New Fluoropyrimidine Toxicity Variants Reported in DPYS and PPARD

A recent paper highlights new variants and genes associated with severe fluoropyrimidine-related toxicity in patients who were genotyped as negative for the four DPYD variants the European Medicines Agency recommends testing for.

Online ahead of print in the Human Genomics journal is De Mattia et al "The burden of rare variants in DPYS gene is a novel predictor of the risk of developing severe fluoropyrimidine-related toxicity" [PMID: 37946254].

The study sequenced 120 patients with fluoropyrimidine induced grade 3-5 toxicity confirming they lacked DPYD*2A, DPYD*13, c.2846A > T, c.1236G > A-HapB3. The paper reports rare and common variants including DPYS rs143004875-T and PPARD rs2016520-T which were associated with increased risk of severe toxicity.

Insights on CYP2C19 and phenoconversion

Phenoconversion in the PGx context is a drug-drug interaction that impacts a drug metabolizing phenotype such that it mimics the effects of a metabolizer genotype (see blog from October 2022). Historically much of the discussion on phenoconversion has focused on CYP2D6.

A new paper in Frontiers in Pharmacology investigates the phenoconversion effects of different CYP2C19 inhibitors [PMID:37361233]. 

Forty donor liver samples were genotyped for CYP2C19 *2, *3 and *17 and the metabolizer phenotypes predicted. Microsomes were assayed with the probe drug s-mephenytoin and then in the presence of strong CYP2C19 inhibitor fluvoxamine, moderate inhibitors omeprazole and voriconazole and weak inhibitor pantoprazole to look at changes in metabolizer status. 

Excerpts from paper:

“Our results demonstrate that the outcome of a DDI is dictated by both inhibitor strength and CYP2C19 activity, which is in turn dependent on genotype and non-genetic factors including comorbidities. … 

Fluvoxamine, a strong inhibitor of CYP2C19, caused 86% of *1/*17 donors to become phenotypically IM, whereas most of genetically-predicted IMs were converted to a PM phenotype (57%). In accordance with unaltered CYP2C19 activity in patients with gastroesophageal reflux disease taking pantoprazole, weak inhibition by pantoprazole did not induce phenoconversion…

However, the outcomes of DDIs with moderate inhibitors (omeprazole/voriconazole) matched less well to the proposed phenoconversion model by Mostafa et al, which predicted that NMs/IMs convert to a PM phenotype upon moderate inhibition of CYP2C19. In our study, voriconazole, which acts as a moderate CYP2C19 inhibitor, significantly reduced the drug metabolizing capabilities of CYP2C19 by approximately one level (i.e., from a phenotypic NM to a IM). As a result, 40% of the donors (12/30) were converted into IM or PM phenotypes by voriconazole. Though, none of the NMs were converted into PMs, except for one donor who already exhibited impaired CYP2C19 activity in the absence of voriconazole treatment (basal phenoconversion). For omeprazole, phenoconversion into IM or PM phenotypes was even less frequently seen, in only 10% of the donors …

Altogether, our data suggest that CYP2C19 inhibition by moderate inhibitors can result in phenoconversion, but it seems unlikely to result into a PM phenotype for wild-type *1/*1 genotypes.”

There are a number of interesting results and discussion points:

• There is phenoconversion from disease phenotype - namely diabetes. 
• The initial concordance for genotype to phenotype with s-mephenytoin was only 40% and the two CYP2C19*17/*17 did not have ultra-rapid UM phenotype (with Vmax in the low normal NM range). The discussion mentions “other (rare) genetic variants within CYP2C19 could also have influenced the mismatch between predicted and observed activities in our study” but it would have been useful to have ruled out *4. The *17/*17 did produce functional mRNA but the *4 is in the start codon and its impact is on translation not transcription [PMID: 9435198]. 
• There were two *1/*1 outliers with very high UM phenotype that would be interesting to see further genetic analysis of especially given the escitalopram UM CYP2C-haplotype defined by rs2860840T and rs11188059G in [PMID: 33759177]. 

Overall, this paper shows that while phenoconversion exists for CYP2C19 based on disease status and DDIs, the impact is not a simple downgrading of phenotype (e.g. from IM to PM) that can be applied in a consistent manner across subjects. The authors show that even for strong inhibitors, phenoconversion happens in 40%-86% of subjects with no clear way to predict which subjects would experience phenoconversion and which wouldn’t. More research on how DDIs alter patient-predicted genotype to phenotype  is needed to enable better prediction of patient phenotype for PGx drug dosing recommendations.

