Navigating Pharmacogenomic Testing in Practice: Who to Test and When to Test

Abstract

There is increasing attention on the clinical utility and value of pharmacogenetic (PGx) testing to individualize medication management. Most clinical practice guidelines from medical professional societies do not recommend routine PGx testing, with a few key exceptions. Inconsistent recommendations across clinical practice guidelines, FDA product labeling, and payer reimbursement policies have hampered widespread adoption of testing. Multiple resources exist to aid in the adoption and use of actionable PGx test results in clinical practice; however, most of these resources do not provide guidance on who should receive PGx testing and when—a critical question the clinical community continues to struggle with. There are multiple considerations when answering this question beyond the clinical validity of the drug–gene interaction itself, such as the actionable result frequency, severity of the adverse clinical outcome, predictive power of the PGx test, suitability of alternative treatments, cost, and turnaround time of test results. This perspective discusses these considerations and models for testing including preemptive screening, pretreatment testing, and reactive testing, highlighting advantages and disadvantages of each approach. The authors provide their perspectives on identifying candidates for PGx testing in the current real‐world environment and how that differs from a clinically ideal scenario.

Pharmacogenomics (PGx) refers to the influence of an individual's genome on medication efficacy or toxicity. For this mini‐review, we will focus on germline genetics, which are shared across tissue types and are stable over the lifetime. Despite several examples of PGx testing improving outcomes in real‐world settings, ¹ , ² , ³ , ⁴ PGx testing remains sparsely used in routine practice in the United States. Several reasons for low utilization have been proposed, including inadequate clinician education, test turnaround time concerns, lack of electronic health record (EHR) integration, and poor reimbursement. ⁵

Another contributor to limited PGx testing uptake is the lack of guidance on which patients to test. The Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines are the primary source for PGx‐guided treatment recommendations in the United States, but the guidelines address how to use existing PGx information—not who or when to test. Similarly, the US Food and Drug Administration (FDA) maintains the Table of Pharmacogenetic Associations, which groups drug–gene interactions by the FDA's opinion on whether the evidence supports treatment modifications for certain genotypes has a potential influence on outcomes or solely affects pharmacokinetics. ⁶ There are very few drugs that require pretreatment PGx testing per FDA labeling, again, leaving who or when to test up to clinician discretion.

Likewise, disease‐specific clinical practice guidelines often do not provide guidance or provide guidance that conflicts with other professional bodies on whom to test. ⁷ In some cases, guidelines or FDA labeling may recommend testing for patients from “at‐risk” populations but the definition of “at‐risk” is complex. For example, the American Society of Rheumatology's gout treatment guidelines recommend testing for HLA‐B*58:01, a predictor of severe cutaneous adverse reactions to allopurinol, in patients of certain racial or ethnic backgrounds before starting allopurinol treatment. ⁸ This is based on the risk allele being more prevalent in certain genetic ancestries. Although grounded in logical rationale, applying the evidence to real‐world practice appropriately and identifying patients that have ancestry from these populations is not a straightforward task.

There is a dearth of guidance on when to recommend PGx testing for patients. Further, existing recommendations can be complicated to implement in practice. Here, we address considerations for “who to test” and “when to test” that clinicians may use to aid their decision making for ordering PGx testing.

IDENTIFYING CANDIDATES FOR TESTING

Individual patients may request PGx testing based on their history of adverse events, a desire to avoid trial‐and‐error prescribing, or perceived benefit of personalized care. At the health system level, identifying opportunities for PGx testing is dependent on several factors that converge to affect clinician and patient demand for testing (Table 1 ).

Table 1.

Factors influencing adoption of PGx testing

Clinician awareness +/− patient engagement Prevalence of actionable PGx results Prevalence of medication use Availability of alternative treatment strategies Severity of the adverse clinical outcome Likelihood of the adverse clinical outcome in those with an actionable PGx result Insurance coverage, cost to the patient, and cost to the health system Quality of supporting evidence for the clinical actionability of the drug–gene interaction and clinical utility of PGx testing Testing/treatment changes supported by clinical practice guidelines and the FDA Availability of testing and ease of integration into existing clinical workflows

