College of American Pathologists
Printable Version

  Facts catching up to
  fuss in PGx


cap today

June 2005
Cover Story

William Check, PhD

Pharmacogenetics testing, or PGx, has been a long time coming to its home in the clinical laboratory, but with work beginning now on a practice guideline for the lab, it may be closer than ever.

The origins of pharmacogenetics date back to 1952, with a report that primaquine, an important antimalarial, could induce hemolytic anemia in patients with glucose-6-phosphate dehydrogenase deficiency. This was the first demonstration that an inherited variation in an enzyme could affect drug activity. Then, in 1977, work on variations in the efficacy of the antihypertensive drug debrisoquine gave rise to the idea that “hydroxylation of debrisoquine is controlled by a single autosomal gene” (Mahgoub A, et al. Lancet. 1977;2:584–586). This hypothesis eventually led to the discovery in 1993 of the CYP2D6 gene, the first of 57 isoforms of the cytochrome P450 system, the primary mechanism for eliminating drugs from the body. Each CYP gene has many alleles that greatly affect drug metabolism, significantly altering therapeutic efficacy and raising the risk of adverse drug reactions. Genotyping CYPs could make drug therapy safer and more efficient.

Determining which polymorphisms an individual carries at one or more of the major CYP loci could be an important adjunct to clinical medicine. Yet, to the frustration of its advocates, the clinical promise of pharmacogenetics testing has always seemed to be just a few years away.

Perhaps that is finally about to change. At the International Association of Therapeutic Drug Monitoring and Clinical Toxicology meeting in April in Louisville, Ky., three half-day sessions were devoted to the presentation and discussion of preliminary recommendations for a laboratory medicine practice guideline for pharmacogenetics testing under the auspices of the National Academy of Clinical Biochemistry, the academy of the American Association for Clinical Chemistry. Recommendations will be posted on the NACB Web site for comment ( A draft guideline will be presented at the 2006 AACC meeting.

“Pharmacogenetics is rapidly evolving and laboratories need guidance,” said Roland Valdes Jr., PhD, chair of the guideline committee. Several driving forces make it timely to push for greater use of pharmacogenetics testing now, in the view of Dr. Valdes, professor and senior vice chairman of pathology and laboratory medicine and director, clinical chemistry and toxicology, University of Louisville Health Sciences Center. One of the most important is rapid advances in genetic technology, particularly as a result of the Human Genome Project, which make it possible to detect many polymorphisms quickly and accurately. “If handled correctly,” Dr. Valdes said at the April meeting, “it is believed that pharmacogenetics will add value to care as an adjunct to therapeutic drug monitoring.” At the same time, he cautioned, “We have to be careful that we don’t overextend our claims.” In addition, physicians are not yet comfortable using pharmacogenetic information, Dr. Valdes said, “which is a major drawback.”

“My reading is that the community is very interested in providing this service to help the many individuals who are facing possible less-than-optimum therapeutics because of lacking genetic information,” Dr. Valdes told CAP TODAY. Despite this enthusiasm, bringing pharmacogenetics into clinical medicine will not be easy. One of the main tasks will be to accumulate convincing evidence. While many retrospective studies show that CYP polymorphisms are statistically associated with dosing effects and adverse drug reactions in patients on such drugs as warfarin, post-transplant immunosuppressants, and psychiatric agents, there is a dearth of studies demonstrating that selecting a drug or dose based on prospectively obtained information about a patient’s CYP alleles improves clinical outcomes. Until such studies are available, pharmacogenetics testing is likely to be limited to helping clinicians understand unusual drug responses in patients already taking a drug.

While this is an important and helpful application, it doesn’t reach pharmacogenetics’ full potential. “In my opinion, the real benefit of this technology will be in predicting clinical response and ADRs [adverse drug reactions] of a particular therapy in a particular patient in the mode of screening every patient prior to initiating therapy,” Werner Steimer, MD, told CAP TODAY. “With regard to that application, it is still research.” Dr. Steimer, assistant medical director at the Institute for Clinical Chemistry and Pathobiochemistry, Munich University of Technology in Germany, would like to see more prospective clinical outcomes data before pharmacogenetics testing is widely adopted for pretherapeutic screening. Nonetheless, he says, “It is definitely necessary to formulate guidelines, even though it is kind of challenging at the moment because there are many questions and very little information. Still, we do need to get started.”

