NEW YORK (GenomeWeb) – As America is struggling with an epidemic of opioid abuse, genetic testing company Proove Biosciences is hoping its tests will help physicians more carefully prescribe opiates for pain management.
Like other pharmacogenetics companies, Proove offers tests that measure a person's SNPs in cytochrome P450 and other genes and inform drug prescribing. However, the company also offers assays that are unique to the market, such as the Opioid Risk test to determine a patient's risk of misusing opioid medications.
Experts in genetics and addiction consulted by GenomeWeb were particularly concerned about this test, which the company markets as a tool to help physicians determine whether or not patients are at risk of abusing opioid medications. The company claims to have performed this test on 50,000 patients already.
There were about 4 million nonmedical users of prescription pain relievers in 2014, and 2 million people with substance use disorders related to pain medication, according to the Substance Abuse and Mental Health Services Administration. Opioid drugs are pleasurable and reinforcing, so the conversion rate from use to dependence is high, yet the difference between an effective dose and a toxic dose can be small, resulting in accidental overdoses.
The crisis has led the US Centers for Disease Control to issue new guidelines on prescribing opioids for chronic pain, and Congress to pass a bill last week to combat opioid abuse, including measures encouraging clinicians to reduce unnecessary prescribing.
Irvine, California-based Proove says its genetic tests can help doctors identify and better manage patients at risk for abusing opioids, and maintains it has the data to support these claims. "Every day in America there are people that are becoming addicted and dying from opioids — if the current standard of care was working, then we wouldn't have the epidemic we're having today," said Proove CEO and founder Brian Meshkin.
Proove claims to be "the research leader investigating and publishing data on the genetics of personalized pain medicine, with clinical research sites across the United States." Meshkin told GenomeWeb, "We want our name to become a verb, where you prove the treatment and prove the diagnosis."
According to Meshkin, Proove has been conducting extensive research on the pharmacogenetic tests it sells. The company has presented 79 posters at scientific and medical conferences in the past four years based on studies of patients, Meshkin noted. It has also conducted research at numerous clinical sites. The company recently submitted a study for peer review, and is now submitting "three or four studies per month," he said. Proove has published only one review article to date.
But as the company is shoring up the evidence on its tests, Proove performed 198,000 tests last year, according to Meshkin. The 11 tests in Proove's portfolio cost between $299 to $2,599, according to the firm, and currently about one in three is reimbursed by insurance. "If a test is not reimbursed, then a patient may have an obligation that has to be met through either a out-of-pocket financial contribution or providing service as a study participant," Meshkin said. Although a number of companies currently offer similar pharmacogenmic lab-developed tests — including Genelex, Pathway Genomics, GeneAlign, PCLS, Kailos, RS Medical, Millennium Health, Quest Diagnostics, Molecular Testing Labs, Exceltox, Kashi Health, and ViaQuest — and many of these companies sell PGx testing for pain medications, Proove's drug metabolism genetic profile claims to be "the most comprehensive test on the market."
The firm also sells genetic tests assessing a patient's pain perception, risk of becoming dependent on opioids, opioid response, risk of side effects with non-steroidal anti-inflammatory drugs, response to medically assisted detoxification, and response to an epidural with the opioid fentanyl.
Last week, Proove launched three more tests: Proove Fibromyalgia, Proove Addiction, and Proove Thromboembolism.
Patients who receive testing from Proove are asked for broad informed consent, according to Meshkin, essentially giving the firm permission to track patient medical records for long periods of time to determine how testing affects outcomes. The company also documents how doctors are using the information, and archives patient samples for future research purposes in "the largest biobank in the world for chronic pain."
By collecting longitudinal data on the impact and utility of its tests, Proove hopes to both demonstrate to payors that its tests result in better patient outcomes for reimbursement and to improve the algorithms that underlie its tests, Meshkin said.
Proove exclusively licensed the rights to informatics algorithms that power its tests from the University of Utah. These algorithms pull in patient physiological characteristics and genetic information obtained in the firm's CLIA and CAP-certified labs using a Life Technologies Quant Studio real-time PCR platform. Meshkin said he is in talks with Affymetrix and Illumina to bring in house a high-density SNP and sequencing platform before year end.
Proove's pain focus
Chronic pain is difficult to manage, often requiring medication as well as physical and psychological therapies, and occasionally more invasive interventions. Patients are usually suffering from other health problems, and pain has impacted their lives and reduced their function.