CYP3A5 genotyping is a more accurate predictor of drug response than race alone

 A new paper in Journal of Clinical Pharmacology from a group at Indiana University [PMID:37042314] implemented genotyping for CYP3A5 in a kidney transplant center.

The team used CPIC guidelines for tacrolimus dosing based on CYP3A5 genotype.

Implementation included provider education and clinical decision support in the electronic medical record.

This study reinforces that CYP3A5 genotype is an important predictor of therapeutic tacrolimus trough concentrations. They demonstrate that CYP3A5 normal and intermediate metabolizers had fewer tacrolimus trough concentrations within the desired range post-transplantation and took longer to achieve therapeutic dose than poor metabolizers. While the authors note they were underpowered to measure outcomes, there was a trend towards transplant rejection or all-cause mortality within the first year of transplant based on CYP3A5 metabolizer phenotype.

The paper highlights how, despite the guidelines from CPIC being published in 2015, the FDA label still currently only has language around race-based dose adjustment rather than giving precise guidance based on genotype:

“The FDA drug label recommends higher starting doses in individuals of African ancestry, but only 70% of African Americans are normal/intermediate metabolizers. CYP3A5 normal/intermediate metabolizers are also found among whites and Asians (East Asian and Central/South Asian) at lower frequencies (14% and 44-55%, respectively).”

“Self-reported African American race is more closely associated with CYP3A5 expresser status than other self-reported race categories, but self-reported race is not an accurate surrogate for genotype.”

The discussion is a reminder that pharmacogenomics can play a key role in reducing bias and fulfilling personalized precision medicine.

“Equality and minimization of bias in healthcare has recently become prioritized by healthcare systems as recognition of racial bias has come to the forefront in many non-healthcare aspects of society”

“One dose standard protocols and using race as a surrogate for genotype can both potentiate racial disparities in tacrolimus dosing. Routine CYP3A5 genotyping is a more accurate predictor of drug response than race alone and deemphasizes race as a biological variable in clinical care”

CYP2C18 and knowledge gaps

Pablo Zubiaur & Andrea Gaedigk have an editorial online ahead of print in Pharmacogenomics calling for the inclusion of CYP2C18 in more studies of drug metabolism [PMID: 36331025].

CYP2C18 is in a cluster on chromosome 10 that has CYP2C18, CYP2C19, CYP2C9 and CYP2C8 that spans 500k bases (NCBI gene browser). The authors comment that CYP2C18 is only included in three PharmGKB pathways (there are actually four: clobazam, diclofenac, warfarin and acenocoumarol), while the other genes of the CYP2C locus are in many. CYP2C19 and CYP2C9 have a volume of data annotated in PharmGKB, CYP2C8 is less populated and CYP2C18 has little (see table below). Similarly, CYP2C19, CYP2C9 and CYP2C8 have haplotypes in PharmVar, while CYP2C18 does not. As Zubiaur and Gaedigk discuss, there are comparatively very few PGx peer-reviewed papers about CYP2C18 and PharmGKB depends on these publications for annotations and pathway creation. We strongly agree with the authors that more studies regarding CYP2C18 would be valuable contributions to the field and look forward to curating them into PharmGKB as they are published.

Like many grant-funded projects, PharmGKB is a small team with limited resources and is unable to manually curate all of PubMed, so sometimes papers are published that we have not yet curated. We encourage PharmGKB users to contact us anytime if they identify important papers for us to curate. If users find knowledge gaps or have recommendations for additional pathway candidate genes, please send the relevant references to feedback@pharmgkb.org. We also encourage users who are interested in collaborating on a drug pathway to contact us.

Curators' Favorite Papers

The growing number of individuals using direct to consumer (DTC) genetic tests has also led to an increase in physicians counseling patients on their results. In an effort to reveal the perspectives of physicians who receive these “unsolicited genomic results” (UGR), the Electronic Medical Records and Genomics (eMERGE) Network has conducted a survey, published in Genetics in Medicine (Physicians’ perspectives on receiving unsolicited genomic results) in which they ask physicians about their perspectives on the actionability of positive test results, what they believe their responsibilities are for UGRs, the impact of UGRs on patients and clinical workflow, and the the process and support involved in returning results. The 25 participants were selected because of their extensive experience as primary care physicians, pediatricians, cardiologists, or oncologists. The primary concern of physicians was that only actionable results should be returned to patients, and that any follow-up should include clinical decision support and be evidence-based. Additional concerns included the additional time that would need to be set aside to integrate results into clinical workflow, increased patient anxiety and confusion, and whose responsibility it was to “own” genetic test results. 