Assessing the current status of pharmacogenetics at the April meeting in Louisville, Leslie Shaw, PhD, said, “We don’t have the tools in hand now to predict the effect of genes on drug activity, but we have leads.” For such agents as warfarin and nortriptyline, where there is a strong correlation of CYP alleles with drug clearance, “Maybe we could use extreme genotypes to predict drug concentrations,” said Dr. Shaw. Extreme genotypes are those that make people poor metabolizers or ultrarapid metabolizers. “However,” said Dr. Shaw, director of the therapeutic drug monitoring and clinical toxicology laboratory and professor of pathology and laboratory medicine at the University of Pennsylvania Health System, “most people do not have extreme genotypes. If the goal is to get to a specific drug concentration, pharmacogenetics will not get us there right now.”

With regard to offering a pretherapeutic screening service to clinicians, Saeed Jortani, PhD, told CAP TODAY: “There is some education to be done at the grassroots level. We need to show physicians the value of pharmacogenetics testing for their patients. Hopefully outcome studies can provide concrete information.” Dr. Jortani, assistant professor of pathology and laboratory medicine and director of the diagnostic reference laboratory at the University of Louisville Health Sciences Center, acknowledges that some studies of this type do exist. “We should make every effort to make the testing available to the clinicians based on the currently available data. But we need many more to show the value of pharmacogenetic testing for overall patient care and cost,” he says.

Dr. Jortani is participating in a trial involving more than a dozen medical examiners that is headed by Steven Wong, PhD, department of pathology, Medical College of Wisconsin, Milwaukee, which is assessing the use of pharmacogenetics as a tool in molecular autopsy. In more than 1,300 cases of death by methadone overdose, the study is asking whether pharmacogenetics data could assist in determining cause of death. “Would pharmacogenetics have been useful in estimating the potential for drug toxicity and overdose?” Dr. Jortani asks. This is important because for methadone the therapeutic and lethal ranges overlap.

Others, however, think a requirement for clinical outcomes will unduly restrict the adoption of pharmacogenetics testing. “[Calling for more evidence] doesn’t excite anyone to use the technology,” said Gualberto Ruaño, MD, PhD. “It sends the message, OK, come back in 10 years.” Dr. Ruaño, president and chief executive officer of Genomas Inc., a newly formed commercial laboratory based in Hartford (Conn.) Hospital that will offer pharmacogenetics testing, advocates this principle: “Start with low-lying fruit.” Labeling on several drugs provides an opening for pharmacogenetics testing, he said, particularly psychiatric drugs. “I have found that psychiatrists like the concept of avoiding titration,” he said.

Looking to labeling could be a fruitful strategy, judging from comments made during the guideline sessions by N.A.M. Atiqur Rahman, PhD, of the Office of Clinical Pharmacology and Biopharmaceutics, Food and Drug Administration. “We [at FDA] have a strong commitment to pharmacogenetics,” Dr. Rahman said. He noted that the FDA is uniquely positioned to support its transition from the research laboratory to the bedside. “We believe if you build [pharmacogenetics testing], people will use it.”

Deborah Payne, PhD, former director of molecular diagnostics at the University of Texas Medical Branch, Galveston, says pharmacogenetics must be implemented as soon as possible. “Its main purpose is to save lives and prevent ADRs. I believe the data are definitely there,” she told CAP TODAY.

Experience with pharmacogenetics in clinical contexts varies. Dr. Shaw directs a laboratory residency training program in clinical toxicology into which he has incorporated study data on pharmacogenetics testing, showing its potential connection with therapeutic drug monitoring. For two drugs, one used in a pre-transplant conditioning regimen and one used in post-transplant regimens, bisulfan and mycophenolic acid respectively, he provides pharmacokinetics-based interpretive reports. He is working with collaborators to incorporate pharmacogenetics testing in clinical studies involving calcineurin inhibitors, sirolimus, and mycophenolic acid in renal transplant patients in the next year.

