The Use of the Nicotine Metabolite Ratio as a Biomarker to Personalize Smoking Cessation Treatment: Current Evidence and Future Directions

The Development of the Nicotine Metabolite Ratio (T0)

Chronic smoking, or dependence, is maintained by the presence of nicotine, which is highly addictive given its underlying neurobiological effects that include the stimulation of neurotransmitters such as dopamine (15). Over the past 15 years, researchers have come to recognize the importance of variability in the rate at which nicotine is metabolized for determining individual smoking behaviors (16, 17). As will be reviewed in greater detail below, it has been hypothesized that relative to slower metabolizers of nicotine, faster metabolizers clear nicotine more quickly than slower metabolizers, exhibit more signs and symptoms of nicotine dependence (e.g., cravings), and titrate their nicotine dose (e.g., smoking more cigarettes per day) to minimize withdrawal symptoms. Thus, the development of a biomarker that noninvasively and reliably assessed smokers' rate of nicotine metabolism would offer clinically relevant information that could further guide smoking cessation interventions.

Nicotine is metabolized by the P450 liver enzyme CYP2A6 (18) into the metabolite cotinine, which is in turn metabolized again by the P450 enzyme into 3′hydroxycotinine. About 60% of the variability in nicotine clearance and about 40% of the variability in cotinine clearance is heritable (19–21). Many functional polymorphisms in the CYP2A6 gene that affect enzyme activity and nicotine clearance have been characterized (ref. 22; http://www.cypalleles.ki.se/cyp2a6.htm]. Reduced-activity and null mutations are associated with slower nicotine clearance (23–25) and are significantly more prevalent among nonsmokers compared with smokers, including within samples of European, Asian, and African descent (25–29). Among smokers, those with reduced-activity or null variants of CYP2A6 are less nicotine dependent (30, 31), are more likely to smoke for shorter durations (28), and are more likely to quit smoking (32), compared with those with wild-type genotypes. However, the feasibility of using CYP2A6 variants to predict therapeutic response to treatments for smoking cessation (and, thus, to personalize treatment) is diminished by: (i) the complexity and costs of genotyping, (ii) CYP2A6 variants yet to be identified, and (iii) that CYP2A6 variants do not account for environmental influences on nicotine metabolism rate, such as oral contraceptive use (33, 34). Furthermore, primary care physicians, who could reach upwards of 70% of smokers with personalized treatment, report a reluctance to adopt genetic testing to individualize treatment for smoking cessation, because of a lack of training in genetics and difficulty in interpreting and conveying test results adequately (24).

To address these feasibility challenges, a genetically informed biomarker of nicotine clearance, the NMR, was developed. The NMR is computed by taking the ratio of the nicotine metabolites derived from smoking (3′hydroxycotinine [3HC]/cotinine) and offers several advantages over genotyping, making it better suited for personalizing smoking cessation treatments. First, the NMR reflects the substantial influence of CYP2A6 genetic variation as well as the influence of other genetic and environmental factors known to influence nicotine clearance in vivo (e.g., age, race, hormonal factors, and smoking itself; refs. 35, 36). Second, the NMR can be assessed noninvasively (i.e., from saliva or plasma, with high reliability, or urine with moderate reliability, without additional drug administration) based on metabolite levels from nicotine derived from the smokers’ usual cigarettes. The NMR is highly reproducible and independent of time since last cigarette (36–39). Although nicotine has a half-life of only 1–2 hours, the half-life of cotinine is relatively long (∼16 hours), resulting in relatively constant levels of cotinine in smokers. While the half-life of 3HC dosed alone is about 5 hours, the levels of 3HC derived from cotinine in vivo are formation-dependent during smoking; thus, the half-life of 3HC is the same as that of cotinine and the ratio of 3HC/cotinine remains constant over time (36, 38). Third, the NMR is strongly correlated with nicotine clearance (r = 0.70–0.95), as well as CYP2A6 genotype, in smokers of European ancestry (36, 40–42) and in African Americans (25, 26). The NMR combines the advantages of pharmacogenetic tests (no restrictions on sample timing) and phenotyping tests (captures environmental and genetic variation) but without the need for extraneous additional drug administration and the complication of timed drug and metabolite sampling, which is required with traditional phenotyping. Finally, while physicians report concern about using genetic data to tailor treatments for smoking cessation (and lower intention to use) because such information could also be stigmatizing and used for employment or insurance discrimination, they indicate a greater willingness to use a test to individualize treatment for smoking cessation if the test is focused on an enzyme like the NMR, because it does not convey such additional risks (43–45). Likewise, data show that patients are highly interested in using genetic information to select treatments for smoking cessation (46, 47) but express concern about genetic information being misused, which would not be the case with the NMR (48, 49). Thus, the NMR represents a pharmacogenetic marker with an established functional phenotype ideal for treatment optimization. In addition, while the issues pertaining to assessing DNA markers in the clinic are being evaluated (i.e., privacy, patient, and clinician acceptance), an easily accessible noninvasive phenotypic marker may have greater patient and clinician acceptance (50), which may ultimately facilitate translation to clinical practice.