Pharmacogenetic algorithms have been demonstrated to improve the dosing of essential life-saving medications such as warfarin, as well as to reduce the incidence of adverse effects. However, dosing algorithms and recommendations often include genetic variants that come from cohorts of individuals who are predominantly of European descent and largely overlook variants that are prevalent in other populations. The authors of a new editorial in the journal Pharmacogenomics (Preventing the exacerbation of health disparities by iatrogenic pharmacogenomic applications: lessons from warfarin) review studies investigating warfarin dosing in underrepresented and admixed populations in Puerto Rico, Brazil, and the United States. In addition to noting the importance of cataloging variants in diverse populations, the authors also stress the complications that may arise when self-reported color or race categories, which vary by country, are used as proxies to infer the presence of pharmacogenetic variants. The authors also note that the 2017 update of the Clinical Pharmacogenetic Implementation Consortium (CPIC) for Warfarin Dosing reflects some of these new findings and it includes self-reported ancestry and variants that are more commonly found in African-American patients in the dosing algorithm.

You can find more information about pharmacogenetic guided dosing for warfarin on cpicpgx.org as well as on the PharmGKB’s annotation of the CPIC Guideline update for warfarin and CYP2C9, VKORC1, CYP4F2 and the single nucleotide polymorphism, rs12777823.

Curators' Favorite Papers

A critical component of pharmacogenetic/pharmacogenomics (PGx) implementation into clinical care is the integration of PGx data in clinical decision support (CDS) and electronic health records (EHRs). A new article in the journal Human Molecular Genetics (Genomics and electronic health record systems) discusses the benefits of, and current challenges to, the integration of genomics and PGx data, into EHRs. In particular, the article discusses how PGx data in EHRs relates to questions of standards and evidence generation, and how data should be represented so as not to be overwhelming for clinicians. The article adeptly outlines specific components of CDS pertaining to whole genome sequencing (WGS) (e.g. ordering and interpreting a test, importing data, trigger alerts and warning in EHRs, and evaluating outcomes) and describes existing tools, such as application programming interfaces (APIs), that link knowledgebases (including PharmGKB) to EHRs to assist with CDS.  The authors conclude that with continued developments in technology, and ambitious research programs with large, diverse cohorts, such as the NIH-sponsored All of Us program, the “goals of generating new knowledge and clinically relevant discoveries using population-based genomics data can someday be achieved by using EHRs”.

A patient’s diagnosis and medical treatment is frequently guided by where that patient’s laboratory values fall along a range of established “normal” values. A new “Viewpoint” (In the Era of Precision Medicine and Big Data, Who Is Normal?) in the Journal of the American Medical Association (JAMA) raises an important question in the context of laboratory values and precision medicine: if medicine is personalized, to what do we compare an individual’s laboratory values if “normal” is actually relative? The authors consider solutions to ensure that test results “be interpreted in reference to a population of ‘similar’, ‘healthy’ individuals”. For example, the authors propose 1) that longitudinal data on individual outcomes be accessible to researchers to determine whether selected reference values are truly useful 2) that large-scale analyses be carried out across data sets 3) that reference values be tailored to patients and delivered at point of care, and 3) that “computationally derived genetic ancestry” be linked to laboratory test values, so that race is not used as a proxy.

Curators' Favorite Papers

The first paper comes from the journal Pharmacogenomics (Clinical pharmacogenetics: how do we ensure a favorable future for patients?) and it discusses the factors that have impeded the implementation of pharmacogenomic (PGx) testing into routine clinical care. Randomized clinical trials (RCT) are the current gold-standard for clinical research for new drug approvals, but the nature of PGx studies is ill-suited to the RCT format. The authors propose alternatives to RCTs such as demonstration of non-inferiority to standard of care, N-of-1 trials for individuals or “hybrid effectiveness-implementation” clinical trials (trials that blend design components investigating clinical effectiveness with implementation research). They also emphasize that implementation of PGx data would require comprehensive, pre-emptive testing, population specific PGx considerations, and storage of results in electronic medical records (EMR) with adequate clinical decision support (CDS) tools. Finally, they make the case for testing for genetic variants with robust evidence of a PGx association, and specifically cite the Clinical Pharmacogenenomics Implementation Consortium (CPIC) as a “promising place to start” in selecting PGx genes to test as well as PharmGKB as a resource for information on gene-drug PGx associations.