Tacrolimus, cyclosporine, and sirolimus are primarily metabolized and cleared by the CYP3A4 and CYP3A5 enzyme isoforms. So far about 20 single nucleotide polymorphisms, or SNPs, have been identified for the CYP3A4 gene and five SNPs within the CYP3A4 promotor region. But the functional significance of these is unclear for many of these SNPs, and in most studies to date there has been little correlation of these genotypes with alterations in immunosuppressant drug clearance.

However, for tacrolimus, a significant correlation between expression of CYP3A5, reflected by the presence of the wild-type allele CYP3A5*1, and significantly increased clearance of the drug has been observed in several retrospective-design studies. Dr. Shaw believes this finding is potentially clinically important. An important clinical problem is the difficulty of getting to sufficiently high doses and steady-state levels of tacrolimus in a subset of renal and lung transplant patients in the early post-transplant period due to apparently very high clearances in these individuals. “If the rapid clearer could be identified prior to transplant and the tacrolimus regimen altered to a higher-than-normal average dose at the beginning of immunosuppressant therapy, therapeutic target concentrations could be reached earlier,” Dr. Shaw explains. This possibility is an example of the type of problem that needs to be studied prospectively to determine the clinical utility of CYP3A5 genotyping. “A real issue for the existing study data is the wide scatter in the dose-adjusted data for each genotype, so what is sorely needed is to test whether or not genotyping will add to our ability to safely titrate each patient within the first week following transplant surgery to the therapeutic target concentration,” he says.

Improved dosing could be valuable, Dr. Shaw told CAP TODAY. “Early acute rejection sets up the organ for long-term graft failure,” he says, “so we want to reach the target levels and dose for anti-rejection therapy early. One approach is to be aggressive and follow levels each day, but we don’t want to overdose because these drugs can be toxic.” Physicians would like to know which patients they can safely dose at a high level. “That would be a great example to prospectively evaluate the benefit of measuring CYP3A5,” Dr. Shaw says.

Kristen Reynolds, PhD, is director of laboratory operations in the Pharmacogenetics Diagnostic Laboratory at the University of Louisville Health Sciences Center ( Like Dr. Shaw, they do therapeutic drug monitoring for mycophenolic acid. The university hospital does international normalized ratios for the hospital’s large coagulation clinic.

“How pharmacogenetics can add to that is the next logical question,” Dr. Reynolds said in an interview with CAP TODAY. “We find that it is really beneficial to focus on certain groups. For instance, we have met with pharmacists and our hem/onc group to let them know that we can genotype CYP2C9 a priori to detect if a patient is more likely to have a hypersensitivity reaction to warfarin.” Physicians have also contacted them for posttherapeutic or explanatory genotyping of CYP2C9 for patients with a high INR whom the clinician can’t stabilize on warfarin. “In several instances, those patients have had genetic deficiencies in CYP2C9,” Dr. Reynolds says.

Many requests for genotyping come from patients who have found the laboratory through its Web site, Dr. Reynolds says. Recently, “a very astute patient,” who was about to have hip replacement surgery, contacted her. This woman had previously had trouble with medications and requested genotyping. She was found to be heterozygotic for a deficient CYP2C9 allele, making her an intermediate metabolizer. Unfortunately, despite the pharmacogenetics report, a physician put the woman on a standard dose of warfarin. Four days later she was in the emergency department for a high INR. She had symptoms but did not have a major bleed. “We still need to do a substantial amount of education,” Dr. Reynolds says.

With regard to pharmacogenetics, “warfarin is one of the most-studied drugs,” according to Dr. Reynolds. “It is one for which we will have specific dosing instructions at some time. However, lots of large studies still need to be done to establish dosing changes.”

Most calls to the laboratory come from patients who are taking antidepressant medication. Dr. Reynolds finds that patients who have had trouble tolerating antidepressants “are desperate to get help any way they can.” One woman was about to go on a new antidepressant that was metabolized by CYP2D6. “Based on her history, she had had trouble with other 2D6 drugs,” Dr. Reynolds says. Genotyping showed she had a slow-metabolizer allele and was at increased risk of adverse reactions. How that finding should affect dosing is not yet established. “There is less information about specific dosing changes with antidepressants [than with warfarin],” Dr. Reynolds says. “That class has even further to go.”