Following determination that NMR is strongly correlated with CYP2A6 variants and nicotine clearance (51), subsequent studies have demonstrated the good reliability and validity of using saliva and plasma, and moderate reliability for urine, to determine NMR across various subgroups of smokers and have shown convergent findings across different laboratories (52). Recent studies have examined the prevalence of the NMR and its association with smoking phenotypes among smokers with HIV (53), among smokers using reduced nicotine-content cigarettes (54), among smokers receiving treatment for alcohol dependence (55), among pregnant smokers (56), and among smokers receiving treatment for opiate dependence (57).

Initial Assessment of the NMR and Its Relationship with Smoking Behavior (T1)

The NMR predicts smoking behavior and related exposure to nicotine and tobacco-specific carcinogens. A prior systematic review found that faster metabolizers of nicotine based on the NMR smoke more cigarettes per day than slower metabolizers (33), results that have been replicated in subsequent international population-based studies (58). A small number of studies, however, which have generally used smaller samples or used less reliable methods of measurement (e.g., urine), have failed to find a relationship between smoking rate and NMR (16).

The number of “cigarettes per day” (CPD) one smokes is a commonly used self-report measure in both research studies and social histories taken in medical settings. However, CPD is not always strongly associated with nicotine or carcinogen exposure (59) and may fail to capture important dimensions of smoking behavior. Smoking topography, or measures of how one smokes a cigarette, including the frequency of puffs, puff volume, and the inter-puff interval, can provide more accurate assessments of nicotine and carcinogen exposure. Laboratory topography studies have demonstrated that compared with slower metabolizers, faster metabolizers take larger puff volumes and consequently have greater exposure to the carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1butanol (60). While Carroll and colleagues (61) found the inverse relationship between NMR and NNAL among a small sample of light smokers of American Indian descent, more recent longitudinal research replicated these findings in naturalistic settings, where smoking topography assessments may represent more valid measures of smoking behavior than laboratory-based studies. Compared with slower metabolizers, faster metabolizers were observed to smoke more CPD, take more puffs per day, and have higher total daily puff volumes, even after adjusting for age and sex (62).

It has been hypothesized that NMR is associated with smoking behaviors because faster metabolizers experience a greater physical dependence on nicotine than slower metabolizers. Studies of the relationship between the NMR and nicotine withdrawal symptoms have been mixed (16). Some but not all studies suggest a relationship between higher NMR (faster metabolism) and greater anxiety, irritability, sleep disturbance, and cognitive changes during periods of abstinence (63, 64). A more recent study by Liakoni and colleagues (65) found that faster metabolizers of nicotine reported more negative affect and cravings within just 6 hours of daytime abstinence, supporting the hypothesis that faster metabolizers smoke to avoid or relieve symptoms of withdrawal. In contrast, faster metabolizers did not report greater satisfaction or reward after smoking a cigarette versus slower metabolizers (but see below for a discussion of the relationship between NMR and rewards of smoking following acute withdrawal), failing to provide support for the idea that faster metabolizers smoke more because they derive greater reward from nicotine use. NMR, in this study, also served as a better predictor of withdrawal than CYP2A6 genotype.

While faster metabolizers clear nicotine more quickly and appear to titrate their nicotine dose to minimize withdrawal symptoms, these general findings need to be qualified by important race and sex differences. African Americans have, on average, slower rates of nicotine metabolism and smoke fewer CPD than White smokers (66). Of note, NMR appears to be less strongly related to nicotine intake for African American smokers compared with White smokers: African American smokers with higher rates of nicotine metabolism do not compensate by smoking more CPD, as is the case for White smokers (67). Women, by contrast, have higher average rates of nicotine metabolism than men, attributable to the effects of estradiol on CYP2A6 activity (68), but smoke fewer CPD (69) and exhibit smaller puff volumes (70) than men. Taken together, the relationship between NMR and smoking behaviors may be stronger for White and male smokers than African American and female smokers.