Using medication to manage pain can be challenging for doctors, because when patients don't metabolize certain analgesics properly they continue to experience pain, but with opioids, they are at risk of addiction.
Opioids can also lead to overdose, and death, said Steven Richeimer, chief of the division of pain medicine at Keck School of Medicine at the University of Southern California, and he and his colleagues are also concerned that opioids can sometimes make pain worse via so-called opioid-induced hyperalgesia.
"Our field is struggling to try to develop an understanding of options and [to know] when the risks are greatest," said Richeimer, who uses Proove's genetic tests and serves on the firm's medical advisory board. "That's where Proove has some potential."
Proove and Richeimer's group at USC are working together on a research study that's also funded by the firm.
"As part of that, we are doing their testing on quite a few of our patients, and examining the process of giving results to patients, the utility of the tests as they are right now, and we're looking to develop new genetic markers for treatment responses," Richeimer said. He noted that his group offers these tests to patients under a research protocol, since Proove will refine its tests based on the research.
Richeimer highlighted that the opioid risk test has benefited some patients. "Like a lot of other pain doctors around the country, we are trying to run very tight ships about how much opioid we use," he said. "If we have a patient on little to no opioid, [the Opioid Risk Test] is less helpful, although sometimes it's helpful even then for the patient to be aware if they are at high genetic risk for addiction."
Proove developed the Opioid Risk Test based on genes in the scientific literature that have associations with opioid abuse related disorders, as well as mental health disorders commonly seen in chronic pain patients. "I chose those particular genes based on a third party literature search and then started a process of clinical development," Meshkin said. A proof-of-concept description of the gene set is published in the International Journal of Biological Sciences, Meshkin said, a review article that is the firm's only published work to date.
The test itself measures genetic variation in 12 genes that may have some relationship to mental health and behavior.
Specifically, it measures SNPs in or near genes for the serotonin 2A receptor (rs7997012), dopamine transporter (rs27072), serotonin transporter (rs140701), dopamine beta-hydroxylase (rs1611115), catechol-O-methyltransferase (rs4680), methylenetetrahydrofolate reductase, or MTHFR (rs1801133), dopamine D2 receptor (rs1800497), opioid receptor kappa 1 (rs1051660), dopamine D1 receptor (rs4532), GABA-A receptor gamma 2 (rs211014), dopamine D4 receptor (rs3758653), and opioid receptor mu 1 (rs1799971).
The test also involves a questionnaire documenting phenotypic risk factors. This is combined with a patient's genetic data and then analyzed by a proprietary algorithm to generate a score on a scale of 0 to 52.
A comparison of 80 patients diagnosed with opioid drug dependence versus 82 controls revealed that a score of greater than or equal to 13 predicted risk of aberrant behavior to opioids, according to posters Proove presented at conferences recently.
Data the firm presented this year at the International Conference on Opioids showed the test had a positive predictive value of approximately 93 percent — a number the firm highlights in its promotional materials for this test. This value held up in a validation study of around 900 patients.
Data in another poster provided by Proove further showed that, compared to published data on the commonly used questionnaire-based Opioid Risk Tool (ORT) for Narcotic Abuse, the test had a negative predictive value of about 83 percent, and a sensitivity and specificity of 80 and 94 percent, respectively.
But some academic researchers who study the genetics of addiction are skeptical about Proove's claims. Joel Gelernter, a professor of psychiatry, genetics, and neuroscience at Yale University School of Medicine, co-authored a GWAS of opioid dependence in Biological Psychiatry in 2014.
To discover risk loci requires unbiased assessments, he said, like GWAS, whole-genome sequencing, or even a genetic linkage, and such studies are "the state of the art in identifying genetic risk factors for opioid dependence."
However, studies Proove lists to support its choice of targets for the opioid risk test are "predominantly small, candidate gene studies that were OK for the time, but with the knowledge of today we recognize them as being quite underpowered," he said.
And although the SNPs Proove selected involve logical choices of neurotransmitter receptors, transporters, and related molecules, "One thing that we've learned from all the complex trait studies over the years is that our knowledge of the biology isn't good enough, by and large, to be able to pick what our hits are going to be in advance," Gelernter said.
Furthermore, the majority of the studies Proove cites in the opioid risk test promotional material to support its choice of targets are more than five years old, with about one quarter being more than 10 years old.
"There are certainly some very good investigators on this list, but many of the leaders in the field are not represented," Gelernter said, adding, "I think most of the leaders of the field, including the ones on this list, would be very reluctant to advocate moving whatever their findings are into the clinic."