Dr. Reynolds did not hear back from this patient, which is not uncommon. “We have had several consults where we have spent significant time with the patient understanding their past history and giving genotyping results,” she says. “Now we need to have a mechanism for following those outcomes.”

As one might infer from Dr. Reynolds’ experiences, much of the demand for pharmacogenetics testing is consumer driven, and patients who order it pay out of their own funds. The University of Louisville’s Pharmacogenetics Diagnostic Laboratory charges $250 for CYP2D6 and $225 for CYP2C9.

Dr. Reynolds believes that pharmacogenetics testing is going to continue in this case-by-case way for a while. ”If we would do a screen, how would we decide which enzymes to screen for?” she asks. “Would we do the whole panel? That would dilute out the message. What would you screen for if not based on a specific individual getting ready to go on a specific medication or a patient with a specific past history with a particular class of drugs? There will still need to be an individualized aspect to it.”

Dr. Steimer has done research into the effects of CYP allelic variation on drug activity and adverse drug reactions. He did one of the few prospective studies showing that a CYP genotype predicts clinical outcomes. In this case, CYP2D6 allelic composition predicted the rate of adverse reactions among patients starting amitriptyline therapy (Steimer W, et al. Clinchem. 2005;5:376–385). The difference in the incidence of adverse reactions between those having two functional alleles versus those having one was substantial—12.1 percent versus 76.5 percent. Dr. Steimer is starting a larger study to verify these findings. “I have no problem interesting my psychiatric colleagues in such a study,” he says. One reason is that in Germany and many other European countries tricyclic antidepressants remain one of the mainstays of therapy, Dr. Steimer says, especially in centers where difficult-to-treat patients are managed. Since tricyclic antidepressants are at least as effective as newer antidepressants, reducing adverse reactions would make these older drugs equivalent to newer drugs often at a fraction of the cost.

For how many drugs is this kind of information available? “I’m afraid for very few,” Dr. Steimer says. “We published one of the first prospective studies reporting a significant correlation between CYP2D6 genotype and adverse drug reactions. There is very little evidence regarding clinical outcome, to say nothing of cost efficiency.” One of the reasons is that such studies are difficult to do. To see larger effects, most studies focus on ultrarapid metabolizers and poor metabolizers. Because these extreme genotypes make up only six to 10 percent of the population, Dr. Steimer says, “you have to test many people to get one who might respond. So you need studies with 1,500 to 2,000 patients, which are extremely difficult to perform and finance, in particular if you look at old cheap generic drugs. So we must show these measures can be cost-efficient for health care payers and providers.”

At the guideline sessions of the Louisville meeting, Dr. Steimer was charged with making recommendations about clinical laboratory services. Some of his conclusions and preliminary recommendations were as follows:

  • There is no established clinical situation for pretherapeutic CYP2D6 screening.
  • At present, there is little economic evidence to support pharmacogenetics screening.
  • It is mandatory to demonstrate not only functional effects but associations between genetic variation and clinical outcome.
  • Proof of cost-effectiveness will probably be necessary to drive the implementation of pharmacogenetics into widespread clinical practice.

Considerable research on clinical pharmacogenetics, particularly with psychiatric drugs, is being carried out by another German scientist, Julia Kirchheiner, MD, clinical pharmacologist and senior lecturer in the Institute of Pharmacology, University of Cologne. Much of this work is summarized in two review articles (Kirchheiner J, et al. Molecular Psychiatry. 2004;9:442–473; Kirchheiner J, Brockmöller J. Clin Pharmacol Ther. 2005;77:1–16).

