Determining a Drug Response

Pharmacogenomic testing can guide dosing and prevent serious adverse reactions.

By Sonya Collins

Photographs by Arturo Rodgriguez

The pain was too intense to ignore. Sixteen years later, Tony Cirrincione still remembers it. He was on a weekend ski trip, and he staggered into the ski lodge, leaving his wife in charge of their son and the five other Cub Scouts they’d brought along. Wincing at the dull ache in his back, he tried to stretch away what had to be a muscle cramp. But the bursts of pain grew only more intense, erupting at more frequent intervals. Soon he was in the passenger seat of their Nissan minivan, the Cub Scouts in back, as his wife raced them to the emergency room at NorthShore University HealthSystem in Chicago. Kidney stones, the nurse declared the moment she saw him doubled over the triage station. Minutes later, he lay on a gurney in the ER, waiting for the prescription opioid Dilaudid (hydromorphone) to take hold. ­But it never did.

As many as 10 percent of people do not receive pain relief from opioids. That’s a staggering statistic in a country where more than 75 million people live with chronic pain. Many individuals, like Cirrincione, who don’t respond to certain prescription pain relievers have no idea why. But a type of genetic testing is bringing answers to a growing number of them.

It’s not only pain medications that are implicated. Gene variants could also interfere with about 100 to 150 of the 1,200 Food and Drug Administration–approved prescription drugs on the market today. That may seem like a small number, but consider the burden of the conditions these drugs treat — depression, heart disease, cancer, and HIV, among others. Some gene variants can render drugs useless. Others may mean a person needs a larger or smaller dose than the standard. Still other variants could put a patient at risk for potentially serious side effects. Genetic variability could explain some of the one million emergency room visits that adverse drug reactions cause every year.

Until recently, we took it for granted that risks of poor effectiveness or toxicity go hand in hand with taking medications. But the growing field of pharmacogenomics is turning that assumption on its head. Although genes won’t explain all of a patient’s variable responses to a medication, a pharmacogenomic test from a simple cheek swab could help predict a person’s individual response to a medication before the doctor ever writes the prescription.

What do genes have to do with it?

Opioid pain medications work by attaching to receptors in the brain, spinal cord, gastrointestinal tract, and other organs to reduce the perception of pain. A number of variations in several different genes can affect how well these drugs work in a given person.

Cirrincione had two excruciating bouts with kidney stones over the course of about 10 years before he learned why certain pain medications didn’t work for him. By this point, he had discovered on his own — painstakingly — that ibuprofen did far more for him than hydromorphone ever had. Just last year, Cirrincione’s wife, who works at NorthShore University HealthSystem, heard about the new pharmacogenomics clinic. People could go in for a cheek swab, she told her husband, and specialists could test their DNA for scores of gene variations that could influence response to certain medications. Cirrincione’s wife thought it sounded interesting, so she and her husband were among the first at the hospital to try it.

A few days later, Mark Dunnenberger, the director of the pharmaco-genomics program, called Cirrincione with information that explained the pain all those years ago. Cirrincione has a variation in the gene OPRM1 that changes the receptors to which morphine-based opioids attach. As a result, the drugs can’t activate the opioid receptor and may bring little to no pain relief.

That’s not the only gene that can interfere with the work of pain relievers. Variations in the CYP2D6 gene can make a person either a poor or an ultra-rapid metabolizer of certain opioids. Take codeine, for example. The body activates codeine by converting it into morphine, and it’s the morphine that delivers most of the pain relief. A poor metabolizer of codeine doesn’t convert the drug into morphine. A rapid metabolizer quickly converts a higher amount of codeine to morphine, putting the patient at risk for toxic levels in the bloodstream.

“There was a case of a breastfeeding mother who got Tylenol with codeine after giving birth. She was an ultra-metabolizer and accumulated so much morphine in her breast milk that the baby died from respiratory depression,” says Dunnenberger. Respiratory depression, when breathing slows to such a degree that the brain does not get enough oxygen, is often the actual cause of death or brain damage in an opioid overdose. While the baby’s death was a vary rare occurrence, the FDA responded by issuing a warning about this risk.

Including pain meds, CYP2D6 variants impact some 40 known medications, such as the metabolism of pimozide for Tourette syndrome and tetrabenazine for Huntington’s disease. As with many genes, such as those that code for eye and hair color, drug-related gene variants can be more prevalent in one subset of the population than another. Up to 10 percent of Caucasian people carry CYP2D6 variants that interfere with drug metabolism. Fewer than 1 percent of Asians carry the variants. In some African populations, carrier rates may be as high as 35 percent.