The second paper, authored by the International Society for Biocuration (Biocuration: Distilling data into knowledge), comes from the journal PLoS Biology. It explains the specific role that biocurators play within teams that manage biological information resources and databases. Beginning with the premise that data is an asset whose value increases each time it is shared, the authors argue that biocurators maximize value by assuring the “accuracy, comprehensiveness, integration, accessibility, and reuse” of data through the process of extracting knowledge (such as data) from unstructured forms (usually publications) into structured and machine readable forms to enhance its usability and sharing. The authors note that there is encouraging development with regard to data reporting tools, an increase in demand and support for data standards and a growth in the use of biocuration tools by researchers, and all of these are expected to facilitate data curation, data sharing and ultimately, scientific progress.

Curator's Favorite Papers

Genetics has been shown to have a profound effect on response to treatment in patients with major depressive disorder (MDD) and may account for as much as 42% of the variability in treatment response, according to some studies. In addition, personality traits, as defined by “the Big-Five" (openness, conscientiousness, extraversion, agreeableness, and neuroticism), which are likely to also be influenced by genetics, may also associate with response to anti-depression treatment. A new study (Association of the Polygenic Scores for Personality Traits and Response to Selective Serotonin Reuptake Inhibitors in Patients with Major Depressive Disorder) intended to uncover gene variants associated with the cross-trait associations between “the Big-Five” and treatment response using data from the Pharmacogenomics Research Network Antidepressant Medication Pharmacogenomic Study (N = 529) and the International SSRI Pharmacogenomics Consortium (N = 865). Evidence points to an overlap in association between specific personality traits and SSRI treatment outcomes and that the association is partly due to genetics. Specifically, the study evaluated the combined effect of multiple genes (polygenic score, PGS) and their association with one of the five personality traits, as well as response and remission in patients with MDD who are prescribed SSRIs. The PGS for openness personality with treatment response was statistically significant in the ISPC cohort and statistically significant with remission in the PGRN-AMPS sample. PGS for conscientiousness was associated with response, but not remission. Cross-trait meta-analysis of GWAS uncovered eight overlapping genetic loci with previously reported associations with response to SSRIs as well as certain personality traits, particularly neuroticism.  Of note, several PharmGKB members were co-authors on this study: Katrin Sangkuhl, Scientific Curator, Ryan Whaley, Technical Lead, Russ Altman, Co-PI of PharmGKB and Teri Klein, Director and Co-PI of PharmGKB. 

You can find more information about pharmacogenetic guided dosing for SSRIs on cpicpgx.org as well as on the PharmGKB annotation of the CPIC Guideline for SSRIs and CYP2D6 and CYP2C19.

Curators' Favorite Papers

The first paper, by A. Ahmed et al. in the journal Clinical Pharmacology and Therapeutics (Benefits of and Barriers to Pharmacogenomics-Guided Treatment for Major Depressive Disorder), discusses pharmacogenetic (PGx) testing in the context of depression. In a pilot study, 1,002 out of 1,013 (99%) subjects had an actionable variant in CYP2D6, CYP2C9, CYP2C19, SLCO1B1 or VKORC1. With regards to anti-depressants, 79% of subjects had actionable variants in CYP2D6 (metabolizes fluoxetine, paroxetine, nortriptyline, and desipramine), 60% in CYP2C19 (metabolizes citalopram and escitalopram) and 36% in CYP2C9 (metabolizes fluoxetine). The authors posit that wider implementation will still require several steps going forward. In addition to making PGx testing pre-emptive, so that genotype information is already present in patient electronic health records (EHRs), physicians will need to be convinced of the clinical utility of PGx testing for patients with mood disorders.  The authors believe this means generating data from larger studies “conducted by disinterested groups”, studies in adolescents, studies investigating multiple genes and the inclusion of metabolomics to inform genomics.


The second paper by Maciel et al. in the journal Neuropsychiatric Disease Treatment (Estimating Cost Savings of Pharmacogenetic Testing for Depression in Real-World Clinical Settings) describes potential cost-savings associated with pharmacogenetic (PGx) testing in patients with depression. The researchers estimated a one-time cost of USD $2,000 for PGx testing and used a published cost-calculator to estimate cost-savings. Data were from a published double-blind, multi-center, randomized clinical trial of 685 adults who had been diagnosed with depression or anxiety. Patients were randomized to PGx-guided treatment or standard of care (control). The study found that those subjects randomized to PGx-guided dosing and prescribing had significant improvements in clinical outcomes as compared to the control group. Using these data, the study calculated that the cost savings for PGx-testing vs standard of care for patients with depression or anxiety totaled USD $3,962 per year after accounting for the one-time cost of PGx testing.

A new meta-analysis of 522 studies that was recently published in The Lancet concluded that at minimum, short-term treatment with anti-depressants was more effective than placebos in treating acute depression. The study made headlines in major media-outlets and found that some anti-depressants worked better than others, which may lead to more interest in PGx-guided dosing.