It is important to estimate how often PGx testing will return an “actionable” result—one that would alter prescribing. An actionable result requires that a patient will be treated with a medication affected by PGx and carries a variant that affects that medication. Identifying the prevalence of prescription orders for the drug affected by PGx or its alternative can be determined via EHR reports detailing order setting, ordering clinicians, and medical specialties. Different methods can estimate the prevalence of actionable variants. One approach, particularly in diverse communities, is to utilize ranged frequencies from CPIC's allele and phenotype data across multiple biogeographic groups. Another method is to evaluate the race and ethnicity of the patients served by the health system and combine this with allele and phenotype frequency tables that are available from CPIC. ⁹ Notably, there are concerns over the conflation of genetic ancestry with race and ethnicity. ¹⁰ The social constructs of race, ethnicity or biogeographic origin are different from genetic ancestry and are often inaccurate surrogates. ¹¹ Clinicians may find this challenging to apply practically to an individual patient as the US population becomes more admixed. ¹²

Another consideration is the severity of the adverse clinical outcome due to the drug–gene interaction. Clopidogrel is less effective in CYP2C19 intermediate or poor metabolizers, increasing the risk of costly and potentially fatal major adverse cardiovascular events after percutaneous coronary intervention (PCI), such as stent thrombosis and myocardial infarction. Alternatively, prescribing metoprolol in CYP2D6 poor metabolizers results in increased metoprolol concentrations and greater decreases in blood pressure and heart rate. This may increase the risk of bradycardia if the drug is not started at low doses. The first example clearly represents a more severe consequence of the drug–gene interaction.

One should also consider how much the test result impacts the likelihood of an adverse clinical outcome. This can be ascertained by measures such as odds ratio, risk ratio, sensitivity, specificity, and predictive values in genetic association studies. ¹³ These measures can vary widely between known drug–gene interactions and clinical outcomes. The odds ratio for the development of severe cutaneous adverse reactions from carbamazepine in HLA‐B*15:02 carriers is >1000, ¹⁴ whereas the high expression genotypes of the serotonin transporter gene SLC6A4 are only weakly associated with the odds of remission in White patients treated with SSRIs (OR ~1.4). ¹⁵ However, PGx associations with lower effect size can still be clinically relevant if the risk allele is common and the outcome is clinically meaningful. The impact of PGx‐guided care in some clinical contexts has been measured in interventional randomized controlled trials or pragmatic clinical trials. Outcomes from these studies, such as relative risk reduction or the number of individuals needed to be treated with genotype‐guided care to avoid one adverse clinical outcome (number needed to genotype), can be used to project the clinical impact of PGx‐guided care vs. standard practice.

The availability and risks of alternative treatment need to be considered. Genotype‐guided dose reductions to prevent ADRs should be supported by data demonstrating non‐inferiority of efficacy or pharmacokinetic parameters. While direct assessment of efficacy outcomes provides the strongest evidence, pharmacokinetic normalization across metabolizer groups can serve as a practical surrogate endpoint when efficacy studies would require prohibitively large sample sizes. At the patient level, alternative treatments may have unacceptable risks or may be unaffordable. For example, in a patient on anticoagulation with multiple bleeding risk factors, a cardiologist may be uncomfortable with prescribing higher potency antiplatelet drugs, such as prasugrel or ticagrelor. PGx testing is most valuable when it leads to modifying care based on the results. It has limited value in cases where alternative treatments are inaccessible or unacceptable for a given patient (e.g., example above with ticagrelor and prasugrel for a patient with high risk of bleeding).

One of the first questions asked by patients and clinicians is as follows: “What does PGx testing cost?” In the United States, as with any laboratory test, the opacity of healthcare expenditures and pricing makes answering this question difficult. Recent updates to Medicare local coverage determinations (LCDs) specify coverage for PGx testing when using medications with CPIC Level A or B designations (> 100 medications). These LCDs are active in 40 states and have been proposed in the remaining 10 states, suggesting that coverage may soon be harmonious for Medicare beneficiaries. Private payers exhibit variable coverage, with very few covering multigene panel testing and none covering fully preemptive screening where the patient is not being prescribed a drug with a potential drug–gene interaction. ¹⁶ The cash price for most multigene panel tests are $300–500, although several laboratories offer financial assistance plans based on household income. Testing in the inpatient setting is often included as care paid for through Medicare's diagnosis‐related group payments, and thus at institution discretion in cost management. The out‐of‐pocket cost is often patient‐specific and difficult to determine a priori. Additional cost‐effectiveness studies, particularly for multigene panels, are needed and efforts could be made in aligning PGx testing with value‐based care or bundled payment models.

