Toward the implementation of genomic applications for smoking cessation and smoking-related diseases

Genetic information about smoking and smoking-related diseases could be used to identify individual disease risk, motivate quit attempts, and optimize smoking cessation treatment.

Abstract

The incorporation of genomic information into routine care settings is a burgeoning area for investigation in behavioral medicine. The past decade has witnessed rapid advancements in knowledge of genetic biomarkers associated with smoking behaviors and tobacco-related morbidity and mortality, providing the basis for promising genomic applications in clinical and community settings. We assessed the current state of readiness for implementing genomic applications involving variation in the α5 nicotinic cholinergic receptor subunit gene CHRNA5 and smoking outcomes (behaviors and related diseases) using a process that could be translatable to a wide range of genomic applications in behavioral medicine. We reviewed the scientific literature involving CHRNA5 genetic variation and smoking cessation, and then summarized and synthesized a chain of evidence according to analytic validity, clinical validity, clinical utility, and ethical, legal, and social implications (ACCE), a well-established set of criteria used to evaluate genomic applications. Our review identified at least three specific genomic applications for which implementation may be considered, including the use of CHRNA5 genetic test results for informing disease risk, optimizing smoking cessation treatment, and motivating smoking behavior change. For these genomic applications, we rated analytic validity as convincing, clinical validity as adequate, and clinical utility and ethical, legal, and social implications as inadequate. For clinical genomic applications involving CHRNA5 variation and smoking outcomes, research efforts now need to focus on establishing clinical utility. This approach is compatible with pre-implementation research, which is also needed to accelerate translation, improve innovation design, and understand and refine system processes involved in implementation. This study informs the readiness to incorporate smoking-related genomic applications in real-world settings and facilitates cross-disciplinary collaboration to accelerate the integration of evidence-based genomics in behavioral medicine.

Implications

Practice: Potential applications of genetic information related to smoking and smoking-related diseases in clinical care include identifying elevated disease risk in individuals who smoke, motivating smoking cessation, and personalizing treatment of smoking cessation.

Policy: Accelerating and maximizing the public health benefit of genomics in behavioral medicine requires ongoing monitoring of a rapidly evolving evidence base and consideration of the contexts in which genomic applications are to be implemented, including organizational factors, existing guidelines, and insurance payments.

Research: Future research should focus on establishing the clinical utility of genetic information related to smoking cessation and on understanding the appropriate and effective implementation of genomic applications in routine settings.

INTRODUCTION

The widespread implementation of genomic advances and other precision medicine innovations into healthcare settings is an exciting yet underexplored frontier [1–3]. Recent large-scale efforts, such as the Genomics and Population Health Action Collaborative (GPHAC) through the National Academies of Sciences, Engineering, and Medicine (nationalacademies.org/hmd/Activities/Research/GenomicBasedResearch/Innovation-Collaboratives/Genomics-and-Population-Health.aspx), the Centers for Disease Control and Prevention (CDC) Public Health Genomics Knowledge Base (phgkb.cdc.gov), the Clinical Pharmacogenetics Implementation Consortium (CPIC; pharmgkb.org/page/cpic) and the National Human Genome Research Institute (NHGRI)-funded Implementing Genomics in Practice (IGNITE) consortium (ignite-genomics.org) [4], have begun to shape the research, practice, and policy landscapes for large-scale genomics implementation. Private companies such as 23andMe are implementing large-scale direct-to-consumer genetic testing, having genotyped more than 2 million individuals (https://mediacenter.23andme.com/about-us/). Systematic reviews and expert consensus panels note that the majority of work to date has focused primarily on using genomic information to identify risk of a select few hereditary diseases including breast and ovarian cancer, colorectal cancer, early-onset heart disease, and congenital diseases through prenatal and newborn screening [5, 6]. Although there is great promise for translating genomic discovery into better population health, a recent systematic review concluded that this potential benefit will not be fully realized until more is known about the optimal integration of genomic advances into clinical care and public health practice. The next frontier is to leverage established implementation science principles and the early successes of genomics implementation to benefit a wider range of diseases and disorders, particularly in behavioral medicine.

Genomics implementation research is increasingly being conducted in behavioral medicine. This research has focused on identifying novel approaches for incorporating personalized genomic information to improve health behaviors [7] and on characterizing the primarily positive attitudes of patients and providers yet practical challenges to implementation [8]. Recent studies have demonstrated that returning genetic information altered perceptions of risk for Type 2 diabetes which may lead to risk-reducing behavior change [9] and was useful for determining treatment and managing risk in family members [10]. Further, studies have shown that the vast majority of patients were interested in receiving personal genomic risk results for cancer and heart disease and recognized the importance of both behavioral and genetic factors contributing to disease, reducing concerns about fatalism in patients following return of results [11, 12].