You can find PGx-guided drug dosing guidelines from the Clinical Pharmacogenetic Implementation Consortium (CPIC) for tricyclic anti-depressants and selective serotonin re-uptake inhibitors on cpicpgx.org. You may also find more information about the following genes and drugs on PharmGKB:

Genes
CYP2D6
CYP2C9
CYP2C19

Drugs
fluoxetine
paroxetine
nortriptyline
desipramine
citalopram
escitalopram
amitriptyline
fluoxetine

Curators' Favorite Papers

The first paper (Implementation of Standardized Clinical Processes for TPMT Testing in a Diverse Multidisciplinary Population: Challenges and Lessons Learned) describes the process of implementing systemwide thiopurine methyltransferase (TPMT) testing at the University of Florida Health Personalized Medicine Program (UF PMP). After the successful implementation of CYP2C19 testing for patients undergoing percutaneous coronary intervention in an inpatient setting, researchers opted to analyze test ordering patterns for TPMT in a large and more diverse group of patients. Pharamacogenetic (PGx) test results were incorporated into Electronic Health Records (EHR) using Clinical Pharmacogenetic Implementation Consortium (CPIC) guidelines. Data came over a period of three years from 834 patients for whom TPMT testing was ordered (February 2014 to February 2017). In general, enzymatic testing was ordered far more frequently than genetic testing (83% vs. 17%) and rates differed significantly (P< 0.0001) between outpatient (580) and inpatient orders (293). Significant differences were discovered between specialties with regards to TPMT test orders - for example,  hematology/oncology service providers ordered genotype testing more frequently in the inpatient setting (95% of tests from hematology/oncology vs. 60% of tests from other specialties; P< 0.0001). The study also discusses differences in testing between populations and breaks down test alert notification preferences by specialties in detail. You may find more information on TPMT, azathioprine, thioguanine,  or mercaptopurine on PharmGKB and PGx-guided dosing guidelines on PharmGKB or CPICpgx.org.

The second paper (Patient understanding of, satisfaction with, and perceived utility of whole-genome sequencing: findings from the MedSeq Project) examines patient-subject understanding of informed consent, satisfaction with disclosure of results, and perception of whole genome sequencing utility (WGS) in cardiology and primary care patients. Patient-subjects were recruited by their cardiologists (already diagnosed with hypertrophic cardiomyopathy or dilated cardiomyopathy) or primary care physicians (PCP) (healthy adults) and were randomized to give family health history (FH) alone or undergo WGS too (FH + WGS). Results were discussed with their respective physicians, who had been counseled in genetics. A majority of subjects (N = 203) were White, non-Hispanic, had completed college and had annual household incomes of over $100K. Most subjects were able to adequately describe the study and demonstrated high genetics and health literacy, as well as an understanding of informed consent and experienced a high level of satisfaction. However, some differences emerged between the FH and the FH + WGS groups: subjects in the FH + WGS group were more likely to report feeling that they had received too much information, and subjects in the FH group reported less satisfaction and more decisional regret than the FH + WGS group. The authors suspect that FH subjects may have been disappointed at not having been randomized to receive WGS results. The primary care group was also more likely to feel that they had received too much information relative to the cardiology group. The authors note that the results are encouraging but suggest that similar efforts in future studies should aim to temper subjects’ expectations regarding WGS and that studies in more diverse cohorts will be necessary.

Curators' Favorite Papers

The first paper from Genetics in Medicine by Khoury et al. (From Public Health Genomics to Precision Public Health: a 20-Year Journey) reviews developments in the field of public health genomics over the last twenty years. Public health genomics deals with the “effective and responsible translation of genomic research into population health benefit” through assessment, policy, and assurance. It summarizes current research projects in the field and describes the role of organizations, including the Centers for Disease Control and Prevention (CDC) in the development and implementation of evidence-based guidelines for genetic testing. The authors recognize that genomics cannot be isolated from other determinants of health including behaviors or socioeconomic factors such as housing, education, and access to care and the need for subsequent developments in “precision public health” to integrate genomics data with other health determinants to improve public health outcomes.


Despite the fact that over 100 GPCRs are targeted by approximately 34% of FDA-approved drugs, the frequency of genetic variation of GPCRs is not known according to a study by Hauser et al. from the December issue of Cell (Pharmacogenomics of GPCR Drug Targets). The study evaluates pharmacogenetic (PGx) variation in 108 G-protein coupled receptors (GPCRs) using datasets from the exome aggregation consortium (ExAC) and the 1000 Genomes Project that include over 60,000 individuals and estimates that there is an average of 128 rare and 3.7 common variants per receptor and that 25% of all positions in each GPCR contains a missense variant. In addition approximately 120 of the 60,706 individuals from the dataset harbored loss of function mutations in a GPCR drug target and each GPCR had approximately two duplications and three deletions. The authors support their findings with an analysis of the molecular literature, including data from PharmGKB Clinical Annotations, functional PGx studies of GPCRs on drug response and efficacy and an economic analysis of how incorporation of GPCR PGx could decrease the UKs National Health Service (NHS) financial burden.