Not all opioids are metabolized by CYP2D6, though. That’s why poor metabolizers of one drug might benefit from a larger dose of that drug or a standard dose of a different drug. Rapid metabolizers might benefit from a lower dose, which would lower the risk of accumulating toxic levels of medication in the bloodstream. Or, like poor metabolizers, they might be better off with another drug.

The gene CYP2C19 also influences drug metabolism. Cirrincione has a variation of CYP2C19 that makes him a rapid metabolizer of a handful of drugs, including some types of antidepressants, proton pump inhibitors for acid reflux, some sedatives and muscle relaxants, and the blood thinner Plavix (clopidogrel). Fortunately, Cirrincione hasn’t ever needed any of these drugs, but the warning about Plavix hit close to home.

“Both my parents had coronary artery disease. That was listed as the cause of death for both of them. And my brother had a couple of heart attacks and had stents,” Cirrincione says. Rapid metabolizers of Plavix, which is often given to those with coronary artery disease, may be at an increased risk for bleeding.

Pharmacogenomic prescribing guidelines

CYP2C19 is a gene that encodes a liver enzyme of the same name. The enzyme plays a role in the metabolism of 5 to 10 percent of drugs currently in use. When doctors know that a patient has a meaningful variation in this gene, they can adjust prescriptions accordingly, if alternatives exist.

Doctors can learn about recommended alternative therapies through the Clinical Pharmacogenetics Implementation Consortium (CPIC). The group compiles evidence-based alternative dosing and prescribing recommendations for patients who carry relevant gene variants. Psychiatrists are particularly enthusiastic about how CPIC guidelines could impact their field.

“I have worked with some patients who have depression and who went through protracted courses on medications that caused side effects or that did not bring them any benefits. After 18 months, their doctors got them on the medications that their genetic results later suggested would be the right medication for them.”

“In CPIC, we’re often contacted by outside people to see which guidelines we’re writing next. The most common requests we get are from the psychiatry community to deal with antidepressants and antipsychotics,” says Mary Relling, chair of Pharmaceutical Sciences at St. Jude Children’s Research Hospital in Memphis, Tennessee. Relling co-leads CPIC. “That’s a very frustrating area, which has traditionally relied on trial and error to decide what drug works. Because it takes so long to evaluate whether the drug is working, it really puts patients at risk of not having their disease controlled for a long time.”

Currently CPIC offers recommendations to avoid certain antidepressants or to adjust the dose for people who have particular genetic variants. These recommendations could save people from spending months or years trying different medications before they find relief.

“I have worked with some patients who have depression and who went through protracted courses on medications that caused side effects or that did not bring them any benefits. After 18 months, their doctors got them on the medications that their genetic results later suggested would be the right medication for them,” says Dunnenberger, who is also a CPIC member.

FDA Warnings

The FDA requires pharmacogenomic warnings on the labels of about 100 drugs. In some cases, the label recommends that prescribers test for a particular gene variant before prescribing the drug. This is true for abacavir, an antiviral used in combination with two other drugs as a first-line treatment for HIV.

Abacavir can cause a severe, potentially life-threatening side effect known as Stevens-Johnson syndrome in people who carry a variant of the HLA-B gene identified as HLA-B*57:01. Stevens-Johnson syndrome causes a fast-spreading, painful purple or red rash that eventually blisters and causes the skin to fall off. “People with Stevens-Johnson are treated like burn patients, and there’s a very high mortality rate,” says Dunnenberger. “Pharmacogenomics has reduced the number of people who get Stevens-Johnson syndrome when given abacavir.”

People who don’t carry the genetic marker run a very low risk of having this severe allergic reaction. “It’s one of the cleanest examples of a genetic test having a high predictive positive value and high negative predictive value,” says Relling.

Another HLA-B variant — B*15:02 — is most common in some Asian populations and raises the risk of Stevens-Johnson syndrome in people who take carbamazepine, an anti-seizure medication and mood stabilizer used in the treatment of epilepsy and bipolar disorder. Allopurinol, a medication for gout, can cause Stevens-Johnson in people with yet another HLA-B variant: B*58:01. Fortunately, alternatives to all three of these drugs exist, so prescribers might opt to genotype patients before prescribing or, when the alternative is safer and equally effective, they might go with a different drug from the start.

No viable alternative

Prescribers and their patients face tough decisions when safe and effective alternatives do not exist. “The tricky thing is, for some of the alternatives, there’s evidence to suggest that [the gene variant] is a marker for Stevens-Johnson syndrome in those alternative drugs as well, so you have to be careful about which ones you recommend. It depends on what you’re trying to treat. It’s not about just blindly picking one drug or another,” says Dunnenberger.

