An algorithm designed to predict the likelihood that a person will experience a problematic pattern of opioid use may not be clinically useful despite receiving premarketing approval from the U.S. Food and Drug Administration (FDA), a study suggests.
Researchers found little evidence demonstrating that the genetic variants used by the algorithm predicted the risk of opioid use disorder (OUD) when tested in over 400,000 individuals.
Independent testing showed high levels of inaccuracy using the algorithm’s 15 single nucleotide variants (SNVs), with just 47 of 100 predicted cases or control individuals correctly classified.
Collectively, these SNVs accounted for less than one percent of the variation in OUD risk, the researchers reported in the journal JAMA Network Open.
“Notably, clinicians could better predict OUD risk using an individual’s age and sex than the 15 genetic variants,” pointed out Christine Davis, PhD, from the Center for Studies of Addiction at the University of Pennsylvania Perelman School of Medicine in Philadelphia, and team.
They added that “although the test approved by the FDA is intended to complement standard clinical assessment, its use is unlikely to confer additional benefits and may instead give clinicians and patients false and potentially harmful information.”
OUD is a mental health condition that revolves around the problematic use of opioids such as fentanyl. In 2022, 6.1 million people in the U.S. aged 12 years or older met the criteria for OUD, with over 90% admitting misuse of prescription analgesics and over 40% reporting receipt of the misused medication from a physician.
Given the surge in opioid overdose deaths, efforts have been made to identify individuals at risk of opioid misuse and there have been attempts to develop and commercialize genetic risk OUD algorithms.
With this in mind, Davis and co-workers examined the value of genetic variants included in the AvertD algorithm, which recently received premarketing approval from U.S. regulators.
The algorithm includes 15 SNVs designed to predict OUD risk and was developed using 1381 U.S. individuals and tested in a multicenter clinical study that enrolled 812 patients, of whom 385 were included in the analyses.
The manufacturer reported sensitivity of 82.76% and specificity of 79.23%.
Davis and the team noted that the package insert for the algorithm states that the “15 detected genetic polymorphisms are involved in the brain reward pathways that are associated with OUD.”
However, they added that it provides no citations to support the associations, “all of which appear to have been identified through candidate gene studies have been identified through candidate gene studies.”
To investigate further, the researchers conducted a case-control study of 452,664 ancestrally diverse U.S. veterans with opioid exposure, including 33,669 with OUD.
Participants had a mean age of 61.2 years, and over 90% were male.
Collectively, the 15 candidate genes accounted for just 0.40% of variation in OUD risk, which was consistent with small individual effects of common genetic variants on complex traits. By comparison, age and sex alone accounted for 3.27% of the variation.
An ensemble machine learning model using the 15 variants as predictive factors correctly classified just 52.83% of individuals in an independent testing sample.
The authors noted the high rate of false positive and negative findings using the SNVs.
“False-positive findings can contribute to stigma, cause patients undue concern, and bias health care decisions,” they commented.
“False-negative findings could give patients and prescribers a false sense of security regarding opioid use and lead to inadequate treatment plans.”
The team added: “Although the AvertD test uses a proprietary algorithm, the issues identified herein suggest that the manufacturer has a fundamental misunderstanding of genetic principles, particularly the impact of differences in population structure and allele frequency.
“Genetics researchers have argued against the use of candidate genes to predict OUD and other psychiatric traits. Most recently, 153 genetics experts indicated concerns about use of the AvertD test in clinical settings.”