Cigarette smoking, and nicotine use disorder, may represent a prime area for genomics implementation efforts in behavioral medicine. Mounting evidence of the potential clinical relevance of emerging genomic applications for populations that smoke, including variation in nicotinic receptor subunits and nicotine metabolism genes that influences smoking behavior, suggests that we may be approaching the point of integrating smoking-related genetic information at the point of care [13–28]. There remains a need, however, to understand the potential utility of integrating genomic advances known to influence cigarette smoking, which remains one of the largest causes of preventable death in USA. For emerging genomic applications such as smoking-related genetic information, we need to first determine the readiness and appropriateness for implementation using processes that account for the evolving evidence base. If considerable supporting evidence suggests that implementation may be imminent, then well-developed frameworks for implementation research, such as the Consolidated Framework for Implementation Research (CFIR) [29], should be leveraged to guide pre-implementation research on smoking-related genomics.

In this article, we expand the early-stage work in genomics implementation by focusing on emerging evidence of the role of genetic variation on smoking behavior as a specific example of a promising genomic innovation in behavioral medicine. We apply existing frameworks to (a) demonstrate the current evidence base for the identified genomic application(s), (b) specify remaining and addressable knowledge gaps and, if warranted by the evidence base, (c) point to key implementation issues (contextual factors, strategies, outcomes) that should be considered prior to use of specific genomic applications in practice. As such, this report advances the work of multiple scientific fields—genetics, by demonstrating a path toward implementation of emerging and potentially beneficial genomic innovations in behavioral medicine, and implementation science, by applying its methodologies to enhance the use of genomic information to improve health care and population health, roles that have remained largely undefined to date.

Present study

In this report, we use the α5 nicotinic cholinergic receptor subunit gene, CHRNA5, as an example of an emerging genetic variant in behavioral medicine. We focused on variation in this gene for its robust connection with smoking heaviness, lung cancer, and chronic obstructive pulmonary disease in high-powered genome-wide association studies (GWAS) [13–23]. Our primary aim was to determine the current state of readiness for implementing genomic applications involving CHRNA5 variation and smoking outcomes, using a process that could be translatable to a wide range of genomic findings in behavioral medicine that have, to date, fallen outside the purview of genomics implementation research and practice.

CHRNA5 and smoking behaviors and diseases (the backdrop)

The past decade has witnessed rapid advancements in knowledge of genetic biomarkers associated with tobacco-related morbidity and mortality. Extensive GWAS research has unequivocally demonstrated that variation in CHRNA5, the gene encoding the α5 nicotinic receptor subunit, predicts smoking heaviness, delayed smoking cessation, lung cancer, chronic obstructive pulmonary disease, and early mortality [13–23].

The first large-scale genetic analysis of nicotine use disorder was conducted in 2007 and identified the CHRNA5 single nucleotide polymorphism (SNP) rs16969968 as increasing risk for nicotine use disorder and heaviness of smoking [13, 30]. The association between this variant and smoking-related phenotypes was subsequently validated at a genome-wide significance level by multiple independent groups [17, 19, 31]. In genome-wide association meta-analyses that included over 73,000 subjects, rs16969968 had a highly significant association with cigarettes smoked per day, p = 5.57 × 10⁻⁷² [32]. Functional studies have indicated that rs16969968 causes an amino acid change from aspartic acid to asparagine in the α5 nicotine receptor subunit [14]. This amino acid change alters receptor function and influences risk for smoking behaviors [33].

It is well-established that this same variant that influence smoking behaviors also contribute to increased risk for lung cancer and chronic obstructive pulmonary disease [16–19, 34, 35]. More recently it has been shown that lung cancer patients with a history of smoking who have the high-risk rs16969968 genotype had a 4-year earlier age of diagnosis of cancer, a clinically important genetic finding given that fewer than 50% of lung cancer patients survive more than 1 year following diagnosis [22]. Importantly, those at highest genetic vulnerability can reduce the increased risk of lung cancer by quitting smoking [36]. These findings suggest that the CHRNA5 variant rs16969968 affects lung cancer diagnosis through its effects on smoking.