Interestingly, two of the references supporting the test are from the journal Medical Hypotheses, which publishes theoretical papers. One describes a pilot study of a method to induce rapid detoxification in chronic methadone users that was performed on 12 patients. It is cited to support relationship of the DRD4 SNP to drug abuse. And use of the dopamine beta hydroxylase SNP target to assess opioid risk is supported by a theoretical Medical Hypotheses study along with a study of Han Chinese children with attention deficit/hyperactivity disorder.
Meanwhile, Gelernter's analysis of more than 12,000 subjects revealed the strongest associations between opioid dependence and genes involved in calcium and potassium signaling in the brain. None of the Proove targets were revealed in this GWAS.
A meta-analysis published this year in Behavioral Genetics, however, did indicate one of Proove's targets, a SNP in the mu1 opioid receptor gene (rs1799971), may contribute generally to addiction, although the authors noted the associations were not statistically significant, possibly because of reduced sample sizes.
Gelernter's work also refutes an element supporting a different marker in the Proove test — a SNP in the dopamine D2 receptor that the firm says is linked to drug addiction based on a study from 1993.
Gelernter and colleagues found no evidence of association of the rs1800497 variant with alcohol or drug dependence, but rather discovered that this marker is in linkage disequilibrium with other variants they found to be associated with nicotine dependence. "Thus, it is apparent that previous reports of association of substance dependence traits that included this marker only could, potentially, be attributable to LD with a more centromeric marker mapping to ANKK1 or TTC12," they wrote in a 2006 study published in Human Molecular Genetics. These results were later extended by another group, in a 2013 study published in JAMA Psychiatry.
There are now several opioid dependence GWAS data sets available in the National Center of Biotechnology Information's database of Genotypes and Phenotypes, or dbGaP, Gelernter said.
"If you wanted to know the relationship of [Proove's targets] to opioid dependence risk ... you can make a polygenic risk score from these 12 SNPs," to test out its predictive value, he said. "My guess is that they would not be predictive in a useful way, but that's just a guess, it is something that you can test today."
He further said that, based on what scientists currently know about complex trait genetics, a polygenic risk score that is predictive would be expected to require many more than 12 markers. "If you were to take the top 12 hits out of our GWAS instead, I would predict that wouldn't be very predictive because that is not enough markers ... but it ought to be more predictive than this set of 12 because their significance was higher in a better-powered GWAS study."
Ultimately, Gelernter's impression was that the six phenotypic items that go into Proove's test algorithm — a personal history of alcoholism, illegal or prescription drug abuse, age between 16 and 45, any mental health disorders, and current depression — might have the most utility in the algorithm to predict opioid misuse.
As far as the genetic markers, he said "I don't want to speak for my colleagues, but I think that most people would not say that we are ready to move into the clinical world based on the currently available published literature, and certainly not on studies of DRD2 variation from 1993."
Henry Kranzler, a researcher in the pharmacogenetics of substance dependence, director of the Center for Studies of Addiction at the University of Pennsylvania, and a co-author on the opioid dependence GWAS with Gelernter, also reviewed Proove's promotional material at the request of GenomeWeb.
"Unfortunately, claims such as the ones made by Proove are not reviewed by the [US] Food and Drug Administration," he said in an email, adding, "If they were, it would likely not withstand the scrutiny. There is no published, peer-reviewed evidence that supports Proove's claim that it is able to identify individuals' susceptibility for opioid misuse."
Kranzler noted that the individual phenotypic markers, based on self-report, that are included in Proove's opioid misuse algorithm are largely established, but in his opinion the effects are of limited clinical utility.
Further, Kranzler said it is unlikely that, even combining phenotypic markers with information from genetic markers — for which he said there is inadequate support in the literature — could yield a positive predictive value of 93 percent. Kranzler believes Proove's claims about the clinical utility of its algorithm need independent evaluation.
GenomeWeb also spoke with Richard Soper, an addiction specialist in Tennessee and a member of American Society of Addiction Medicine, and Thomas Kosten, director of the division of alcohol and addiction psychiatry at Baylor College of Medicine. These experts concurred that, without support in peer-reviewed studies or outside validation, it is hard to believe that this test would have clinical validity based on the list of markers in the product brochure.
Kosten's own pharmacogenetic studies, for example, have implicated the mu opioid receptor SNP very narrowly in responsiveness to the drug naltrexone among people with alcohol dependence. He said there is not much evidence it is related to naltrexone response in opioid dependence, much less whether a patient will develop dependence to begin with.