Translating the findings on NMR and smoking behavior to lung cancer risk, Tanner and colleagues (71) examined rates of nicotine metabolism between a Southwestern American Indian tribal population, who have a smoking prevalence comparable with national rates (14%), and a Northern Plains American Indian tribal population, who exhibit considerably higher smoking rates (50%) and 6-fold higher lung cancer incidence and mortality rates. As hypothesized, the Northern Plains tribal population had higher mean levels of NMR than the Southwestern tribal population. This effect could only be explained, in part, by differences in CYP2A6 genotype, suggesting that there may be other factors that account for higher rates of NMR in the Northern Plains tribal population compared with the Southwestern tribal population. With respect to lung cancer risk, Yuan and colleagues (72) investigated the relationship between CYP2A6 variants, NMR, and biomarkers of tobacco consumption in a nested case–control study of lung cancer cases and matched controls within a prospective cohort from Singapore. Consistent with prior research, faster metabolizers had higher intake of tobacco and greater NNAL exposure than slower metabolizers. In addition, slower metabolizers had a 3-fold risk reduction for developing lung cancer versus faster metabolizers. It is important to note that while the relationship between a CYP2A6 genotype–determined measure of nicotine metabolism and lung cancer risk remained even after accounting for smoking intensity and duration, a similar relationship observed between NMR and lung cancer risk became nonsignificant when adjusting for these covariates. This is likely because CYP2A6 genotype also mediates the metabolic activation of the carcinogen NNAL found in cigarette smoke (73). Thus, nicotine metabolism rate may influence carcinogenesis via two mechanisms: (i) faster nicotine metabolism contributes to higher levels of tobacco intake and therefore greater exposure to carcinogens and (ii) greater risk of carcinogenesis per unit of carcinogen exposure.

The NMR and Response to Treatments for Smoking Cessation (T2)

The relationship between the NMR and response to treatments for smoking cessation has been examined in a growing number of studies (See Table 1; refs. 74, 75). Four of these studies examined the relationship between baseline NMR and response to NRT and each showed that slow metabolizers of nicotine have significantly higher quit rates versus fast metabolizers (64, 76–78). A separate placebo-controlled study showed that simply increasing the dose of nicotine to 42 mg among fast metabolizers of nicotine does not increase quit rates significantly, versus the standard 21-mg dose (79). In a fifth study, the nonnicotine medication bupropion, which enhances dopaminergic neurotransmission, significantly enhanced quit rates for fast metabolizers of nicotine, but did not provide an added benefit for slow metabolizers (80). Similarly, Glatard and colleagues (81) showed, in a nonrandomized longitudinal study, that varenicline was a more effective treatment for smoking cessation than NRT for faster nicotine metabolizers. Varenicline, like bupropion, enhances dopaminergic neurotransmission by acting as a nicotinic partial agonist. The authors also found that women overall benefitted more from varenicline than men, perhaps because women have higher average nicotine metabolism. In a study of smokeless tobacco users treated with nicotine lozenges, NMR did not predict abstinence at 3 months but was positively correlated with lozenge use (82). These findings suggest that NRT is more effective for slower metabolizers, whereas nonnicotine medications are more effective for faster metabolizers. As will be discussed in greater detail below, NRT may prove less beneficial for faster metabolizers relative to slower metabolizers because replacement nicotine may not produce the same dopaminergic effect for these smokers, in part, because it is cleared more quickly; varenicline, in contrast, may better address dopaminergic activity that is more critical for fast metabolizers to quit successfully.

Studies cited above using retrospective analysis linking NMR to treatment response led to the first prospective NMR-stratified pharmacogenetic trial of treatments for smoking cessation (83); 1,246 smokers characterized as slow or fast metabolizers of nicotine were randomized to placebo patch and placebo pill, nicotine patch and placebo pill, or varenicline and placebo patch. Varenicline was selected as the nonnicotine medication that was hypothesized to be efficacious for fast metabolizers because it was (and remains) the most efficacious medication for smoking cessation, even versus bupropion (12). The results showed that at both end-of-treatment and 6 months following the target quit date, faster metabolizers had significantly higher quit rates if treated with varenicline versus nicotine patch and that slow metabolizers exhibited similar quit rates across the two treatments but reported more severe side effects if treated with varenicline. In a number needed to treat (NNT) analysis, among slow metabolizers, there was little difference in the NNT to yield one successful quitter (10.3 for patch vs. 8.1 for varenicline). But among fast metabolizers, the NNT to yield one successful quitter is 26 for the patch versus 4.9 for varenicline. Thus, treating slow nicotine metabolizers with the patch and fast nicotine metabolizers with varenicline increases effectiveness, minimizes side effects, and saves costs.