Dr. Kirchheiner’s work illustrates the fallacy of extrapolating from associations between CYP alleles and drug concentrations to clinical endpoints. For instance, in one study she and her coworkers found “a linear relationship between the number of active CYP2D6 genes and metabolic clearance of metoprolol,” with a tenfold difference between poor metabolizers and ultrarapid metabolizers. Yet, they reported, “Metoprolol pharmacodynamics differed only by less than twofold, and there was only a marginal difference in metoprolol efficacy on heart rate between the [extensive metabolizer] and UM [ultra rapid metabolizer] groups” (Kirchheiner J, et al. Clin Pharmacol Ther. 2004;76:302–312). Another study found that “[H]igh CYP2D6 activity may only explain a very small fraction of the cases with therapeutic failure in treatment with [the antidepressant] mirtazapine” (Kirchheiner J, et al. J Clin Psychopharmacol. 2004;24:647–652).

Based on finding strong associations between CYP2D6 and CYP2C19 alleles and drug response or adverse reactions for psychiatric drugs, Dr. Kirchheiner and her colleagues have embarked on a study of dose adjustment for CYP genotype. In an e-mail communication, Dr. Kirchheiner said, “We developed these adjustments for dosage according to genetically caused variability in drug metabolism. However, these adjustments are not yet ready for clinical practice. Before they are adopted, a validation of the cost-benefit ratio and a prospective study has to be performed.” Thus, she said, it would be incorrect to imply that they are using these dose recommendations clinically.

Dr. Ruaño, applying his suggestion to “look into drug labels and find justification for why pharmacogenetic profiling makes sense,” specified three pharmaceuticals. First is the proton pump inhibitor omeprazole (Prilosec), used to treat gastroesophageal reflux disease. Subtherapeutic levels of omeprazole in ultrarapid metabolizers increase the risk of relapse. Thus the comment in the package insert: “Dose adjustment for healing erosive esophagitis should be considered.”

Dr. Ruaño called warfarin “by far the poster child for why we should be doing pharmacogenetic testing.” The package insert suggests lowering the dose for poor metabolizers, which Dr. Ruaño interpreted as a call for pharmacogenetics testing. “We will keep doing INRs,” he told CAP TODAY. “There is no expectation that CYP testing will make INR obsolete. All we are going to say is that we can enhance the level of individualization even further by providing knowledge about whether a patient can metabolize this drug.”

Third is atomoxetine (Strattera), a selective norepinephrine reuptake inhibitor used to treat attention deficit hyperactivity disorder. For this drug the package insert says, “Laboratory tests are available to identify CYP2D6 poor metabolizers.”

Dr. Ruaño is planning to launch his laboratory service this fall. Initially it will offer genotyping for CYPs 2D6, 2C9, and 2C19. “The mutations we are going to be reporting are very clear-cut,” he says. “For instance, a deletion in 2D6 that implies no functional gene. We will report that some polymorphisms are frameshifts that make a dysfunctional sequence and some are splicing defects in which parts of the protein are deleted. We will not include any clinical literature in the report, only molecular biological and biochemical literature about the effects of CYP mutations—that will be the extent of it.”

Even with pharmacogenetics testing, Dr. Ruaño told CAP TODAY, “It doesn’t mean we are going to avoid drug-drug interactions. We will not try to modify dosages and prevent problems—frankly we do not have the data for that right now. But we can flag the problem and alert the clinician.” And the treating physicians can decide how to use the information. For now, Dr. Ruaño says, much of pharmacogenetics will be done in this observational way. “It’s not the job of the laboratory to tell clinicians what to do, but we can provide them with new tools.”

Like all molecular testing, pharmacogenetics poses challenges for quality control, and it was Dr. Payne’s task to address this for the guidelines. She proposed adopting the recommendations in CAP’s molecular pathology checklist as well as recommendations for HIV sequencing. “Most of the QC/QA issues regarding pharmacogenetics have already been addressed and are in active practice in other areas of molecular diagnostics,” Dr. Payne told CAP TODAY. “Pharmacogenetics essentially mirrors the same issues that molecular pathology has been dealing with for years now.” For instance, Dr. Payne suggested adopting the practice from HIV resistance testing of redundancy in covering a particular variant.

She raised the question of whether pharmacogenetics data should be in the electronic medical record, like allergies. Critical variants might be flagged. “Should genetic variants be treated as critical values, where the laboratory calls the clinician?” she asked.

