Updates to Drug Label Annotation Tags and Criteria for PGx levels

PharmGKB has made some changes in the Drug Label Annotation process to clarify the pharmacogenomic (PGx) relationships of biomarkers. PharmGKB curates all drug labels on the FDA’s Table of Pharmacogenomic Biomarkers in Drug Labeling and previously attempted to classify all labels with PharmGKB-created “PGx levels” . However many of the relationships do not fit standard pharmacogenomic definitions. It is clear from over a decade of label curation that many labels, including many on the FDA Biomarker list, mention genes, variants, chromosomal rearrangements, testing, drug-drug interactions and other incidental information but do not explicitly provide pharmacogenomic clinical guidance or information. Accordingly, we have altered the PGx level definition for “Informative PGx”, and created a new PGx level “Criteria Not Met” and a new drug label annotation tag "Indication”.

PharmGKB also attempts to identify and curate drug labels from other regulatory agencies for drugs with annotated FDA labels. Labels from other regulatory agencies may or may have similar information to the FDA-approved label, but sometimes the label cannot be found after a search. Previously, if a label was not found, the entry on the Drug Label Annotations Landing page was left blank. However, if a search has not yet been done, the entry is also blank, which led to ambiguity. PharmGKB now states when a label is not found.  

The main changes we are implementing are:

1. to define where the testing requirement is part of the indication of the drug.

The Indication tag is defined as “The gene or variant is an Indication for the drug, or the drug is for use in a specific population defined by the gene or variant(s), according to the label.” There are many anti-cancer drug examples of this including adagrasib and afatinib, and currently 128 out of 182 indication-tagged drug labels regard variants and genes in the cancer genome. Additional non-cancer examples are alglucosidase alfa, a hydrolytic lysosomal glycogen-specific enzyme indicated for patients with Pompe disease, and desmopressin, a synthetic vasopressin analog that is FDA approved to increase plasma levels of factor VIII activity in patients with hemophilia A (F8 deficiency) and von Willebrand’s disease Type I. 

2. to update the criteria for "informative PGx" level and to indicate where the information on the label does not meet our criteria for pharmacogenomics with tag "Criteria not met". Previously, these would have been lumped into “Testing required” and “Informative PGx” tags, respectively.

The Informative PGx level is now defined as “The label contains information stating that particular gene/protein/chromosomal variants or metabolizer phenotypes do NOT affect a drug’s efficacy, dosage, metabolism or toxicity. Or, the label states that particular variants or phenotypes affect a drug’s efficacy, dosage, metabolism or toxicity, but this effect is NOT “clinically” significant”.

The Criteria not met level is defined as "the labels appear or appeared on the FDA Biomarker List but do not currently meet the requirements to be assigned as “Testing required”, “Testing recommended”, “Actionable PGx” or “Informative PGx”.  PharmGKB annotates every label that appears on the FDA Biomarker list, regardless of whether we would otherwise annotate the label. PharmGKB also checks European Medicines Agency (EMA), and Health Canada/Santé Canada (HCSC) databases for the same drug label and annotates them if they contain pharmacogenomics information and applies the PGx level tags as appropriate.”

An example that changed from "informative PGx" to "criteria not met" is blinatumomab, which is listed on the FDA table of biomarkers with BCR-ABL and CD19. The mechanism of action is via CD19 but the label does not have information on variants of CD19 and any changes in efficacy or toxicity. The label includes information on clinical studies that included patients with Philadelphia chromosome negative Refractory B-cell Precursor ALL.

Example where labels language is different across the different agencies:

Lomitapide (https://www.pharmgkb.org/chemical/PA166114922/labelAnnotation) - FDA and HCSC criteria not met, EMA testing required.

Mavacamten (https://www.pharmgkb.org/chemical/PA166272922/labelAnnotation) -  FDA and HCSC label for Mavacamten has actionable PGx whereas the EMA label requires CYP2C19 genotyping. 