Consistent with the general trend that treatments validated in RCTs do not always translate to broader populations and settings (84), real-world evaluations of NMR to personalize treatment for smoking cessation (see Fig. 1, T3) have produced mixed findings thus far. In this context, “real-world” refers to research conducted in more naturalistic settings such as clinics without the same level of methodologic control, rigorous measurement, or stringent eligibility criteria (84). For example, a feasibility pilot RCT conducted by Wells and colleagues (85) observed that >90% of participants endorsed the practice of “metabolism-informed care” that matches faster metabolizers with varenicline and slower metabolizers with NRT. Furthermore, compared with standard care, metabolism-informed care more than tripled the odds of being prescribed NMR-matched medication. This study was not designed to detect differences in abstinence rates but does support the feasibility of incorporating NMR into routine clinical practice.

In contrast, a prospective observational study of 1,556 treatment-seeking smokers that examined abstinence outcomes in a natural treatment setting found that NMR did not moderate the effectiveness of varenicline and NRT (86). When interpreting these findings, it is important to note that participants self-selected their treatment without being informed of their NMR and only participants who were abstinent at 4 weeks remained in the study. In another trial (87), smokers with or without psychiatric comorbidity were randomized to varenicline, nicotine patch, or long-term and dual NRT. While NMR did not predict abstinence outcomes overall or by treatment condition, this study characterized the rate of nicotine metabolism as quartiles, compared with the cut-off point used in Lerman and colleagues (83), had very few slow metabolizers, and did not formally test the interaction between NMR and NRT versus varenicline. Finally, Fix and colleagues (58) prospectively examined the relationship between NMR and smoking cessation among 874 participants from five countries (United States, United Kingdom, Mauritius, Mexico, and Thailand). Faster metabolizers of nicotine were more likely to have quit smoking and to remain abstinent for a longer duration than slower metabolizers. While these results differ from the studies reviewed here, this analysis did not consider whether participants used any treatments in their quit attempt.

Importantly, simple smoking characteristics (that are more easily assessed), such as rate of smoking or level of nicotine dependence, tend not to be sufficient to guide treatment personalization to improve cessation outcomes or reduce side effects (47). Despite some mixed findings from more recent translational studies of the NMR, the replication of findings across independent efficacy studies and in a prospectively stratified trial, coupled with the psychometric, practical, and clinical advantages of the NMR, support its use to personalize the treatment for smoking cessation. That said, while emerging findings from real-world evaluations support the feasibility of utilizing the NMR to personalize treatment within routine clinical practice, the available findings are limited and additional studies are needed to better characterize the translational potential of the NMR to improve smoking cessation treatments in real-world clinical settings. For example, studies comparing a “test-and-treat” approach, using the NMR, compared with standard nonpersonalized treatment would represent an important scientific step.

What are Potential Mechanisms Through Which the NMR may Operate to Personalize Smoking Cessation Treatments?

While we do not have sufficient data to explain why faster nicotine metabolism may lead to better responsiveness to varenicline, compared with NRT, the available literature suggests that this relationship is less likely to be mediated by standard smoking phenotypes such as nicotine dependence or smoking rate than with neurobiological differences, nicotine reward, and cue reactivity. With regard to the former, many of the studies described above, which demonstrated an effect of the NMR on treatment response, controlled for rate of smoking and level of nicotine dependence, suggesting that that such factors are less likely to be the mechanisms of NMR effects on treatment response (78, 80, 88). Furthermore, nicotine levels achieved during transdermal nicotine treatment are correlated with cessation rate, but only weakly (22), and there is a clear reduction in quit rates with increasing NMR in placebo-treated smokers (89), suggesting an effect even in the absence of nicotine from smoking or NRT. Finally, the association between the NMR and cessation is observed after subjects have discontinued medication; nicotine's half-life is short (2 hours) but the effects of the differing rates of nicotine clearance on abstinence are observable at 6-month follow-up (22, 90).