Although the majority of research has been conducted with populations of European ancestry, the association between CHRNA5 genetic variation and smoking-related behaviors has been examined across populations. Studies have confirmed that α5 nicotinic receptor subunit variation is associated with risk for heavy smoking in African-ancestry subjects as well as in European-ancestry subjects [37–40]. Despite diverse genetic backgrounds, a meta-analysis of over 22,000 individuals of European, Asian, and African descent who smoke found that rs16969968 was associated with heavy smoking in all three populations [41]. These cross population findings strengthen the evidence from functional studies supporting that the rs16969968 variant alters susceptibility to nicotine use disorder.

CHRNA5 and smoking cessation (focus of our review)

We conducted a review of the scientific literature involving CHRNA5 genetic variation and smoking cessation to examine another aspect of smoking behavior which is clinically important. We searched the PubMed database for all articles using both the terms “CHRNA5” and “smoking cessation,” which yielded 57 results. We limited results to human studies reporting on primary research that mentioned these search terms in the Title, Abstract, or Subject fields, which narrowed the results to 33 articles. We added one article [42] which involved the return of CHRNA5 results but did not include this term in the Title, Abstract, or Subject fields. These 34 articles were reviewed full-text for inclusion; 13 articles did not report any data on a potential relationship between CHRNA5 variation and smoking cessation and thus were excluded, which left 21 articles for inclusion in this review [23–28, 42–56]. Supplemental File 1 presents relevant effect sizes, sample sizes, study designs, and variants highlighted for each of these articles, as well as a summary statement on the genomic application(s) supported, if any, by the findings.

Though variation in CHRNA5 predicts delayed smoking cessation at a population-based level, more equivocal at this point is whether or not CHRNA5 variation is associated with more favorable response to pharmacological treatment and success of pharmacotherapy for smoking cessation [24, 25, 27, 28, 49, 57]. For example, some studies have reported a genotype-by-treatment interaction, whereby those with the high-risk genetic variants are more predisposed to have difficulty quitting without treatment, and this genetic risk can be ameliorated by pharmacological treatment [25, 26]. However, other studies have found no evidence of an association of variation in CHRNA5 and smoking cessation nor a genotype-by-treatment interaction and have concluded that CHRNA5 variants are unlikely to be useful for “personalizing” treatment for smoking cessation [48]. Further complicating this work is that multiple variants marking variation in CHRNA5 have been studied.

As informed by our review, however, emerging evidence suggests that CHRNA5 genetic variation influences smoking cessation, which may have implications for the effectiveness and side effects of specific smoking cessation medications [24–28]. The majority of these studies were retrospective in nature, however, adding some methodological concern that limits the impact of the findings. Consequently, the clinical validity and utility of using CHRNA5 variation to predict the likelihood that a patient will successfully quit smoking with the use of pharmacological treatments—including the Food and Drug Administration (FDA)-approved medications of nicotine replacement therapy, varenicline, and bupropion—is currently unclear. This comprises a ripe area for investigation moving forward, including adequately-powered prospective clinical and pharmacogenetics trials.

Prior meta-analyses [58, 59] found that communicating genetic risk estimates for downstream health conditions has little or no impact on reducing smoking behaviors. Previous trials have presented smokers with categorical disease risk estimates for various conditions, including respiratory illness and lung cancer, based on genomic risk profiles and found no behavior change benefit. However, these meta-analyses reviewed studies of how returning information on genetic variation associated with health outcomes other than smoking behavior may in turn alter a person’s likelihood to quit smoking. Unlike these previous meta-analyses which studied genetic variation associated with other medical conditions, variation in CHRNA5 is directly associated with the behavior of smoking (i.e. heaviness). Therefore, it is possible that genetic information about variation in this gene would be more salient and actionable as it allows for new types of messaging specific to the benefits of smoking cessation. For instance, return of CHRNA5 results enables interventions to convey more specific risk of disease associated with smoking as well as potential benefit of pharmacotherapy for smoking cessation, which may be a more compelling message. In one clinical trial of college students who smoke, receipt of CHRNA5 genetic results indicating above average risk of nicotine use disorder increased perceived risk of smoking and appeared to motivate smoking cessation [42]. While these data yield some tentative support for the potential benefit of returning CHRNA5 results, it should be noted that heightened perceived risk and enhanced motivation to quit smoking do not inevitably lead to smoking cessation. An adequate trial to test this hypothesis would involve a comparison of the risk magnitudes between CHRNA5 and other previously trialed genetic markers on smoking-related disease (e.g. lung cancer) and the degree to which these values are diminished by quitting.

