It’s national poetry month again!

I am not a mouse.
It’s the reference allele
(hg38).

I am not a mouse.
It's the common variant
In these folks tested.


Saw on minus strand
G is the major allele
In this group tested.


C is the minor
It’s the reference allele
(hg38).


Last year I wrote a haiku to highlight my gripe about authors declaring a patient a *1 but not stating which alleles were tested so not conclusively ruling out the presence of variants. This year I have a series of haiku on another gripe of mine - authors using the term “wild-type” when talking about humans. I’ve been guilty of it in the past but when we know better we do better. The one upside of it is it's a way one can avoid using terms major/minor allele when those might be different in different populations and thus ambiguous, but I think most study participants would not want to be categorized as wild-type, certainly most would not want to be called mutant (unless it comes with some X-men type powers). The terms I would recommend are “reference allele/genotype” (on genome build …) and “comparison allele/genotype”. If using major/minor allele then state explicitly what base that was in the given population, and describe the population, and which strand the allele was measured on (especially for C/G and A/T variants and genes on minus chromosomal strand). We have been seeing problems lately when authors assume the reference allele on the genome build hg38 is the major allele in most populations and that is not always the case. If in doubt give as much information as possible.

What does *1 really mean? And why does that matter?

A Haiku about PGx for National Poetry Month.

Diagnosed star one,

Yet severe toxicity.

What did the lab test?

What does *1 really mean? And why does that matter?

The star alleles of the drug metabolizing enzyme genes are an unusual way of defining variation and are perhaps one of the most misunderstood aspects of pharmacogenomics. Arising from attempts to describe and standardize the molecular basis of different drug phenotypes; debrisoquine poor metabolizer phenotype, etc [PMID: 7773298, PMID: 8807658].  Sometimes authors use the term “wild type”, even for humans, to describe the most common form of the gene or protein in a given population where it displays the expected drug metabolism phenotype. The *1 allele is generally used to denote the absence of the variants tested. However, it is not a stable assignment; a *1/*1 individual only tested at one locus may not have the same genetic sequence as a *1/*1 individual tested for a panel of 10 variants in the same gene. This really matters when evaluating the likelihood of drug phenotype - the "reference" is only as good as the number of variants that were checked. *1 is rather a placeholder, it is the absence of certainty, because even with a panel that covers variants that represent 99% of known variation (in the populations examined so far which are not representative of the full global population) there may still be rare variants with significant impact on protein function that we may not yet have documented. While a *1/*1 may not be at increased risk of toxicity (or other phenotypes), or require a change of dosage immediately, they still have the baseline level of risk and it shouldn’t be a huge surprise if they exhibit an adverse reaction to a drug. 

Recent publications of case studies that concluded a lack of involvement of known pharmacogenes without documenting what was tested: 

PMID:35180762 - “Acral Skin Rash Caused by Altered Mercaptopurine

Metabolism in Maintenance Therapy for B-Cell

Acute Lymphoblastic Leukemia”  excerpt “TPMT and NUDT15 genotype at diagnosis were wildtype alleles and therefore we started with full 6-MP dosing.”

This issue is not limited to pharmacogenes using star allele nomenclature: “5-Fluorouracil Neurotoxicity in the Absence of Dihydropyrimidine Dehydrogenase Deficiency Case Report” [PMID:35419161] excerpt “Our patient tested negative for DPD mutations, but it remains possible she harbored a genetic variant not accounted for in the genetic testing panel. Other known risk factors include mutations in the orotate phosphoribosyltransferase and thymidylate synthase genes… “

We recommend authors always include the list of all variants that were tested, and other features see [PMID:30406943].