A Decision-Theoretic Approach to Panel-Based, Preemptive Genotyping

Advancing Pharmacogenomics from Single-Gene to Preemptive Testing

Pharmacogenomic testing can be an effective tool to enhance medication safety and efficacy. Pharmacogenomically actionable medications are widely used, and approximately 90-95% of individuals have an actionable genotype for at least one pharmacogene. For pharmacogenomic testing to have the greatest impact on medication safety and clinical care, genetic information should be made available at the time of prescribing (preemptive testing). However, the use of preemptive pharmacogenomic testing is associated with some logistical concerns, such as consistent reimbursement, processes for reporting preemptive results over an individual's lifetime, and result portability. Lessons can be learned from institutions that have implemented preemptive pharmacogenomic testing. In this review, we discuss the rationale and best practices for implementing pharmacogenomics preemptively.

Assessment of a Manual Method versus an Automated, Probability-Based Algorithm to Identify Patients at High Risk for Pharmacogenomic Adverse Drug Outcomes in a University-Based Health Insurance Program

We compared patient cohorts selected for pharmacogenomic testing using a manual method or automated algorithm in a university-based health insurance network. The medication list was compiled from claims data during 4th quarter 2018. The manual method selected patients by number of medications by the health system’s list of medications for pharmacogenomic testing. The automated method used YouScript’s pharmacogenetic interaction probability (PIP) algorithm to select patients based on the probability that testing would result in detection of one or more clinically significant pharmacogenetic interactions. A total of 6916 patients were included. Patient cohorts selected by each method differed substantially, including size (manual n = 218, automated n = 286) and overlap (n = 41). The automated method was over twice as likely to identify patients where testing may reveal a clinically significant pharmacogenetic interaction than the manual method (62% vs. 29%, p < 0.0001). The manual method captured more patients with significant drug–drug or multi-drug interactions (80.3% vs. 40.2%, respectively, p < 0.0001), higher average number of significant drug interactions per patient (3.3 vs. 1.1, p < 0.0001), and higher average number of unique medications per patient (9.8 vs. 7.4, p < 0.0001). It is possible to identify a cohort of patients who would likely benefit from pharmacogenomic testing using manual or automated methods.

A Cost-Consequence Analysis of Preemptive SLCO1B1 Testing for Statin Myopathy Risk Compared to Usual Care

There is a well-validated association between SLCO1B1 (rs4149056) and statin-associated muscle symptoms (SAMS). Preemptive SLCO1B1 pharmacogenetic (PGx) testing may diminish the incidence of SAMS by identifying individuals with increased genetic risk before statin initiation. Despite its potential clinical application, the cost implications of SLCO1B1 testing are largely unknown. We conducted a cost-consequence analysis of preemptive SLCO1B1 testing (PGx+) versus usual care (PGx-) among Veteran patients enrolled in the Integrating Pharmacogenetics in Clinical Care (I-PICC) Study. The assessment was conducted using a health system perspective and 12-month time horizon. Incremental costs of SLCO1B1 testing and downstream medical care were estimated using data from the U.S. Department of Veterans Affairs' Managerial Cost Accounting System. A decision analytic model was also developed to model 1-month cost and SAMS-related outcomes in a hypothetical cohort of 10,000 Veteran patients, where all patients were initiated on simvastatin. Over 12 months, 13.5% of PGx+ (26/193) and 11.2% of PGx- (24/215) participants in the I-PICC Study were prescribed Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline-concordant statins (Δ2.9%, 95% CI -4.0% to 10.0%). Differences in mean per-patient costs for lipid therapy prescriptions, including statins, for PGx+ compared to PGx- participants were not statistically significant (Δ USD 9.53, 95% CI -0.86 to 22.80 USD). Differences in per-patient costs attributable to the intervention, including PGx testing, lipid-lowering prescriptions, SAMS, laboratory and imaging expenses, and primary care and cardiology services, were also non-significant (Δ- USD 1004, 95% CI -2684 to 1009 USD). In the hypothetical cohort, SLCO1B1-informed statin therapy averted 109 myalgias and 3 myopathies at 1-month follow up. Fewer statin discontinuations (78 vs. 109) were also observed, but the SLCO1B1 testing strategy was 96 USD more costly per patient compared to no testing (124 vs. 28 USD). The implementation of SLCO1B1 testing resulted in small, non-significant increases in the proportion of patients receiving CPIC-concordant statin prescriptions within a real-world primary care context, diminished the incidence of SAMS, and reduced statin discontinuations in a hypothetical cohort of 10,000 patients. Despite these effects, SLCO1B1 testing administered as a standalone test did not result in lower per-patient health care costs at 1 month or over 1 year of treatment. The inclusion of SLCO1B1, among other well-validated pharmacogenes, into preemptive panel-based testing strategies may provide a better balance of clinical benefit and cost.