Curators' Favorite Papers

The first of two papers selected for the November edition of ”Curators' Favorite Papers" is from Nature Reviews Genetics (“Prioritizing diversity in human genomics research”). It highlights the necessity of including individuals from diverse backgrounds in genomic research, both as subjects and as researchers. The authors discuss several proposals to achieve these goals beginning with awareness of genetic and environmental factors that contribute to disparities in health outcomes, establishing sources of dedicated funding, recruiting subjects, researchers and clinicians from diverse backgrounds, and the integration of genomics into existing healthcare systems in underserved communities. The authors mention two pharmacogenomic (PGx) examples to remark on the importance of genetic and geographic diversity for clinical genomics: the risk of Stevens-Johnsons Syndrome/ toxic epidermal necrolysis in individuals of Asian ancestry who carry the HLA-B*15:02 allele that are administered carbamazepine as well as the risk of hemolysis in African-American males harboring G6PD alleles that are administered quinine. 

According to a new paper from the journal Oncotarget (“Moving forward with actionable therapeutic targets and opportunities in endometrial cancer: NCI clinical trials planning meeting report on identifying key genes and molecular pathways for targeted endometrial cancer trials.”), metastatic endometrial cancer (EC) is the fourth most common cancer affecting women, with increased incidence and relatively poor prognosis but no new treatments have been approved in approximately two decades. The paper summarizes the findings from a recent meeting of Gynecologic Cancer Steering Committee (GCSC) and the National Cancer Institute (NCI). Experts gathered to review the literature and generate reports to design early phase clinical trials based on molecular pathway research in EC to improve treatment outcomes in women with EC. The authors generated reports for therapies targeting mutations in those pathways that are commonly implicated in a variety of cancers including DNA-damage repair and cell-cycle checkpoint pathways including ERBB2/HER2, PI3K/ATK/mTOR, WNT pathways as well as the dysregulation of those pathways involving ubiquitin-ligase complex, the immune system and metabolic disorders. 

You can read more about HLA-B and carbamazepine, G6PD and quinine, ERBB2/HER2 and other targeted cancer therapies at the Cancer Pharmacogenomics portal on PharmGKB and CPIC. 

Curators' Favorite Papers

October’s edition of "Curators' Favorite Papers" features “Introducing personalized health for the family: the experience of a single hospital system”, a discussion about a preemptive pharmacogenomic (PGx) testing program in newborns at a Virginia hospital. The program, called MediMap Baby was initiated in 2014 with with the recruitment of participants from a longitudinal family-based whole genome sequence study at the Inova Translational Medicine Institute (ITMI). Initial efforts included the formation of small focus groups to gauge patient interest in the project. The MediMap project began implementation in 2016 where preemptive PGx testing was offered to all families of the newborns born at the hospital. 4,257 PGx tests were conducted at no additional cost and results were entered into the patient’s electronic health record. The program necessitated the training of a multidisciplinary staff and the development of patient educational materials. Genes assayed included TPMT, CYP2C9, VKORC1, CYP2C19, SLCO1B1, CYP2D6 and CYP3A5, which the authors describe as having potential utility for 24 prescription medications.

Dosing and prescribing guidelines involving these and other genes are available on PharmGKB and CPIC.

Curators' Favorite Papers

For this month's edition of "Curators' Favorite Papers" we present two articles discussing how the recommendations from Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines have been used at two different healthcare settings.

Implementation of Clinical Pharmacogenomics within a Large Health System: From Electronic Health Record Decision Support to Consultation Services by Sissung et al. discusses the experience of the Cleveland Clinic with pharmacogenetic (PGx) implementation as well as how a collaboration between pharmacists and physician-geneticists established an ambulatory PGx clinic to provide testing, interpretation, prescribing recommendations, and patient education. The authors go on to discuss how the implementation of three CPIC guidelines (HLA-B*57:01 and abacavir and HLA-B*15:02 and carbamazepine and TPMT and thiopurines) was managed by pharmacists that incorporated patient data into electronic health records (EHR) who developed clinical decision support tools (CDS) based on CPIC guidelines and included custom pre- and post-test alerts. The selection of which guidelines to implement was based on third-party payment reimbursement for genetic tests, the life-threatening nature of potential adverse events as recognized by the Food and Drug Administration (FDA) in drug labels and physician support.

