, the data abstracted from the electronic health records may not have adequately captured clinical contraindications for particular medication choices; such factors may have contributed to decisions against the optimal tobacco cessation treatment in some cases. The study included only individuals who smoked cigarettes but who did not chew or use other forms of tobacco (ie, mono-users), and results may not generalize to individuals who use tobacco in multiple forms. The sample of ANAI people was drawn from a single region in Alaska and does not represent all Indigenous people in Alaska, in other regions, or in the United States generally.

Conclusion

An integrated Tribal healthcare organization provided a unique setting for the present observational cohort study among ANAI individuals attempting to quit smoking. Modifiable clinical factors such as medication type and medication and phenotype concordance appear to increase likelihood of quitting smoking. These results broadly support development of community-engaged research to improve medication and phenotype concordance, such as a pilot intervention to inform individuals of optimal medication when they are selecting treatment options. Such an intervention on implementing meditcation and phenotype concordance holds promise to improve expectations, quit success, and health outcomes for most attempting to quit smoking.

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

This work was supported by the National Institute of General Medical Services, Native American Research Centers for Health (grant numbers U261IHS0079, S06GM123545, and S06GM142122-02). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors thank funding from the Canada Research Chairs program (RFT).

Declaration of Interests

RFT has consulted for Quinn Emanuel and Ethismos Research Inc, which ended in 2020; other authors have no conflicts of interest to report.

Author Contributions

Jaedon Avey (Conceptualization [Supporting], Data curation [Equal], Formal analysis [Supporting], Investigation [Lead], Methodology [Lead], Project administration [Equal], Writing—original draft [Lead], Writing—review & editing [Supporting]), Krista Schaefer (Data curation [Equal], Formal analysis [Lead], Investigation [Supporting], Methodology [Supporting], Project administration [Equal], Writing—original draft [Lead], Writing—review & editing [Lead]), Carolyn Noonan (Formal analysis [Lead], Methodology [Supporting], Writing—review & editing [Supporting]), Susan Trinidad (Conceptualization [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Clemma Muller (Formal analysis [Supporting], Writing—review & editing [Supporting]), Katrina Claw (Conceptualization [Supporting], Data curation [Equal], Formal analysis [Supporting], Investigation [Supporting], Methodology [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Denise Dillard (Conceptualization [Lead], Funding acquisition [Equal], Investigation [Supporting], Methodology [Supporting], Writing—review & editing [Supporting]), Michael Todd (Data curation [Equal], Investigation [Supporting], Project administration [Equal], Writing—review & editing [Supporting]), Julie Beans (Conceptualization [Supporting], Investigation [Supporting], Methodology [Supporting], Writing—original draft [Supporting], Writing—review & editing [Supporting]), Rachel Tyndale (Investigation [Supporting], Methodology [Supporting], Writing—review & editing [Supporting]), Renee Robinson (Conceptualization [Lead], Funding acquisition [Equal], Investigation [Lead], Methodology [Lead], Writing—original draft [Supporting], Writing—review & editing [Supporting]), and Kenneth Thummel (Conceptualization [Lead], Formal analysis [Supporting], Funding acquisition [Equal], Investigation [Lead], Methodology [Lead], Writing—original draft [Lead], Writing—review & editing [Supporting]).

Data Availability

Data not publicly available.