Targeting Problems and Adhering to Standards: A Pathway to Uniform Clinical Laboratory Processes that Maximize the Benefits of Pharmacogenomics

Abstract

Commercial laboratories often have differing approaches to pharmacogenomic testing, which detects the presence of germline genetic variants associated with drug response. Setting standards for clinical laboratories and defining criteria for clinicians to consider when choosing a testing laboratory can help maximize the benefits of pharmacogenomics. Here, we propose specific laboratory practices that should be discontinued or established and followed as standard practice. We also propose ideal practices that would advance pharmacogenomics in clinical care.

CONSIDERATIONS FOR REPORTING

The value of pharmacogenomic testing is becoming increasingly recognized by clinicians and patients, and testing is being adopted across clinical specialties and settings to improve patient outcomes. ¹ At least 76 clinical laboratories in the United States offer pharmacogenomic testing (https://www.ncbi.nlm.nih.gov/gtr/), though the gene content, technology, and method of interpreting and reporting out results vary considerably. ² , ³ , ⁴ This lack of consistency can cause misunderstanding and misapplication of pharmacogenomic results and potentially harm patients. ⁴ , ⁵ Our objective is to discuss four specific elements of pharmacogenomic laboratory practices that are informed by our experiences in the United States but are relevant across international settings: genotype–phenotype interpretation, level of gene‐drug evidence, laboratory report transparency, and laboratory test technology. These elements are also summarized in Table 1 .

Table 1.

Lab practices and recommendations for ‘do's’ and ‘don'ts’

Area of concern	“The do's”: Practices to follow	“The don'ts”: Practices to avoid
Genotype–phenotype interpretation	Follow PharmVar/CPIC standards for nomenclature, provide information in test report to define variant to allele, genotype to phenotype, and report gene activity score when appropriate	Make recommendations and interpretations based on varied sources and internal proprietary algorithms
Level of gene‐drug evidence	Focus on high evidence gene‐drug pairs and clearly distinguish between high evidence, clinically actionablea results and low evidence, not actionable results	List medications with no FDA labeling, CPIC, or DPWG guidelines
Lab report transparency	Address test limitations and caveats, such as reference (*1/*1) result not guaranteeing a normal function/response	Overly simplify results through ‘bucket’ system of coded categories, such as traffic lights (green, yellow, or red) or icons (stop sign, alarm bell)
Lab test technology	Detect AMP recommended allelesb (tier 1 at minimum, tier 2 as ideal) and in pharmacogenes without AMP guidance, detect PharmVar defined allelesb with CPIC and/or DPWG defined clinical function	Have limited panel of allelesb that are not representative of diverse geoancestral populations

CPIC, Clinical Pharmacogenetic Implementation Consortium; DWPG, Dutch Pharmacogenetics Working Group; FDA, Food and Drug Administration; PharmVar, Pharmacogene Variation Consortium.

^a Gene with CPIC guideline, FDA labeling, and/or DPWG labeling.

^b Allele could refer to a single variant or a haplotype, which includes multiple variants that might be defined in star allele nomenclature.

Level of gene‐drug evidence

While many gene–drug associations are described in the literature, not all are clinically actionable. ¹ , ³ Pharmacogenomic guidance is available through several international organizations, though this paper will focus on the U.S. Food and Drug Administration (FDA), international peer reviewed guidelines created by Clinical Pharmacogenetics Implementation Consortium (CPIC) (https://www.clinpgx.org/cpic), and the Dutch Pharmacogenetics Working Group (DPWG). ⁶ , ⁷ CPIC rates the strength of evidence for each gene–drug association on a scale from A through D, with level A or B having sufficient data to warrant clinical action. Only results in gene–drug pairs with clinical guidelines should be used to guide medication decisions, and laboratories should focus their test offerings on actionable pharmacogenes as defined by these organizations. However, many tests include genes and/or medications that are not actionable, which risks the results being misunderstood, misinterpreted, and inappropriately acted on. ⁴ , ⁵ Laboratories should distinguish between pharmacogenes and/or medications that are clinically actionable from those that are not, as the burden is otherwise on clinicians to identify which results can be used in patient care. ¹ , ³ , ⁵ This can exacerbate barriers in the uptake of pharmacogenomics and diminish the perceived validity and value of the field.

Interpretation of genotype–phenotype results

There is a lack of transparency across laboratories in translation of variant to allele, as well as lack of consistency following standards in defining alleles and translating genotype to phenotype. ³ , ⁴ , ⁵ While many laboratories follow the definitions and guidelines set by CPIC and/or DPWG, other laboratories make interpretations based on internal proprietary algorithms. We advise against this practice and encourage use of consensus phenotype interpretations, when available, so that interpretations are consistent across laboratories. Laboratories should also define variant to allele, genotype to phenotype, report gene activity score when applicable, and clearly state the consensus body used in making phenotype interpretations. It is essential to use standardized nomenclature in this process, as inappropriate translation may lead to harm.