Pharmacogenomics Implementation at the National Institutes of Health Clinical Center also by Sissung et al. reviews the PGx implementation process at the National Institutes of Health Clinical Center (NIH CC), which uses the recommendations of CPIC guidelines to inform the selection of gene-drug pairs to be implemented.  The NIH CC has already implemented PGx testing and prescribing recommendations for HLA-B alleles with allopurinol, abacavir, and carbamazepine, and it is currently in the process of PGx implementation for genes involved in the absorption, distribution, metabolism, or excretion (ADME) of drugs using the DMET genotyping platform.  The NIH CC model makes PGx testing, results and recommendations available to all NIH clinicians and also makes test results available to patients and their personal care physicians. The NIH CC expects to cover all CPIC gene-drug pairs within the next five years.

Additional guidelines can be found on the CPIC and PharmGKB and additional information about drugs and genes can also be found on PharmGKB.

Curators' Favorite Papers

The Impact of Whole-Genome Sequencing on the Primary Care and Outcomes of Healthy Adult Patients: A Pilot Randomized Trial

Jason L. Vassy, MD, MPH, SM; Kurt D. Christensen, PhD, MPH; Erica F. Schonman, MPH; Carrie L. Blout, MS, CGC; Jill O. Robinson, MA; Joel B. Krier, MD; Pamela M. Diamond, PhD; Matthew Lebo, PhD; Kalotina Machini, PhD; Danielle R. Azzariti, MS, CGC; Dmitry Dukhovny, MD, MPH; David W. Bates, MD, MSc; Calum A. MacRae, MD, PhD; Michael F. Murray, MD; Heidi L. Rehm, PhD; Amy L. McGuire, JD, PhD; and Robert C. Green, MD, MPH for the MedSeq Project

This study, published  in the Annals of Internal Medicine, received a lot of attention in prominent media outlets including Wired, STAT, NPR, Washington Post and Science Magazine. As part of the MedSeq project, 100 healthy adults were recruited by 9 primary care providers (PCP), themselves briefed on genomics and how to refer patients to genetics experts. All patients gave a detailed family history (FH) but were randomized to get either whole genome sequencing (WGS) (WGS +FH) or not (FH) and were surveyed at 6 months regarding follow-up care and well-being. The study authors uncovered monogeneic disease risk (MDR) variants in 11 patients, but only 2 manifested the phenotype (fundus albipunctatus and subclinical porphyria). 48 of 50 subjects had a pharmacogenetic variant in a gene affecting one of five drugs (warfarin, clopidogrel, simvastatin, metformin, and digoxin) and results were added to electronic health records (EHR). PCPs recommended new clinical actions in 16% and 34% of FH and WGS+ FH patients, respectively. At 6 months, 30% and 41% of patients made health behavior changes in the FH and WGS-FH groups, respectively, but no significant differences in self-reported health, anxiety or depression scores emerged between groups. Overall, the authors conclude that identification of MDR variants did not improve short-term health outcomes, and results do not support the routine use of WGS in healthy patients. However, the authors note that the ability of PCPs to adequately manage patient results, their use of EHR, and their appropriate referrals to genetic specialists, as well as self-reported patient well-being after receiving results are encouraging signs that indicate a favorable environment for WGS in future studies and specific clinical care situations.

You can read the study here and more information about the pharmacogenetics of warfarin, clopidogrel, simvastatin, metformin, and digoxin is available at PharmGKB. 

Curators' Favorite Papers

Electronic medical record-integrated pharmacogenomics and related clinical decision support concepts
Authors: Caraballo PJ Bielinski SJ St Sauver JL Weinshilboum RM

The authors of this article from Clinical Pharmacology and Therapeutics, discuss the challenges to successful integration of pharmacogenetic/pharmacogenomic (PGx) clinical decision support (CDS) tools into electronic health and medical records (EMR). They provide examples of medical systems that have already done so and they specifically highlight the success of the Mayo Clinic where CDS tools for eleven PGx genes and nineteen drug-gene interactions have been successfully integrated into EMRs and patient care.

https://www.ncbi.nlm.nih.gov/pubmed/28390138


Celebrating parasites
Authors: Greene CS, Garmire LX, Gilbert JA, Ritchie MD, Hunter LE.