The relationship between NMR and tobacco use treatment response may be mediated by variability in neurobiological nicotine reward. Studies have shown that faster nicotine metabolism is associated with greater subjective nicotine reward, increased nicotinic receptor availability, and increased brain activity in response to smoking cues. Sofuoglo and colleagues (91) in an intravenous nicotine study showed that faster nicotine metabolism was associated with increased reports of subjective and physiologic reward to nicotine. Dubroff and colleagues (92), using PET imaging, showed that increased α4β2* nAChR receptor availability in the thalamus was associated with faster nicotine metabolism following overnight abstinence, and two neuro-imaging studies found that, compared with slow metabolizers, fast metabolizers show greater brain activation to smoking cues (93, 94). More recently, Li and colleagues (95) suggested that different rates of nicotine metabolism alter striatal–cingulate circuits within the brain in ways that persist after nicotine clears the body. Specifically, neuroanatomic and functional connectivity differences have been observed between slower and faster metabolizers within two brain regions consistently related to drug dependence: the ventral striatum and dorsal anterior cingulate cortex. These differences, however, were not observed among participants without a history of smoking, underscoring the possible gene (fast vs. slow nicotine metabolism) × environment (exposure to nicotine) interaction. In addition, PET imaging research by Di Ciano and colleagues (96) showed that slower metabolizers had fewer D2-type dopamine receptors than faster metabolizers, drawing a link between NMR and dopaminergic signaling. While further research is required to fully delineate the relationships between the NMR and brain circuitry, the existing evidence strongly suggests that personalized approaches have implications for smoking cessation treatments that extend beyond acute withdrawal.

Challenges and Future Directions (T3 and T4)

The next steps for utilizing the NMR as a method to personalize treatment for smoking cessation involve addressing the challenges that impede the application of scientific research findings concerning the NMR to the broadest population possible and ensuring that we overcome the chasm between scientific discovery and clinical practice. To that end, future work is needed to address issues related to methodology, generalizability, expanded use, and implementation.

Methodologically, two key challenges need addressing. First, there is not an established NMR cut-off point for classifying slower versus faster metabolizers of nicotine. Schnoll and colleagues (78) previously used a cutoff of 0.26 in an NRT trial based on a ROC analysis, which corresponded to the first quartile for their sample. However, in a subsequent trial (88), the cutoff was increased to 0.31 based on a reanalysis of the ROC. Other studies (87) have employed a relative approach, such as classifying participants based on quartiles and treating those in the first quartile as “slower metabolizers,” those in the second quartile as “moderate metabolizers,” and those in the top two quartiles as “faster metabolizers.” However, this approach is affected by the range of the NMR within a given sample, leading to inconsistency across studies. This variability in cut-off point creates challenges for both comparing results across studies and informing clinical practice. We argue strongly in favor of conducting secondary data analyses on NMR values pooled from studies that assessed NMR and collected clinical outcomes to establish cutoffs, a relatively efficient and readily feasible approach to addressing this key methodologic challenge. A second key methodologic challenge is the establishment of cut-off points across blood/plasma, urine, and saliva assays. While the evidence suggests that plasma and saliva results are highly correlated (97), other research (52, 98) suggests that results from urine are less reliable. More specifically, urine NMR results are less consistent and appear to be less robust to analytic method and laboratory than NMR values obtained from blood and saliva assays. While urine NMRs have been described as a “relatively good proxy” (ref. 98; p.1110), where possible it may be preferable to utilize blood or saliva samples for assessing NMR. It will be important to resolve these two key methodologic challenges because studies have shown (85, 99) that without accurate metabolism-informed care, smokers are unlikely to be prescribed the NMR-matched medication.

A second important future direction will be to generalize the personalized approach to population subgroups with elevated smoking rates and altered NMR. For example, the mean NMR may be substantially higher among certain Native American/Indian tribes (71), among pregnant women (56, 100), among individuals who are HIV⁺ (53), and among those who consume heavy amounts of alcohol (55). Furthermore, while rates of smoking among people with serious mental illness (101), substance use conditions (57), and among those from the LGBTQ community (102) are 2–3 times the rate of the general population, no data on the NMR are currently available for these communities, making it challenging to guide NMR-based treatment. As such, documenting the rates of NMR and validating the relationship between the NMR and smoking phenotypes in these communities is essential for generalizing the NMR approach to treatment to the most important subgroups of smokers.