There is at least one ongoing effort to systematically quantify the benefit of pretreatment genotyping for individual drug–gene interactions. Recent guidelines from the Dutch Pharmacogenomics Working Group include a “Clinical Implementation Score,” which involves an assessment of clinical consequence of the drug–gene interaction, level of supporting evidence, the number needed to genotype, and availability of PGx information in the European Medicine Agency (EMA) labeling. ¹⁷ The score is intended to characterize testing as “Potentially Beneficial,” “Beneficial,” or “Essential”; however, the EMA labeling criteria and the framing of the allele frequencies to the Dutch population makes the score less applicable to US institutions. Nonetheless, these scores illustrate critical factors that should be considered when evaluating the clinical utility of a test.

At a minimum, PGx tests that have demonstrated clinical utility in prospective trials should be considered for adoption in clinical practice, including those required or recommended for testing based on FDA labels or clinical practice guidelines, and those associated with FDA Boxed Warnings (e.g., abacavir, carbamazepine, clopidogrel, codeine, and rasburicase/pegloticase). In some instances, forgoing testing has led to concerns for malpractice. ¹⁸ In the future, it is theorized that misinterpretation or misapplication of PGx information resulting in administration of the wrong drug or dose is of tangible concern. ¹⁹ Testing may be considered for patients receiving any drug with a CPIC guideline ⁹ or on the FDA Table Section 1 ⁶ (PGx associations for which the data support therapeutic management recommendations), although uptake may be limited by the absence of strong recommendations from clinical practice guidelines ⁷ and the FDA, and variable reimbursement. ¹⁶ Table S1 summarizes drug–gene pairs, whether there is a published CPIC guideline with therapy recommendations and/or FDA guidance (FDA Table and/or drug product labeling) as of March 2025.

CURRENT PGX TESTING MODELS

Table 2 summarizes three testing models, including examples, advantages, and disadvantages.

Table 2.

Comparison of PGx testing models

PGx testing model	Examples of PGx testing model	Advantages	Disadvantages
Preemptive screening	Preemptively screen patients upon admission to the hospital using a multigene panel with downstream clinical decision support for actionable drug–gene interactions (e.g., St. Jude Children's Research Hospital PG4KDS)	Valuable for managing chronic conditions where patients may require multiple medications over time Real‐time access to PGx results at the point of care (no turnaround time) Clinical decision support can facilitate downstream use of results Generation of large datasets for research Greater long‐term cost‐effectiveness	Understanding and interpretation of known functional variants can evolve over time Mostly not reimbursed Typically requires research funding and consenting of patients Challenging to evaluate immediate outcomes (clinical and cost) Larger panels or sequencing may require robust/costly bioinformatics tools
Pre‐treatment testing	Testing of a specific gene or set of genes prior to treatment but based on a diagnosis or medication order in which PGx results may be useful to guide drug prescribing for an index drug or drugs of interest (e.g., DPYD testing prior to fluoropyrimidine dosing)	More likely to be reimbursed than preemptive screening Easier to measure immediate outcomes (clinical and cost) from testing Clinical decision support can facilitate downstream use of results, especially if using a multigene panel	Potential delay in returning results Requires the establishment of a standard operating procedure, protocol, or clinical decision support system to accurately identify candidates for testing Relying on the provider to identify patients may lead to missed opportunities
Reactive testing	Ordering a test after a patient has initiated treatment, often in reaction to a poor or adverse outcome (e.g., PGx testing after failure of first‐line selective serotonin reuptake inhibitor therapy)	PGx results may provide insights that inform alternative treatment options More likely to be reimbursed than preemptive screening	Actionability of the results to guide prescribing are limited Pharmacokinetic‐related PGx results are most beneficial for initial dose selection and titration Applying multigene panel results for preemptive use in the future is possible, but turnaround times required for reactive testing may impede use of larger panels Less cost‐effective since the adverse outcome (therapy failure, toxicity, hospitalization, etc.) has already occurred

Preemptive screening all comers

Preemptive PGx screening refers to the proactive assessment of an individual's genetic profile to predict how they will respond to specific medications before those medications are needed. Several notable programs have successfully implemented panel‐based preemptive screening as part of their research programs (e.g., St Jude Children's Research Hospital, ²⁰ Vanderbilt, ²¹ and Mayo Clinic ²² ). Preemptive screening is particularly valuable in managing chronic conditions where patients may require multiple medications over time. Real‐time access to PGx results for multiple potentially actionable genes at the point of care eliminates the turnaround time associated with testing. This immediacy is particularly beneficial within integrated health systems that share EHRs. The effective use of these results in real time necessitates the implementation of clinical decision support (CDS) systems, which, while available from major EHR vendors, vary in their robustness. Preemptively screening patients on large panels also enables the accumulation of large datasets rich in genomic data for research. Moreover, many preemptive PGx panels have extensive coverage, providing substantial amounts of genomic information at minimal additional costs compared with single gene or smaller multigene panels.