Process for ongoing monitoring of rapidly evolving genomic evidence for smoking

In the absence of randomized controlled trials to establish effectiveness data in genomics, the Office of Public Health Genomics of the CDC has emphasized the importance of constructing a chain of evidence for distinct genomic applications whereby key questions are answered to inform treatment and management decisions [60]. While individual studies and systematic reviews have contributed to and synthesized the knowledge base related to CHRNA5 genetic variation and smoking outcomes, to date no such chain of evidence has been created for genomic applications related to these associations. To identify existing evidence gaps and determine the state of readiness for implementation of this genomic application, we constructed a chain of evidence using the well-established ACCE model, which refers to the evidence evaluation components of analytic validity (i.e. how accurately the laboratory test measures the biomarker), clinical validity (i.e. how strongly the biomarker is associated with disease or response to treatment), clinical utility (i.e. how useful the test is in improving clinical care or health behaviors), and ethical, legal, and social implications. Table 1 details the synthesis and adequacy of the evidence for testing genetic variation in CHRNA5. This chain of evidence is the process by which we evaluate the state of readiness for implementing genomic applications involving CHRNA5 genetic variation and smoking outcomes. We also use a system created by the EGAPP Working Group for rating the adequacy of information as convincing, adequate, or inadequate based on the quality of synthesized evidence, based on the proportion and extent to which each ACCE question is addressed, for each component of the chain of evidence [60].

Table 1.

Chain of evidence for genomic applications involving CHRNA5 genetic variation and smoking-related outcomes

	Key question	Synthesized evidence	Sources	Adequacy of evidence
Analytic Validity	How often is the test positive when the variant is present?	>99.9%	[76–79]	Adequacy Rating: Convincing Rationale: All elements of analytic validity indicate exceptionally high reliability and validity of genetic testing
	How often is the test negative when the variant is not present?	>99.9%
	Is an internal quality control program defined and externally monitored?	Yes, in CAP CLIA certified laboratories
	Have repeated measurements been made on specimens?	Yes, with >99.9% reproducibility
	How similar are results obtained in multiple laboratories using the same, or different technology?	>99.7% similar between HapMap and NHLBI GO Exome Sequencing Project
	What range of specimens have been tested?	Millions
	How often does the test fail to give a useable result?	<1%
Clinical Validity	How often is the test positive when the disorder (smoking, lung cancer, chronic obstructive pulmonary disease) is present? How often is the test negative when the disorder (smoking, lung cancer, chronic obstructive pulmonary disease) is not present?	The presence of each additional risk allele is associated with 20–30% higher odds of nicotine use disorder, lung cancer, and chronic obstructive pulmonary disease	[16–19, 23]	Adequacy Rating: Adequate Rationale: Nicotine use disorder is a common disorder with fairly high heritability, and CHRNA5 has been found through multiple GWAS to be associated with smoking heaviness and several smoking-related diseases and mortality across diverse populations
	What is the prevalence of smoking, lung cancer, and chronic obstructive pulmonary disease in this setting?	Common in the general population (~20%)	[80, 81]
	Has the test been adequately validated on all populations to which it may be offered?	Validated in African, East Asian, European populations; this work is ongoing	[37, 39, 41, 76]
	What are the genotype/phenotype relationships?	CHRNA5 variation and nicotine use disorder, lung cancer, chronic obstructive pulmonary disease, morbidity/ mortality	[13–23]
Clinical Utility	What is the impact of a positive (or negative) test on patient care?	Expected impact is to drive pharmacotherapy and optimize treatment effectiveness; further testing is necessary	[24, 26, 27]	Adequacy Rating: Inadequate Rationale: Evidence is lacking but there is a strong premise for utility: test results are actionable and have applications in both clinical and direct-to- consumer contexts; reach is increasing, and cost is decreasing. Effectiveness trials are needed at this time.
	Is there an effective remedy, acceptable action, or other measurable benefit?	Yes, the genomic applications of disease risk stratification, optimizing smoking cessation treatment, and motivating smoking cessation and reduction	[13–23, 26, 27, 57]
	Is there general access to that remedy or action?	Effective smoking cessation medications (nicotine replacement therapy, varenicline, bupropion) and counseling exist; disease risk estimates can be modeled	[82, 83]
	Is the test being offered to a socially vulnerable population?	Individuals who smoke tend to have lower income, lower education; More research is needed on managing risks of test introduction.	[80, 81]
	What are the financial costs associated with testing?	Cost is unknown. In the example of 23andMe, the potential price of genome array is $199	[77]
	What are the economic benefits associated with actions resulting from testing?	Currently unknown	—
Ethical, Legal, and Social Implications (ELSI)	What is known about stigmatization, discrimination, privacy/confidentiality and personal/family social issues?Are there legal issues regarding consent, ownership of data and/or samples, patents, licensing, proprietary testing, obligation to disclose, or reporting requirements?What safeguards have been described and are these safeguards in place and effective?	No existing work on ELSI for the current genomic applications. This area is wide open for study, creating an opportunity for research. May be particularly difficult to assess these issues for specific genetic tests.	[61]	Adequacy Rating: Inadequate Rationale: Evidence is needed to inform the questions. There may be a need for additional questions to address ELSI more comprehensively