Laboratory report transparency

There are various strategies that laboratories use when reporting out test results, some of which risk misunderstanding and misapplication of results in clinical decision making. ⁸ , ⁹ Pharmacogenomic test results are not a deterministic finding for medication response. There are many non‐genetic factors that can impact response, such as diet and underlying medical problems. Laboratories should emphasize that pharmacogenomic results are one of many factors to consider when optimizing medication use.

Test reports should address the caveats associated with “negative” or *1 results, which laboratories could consider more accurately describing as “no variants identified” or “negative for select variants analyzed” when testing involves a genotype assay. The absence of a queried variant in a pharmacogene does not guarantee normal gene function. Performing full gene analysis and detailed classification of variant function can help reduce the risk of missing an informative genotype in a pharmacogene. Patients with a *1/*1 result may still possess actionable findings that remain undetected, especially if allele coverage is not broad enough.

Many laboratories have report structures that oversimplify results and the complexity of medication response. Some strategies of organizing report templates in an intuitive and user‐friendly way can be effective, such as grouping medications by indication and summarizing prescribing considerations in each medication section. Laboratories should avoid reports that “bucket” results into coded categories, such as traffic lights (green, yellow, or red) or icons (e.g., stop sign, alarm bell). Such terminology and coded systems may be intended to succinctly indicate the level of evidence associated with a gene–drug pair or the severity of risk associated with the variant. However, there is no standard definition for what these coded systems mean across labs and how they translate to different guideline levels of evidence (e.g., CPIC Level A recommendation), making them easy to misunderstand and misinterpret as prescription guidance. ⁴ , ⁸ , ⁹ For example, classifying reference (*1/*1) results in green could provide false reassurance that a patient has normal function and will have a normal response to the medication(s) without considering testing limitations and non‐genetic factors. Similarly, flagging a result as impacting gene function and bucketing as “yellow” or “red” could be assumed to have an increased risk for an adverse drug response, which is not always correct. Laboratories should not oversimplify test results in a way that may cause the clinician or patient to avoid certain medications inappropriately. It would be ideal for laboratories to invest in clinical support staffed by pharmacogenomic experts who could be consulted by clinicians with questions.

Laboratory test technology

Most pharmacogenomic laboratories utilize a genotype assay designed to detect the presence of specific variants in a pharmacogene. This strategy limits sequencing costs and reduces the burden of interpreting rare variants with an uncertain impact on gene function. Not all genotype assays are the same; some include broad alleles from diverse geoancestral population groups. Testing should, at a minimum, include known alleles recommended by the Association for Molecular Pathology (AMP), and ideally include tier 2 alleles (https://www.clinpgx.org/ampAllelesToTest) when AMP guidance is available for a gene. We recommend that laboratories clearly state in reports when AMP tier guidance is followed and specify alleles included. Laboratories should consider testing rare alleles that are known to impact gene function but do not meet criteria for tier 1 or 2 status because they are below the population frequency cut‐off. For pharmacogenes without AMP guidance, laboratories should have a consistent approach to selecting well‐characterized alleles that alter gene function. Pharmacogenomic tests with broader allele coverage promote health equity and maximize opportunities for evidence‐based prescribing.

“THE ASPIRATIONAL”: PRACTICES TO AIM FOR

Sequencing & variant detection

When pharmacogenetic tests are designed to look only for specific variants that have a higher frequency in the general population, they can miss other variants that are more rare but clinically relevant. Some actionable genotypes occur more frequently in specific biogeographical ancestry groups, and genotype tests have the potential to exacerbate racial disparities and inequities in medicine if the assay does not detect variants that are present in diverse groups. ¹ , ¹⁰ It is ideal for laboratories to utilize more comprehensive testing methods, specifically full gene sequencing, though additional methods may be necessary to address technical limitations for genes with homologous pseudogenes or complex regions. Laboratories should be transparent about the methodologies used for each gene, including whether genotyping or sequencing is performed, what variants are reported, if other methods are used to improve assessment, and any known limitations to the approaches used. Laboratories may opt to only report out known impactful genotypes defined by consensus organizations such as PharmVar and characterized by CPIC and/or DPWG at the time of testing, but collecting comprehensive genotype data through sequencing allows laboratories to start reporting genotypes that are newly identified to impact gene function and facilitates opportunities for future reanalysis. As the number of clinically actionable genotypes for established pharmacogenes continues to increase, some laboratories may offer provider‐initiated updated reports, similar to reanalysis practices by diagnostic genetic laboratories. These test enhancements and offerings could incentivize clinicians when selecting a laboratory.