3. To show when a search was performed for the label but the curator was unable to find the document at the stated repository. 

Entries in the Drug Label Annotations landing page now show “No document available” if a label could not be found. See an example screen shot from the landing page below.

4. To enable accession to the label source directly for FDA and EMA labels.

At the bottom of the drug label annotation page, there is a section titled “Source” that has a link to download the drug label that is the source of the information in the annotation. This label is typically also available in the body of the annotation. The downloadable PDF file is usually highlighted with relevant PGx and drug-drug interaction information. See an example screen shot from the label annotation for abacavir below.

When we began annotating FDA-approved drug labels over a decade ago, we did not anticipate how widespread the use of our label annotation tags would become. Many groups have published papers that discuss PGx and FDA labels using the relative numbers of our tags (e.g. PMIDs 26693845, 25317785, 25521333, 29205206, 29722046, 34412102, 28073152). In 2012 we had 113 label annotations from the FDA; we now have over a thousand label annotations across global sources. As stated in our blog from 2013, “Curators have tried to define clearly how they interpret the level of PGx information on labels, resulting in definitions that are really a rather long list of criteria. These definitions may change over time as new label wording comes up and definitions need to adjust for these new cases.” Indeed, the PGx level definitions have been tweaked over the years to accommodate the unique nuances of labels encountered during the curation process, and we continually work to apply the changes across labels from the FDA and then cross-check against other international regulatory bodies, such as the EMA and HCSC. We are pleased that the community had found our label annotations useful and strive to improve and expand the annotations for future use. We hope these updates make the drug label annotations more consistent and evidence more transparent. Please contact us at feedback@pharmgkb.org with questions.

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”

Update to FDA-approved drug label annotations

In February 2020, we blogged about the FDA's newly released Table of Pharmacogenetic Associations. Since then, there has been much interest in understanding how that table was created, how it compares to the information on the drug labels, and how it compares to the FDA's Table of Pharmacogenomic Biomarkers in Drug Labeling, which has existed for many years and is routinely curated by PharmGKB.  With this in mind, PharmGKB has created a section on its FDA-approved drug label annotations for information from the Table of Pharmacogenetic Associations (Figure 1). 

Figure 1. Screenshot of part of the FDA-approved drug label annotation for codeine.

We also have a new landing page specifically for FDA-approved drug label annotations that can be sorted and filtered by different criteria in the column headings, including the category of the drug from the Table of Pharmacogenetic Associations ("FDA PGx Association"). The table can be downloaded in TSV format as either the full or filtered version (Figure 2).

Figure 2. Screenshot of the FDA Drug Label Annotations table.

This table can be found on the PharmGKB homepage under the "Annotation" and "Clinical" section (Figure 3) and is in addition to our Drug Label Annotations table that includes labels from multiple regulatory agencies found at the top left corner of the homepage.

Figure 3. PharmGKB homepage.

As a reminder, PharmGKB drug label annotations provide (1) a brief summary of the PGx in the label, (2) an excerpt from the label, including any guidance from the label for patients with a particular genotype/metabolizer phenotype if it exists, (3) specific variants discussed on the label, particularly if there is prescribing guidance for them, and (4) a downloadable highlighted label PDF file.  PharmGKB also "tags" labels to indicate certain information, including: 

    (5) the "PGx Level" tag ("Testing required", "Testing recommended", "Actionable PGx" and "Informative PGx") which is the PharmGKB interpretation of the level of action implied in each label

    (6) the "Dosing Info" tag which indicates dosing information based on genotype/metabolizer phenotype exists on the label

    (7) the "Alternate Drug" tag which indicates if a drug is either indicated or contraindicated based on genotype/metabolizer status on the label

    (8) the "Prescribing Info" tag which indicates if any guidance from the label for patients with a particular genotype/metabolizer phenotype exists on the label

    (9) the "Cancer Genome" tag which indicates if the label discusses a gene or variant present in a tumor/cancer cell

    (10) the "On FDA Biomarker List" tag if the label is on the FDA's Table of Pharmacogenomic Biomarkers in Drug Labels.