Analytic validity

The genomics field has reached a point where genetic testing for low frequency (0.5%–5%) and common (>5%) variants yields highly reproducible results, indicating strong analytic validity. For analytic validity, the issues of sensitivity (How often the test is positive when a variant is present) and specificity (How often the test is negative when a variant is not present) are particularly important; both of these components are exceptionally high as they approach values of 100%. Altogether, we deemed analytic validity for genomic applications to be convincing, as the evidence stems from collaborative studies that used large panels of well-characterized samples, with consistent and generalizable findings [60].

Clinical validity

For clinical validity, the issues of sensitivity (How often the test is positive when the disorder is present) and specificity (How often the test is negative when a disorder is not present) are again of central importance. Given that CHRNA5 informs probabilities of smoking-related behaviors and outcomes, these values are lower and less robust than those related to analytic validity yet still fairly strong. Importantly, high-risk genotypes have been validated as contributing to disease risk across populations, and multiple genotype/phenotype relationships have been established as genetic variation in CHRNA5 is associated with increased risk of heavy smoking, delayed smoking cessation, smoking-related disease, and mortality. Altogether, we deemed clinical validity for these genomic applications to be adequate, as the evidence stems primarily from well-designed case-control studies and systematic reviews [60].

Clinical utility

For clinical utility, the existing evidence is sparser. Genetic variation in CHRNA5 is associated with response to smoking cessation pharmacotherapy in some studies [24, 26, 27], suggesting that a positive or negative test may impact patient care by driving pharmacotherapy decisions by patients and health care professionals. However, the financial costs, economic benefits, and degree to which the probabilities of smoking-related behaviors and diseases generated from genetic test results are actionable in clinical settings remains unclear. Therefore, we deemed clinical utility for these genomic applications to be inadequate and largely a function of preliminary or unpublished data and expert opinion [60]. As we discuss in a later section, future research should focus on addressing this question of clinical utility.

Ethical, legal, and social implications

The primary issues for Ethical, legal, and social implications (ELSI) involve concerns about stigmatization, discrimination, and privacy and confidentiality associated with the genetic test; legal issues about consent, ownership of data and/or samples, and obligation to disclose; and the presence and effectiveness of any safeguards. Although ELSI questions have historically been understudied relative to the other ACCE components [61], efforts are currently underway to address these issues more comprehensively within the genomics field. For example, the Genomics and Population Health Action Collaborative under the National Academies of Sciences, Engineering, and Medicine is leading the public health genomics field in understanding health inequalities related to genetic testing and return of results. Very little, if any, work has been done to address specific issues for individual genetic tests such as CHRNA5, however. As such, we deemed ELSI for these genomic applications to be inadequate, reflecting a key opportunity for future research.

Key applications and targets of genomic information for smoking

Significant genomic advances have been made in behavioral medicine, including the well validated association of CHRNA5 genetic variation and smoking-related behaviors and diseases. It is important to explore whether this considerable body of evidence should be translated into practical genomic applications that are integrated routinely in direct-to-consumer and clinical contexts. There are potentially at least three applications in which the field of behavioral medicine could readily apply genomics-informed probabilities involving smoking behaviors and diseases based on CHRNA5 genetic variation.

Application 1: identifying elevated disease risk in individuals who smoke

Though all individuals who smoke are at increased risk for smoking-related diseases, variation in CHRNA5 alters this risk. Risk stratification, a goal of precision medicine, at the clinical and public health level is a key application of information on smoking-related genetic variation for lung cancer, chronic obstructive pulmonary disease, and early mortality. Incorporating this key genomic information into patient care may facilitate the identification of increased risk level for smoking-related diseases to inform disease prevention interventions and prioritize care management efforts such as increased screening (e.g. lung cancer screening) for those at elevated risk. Further, delivering this type of personalized health information directly to individuals more broadly is important in its own right to inform health-related decision making and is well-aligned with patient-centered approaches to care.

Potential health impact

Improve risk stratification; Enhance precision care and population health management; Reduce risk of morbidity/mortality.

Target of implementation

Individuals, Clinical providers, Public health practitioners.