Predicting Survival in Veterans with Follicular Lymphoma Using Structured Electronic Health Record Information and Machine Learning

The most accurate prognostic approach for follicular lymphoma (FL), progression of disease at 24 months (POD24), requires two years’ observation after initiating first-line therapy (L1) to predict outcomes. We applied machine learning to structured electronic health record (EHR) data to predict individual survival at L1 initiation. We grouped 523 observations and 1933 variables from a nationwide cohort of FL patients diagnosed 2006–2014 in the Veterans Health Administration into traditionally used prognostic variables (“curated”), commonly measured labs (“labs”), and International Classification of Diseases diagnostic codes (“ICD”) sets. We compared performance of random survival forests (RSF) vs. traditional Cox model using four datasets: curated, curated + labs, curated + ICD, and curated + ICD + labs, also using Cox on curated + POD24. We evaluated variable importance and partial dependence plots with area under the receiver operating characteristic curve (AUC). RSF with curated + labs performed best, with mean AUC 0.73 (95% CI: 0.71–0.75). It approximated, but did not surpass, Cox with POD24 (mean AUC 0.74 [95% CI: 0.71–0.77]). RSF using EHR data achieved better performance than traditional prognostic variables, setting the foundation for the incorporation of our algorithm into the EHR. It also provides for possible future scenarios in which clinicians could be provided an EHR-based tool which approximates the predictive ability of the most accurate known indicator, using information available 24 months earlier.

Retrospective Review of Pharmacogenetic Testing at an Academic Children's Hospital

There is limited evidence to support pharmacogenetic (PGx) testing in children. We conducted a retrospective review of PGx testing among 452 patients at an academic children's hospital to determine the potential utility of PGx in diseases of childhood and to identify targets for future pediatric pharmacogenetic research. An actionable gene-drug pair associated with the 28 genes tested (Clinical Pharmacogenetics Implementation Consortium (CPIC) level A or B, Pharmacogenomics Knowledge Base (PharmGKB) level 1A or B, or US Food and Drug Administration (FDA) recommendation and a PharmGKB level) was present in 98.7% of patients. We identified 203 actionable gene-drug-diagnosis groups based on the indications for each actionable drug listed in Lexicomp. Among patients with an actionable gene-drug-diagnosis group, 49.3% had a diagnosis where the drug was a therapeutic option and PGx could be used to guide treatment selection. Among patients with an associated diagnosis, 30.9% had a prescription for the actionable drug allowing PGx guided dosing. Three genes (CYP2C19, CYP2D6, and CYP3A5) accounted for all the gene-drug-diagnosis groups with matching diagnoses and prescriptions. The most common gene-drug-diagnosis groups with matching diagnoses and prescriptions were CYP2C19-citalopram-escitalopram-depression 3.3% of patients tested; CYP2C19-dexlansoprazole-gastritis-esophagitis 3.1%; CYP2C19-omeprazole-gastritis-esophagitis 2.4%; CYP2D6-atomoxetine-attention deficit hyperactivity disorder 2.2%; and CYP2C19-citalopram-escitalopram-obsessive-compulsive disorder 1.5%. PGx could be used to guide selection of current treatment options or medication dosing in almost half (48.7%) of pediatric patients tested. Mood disorders and gastritis/esophagitis are promising targets for future study of PGx testing because of the high prevalence of these diagnoses and associated actionable gene-drug pairs in the pediatric population.

How to Transition from Single-Gene Pharmacogenetic Testing to Preemptive Panel-Based Testing: A Tutorial

There have been significant advancements in precision medicine and approaches to medication selection based on pharmacogenetic results. With the availability of direct-to-consumer genetic testing and growing awareness of genetic interindividual variability, patient demand for more precise, individually tailored drug regimens is increasing. The University of Florida (UF) Health Precision Medicine Program (PMP) was established in 2011 to improve integration of genomic data into clinical practice. In the ensuing years, the UF Health PMP has successfully implemented several single-gene tests to optimize the precision of medication prescribing across a variety of clinical settings. Most recently, the UF Health PMP launched a custom-designed pharmacogenetic panel, including pharmacogenes relevant to supportive care medications commonly prescribed to patients undergoing chemotherapy treatment, referred to as "GatorPGx." This tutorial provides guidance and information to institutions on how to transition from the implementation of single-gene pharmacogenetic testing to a preemptive panel-based testing approach. Here, we demonstrate application of the preemptive panel in the setting of an adult solid tumor oncology clinic. Importantly, the information included herein can be applied to other clinical practice settings.

Use of Real-World Evidence in US Payer Coverage Decision-Making for Next-Generation Sequencing-Based Tests: Challenges, Opportunities, and Potential Solutions

OBJECTIVES

Given the potential of real-world evidence (RWE) to inform understanding of the risk-benefit profile of next-generation sequencing (NGS)-based testing, we undertook a study to describe the current landscape of whether and how payers use RWE as part of their coverage decision making and potential solutions for overcoming barriers.

METHODS

We performed a scoping literature review of existing RWE evidentiary frameworks for evaluating new technologies and identified barriers to clinical integration and evidence gaps for NGS. We synthesized findings as potential solutions for improving the relevance and utility of RWE for payer decision-making.

RESULTS

Payers require evidence of clinical utility to inform coverage decisions, yet we found a relatively small number of published RWE studies, and these are predominately focused on oncology, pharmacogenomics, and perinatal/pediatric testing. We identified 3 categories of innovation that may help address the current undersupply of RWE studies for NGS: (1) increasing use of RWE to inform outcomes-based contracting for new technologies, (2) precision medicine initiatives that integrate clinical and genomic data and enable data sharing, and (3) Food and Drug Administration reforms to encourage the use of RWE. Potential solutions include development of data and evidence review standards, payer engagement in RWE study design, use of incentives and partnerships to lower the barriers to RWE generation, education of payers and providers concerning the use of RWE and NGS, and frameworks for conducting outcomes-based contracting for NGS.

CONCLUSIONS

We provide numerous suggestions to overcome the data, methodologic, infrastructure, and policy challenges constraining greater integration of RWE in assessments of NGS.