Figure 4. Screenshot of the FDA-approved drug label annotation for irinotecan to illustrate the types of information found in a label annotation.

PharmGKB response to the FDA Table for Pharmacogenetic Associations

In February, the FDA posted the Table for Pharmacogenetic Associations (PGx Table), and PharmGKB and CPIC blogged about the table a few days later.  We identified substantial areas of overlap and some differences between the gene-drug pairs listed on the FDA table and those with published CPIC guidelines. It is important to note that the FDA PGx Table, CPIC, and PharmGKB may update their content at any time. These comments reflect the PGx table as of May 2020,

As of the current posting, the evidence the FDA used to construct the PGx Table is not explicitly posted, although most of the gene-drug pairs with therapeutic management recommendations on the PGx Table have labels also listed on FDA’s Table of Pharmacogenomic Biomarkers in Drug Labeling (Biomarker Table) containing some kind of prescribing information as defined by PharmGKB. The CPIC process for creating guidelines has been published and involves field experts conducting extensive reviews of the peer-reviewed literature, all of which is cited or listed as supporting evidence in the guideline publications.

Direct comparison of which gene-drug pairs have CPIC guidelines versus which are listed on the FDA PGx Table is difficult due to the completely different manners in which the pairs are selected and ranked, as well as the different content and purposes of CPIC and the FDA PGx Table. 

Here we present our compilation tables of the gene-drug pairs on FDA’s PGx table with CPIC gene-drug pairs (some of which have guidelines, some are pending, and some are not considered actionable), PharmGKB annotations of drug labels found on FDA’s Biomarker Table, PharmGKB clinical annotations, and PharmGKB annotations of clinical guidelines published by several groups, including CPIC.  These tables are updated manually and therefore may not be current with recent changes in clinical guidelines, the FDA Biomarker Table or PharmGKB clinical annotations at the time of download.

This blog post has also been submitted as a comment to the open docket at FDA.

Call for clarification on antidepressant pharmacogenomics

A new Perspective in Clinical Pharmacology and Therapeutics outlines the need for improved communication from pharmacogenomic testing companies and regulators on the role of pharmacogenomics in antidepressant therapy.

Beginning antidepressant treatment can be an arduous experience for patients. Some patients fail to respond to first-line antidepressants, leading to an iterative process of trying different medications until depressive symptoms improve, while others experience antidepressant-induced adverse effects. This current situation highlights the need to make evidence-based improvements to the drug selection and dosing process.

Some pharmacogenomic information is included on the US Food and Drug Administration (FDA)-approved labels of a number of antidepressants. However, the actionability of this information can vary widely between labels and there is no guidance to clinicians on when to seek pharmacogenomic testing for a patient. PharmGKB assigns a PGx Level and tags to our drug label annotations to help users easily recognize labels containing actionable pharmacogenomic information such as recommendations for dose or altered drugs.

The publication’s authors, including Dr. Teri Klein and Dr. Katrin Sangkuhl (PharmGKB and CPIC), Dr. Andrea Gaedigk (PharmVar), as well as Dr. Kelly Caudle and Dr. Roseann Gammal (CPIC), argue that the inconsistencies in drug labels combined with the FDA’s recent communications on the safety and validity of pharmacogenomic testing has created confusion among clinicians and patients about pharmacogenomics in antidepressant prescribing. They also highlight the work of CPIC in producing evidence-based clinical guidelines that can be used to guide drug selection and dosing. In particular, the CPIC guidelines for selective serotonin reuptake inhibitors and tricyclic antidepressants are based on the critical review of decades of scientific evidence.

The article ends with a call for clarity on the level of evidence required to implement pharmacogenomic-guided prescribing in the clinic, particularly with reference to sertraline and escitalopram.