Summary

This specific application of the CHRNA5 variant and heaviness of smoking and risk of lung cancer and chronic obstructive pulmonary disease has the strongest supporting evidence base and thus carries the highest readiness for implementation.

Application 2: personalizing treatment of smoking cessation

The clinical use of genetic results to inform and optimize smoking cessation pharmacotherapy reflects a key application of precision medicine. In our review of the literature, 16 out of 21 (76%) of included studies provided some degree of support for the potential application of either personalizing smoking cessation treatment or motivating smoking cessation. Of the 21 included articles, 15 reported on cohort designs, 5 reported on case-control studies, and 1 reported on a randomized controlled trial (Supplemental File 1). Given the limitations of the study designs used to date and the lack of intervention evidence (e.g. RCT) supporting the personalization of smoking cessation treatment using CHRNA5, the support indicated in these studies must be considered tentative at this time. Currently, using CHRNA5 results to stratify those at higher risk of difficulty with smoking cessation as well as potentially guide smoking cessation treatment is a promising area for further investigation but not one with sufficient evidence yet. Genomic information on CHRNA5 predicts an increased risk of failed smoking cessation and thus increased need for pharmacotherapy (i.e. nicotine replacement therapy, varenicline, bupropion) and counseling. Genetic variation in CHRNA5 may inform the likelihood of pharmacotherapy to yield maximal effectiveness and minimal side effects. Currently, there is a lack of evidence that CHRNA5 variation is associated with smoking cessation after nicotine replacement therapy; however, this represents only one aspect to personalizing smoking cessation treatment based on smoking-related genetics [62, 63]. Further research will be needed to clarify the potential utility of genomic results on effectiveness and side effects to smoking cessation medications.

Potential health impact

Improve smoking cessation as well as and reduce side effects of smoking cessation medications by optimizing treatment.

Target of implementation

Patients, Clinical providers.

Summary

This specific application of the CHRNA5 variant and smoking cessation has a moderate amount of support but evidence of clinical utility is needed before implementation research is warranted.

Application 3: motivating behavior change and smoking cessation

Previous meta-analyses [58, 59] have shown no benefit to providing individuals who smoke with disease risk estimates for respiratory illness and lung cancer to motivate risk-reducing behavior. Therefore, caution must be taken when considering the viability of other specific genomic applications to motivate smoking cessation. However, the return of genetic test results for CHRNA5 variation, which would communicate risk information specific to smoking behaviors, could potentially help to motivate cessation. As a tangible example, individuals who smoke could be presented, either direct to consumer or in coordination with a care provider, with personalized and targeted genomic profiles that communicate categorical disease risk (e.g. not elevated, elevated, highly elevated) for lung cancer and chronic obstructive pulmonary disease, based on their CHRNA5 genotyping results. As others have previously noted [7], the greatest benefits both in pharmacogenomics and in behavioral health approaches may be using an individual’s genetics to tailor interventions or provide novel recommendations for how to approach smoking cessation treatment.

Potential health impact

Enhance likelihood of smoking cessation or reduction; Encourage use of smoking cessation medication; Reduce risk of morbidity/mortality; Satisfaction and empowerment.

Target of implementation

Patients, Smoking populations in the community.

Summary

Consistent with prior meta-analyses [58, 59], this specific application of the CHRNA5 variant and smoking cessation has the weakest supporting evidence base and is currently inadequate for recommending implementation.

Overall summary

Ultimately, the effectiveness or ineffectiveness of returning genetic results on changing smoking behaviors should not alone be viewed as a deterrent to implementation for other purposes. If found to be clinically useful, incorporating CHRNA5 genetic results for informing smoking-related disease risk and/or personalizing smoking cessation treatment remains an important goal.

Action steps needed to move genomics in behavioral medicine forward

Given that analytic validity and clinical validity have been reasonably established for genomic applications involving CHRNA5 variation and smoking behaviors, the field of genomics in behavioral medicine should focus its efforts on establishing clinical utility through two types of research: (a) Intervention trials and (b) pre-implementation research. Intervention research (i.e. clinical trial) is needed to establish clinical utility to support recommendations for large-scale return of CHRNA5 results to patients, integration and use of specific genomic applications in routine care settings, and to develop useful decision aids that help patients and providers make informed healthcare decisions. Separately, pre-implementation research helps ensure consideration of the feasibility of incorporating genomics-informed probabilities into treatment workflows, appropriate messaging strategies to patients (e.g. how to convey risk to optimize benefit and minimize risk of harm), adapting and refining treatment algorithms as new data emerge, and implications for modifying care according to the genomic information.