By weighting the SNPs and phenotypic variables in an algorithm, Kosten envisioned it could be made to separate a sample of dependent versus non-dependent patients. But this equation applied to a second sample would not necessarily be predictive.
So far, Meshkin has funded Proove himself, and the firm is now in the midst of raising its first round of private equity.
Before launching Proove in late 2009, Meshkin worked in sales and marketing at Eli Lilly and Prometheus Laboratories, and started a nutrigenomics firm, called Salugen, in 2005. Salugen performed about 20,000 direct-to-consumer tests, he said, but he ultimately couldn't make the company profitable. California regulators also issued Salugen a warning letter about DTC marketing, along with a slew of other firms. Meshkin sold Salugen to a European equity firm in mid-2009.
Proove's medical advisory board has 11 members. The majority are pain medicine specialists, both clinicians and researchers, and some of them like Richeimer are early adopters of the company's genetic tests.
One medical advisory board member, Katrina Lewis, a physician who is board-certified in anesthesia and pain management, has ordered nearly 2,000 tests from Proove for her patients. "It has made my job so much easier, and I don't feel like I'm stabbing in the dark," she said.
Lewis noted Proove paid her for the time she spent filling out paperwork as part of an IRB-approved clinical trial, a practice that drew criticism in a recent article on lab-developed testing for psychiatric conditions.
"There is a little bit of paperwork ... which takes about five or ten minutes each time, and they pay for that block," she explained. The effort adds up to between $500 and $1,000 a month, Lewis estimated. Currently, the hospital where she works gets those funds, but at a previous location where she interacted with Proove directly, she received payment. "It's not like it's huge amounts of money," she said, adding that this is standard for research protocols and that she would walk away from the money if she thought it in any way compromised her practice.
Meshkin said the firm compensates clinicians participating in studies for their time filling out assessments based on fair market value, as determined by a valuation firm. The clinicians fill out time sheets, and the compensation is not linked to the number of tests they order. The firm also puts a "patient engagement representative" at some sites where it collects data. Meshkin said the suggestion that these payments are influencing doctors is "absurd," and that the practice is quite standard and legal.
Moreover, in response to those criticizing the scientific basis of Proove's tests, Meshkin said they are "reacting to a concept, without seeing the data," and that it is the genetics combined with the phenotypic markers that gives the tests such a high level of predictive value.
The opioid risk test that is currently commercially available was developed about six years ago. According to Meshkin, the firm added the phenotypic data to its algorithm because "there really isn't a combination of polymorphisms that can 100 percent predict opioid dependence." He estimated genetics alone accounts for only 60 percent of the test's predictive capability.
Meshkin highlighted that Proove has enrolled more than 34,000 patients in prospective, IRB-approved, clinical studies to look at chronic pain outcomes in 2015, and it has more than 100,000 samples of patient DNA in its biobank. The company has also enhanced the underlying test algorithm several times over six years.
"As we collect more data and other people publish more, we will refine it," Meshkin said.
Specifically, for the opioid risk test the firm plans to publish results from three clinical studies. One that has already been submitted for publication is a proof of concept describing the algorithm's predictability for opioid abuse. It evaluated 335 patients across 15 sites, 80 with opioid dependence and three different control groups.
A retrospective observational study is expected to confirm the validity of the algorithm's determination of low, moderate, and high risk of opioid abuse. That study looked at 650 patients, across 28 research sites in one cohort and 31 research sites in a second cohort, with a manuscript expected to be submitted in September.
A third study, that will be submitted this month, is a "retrospective analysis of prospective data" that looked at 100 patients with opioid dependence versus 114 controls to validate the algorithm's predictability.
In addition, the firm is submitting clinical utility studies of more than 1,000 patients in the next few months, to show that patients have improved functioning and there is also an economic benefit to using the tests.
"I appreciate where those guys are coming from," Meshkin said of the firm's critics, "but the discussion we would have in six to eight months, if they see all of the data we have published, will dramatically change their attitude."
The Gelernter and Kranzler GWAS is "a great study," he noted, and the firm is actually using that study, along with others, to expand its test. A limited version of a chip-based test has already validated the calcium and potassium pathways picked up in the GWAS, Meshkin said. The company is also planning to launch a methylation-based test by the end of 2017, which will analyze epigenetic signatures related to opioid dependence and response, and pain perception.
"We wouldn't be doing what we are doing if we didn't have the data to support it," Meshkin said. "When the name of your company is Proove, you have to prove it."