Annotated Swissmedic Drug labels now available on PharmGKB

PharmGKB now has annotated drug labels available from the Swiss Agency of Therapeutic Products (Swissmedic), a regulatory authority responsible for the authorization and supervision of therapeutic products in Switzerland.

The Swissmedic drug label annotations are sourced through a collaboration with the Pharmaceutical Care Research Group (PCRG), Department of Pharmaceutical Sciences, University of Basel. The PCRG group screened and translated pharmacogenomic (PGx) relevant information in the Summaries of Product Characteristics (SmPCs) of all Swissmedic labels through natural language processing (NLP). Using these translations and analysis, PharmGKB annotated Swissmedic drug labels based on our criteria for PGx information.

Like our FDA drug label information, we provide a summary and excerpt of PGx information from the Swissmedic SmPC. Each Swissmedic drug label annotation is also given a PGx level of evidence, and a link to the original full text label on the AmiKo Web (https://amiko.oddb.org/). If the label provides dosing adjustments based on genetic information or metabolizer phenotypes, or states that a drug is either indicated or contraindicated for a particular set of patients, the label annotation will be tagged with "dosing info" or "alternative drug" tag.

It is likely that there are other Swissmedic drug labels that contain PGx information, and PharmGKB welcomes any feedback regarding PGx information within the labels from Swissmedic or other medicine agencies around the world.

View our complete list of drug labels with PGx information.

New and Updated PMDA label translations now available

New translations of package inserts from the Pharmaceutical and Medical Devices Agency (PMDA), Japan, as well as updated translations of existing package inserts, are now available on PharmGKB.

The PMDA is a regulatory agency responsible for scientific reviews for the approval of drugs or medical devices, as well as safety monitoring after approval. The PMDA website provides PDF copies of package inserts for approved drugs, though these inserts are only available in Japanese. Previously, PharmGKB had used a 2013 paper by Shimazawa and Ikeda that selected PMDA inserts to examine for PGx information based on the FDA's Table of Pharmacogenomic Biomarkers in Drug Labeling as it existed on October 2012, and then provided translations of any PGx information present in the PMDA package inserts. Though this paper was useful, additional drug labels with PGx information have been added to the FDA table in the past 4 years, and PharmGKB has also identified drug labels with PGx information independently. 

Now, through a collaboration with the Japanese Society of Pharmacogenomics, as well as Silicon Valley Tech KK, PharmGKB is able to provide translations of a wider set of PMDA labels, as well as more extensive translations of some labels. 

Some of the new labels available include:

Allopurinol
Carbamazepine
Efavirenz
Fesoterodine
Methylene Blue

An overview of all the drug labels on PharmGKB can be viewed here. 

PharmGKB thanks the Japanese Society of Pharmacogenomics for its diligent work in translating these labels, and helping to make them available to the public.

Introducing the new PharmGKB Cancer PGx Portal

PharmGKB has collected a number of resources for Cancer PGx into one easy location. There are tables with direct links to genes important for cancer drug response both for PD and PK, to cancer drug pathways, particular cancers that have PGx data, types of toxicities common to cancer drugs, and external resources.

Eight new VIP gene pages give a short text based summary of important genes for cancer drug response. These are for the genes ALK, ABL1, BCR, BRAF, ERBB2 (HER2), KIT, KRAS and NRAS. Anyone with expertise in the genes who wishes to develop these with us for publication in PG&G, please contact feedback.

There is a shortlist of drug labels for cancer drugs with biomarker PGx.

We currently have 34 anti-cancer agent drug pathways with 8 new pathways in development. The portal gives shortcuts to a selection.

PharmGKB currently uses a flat ontology for diseases, which means that the Neoplasms disease page does not link to the many different cancers we have data for. The cancer portal has direct links to the cancers for which there is the most PGx information in the knowledgebase, such as pediatric ALL, CML, colorectal, breast, renal and non-small cell lung cancers. The portal also has links to the common types of toxicities with PGx data.

Finally there is a collection of external links that are useful for Cancer PGx.