Standard test comparables

In order to help providers make informed decisions regarding what laboratory and what test to order, laboratories should provide minimum standard test details. An example of standard information to include is summarized in Table 2 as an overview of what criteria clinicians should consider when comparing laboratories and test options. Laboratory details would ideally be available for comparison through centralized platforms such as the Genetic Testing Registry (GTR). Providing transparent and consistent elements across laboratories would help clinicians assess and compare options to inform ordering decisions. Encouraging a more uniform laboratory comparison structure will hopefully create a “race to the top” model in which laboratories prioritize clinically impactful data and policies that advance patient care.

Table 2.

Proposed test summary information to be provided by commercial labs

Test content	Number of genes on panel
	List of genes on panel
	Number of clinically actionablea genes
	Number of non‐clinically actionablea genes
	Can the lab provide a customized panel of genes?
	Does the test include all AMP tier 1 allelesb?
	Does the test include all AMP tier 2 allelesb?
Report structure	Does the lab follow CPIC standards for reporting genotype nomenclature and predicted phenotypes?
	How are different result types (reference, actionable, etc.) shown in the report?
	Are actionable results and/or clinical recommendations cited?
	Are medications listed on the test report?
	Are medications on test report limited to only clinically actionablea gene‐drug pairs?
	Are testing methods clearly outlined in the report?
	Are the tested allelesb clearly listed in the report?
Lab operations & support	What are sample types accepted for testing?
	Does the lab have a sample report available?
	Test turn‐around time in days
	Does the lab have a clinical pharmacist and/or genetic counselor available to answer questions?
	Does the lab have EHR integration?
	What is the lab's policy for making updates to panels when new guidelines become available?
Billing	Does the lab bill patient insurance directly?
	Does the lab have a financial assistance program?
	Is there a self‐pay price and/or maximum out‐of‐pocket patient cost?

AMP, Association for Molecular Pathology; CPIC, Clinical Pharmacogenetic Implementation Consortium; EHR, Electronic Health Record.

^a Gene with CPIC guideline, FDA labeling, and/or DPWG labeling.

^b Allele could refer to a single variant or a haplotype, which includes multiple variants that might be defined in star allele nomenclature.

Test billing and costs

Providers should consider how a laboratory's billing policies may impact patient access and uptake of pharmacogenomic testing. Laboratories should make efforts to implement transparent billing policies and create financial assistance programs that help reduce out‐of‐pocket costs. Additional research and collaboration between laboratories, payors, and professional organizations could help improve and expand insurance coverage policies by demonstrating the clinical and cost‐saving benefits of pharmacogenomic testing.

Ongoing laboratory support

In addition to creating client‐facing materials with consistent and standard test details (Table 2 ), laboratories should consider providing clinical support and resources to ordering clinicians, many of whom may have limited pharmacogenomic experience and knowledge. This includes post‐analysis support in the form of trained pharmacists and genetic counselors who would be available to assist with clinical consultations. The availability of trained experts to support clinical interpretation and application of test results could maximize opportunities to leverage pharmacogenomics in clinical care settings and reduce burden on healthcare providers.

EHR live alerts and clinical decision support

Ideally, laboratories should have interoperability capabilities with EHR systems to allow for transfer of pharmacogenomic results as discrete data elements. This would allow health systems to build and trigger clinical decision support derived from evidence‐based guideline recommendations in real time at the point of prescribing. Linking results to a live platform such as the ClinPGx Genotype Selection Interface (https://www.clinpgx.org/genotype) would allow users to access and compare ClinPGx‐curated recommendations based on individual genotypes and phenotypes. Interpretations of genotypes and implications for related medications are changing quickly as research progresses, and while it may not be feasible for laboratories to continually reissue reports to include updated medications, clinicians could utilize live platforms to integrate results into electronic medical systems and decision support tools.

CONCLUSION

As demand from clinicians and patients for pharmacogenomic testing to inform broad medication selection and dosing grows, so does the volume of labs offering tests. While it may be challenging for labs to balance following evidence‐based best practices with maintaining a competitive commercial product, there must be a high standard required from pharmacogenomic testing labs to drive quality patient care and positive outcomes. Clinicians should prioritize ordering pharmacogenomic testing from labs with practices that meet the criteria described here related to genotype–phenotype interpretation, level of gene‐drug evidence, lab report transparency, and testing technology. This will help incentivize alignment of lab policies and adoption of established standards in the field. Meeting these standards can improve widespread implementation and successful application of pharmacogenomics into routine clinical care.

FUNDING

No funding was received for this work.

CONFLICTS OF INTEREST

Elizabeth Fieg and Lauren Marcath are full‐time employees at Baylor Genetics. The other authors declare no competing financial interests.