This correspondence to Nature Genetics is a response to a controversial 2016 editorial from the the New England Journal of Medicine in which the term, “research parasites” was coined. The term described those researchers who do secondary analyses of data generated by other researchers. The authors argue that “parasites” serve an important purpose in research and that the critical re-analysis of data is crucial for the practice of science. To honor such research, they announce two inaugural award categories at this year's Pacific Symposium on Biocomputing: the Junior Parasite Award and the Sustained Parasite Award.

https://www.ncbi.nlm.nih.gov/pubmed/28358134

Constellation: tool for automated CYP2D6 phenotype assignment from WGS

A team from Children’s Mercy and the University of Missouri in Kanas City including Andrea Gaedigk and Greyson Twist published an article about a CYP2D6 phenotype assignment tool in Genomic Medicine (http://www.nature.com/articles/npjgenmed20157). Constellation, a probabilistic scoring system, enables automated ascertainment of CYP2D6 activity scores based on CYP2D6 diplotypes from whole-genome sequences (WGS).

CYP2D6 is involved in the metabolism of about 25% of drugs in clinical use and genetic variations leading to functional consequences affecting drug efficacy and risk of adverse events.  

The gene is highly polymorphic with over 100 allelic variants (star alleles) assigned including CYP2D6 copy number variations and rearrangements with the neighboring CYP2D7. This high degree of variation, high sequence similarity to CYP2D7 and CYP2D8, GC content, repetitive and low-complexity sequences are challenges in analyzing this locus.

The performance of the developed algorithm is evaluated by comparing the CYP2D6 diplotype assigned by the probabilistic WGS analysis using Constellation with the diplotype determined by manual integration (consensus reference) of results obtained by quantitative copy-number assessment, a panel of TaqMan genotype assays, and Sanger sequencing of long-range genomic PCR in 61 samples. Phenotype prediction is consistent between the consensus reference and Constellation calls with the exception of three cases. Constellation was able to accurately identify all poor and ultrarapid metabolizers in WGS data. 

The authors anticipate Constellation to be extensible to identify variations in other pharmacogenomic-relevant genes, enabling future uses of WGS data.

Combining large PGx datasets from cancer cell lines

Testing cancer cell lines in vitro for drug sensitivity is a cornerstone of preclinical drug development. Large publically available datasets can be found at The Genomics of Drug Sensitivity in Cancer Project (GDSE) and The Cancer Cell Line Encyclopedia (CCLE).

Studies attempting to combine large public datasets and analyzing for correlation questioned the reliability of the data due to limited concordance, reported in [PMID: 24284626], discussed in [PMID:24284624] and a confirmation study here.

A new report in Nature describes different methods to analyze the data from CCLE and GDSE and concludes that “data from either study yields similar predictors of drug response” [PMID:26570998].

These papers demonstrate the continuing difficulty trying to compare across large datasets. Such problems include comparing different experimental protocols and measurements for drug sensitivity across studies, trouble matching the drug and cell line names to ensure like comparison, discordance in the genotyping data, and drugs that had few examples of cell lines that were drug sensitive.  As always, attention to detail in the documentation and description of the experiments can help mitigate some of these difficulties. While development of standard testing protocols and data curation and reporting frameworks may lead to better validation of drug response predictors going forward there will always be the need for methods to filter the noise that is inevitable in large datasets.

Curators' Favorite Papers

The current Curators' Favorite Papers' highlights the higher frequency of genetic variants conferring increased risk for adverse drug reactions (ADRs) for many commonly used drugs in persons of African descent. Differences in drug response, cure rates and survival outcomes between different ethnic populations exist. To assess this, Aminkeng et al. genotyped 1330 individuals of African and European descent over 4,000 SNPs in 350 key drug absorption, distribution, metabolism, elimination and toxicity genes. Publicly available databases including PharmGKB were utilized to prioritize the selection of functional SNPs. The distribution of many variants were significantly different between African and European populations. These results were presented in the context of clinical annotations (toxicity, efficacy, ADR and drug response pathway) that were curated from PharmGKB and published studies. Important differences among the African populations were also observed in the frequency of variants. Results presented in this paper may translate to significant differences in drug efficacy and safety profiles in different populations.

Curators' Favorite Papers

The current Curators' Favorite Papers' highlight an article describing a new method in identifying the genetic variation of gene expression using gene sets. To test their novel approach, Abo et al. used KEGG and PharmGKB pathways to define gene sets. In their method, SNPs and mRNA expression were initially grouped as gene sets, then principal components analysis was conducted to collapse the variation and reduce the dimensionality within the gene sets. They proposed that the multiple testing limitations in eQTL studies can be addressed effectively by their method. Many significant associations are found using publicly available data sets. The most significant associations were found between genetic variation and mRNA expression from the same gene sets and provides biological context for SNP-expression associations.