With regard to proficiency testing, Dr. Payne says, “People will have to work with colleagues to do that. Lack of QC materials is something that everyone in diagnostic molecular pathology is working on across the board.”

Dr. Jortani is AACC’s liaison to the CAP’s Therapeutic Drug Monitoring/Endocrinology Resource Committee, which he says has discussed providing proficiency testing materials for pharmacogenetics, perhaps in collaboration with other interested groups. “There will be challenges,” he notes, “such as whether to provide cell lines or current cases and to maintain a constant supply.”

Another issue that arises frequently in molecular pathology is what to include in a report. The Montreal-based company Seryx (makers of the Signature Genetics Interpretive Report, ( was created to provide a pharmacogenetics reporting service from the laboratory to the clinician. In essence, they have developed tools to analyze an individual’s CYP genotype results against his or her medical history and current drug regimen so as to provide a personalized interpretive report for the physician, said Jean-Pierre Morello, PhD, senior scientific research and development executive at Seryx. “Interpretation is the key, as a poor metabolizer genotype may variably impact clinical outcome depending on the drug, medical history, and co-medication,” Dr. Morello says. For example, a poor metabolizer genotype may be associated with decreased drug clearance ranging from as low as 17 percent for some drugs to close to 80 percent for others. This variation is seen even with drugs of the same class. Therefore, Dr. Morello says, “an across-the-board approach for all poor metabolizers to reduce the prescribed drug dose by half, for example, is overly simplistic and often misleading.”

There is also the issue of co-medication to consider. The metabolism of the anti-tumor agent tamoxifen to its more active metabolite is mediated in part by CYP2D6. Many antidepressants are not only metabolized by CYP2D6, but also can inactivate or inhibit this enzyme. Therefore, patients taking tamoxifen and co-medicated with an antidepressant that inhibits CYP2D6 may be at risk of inadvertently losing the benefit of chemotherapy.

“We spent the last four years developing information technology and a pharmacogenetics database correlating genotype with drug response and patient characteristics,” Dr. Morello said. The company’s report is based on this database. For each gene, the report gives the patient’s genotype with its classification (poor, extensive, or ultrarapid metabolizer) and its effect on enzymatic activity (“severely decreased,” “greatly increased”). It also makes therapy recommendations for that specific patient’s drug regimen based on the scientific literature with alternative drugs or dosing information or both when included in the literature. These tailored regimen considerations provide direction for the physician facing “therapy-resistant” patients, Dr. Morello says.

Signature Genetics will launch its service later this summer and charge $275 for a one-gene report and $750 for a five-gene, all-inclusive report.

Roche’s AmpliChip is the only device for detecting CYP variants that has been cleared by the FDA for in vitro diagnostic use ( It detects 29 polymorphisms and mutations for the 2D6 gene and two polymorphisms for the 2C19 gene. TM Bioscience of Toronto has separate kits undergoing laboratory evaluation now for detecting variants in the CYP 2D6, 2C9, and 2C19 genes. All detected SNPs for each gene in each kit are determined in one tube using the double-laser Luminex detection system. The company’s goal is FDA submissions for all of the assays within 12 months, says Stephen Weiss, PhD, VP of market development.

A review article on pharmacogenetics was published in 2003 in the New England Journal of Medicine (Weinshilboum R. N Engl JMed. 2003;348:529–553). In an accompanying editorial, David Goldstein, MD, of University College, London, addressed the tension between continuing to gather clinical data and introducing pharmacogenetics into clinical practice. “Although the prospects for basic research in pharmacogenetics look very promising,” Dr. Goldstein wrote, “the incorporation into clinical practice of the data it generates presents considerable challenges.” He noted, “Basic research in pharmacogenetics deserves the support and the excitement it has generated,” but he cautioned, “[T]his excitement should not lead to unrealistic expectations about the rate at which medicines can be personalized according to genotype.”

For those laboratorians who choose to be the first to offer the limited form of clinical pharmacogenetics service that the data now allow, the guidelines under development will be a useful and welcome form of support.

William Check is a medical writer in Wilmette Ill.