Variant Interpretation in Current Pharmacogenetic Testing

Sally Luvsantseren, Michelle Whirl‐Carrillo, Katrin Sangkuhl, Nancy Shin, Alice Wen, Philip E. Empey, Benish Alam, Sean P. David, Henry M. Dunnenberger, Lori A. Orlando, Russ B. Altman, Latha PalaniappanJournal of Personalized Medicine 2020 · DOI

TL;DR

This paper highlights significant discrepancies in pharmacogenetic test reporting between two commercial companies, leading to patient anxiety and potentially suboptimal clinical decisions. It underscores the urgent need for standardization in interpreting genetic variants and reporting phenotypes to ensure reliable and consistent guidance for precision medicine.

What Problem Does This Paper Address?

The field of pharmacogenetics promises to personalize medicine by predicting how an individual will respond to certain drugs based on their genetic makeup. This is achieved by analyzing specific gene variants and translating them into a "phenotype" (e.g., "poor metabolizer," "rapid metabolizer"). However, the current marketplace for pharmacogenetic testing is fragmented. Numerous commercial companies offer these tests, but they differ in the genes they analyze and, crucially, in how they interpret the genetic variants into actionable phenotypic information. This lack of standardization means that the same genetic data can lead to different, even contradictory, clinical recommendations from different providers. This ambiguity creates confusion for both patients and clinicians, potentially leading to unnecessary distress and impacting treatment decisions. The paper addresses this problem by presenting a real-world case that exemplifies these reporting discrepancies and their consequences.

Methods

The paper uses a case study approach to illustrate the problem. The authors present the experience of a 65-year-old woman who received pharmacogenetic testing from two different commercial companies (Company A and Company B). They then compare the reports generated by these companies, focusing on the interpretation of specific gene variants (CYP2D6, CYP2C19, and CYP1A2) and their resulting phenotypic classifications. The authors also reference established guidelines from the Clinical Pharmacogenetics Implementation Consortium (CPIC) as a benchmark for comparison. The core "method" here is the comparative analysis of commercial test reports against a recognized standard and against each other, highlighting the discrepancies.

Datasets

The paper does not use traditional datasets for training or evaluation. Instead, it relies on:

  • Pharmacogenetic test reports: Specifically, the reports generated by two commercial companies (Company A and Company B) for a single patient.
  • Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines: These are publicly available, evidence-based guidelines used as a reference standard for interpreting pharmacogenetic variants and phenotypes.

Key Insights and Ideas

  • Commercial interpretation variability is a significant issue: The same genetic variants can be interpreted into different phenotypes by different commercial testing companies, leading to conflicting clinical advice.
  • Patient anxiety and distress are direct consequences: Discrepant reports can cause significant worry for patients about their medication response and future treatment options.
  • Lack of transparency in proprietary algorithms: Commercial companies use their own curated data and proprietary algorithms for interpretation, making it difficult for clinicians to scrutinize the basis of their recommendations.
  • CPIC guidelines offer a valuable, standardized approach: The paper advocates for greater adoption of CPIC guidelines as a benchmark for consistency and reliability in pharmacogenetic reporting.
  • Standardization is crucial for clinical adoption: For pharmacogenomics to be widely and confidently adopted in clinical practice, there needs to be a consensus on how genetic variants are interpreted and reported.

Key Results

The case study revealed the following discrepancies:

  • CYP2D6: Company A classified the patient as a "Poor Metabolizer," while Company B classified her as an "Intermediate Metabolizer," despite reporting the same genotypes (4/41). This difference has significant implications for drug dosing, such as paroxetine, and advice regarding codeine.
  • CYP2C19: Company A reported an "Extensive (Normal) Metabolizer" phenotype, while Company B reported a "Rapid Metabolizer" phenotype for the 1/17 genotype. This distinction affects recommendations for drugs like escitalopram.
  • CYP1A2: The companies tested for different variants and consequently reported different genotypes and phenotypes. Company A identified variants associated with its 1B allele, while Company B focused on the 1F allele.

The authors ultimately opted to follow Company B's interpretations for CYP2C19 and CYP2D6 as they aligned with CPIC guidelines.

Limitations and Future Work

Limitations:

  • Single Case Study: The paper is based on a single patient's experience, which, while illustrative, may not represent the full spectrum of discrepancies encountered.
  • Proprietary Nature of Commercial Tests: The exact algorithms and data curation methods used by the commercial companies are not fully disclosed, limiting a complete understanding of the discrepancies.
  • Limited FDA Oversight: Lab-developed tests, like those discussed, are not reviewed by the FDA, contributing to the variability in reporting standards.

Future Work:

  • Industry-wide Standardization: The authors strongly advocate for greater convergence towards standardized interpretation and reporting, ideally aligning with CPIC guidelines.
  • Transparency in Reporting: Encouraging commercial vendors to be more transparent about their data curation and interpretation methodologies.
  • "Fully Compliant with CPIC" Label: Suggesting a labeling system for genetic testing vendors that indicates adherence to CPIC guidelines, helping clinicians and patients make informed choices.
  • Addressing Consumer Accessibility: As pharmacogenetic tests become more accessible to consumers, the need for clear and consistent reporting will only increase.

Summary updated

Abstract

In the current marketplace, there are now more than a dozen commercial companies providing pharmacogenetic tests. Each company varies in the panel of genes they test and the variants they are able to screen for. The reports generated by these companies provide phenotypic interpretations of pharmacogenes and clinically actionable gene-drug interactions based on internally curated data and proprietary algorithms. The freedom to choose the types of evidence to include versus exclude in interpreting genomics has created reporting discrepancies in the industry. The case report presented here reveals the discordant phenotype analysis provided by two pharmacogenetic testing companies. The uncertainty and unnecessary distress experienced by the patient highlights the need for consensus in phenotype reporting within the industry.
BibTeX
@article{luvsantseren2020variant,
  author = {Sally Luvsantseren and Michelle Whirl‐Carrillo and Katrin Sangkuhl and Nancy Shin and Alice Wen and Philip E. Empey and Benish Alam and Sean P. David and Henry M. Dunnenberger and Lori A. Orlando and Russ B. Altman and Latha Palaniappan},
  title = {{Variant Interpretation in Current Pharmacogenetic Testing}},
  journal = {Journal of Personalized Medicine},
  year = {2020},
  doi = {10.3390/jpm10040204},
}
Figure