Although an individual's genotype remains constant throughout their life, the understanding and interpretation of known functional variants can evolve over time. Currently, the process for updating these results and ensuring their relevance for the patient over time is not well‐established. Additionally, fully preemptive panel‐based screening (e.g., no index drug or disease that triggers testing) is unlikely to be reimbursed by insurance. Most programs that have initiated preemptive screening have done so under a research protocol in which patients and/or insurance are not billed; thus, grant funding or institutional support is required. Evaluating the outcomes of testing is challenging if the results are not applied immediately. Depending on the technology and complexity of the testing panel, high‐performing bioinformatics tools may be necessary to analyze the data and translate it into actionable genotype and phenotype results. Moreover, secondary findings may arise when using large sequencing or array panels for population health screening, which could necessitate patient informed consent and interdisciplinary collaboration (e.g., genetic counselors and geneticists). There may also be situations in which PGx results for drug–gene pairs deemed non‐actionable by CPIC and/or FDA are available from panel testing, and sites should make a concerted effort to report only those with the highest evidence and actionable recommendations. Overall, there is limited evidence regarding the cost‐effectiveness of fully preemptive screening, highlighting the need for ongoing research in this area.

Pretreatment testing for specific populations

Pretreatment PGx testing refers to the testing of a specific gene or set of genes prior to treatment but based on a diagnosis or medication order in which PGx results may be useful to guide drug prescribing for an index drug or drugs of interest. For example, a pretreatment DPYD genotype test may be ordered upon diagnosis of metastatic colorectal cancer or initiation of a fluoropyrimidine‐based chemotherapy order set. PGx testing that focuses on specific drug–gene interactions is more likely to be reimbursed, particularly when clinical actionability and/or utility has been demonstrated. This may encourage providers to order these tests more frequently than a preemptive panel. This approach facilitates easier measurement of acute outcomes, as results may be utilized imminently and are directly linked to a present or considered medication. CDS can enhance this process by alerting providers when PGx testing is required or recommended based on a patient's diagnosis or the medications being prescribed.

Importantly, numerous drug–gene associations have already demonstrated cost‐effectiveness, ²³ reinforcing the clinical and economic value of targeted PGx testing. Although a specific diagnosis or medication order may prompt the PGx test for a specific gene, these tests can be designed as panel‐based assays, allowing for the preemptive return of other results for potentially actionable genes that may inform downstream treatment decisions, resulting in even greater clinical and economic value. The first randomized study of the impact of multigene panel testing triggered by prescribing of an index drug—performed across seven European countries—demonstrated a 30% reduction in ADRs. ³ Importantly, robust CDS is needed to facilitate the evidence‐based use of results by other providers who may not have ordered the initial test.

One of the primary disadvantages of pretreatment PGx testing is the potential delay in obtaining results if testing is initiated only after a patient is prescribed a treatment or has a documented diagnosis. For certain indications that require prompt treatment, such as PCI with clopidogrel or the treatment of gastrointestinal cancers with fluoropyrimidines, the turnaround time for test results poses a challenge to receiving and applying results prior to administering the first dose. Additionally, effective pretreatment PGx testing necessitates the establishment of a standard operating procedure, protocol, or CDS system to accurately identify candidates for testing. Relying solely on providers to initiate testing could lead to missed opportunities, as this approach demands a comprehensive understanding of which specific tests to order and when to order them.