Given the strong and rapidly growing evidence base for CHRNA5 genetic variation and smoking-related outcomes, it is not premature to consider the eventual process of integrating related genomic applications into clinical and community settings via pre-implementation research [1, 6]. Further, as clinical utility in part concerns the usability and usefulness of genomic applications within real-world settings, pre-implementation research can be conducted concurrently with intervention research and help to establish clinical utility. As a tangible example, a recent study found that providers agreed with the viability and usefulness of genomics implementation but disagreed on who was responsible for acting on a result, whether patients should be notified, and how to assign clinical responsibility. These are system-level issues that would be appropriately addressed through pre-implementation research [64].

We acknowledge the rationale of waiting for sufficient evidence of clinical utility prior to conducting implementation research. We present Fig. 1 to conceptualize our view of how pre-implementation and implementation research can be appropriately considered when the evidence base for a genomic innovation is rapidly evolving. Certainly, large-scale implementation research would not be appropriate without clear indication of clinical utility, such as with the standard of randomized controlled trials demonstrating effectiveness. To speed the translation of findings into clinical care, intervention trials and pre-implementation research can be done in parallel so that when a threshold is passed demonstrating clinical utility, large scale implementation research can then be undertaken. With this model, pre-implementation research need not wait for sufficient effectiveness data, as in the case of hybrid designs where early-stage implementation research is deemed appropriate despite insufficient evidence of effectiveness [65]. Consistent with this approach, it may be worth considering the timing of implementation research along a “continuum of readiness” based on the current evidence base, with pre-implementation research being more appropriate for genomic innovations with promising yet insufficient evidence for implementation (e.g. clear clinical validity but unclear clinical utility), and larger-scale and more active implementation research being reserved for genomic innovations with clearer clinical utility.

Fig 1.

Linking the chain of evidence for genomic innovations to advance the readiness for implementation.

Additionally, circumstances exist in which current evidence appears promising or where real-world implementation is likely to outpace synthesis of effectiveness data, as is the case with other innovations such as mobile health technologies. Early-stage implementation research may help to ensure that the system adopting the genomic innovation is prepared to apply it in a well-planned and appropriate way that engages and incorporates input from multiple stakeholders. In fact, a significant amount of implementation research on smoking-related genomics has already been conducted, particularly regarding perceptions and expectations of patients and providers [8, 66, 67]. Rather than arguing that the field needs to move into implementation research, we advocate consideration of a broader range of implementation issues. In addition, pre-implementation research can inform other genomic applications more broadly, which reduces the risks associated with the opportunity costs of investing limited resources for this purpose.

Building on existing implementation frameworks for genomics in behavioral medicine

Roberts and colleagues recently reviewed the current state of implementation science in genomic medicine, finding that the use of implementation science frameworks, measures, and strategies are “severely limited” in genomic medicine studies, thwarting progress toward precision medicine [6]. In line with this review, we advance the notion that existing implementation frameworks, including the CFIR and Proctor’s framework for implementation research [29, 68], provide useful starting points for implementing genomics in behavioral medicine. Proctor’s framework encourages the specification and tailored use of implementation strategies [69], which represent the actions that enable the integration and use of evidence-based genomic innovations in clinical and community settings; these strategies can be grouped into six categories: planning, educating, financing, restructuring, managing quality, and attending to the policy context [70]. The Proctor framework also encourages the measurement of implementation outcomes such as the acceptability, feasibility, costs, and sustainability of the genomic innovation; in combination with effectiveness outcomes, these can be considered key determinants of implementation success [71].

Using the CFIR, the context for incorporating genomic applications in practice could be characterized through the broad categories of Innovation Characteristics (current strength and quality of the evidence for CHRNA5 and smoking cessation), Outer Setting (social, political, and economic climate for genomics implementation), Inner Setting (organization-level factors influencing the use of genomic information), and Characteristics of Individuals (knowledge, attitudes, and other personal factors regarding the use of genomic information) [29]. Although not explicitly using CFIR, Oishi and colleagues presented a conceptual model of genomics implementation that focused primarily on organization-level factors (e.g. organization innovativeness) that influence the adoption and delivery of genomic medicine [72]. This framework could be further specified to also attend closely to characteristics of the innovation itself (e.g. interface with the end user) and external environment (e.g. regulatory bodies, payment models) that are key to successful implementation.