In some cases, the alternative is known to be far less effective. Patients and their families must weigh the risks of taking the medication against those of not taking it.

The drug rasburicase can pose this challenge. People newly diagnosed with leukemia or lymphoma may receive this drug for tumor lysis syndrome. This condition occurs when a large number of tumor cells are killed off in chemotherapy and they release a high level of uric acid into the bloodstream in their wake. Rasburicase can eliminate the excess uric acid, which could otherwise cause renal failure. But people with a genetic deficiency in glucose-6-phosphate dehydrogenase (G6PD) — about 400 million people worldwide, most commonly males from Africa, the Mediterranean, and parts of Asia — who take rasburicase are at high risk for hemolytic anemia. This condition kills red blood cells faster than bone marrow can replace them.

“It’s a tough situation to be in. There’s really no other drug that’s as good at dealing with high uric acid as rasburicase is,” says Relling. “There’s an alternative, but it’s not optimal. We try to carefully word our [CPIC] recommendations to say that in situations like this, it’s a life-threatening adverse effect, but [tumor lysis syndrome] is a life-threatening condition. The clinician really needs to weigh that.”

Lack of safe and effective alternatives can stall recommendations on what to do for patients who carry relevant gene variants. “One of the criteria that we have in trying to decide whether to write a CPIC guideline is whether there’s evidence for alternative therapy. Sometimes the controversy in picking the alternative therapy is what slows down our ability to provide good guidelines,” says Relling.

The present and future of pharmacogenomics

Researchers have the data to predict how people might respond to certain drugs. But healthcare system infrastructure lags far behind science in its ability to put pharmacogenomics into practice. Relatively few hospitals use pharmacogenomic testing today. Some estimates say it is fewer than 10 percent.

Payers may be partially to blame for the slow uptake. “To actually get reimbursed for a genetic test for pharmacogenetics, there have to be very specific indications for doing that test. That’s a real challenge,” says Stuart Scott, an assistant professor of genetics and genomic sciences at Mount Sinai Hospital’s Icahn School of Medicine in New York City. Scott’s research includes clinical implementation of genetic testing and personalized medicine.

Time is also a barrier to implementation of pharmacogenomics. Most pharmacogenomic test results are only useful if the prescriber gets them immediately. Whether in a hospital or during an office visit, patients and doctors typically want to start treatment right away. “It doesn’t really help to get the results a week or two weeks later, when the patient is already gone,” says Scott. “It’s harder to change their therapy at that point. So you have to be at a medical institution that has the capacity to get results immediately.”

Or patients would need to have test results on file already. As the cost of genetic testing falls and the tests become more accessible, an increasing number of people may go for pre-emptive cheek swabs, as Cirrincione and his wife did. “Currently only a few places are doing that type of pre-emptive testing. You have to have an infrastructure and financial support to do it, because insurance will not cover pre-emptive genetic testing,” says Scott.

Scott believes that an informatics system that would make the genetic data useful both immediately and later is needed. Such an infrastructure requires an electronic medical record that healthcare providers across specialties and facilities could access, including pharmacists not connected to a hospital but dispensing potentially risky medications at local drug stores. The system would need to alert those prescribing and dispensing drugs about the potential risk.

“If you’re the person who can’t get any pain relief from that codeine that your dentist prescribes you, or if it’s your mother that’s crying in pain from osteoarthritis, you care about it a lot.”

And while one prescriber might order a test to see how a patient will respond, for example, to an antidepressant, the test result could have implications for drugs that other physicians may prescribe years later. For that reason, the information needs to remain in the electronic system for patients’ entire lives and follow them wherever they go.

“Without that system, how would a clinician that didn’t order a test know that the test result even existed?” Dunnenberger asks.

Clinicians are hopeful, though, that the healthcare system will soon be more conducive to preventive pharmacogenomic testing and even provide informatics and financial support for it. “I think a lot of people’s minds change in health institutions and insurance companies when you can prove that this pre-emptive genetic testing actually saves money in the long run by avoiding adverse events that are extremely expensive for the healthcare system,” says Scott.

And the tide may be turning already. The American Society for Health-System Pharmacists predicts that most U.S. regions will have a formal hospital pharmacogenomics program within the next five years.

“We have to be willing to do precision medicine, even if it has a relatively small incremental impact on the overall health of the population,” says Relling. “If you’re the person who can’t get any pain relief from that codeine that your dentist prescribes you, or if it’s your mother that’s crying in pain from osteoarthritis, you care about it a lot.”

  • Garrett A. Hughes

    Yikes! (on learning of this potential threat). Keep up the good work.