Reactive testing for specific populations based on poor/adverse outcomes

Reactive PGx testing refers to ordering a test after a patient has initiated treatment, often in reaction to a suboptimal or adverse outcome. An example includes ordering a multigene PGx test for major depressive disorder in a patient who has already failed two medication trials. One advantage is that while the intention was to explain the response to a specific failed medication, the PGx results may provide insights that inform alternative treatment options (e.g., a patient failing citalopram who is a CYP2C19 ultrarapid metabolizer may also inform the provider to avoid other CYP2C19‐metabolized antidepressants). When utilizing a multigene test reactively, additional test results may be available for future preemptive use. Although less informative, single‐gene reactive testing is more likely to be reimbursed than that of preemptive screening, making it a more accessible option for patients. ¹⁶

While reactive testing can help explain why a specific adverse or poor outcome was observed, the actionability of the results to guide prescribing is limited. Furthermore, many pharmacokinetic‐related PGx results are most beneficial for initial dose selection and titration; thus, if initial dosing has already occurred, the information gained from reactive testing is less useful. If an adverse or poor outcome has already occurred prior to testing, this can lead to increased healthcare utilization due to the management of these adverse events, hospitalization, or use of ineffective treatments. Consequently, pretreatment or preemptive PGx testing is more likely to be cost‐effective than reactive testing.

CONCLUSION AND PERSPECTIVES ON THE IDEAL MODEL FOR PGx TESTING

We envision an ideal scenario for PGx testing in which patients undergo comprehensive preemptive screening, such as whole exome sequencing, whole genome sequencing, genotyping microarray, or other comprehensive panels supported by robust bioinformatics and CDS. Given that nearly 99% of patients are expected to carry at least one potentially actionable result from such tests, ²⁴ this approach facilitates evidence‐based downstream utilization of results, as most patients are likely to receive a PGx‐related medication during their lifetime. ²⁵ However, due to various challenges described earlier, this scenario is not currently feasible in most health systems or clinics. A more attainable strategy involves pretreatment PGx testing triggered by a diagnosis or the prescription of a drug with established drug–gene interactions. This method narrows the patient population to those who would potentially benefit immediately from the test results and enhances the likelihood of reimbursement. Furthermore, panel testing initiated by an index drug or drugs with PGx information—akin to the design of the PREPARE trial ³ —would still allow for the preemptive application of additional results when future drugs with potential drug‐gene interactions are considered.

Shared decision making is essential when deciding who and when to test; patients must be informed about the availability and implications of testing, including potential impact on therapeutic response, toxicity, and cost, allowing them to opt out if they choose (exceptions may apply such as when the FDA label or institutional policy requires testing prior to drug prescribing). Expectedly, many patients will rely on their provider's expertise and opinion on whether they recommend testing. Thus, educating healthcare providers is essential to facilitating informed patient choices. Pharmacists are seen as key stakeholders and knowledgeable on drug pharmacology, including the implications of PGx testing. ²⁶ Including a PGx‐trained pharmacist on the care team who is responsible for providing pre‐ and posttest counseling is ideal. Achieving the ideal testing scenario will require several key elements: provider education, inclusion of PGx test recommendations in clinical practice guidelines, standardized payer policies, robust EHR infrastructure to guide PGx‐based recommendations at the point of prescribing, and updates to FDA labels that align with the FDA Table of Pharmacogenetics and CPIC guidelines. ⁶ , ⁹

A tiered approach for testing may be taken depending on the strength of evidence or severity of the drug–gene interaction. For example, when initiating drugs that have an FDA Boxed warning for a drug–gene interaction, an automated PGx test order may accompany the drug order, like a companion diagnostic. Similarly, drugs with an elevated risk of severe toxicities such as fluoropyrimidines may incorporate a PGx test (e.g., DPYD) as part of the order set. Initiating other drugs with a CPIC guideline, testing recommended in FDA labeling, or with data supporting therapeutic management recommendations in the FDA Table may trigger CDS to consider ordering a PGx test or, at a minimum, consider discussing with the patient the implications of testing.

While tremendous progress has been made through efforts such as CPIC and the Pharmacogenomics Knowledgebase (PharmGKB) ⁰⁹ to establish the clinical validity and actionability of drug‐gene interactions, there is a dire need for improved, evidence‐based guidance on who and when to test. This mini‐review provided critical considerations to aid health systems and providers in deciding how to best adopt PGx testing in clinical practice. Patient, clinical, and environmental factors can each influence the decision of who should receive PGx testing, when it should be done, and which testing model should be used.

FUNDING

No funding was received for this work.

CONFLICT OF INTEREST

J.N.P. has served as a paid consultant for VieCure and Clarified Precision Medicine and has received honoraria from Illumina, Inc. D.M.S. has received research funds to the institution from Kailos Genetics Inc. All other authors declared no competing interests for this work.

Supporting information

Table S1