Third, the past decade has seen a dramatic alteration to the landscape of tobacco products. Whereas before, combustible cigarettes accounted for the vast majority of tobacco use in the United States, now we have electronic cigarettes and cigarillos and we may soon have low nicotine content cigarettes as an FDA policy to support a national tobacco control effort. One recent study examined the NMR as a moderator of smoking rate (CPD and puff volume) and biomarkers of harm exposure, including carbon monoxide and 4-[methylnitrosamino]-1-[3-pyridyl]-1-butanol (NNAL), when provided with low nicotine content cigarettes (54). One concern is that faster metabolizers of nicotine may increase smoking rate and puff volume with low nicotine content cigarettes, increasing their exposure to biomarkers of harm, and undermining this potential health policy. However, the results of this study showed that NMR was not associated with changes in smoking rate or biomarkers of harm when smokers switched to low nicotine content cigarettes. With the substantial rise in e-cigarette use over the past decade and the absence of data on how the NMR related to e-cigarette use, this would be a priority for future research.

Fourth, it may also be important to consider smoking context to explore potential gene × environment interactions in the context of an NMR-based treatment approach. For example, low-income and predominant racial/ethnic minority neighborhoods are characterized by higher rates of tobacco outlet density, or the number of retail locations licensed to sell tobacco products for a given population size within a defined geographic area (103). Higher rates of tobacco outlet density are associated with more tobacco marketing, increased smoking cues, and social norms that are more permissive of smoking. Closer proximity to a tobacco outlet is associated with smoking urges (104) and poorer abstinence rates (105), whereas increases in distance to outlets over time have been associated with improved within-person odds of quitting (106). In addition to optimizing treatment for the subgroups of smokers described above, it may also be important to consider personalized smoking cessation treatments for smokers residing who reside in environments that undermine such cessation efforts. Future research can evaluate whether varenicline mitigates cue-provoked craving among fast metabolizers who reside in neighborhoods with higher rates of tobacco outlet density.

Fifth, a priority for future research in the context of NMR-based treatment is how this approach could be integrated with novel behavioral interventions to more effectively promote smoking cessation. Brandon and colleagues (107) evaluated the feasibility of adding facilitated extinction training to enhance varenicline efficacy. This behavioral approach encourages participants to continue smoking and maintain exposure to smoking cues during the run-in period that precedes an active quit attempt. By pairing smoking and cue exposure with varenicline, which acts as a nicotine antagonist, the behavioral process of extinction will be facilitated; smoking and smoking cues will become disassociated from reward and will become less reinforcing. Whether this novel behavioral approach is particularly effective for fast metabolizers of nicotine is a worthwhile question to pursue. Likewise, it has become increasingly clear that nonadherence to medications for smoking cessation is a major problem that likely underlies variability in treatment outcome, and that interventions are needed to enhance medication adherence (108). Using the NMR-based approach to promote smoking cessation can reduce treatment-related side effects, which are often cited as reasons for treatment discontinuation (109, 110), and thus may improve cessation outcomes through this process. In fact, the effect of personalizing treatment on cessation using the NMR may be enhanced when medication adherence increases (111). As such, using the NMR along with novel behavioral approaches to increase treatment adherence may have a synergistic effect on cessation and should be the focus of future research in this area.

Finally, additional research is needed to demonstrate that integrating personalized treatment into real-world settings is feasible, will improve clinical outcomes, and will not substantially increase costs. From a feasibility standpoint, successful integration of the NMR testing approach will depend on patient, clinician, and system-level factors and, thus, studies are needed to determine the rate at which the provision of effective and appropriate smoking cessation treatment increases when NMR testing is integrated. The study by Wells and colleagues (85) provides encouraging signs about feasibility but additional work is needed to demonstrate that barriers to implementation can be overcome. Likewise, while the Shahab and colleagues (86) study failed to show the real-world effectiveness of the NMR treatment model, methodologic aspects of that study may have undermined its utility to test this question. Large-scale, randomized T3 studies are needed, where smokers are randomized to personalized treatment based on the NMR or nonpersonalized treatment to fully understand the translational potential of this approach. Finally, while the cost-effectiveness of smoking cessation treatments, generally (112), and in the context of pharmacogenetic treatment models under certain circumstances (113), has been demonstrated, the cost-effectiveness of the NMR treatment model has yet to be evaluated. Although the benefits of a biomarker versus gene approach and the low cost of the NMR test will help boost its potential, the integration of the NMR within clinical practice will depend on formal cost-effectiveness analyses.

Disclosure of Potential Conflicts of Interest

R.A. Schnoll receives medication and placebo free from Pfizer and has provided consultation to Pfizer. Dr. Schnoll has provided consultation to GlaxoSmithKline and provides consultation to CuraLeaf. No potential conflicts of interest were disclosed by the other authors.