For maximal applicability, genomics-focused examples of implementation factors may be important to highlight within these existing frameworks. For instance, genomics implementation research should consider the rapidly evolving evidence base for many genomic applications and the ability to personalize interventions. It may also be necessary to account for commonly held beliefs of genetics exceptionalism, which may lead potential users of genomic information to hold these innovations to a higher standard (e.g. require more evidence of clinical utility) than other health innovations. However, genomics implementation efforts can be enhanced and accelerated by leveraging existing implementation frameworks originally conceptualized for more traditional evidence-based innovations. Building on the findings from the Roberts et al. [6] systematic review, we encourage greater consideration of contextual factors, implementation strategies, and implementation outcomes as central elements of pre-implementation research in behavioral medicine genomics.

The value of pre-implementation research for emerging genomic applications

An important next step is to move beyond studying perceptions of genomic application in the abstract or through hypothetical scenarios and instead to trial these innovations, such as providing CHRNA5 testing and return of results to individuals who smoke. This testing will provide a more grounded and relevant study of the context for implementing smoking-related genomic applications. There are at least three potential benefits to considering pre-implementation and implementation issues while data on innovation effectiveness are still evolving.

• 1) Accelerating the transfer of genomic innovations into practice: Collecting data on implementation perceptions (e.g. patient and provider attitudes) and resources (e.g. staffing needs) alongside effectiveness and utility testing, as in the case of hybrid effectiveness-implementation studies [65], will bring us to the point of system-ready genomic applications sooner than addressing effectiveness and implementation issues separately in a stepwise approach. We cannot assume that the context in which genomic innovations are being implemented will simply conform to and readily absorb these innovations. Furthermore, waiting until the effectiveness and utility testing is considered complete only further delays the process of transfer into practice.

• 2) Improving the design and packaging of innovations: Rather than attempting to address feasibility concerns long after a genomic application is ready for implementation, we should approach point of care delivery of genomic information using a design for dissemination lens [73]. This is best achieved when multiple stakeholders (e.g. geneticists, technology developers, clinical leaders and staff, systems scientists) work together to align the content and format of the innovation with the goals and characteristics of the prospective adopters prior to implementation. Pilot tests and rapid small tests of change that identify implementation barriers may pinpoint new ways of presenting the genomic data or integrating it into existing systems and processes to make this information more usable for patient and provider stakeholders [74].

• 3) Refining organizational and system processes involved in implementation: Trialing implementation efforts, pilot testing implementation strategies, and modifying these approaches as the evidence base for CHRNA5 evolves will better prepare healthcare settings for implementation. These processes may create a “warming up” phenomenon, ensuring that the context within healthcare organizations is more “nimble” and receptive for timely uptake of the genomic applications, as envisioned in a learning health care system [1, 75]. To the extent that the healthcare field agrees that genomic innovations will increasingly be used to inform and improve care—a widely accepted view—we must also prioritize health services and systems research that will yield translatable knowledge about the most effective, efficient, and acceptable ways to integrate genomics into care.

CONCLUSION

Genomics implementation, as exemplified by smoking-related genomic applications in behavioral medicine, is a burgeoning area of research focused on optimizing the integration of genomic findings into community and clinical settings. This line of research may facilitate the translation of basic science findings on established genetic variants that alter behavior and risk of disease (e.g. genetic variation in CHRNA5) to inform care management decisions, personalize treatment, and potentially help to motivate smoking cessation. This review demonstrates support for genomic applications involving testing CHRNA5 genetic variation and health related outcomes associated with smoking. We also recognize that genetic testing of CHRNA5 is the tip of the iceberg with smoking behaviors and smoking-related diseases. For instance, testing of genetic variation in nicotine metabolism genes (e.g. CYP2A6) also represents an important avenue to personalize treatment for nicotine use disorder and target pharmacologic treatment for smoking cessation. In a prospective randomized controlled trial, investigators validated an algorithm for treating nicotine use disorder which increases efficacy and reduces side effects using the nicotine metabolite ratio (NMR), a proxy for CYP2A6 function [20, 57]. Discovery of new genetic variants with potential clinical utility continues to advance as well, signaling a need to begin examining implementation research questions to address the integration of imminent genomic applications. This review demonstrates the current state of readiness for incorporating CHRNA5 genetic test results to reduce smoking behaviors and associated diseases and highlights key research and practice gaps. In sum, this report brings the behavioral medicine field closer to implementation of promising genomic innovations for optimal population health.

Supplementary Material:

Supplementary material is available at Translational Behavioral Medicine online.

Compliance with Ethical Standards

Conflict of Interest: The authors do not have any interests that might be interpreted as influencing the research.

Ethical Approval: This article does not contain any studies with human participants or animals performed by any of the authors; as a result, it was not necessary to obtain informed consent as part of this study.