Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
6 Screening Technologies IV: Pharmacogenetics T he screening technologies discussed in the previous three chapters are used mainly to address the issue of whether a particular com- pound is toxic and if so, why. Dr. Lai raised a different issue: Given a useful drug that is toxic in only a subset of patients, how can those patients be identified so the toxicity can be prevented or at least antici- pated? Lai described how pharmacogenetics provided an answer to that question in the case of the anti-AIDS drug abacavir. Abacavir and the hypersensitivity reaction Abacavir is a reverse transcriptase inhibitor used against HIV. It is the sole ingredient in Ziagen, an anti-AIDS drug marketed by GlaxoÂSmith Kline (GSK), and it is also used in combination drugs such as  Trizivir, which contains abacavir, zidovudine, and lamivudine. Abacavir is a highly effective medication and is well tolerated in most patients, but a small percentage of people who take it experience hypersensitivity reaction (HSR). HSR is a multiorgan syndrome whose most common symptoms are fever, rash, nausea and vomiting, and malaise or fatigue. The overall rate of HSR among abacavir users is about 5 percent, and most of these are nonserious episodes that are resolved by discontinuing use of the drug. However, the discontinuation must be permanent. If a This chapter is based on the presentation of Eric Lai, Vice President, PGx Experimental Project Coordination and Analysis, GlaxoSmithKline. 59
60 EMERGING SAFETY SCIENCE patient who has once experienced HSR starts taking abacavir again, the HSR returns very quicklyâin a matter of hours to a day or soâand this time it is lethal. The HSR phenotype is complex. About 78 percent of HSR patients have fever, about 65 percent have rash, and about 96 percent exhibit fever or rash or both. There is a long list of other symptoms that appear in at least 10 percent of HSR cases: nausea/vomiting, malaise/fatigue, muscle or joint pain, headache, diarrhea, itching, abdominal pain, dyspnea, and cough. Most patients have three or more of these symptoms in varying combinations. The time of onset is also variable. A number of patients experience HSR within the first week of taking abacavir, sometimes within 24 hours, but for others it takes longer, and the median time to onset is about 11 days. About 93 percent of reported cases occur within 6 weeks of start- ing abacavir, so one of the exclusion criteria in the GSK studies is that a patient must experience HSR within the first 6 weeks. In 1999, at the time of abacavirâs approval, a two-part postmarket risk management program was established. The first part was aimed at educating health care providers; this included updating labeling infor- mation on a regular basis and monitoring the occurrence of HSR among abacavir users. Monitoring data have revealed that although the number of people taking abacavir has increased steadily over the past 8 yearsâto more than 1 million in 2006âthe number of deaths caused by the drugs has remained relatively stable since 2002 (see Figure 6-1). Thus one can infer that physicians now know that once any kind of HSR-related symp- toms appear in a patient taking abacavir, the patient must be taken off the drug and never given it again. Despite physician awareness, however, the rate of spontaneous HSR has not decreased, as it is not possible to predict whether a patient will exhibit HSR until abacavir is taken. The second part of the postmarket risk management program included a pharmacogenetics study designed to look for genetic factors associated with abacavir-related HSR. the Abacavir pharmacogenetics program The goal of GSKâs pharmacogenetics program was to identify genetic markers that could predict patients at risk of developing HSR from aba- cavir and prevent them from taking the drug, thereby improving its benefit-risk balance. The study would involve gathering patients who had developed HSR; matching them with patients who had not; and then performing association studies, first with candidate genes and then laterâas it became possibleâwith whole-genome analysis.
SCREENING TECHNOLOGIES IV: PHARMACOGENETICS 61 Mortality Patient/years (x103) 1200 1000 800 600 400 200 0 1999 2000 2001 2002 2003 2004 2005 2006 FIGURE 6-1â Cumulative patient/years of exposure to abacavir products and spontaneous reports of HSR-associated mortality among those taking abacavir. These data show that while abacavir use greatly increased, HSR-related mortality remained low. SOURCE: Lai, 2007. Figure 6 -1 The Identification of HLA-B57 Initial calculations, derived from such assumptions as the allele fre- quency of the causative locus and the effect of the locusâs being variably penetrant, implied that it would take 500 case-control pairs to power the study adequately. Since the GSK trial included only 44 cases and 78 controls, the researchers did not expect to identify any candidates. In July 2001, however, they discovered that one of the candidate genes, HLA-B, was playing a major role in abacavir-related HSR. Of the 44 cases in the study, 25 (57Â percent) had the HLA-B57 variant, while of the 78 controls, only 3 (4 percent) had that same variant. After receiving confirmation of the results of the assays from two other laboratories, the GSK researchers were confident that they had identified a predictive biomarker for HSR. To follow up on this conclusion, the group continued to accumulate data; they currently have data from 444 cases and 486 controls. They have further zeroed in on the markerâthe HLA-B*5701 subtype of HLA- B57âand in their studies, this marker predicted HSR with a sensitivity of 50 percent and a specificity of 98 percent.
62 EMERGING SAFETY SCIENCE These results have been confirmed by a group of researchers headed by Simon Mallal at the Royal Perth Hospital in Perth, Australia, who performed a study with 18 cases and 230 controls and obtained the same results, except with a much higher sensitivity. In 14 of the 18 cases, they found the HLA-B*5701 allele, yielding a sensitivity of 94 percent. Lai hypothesized that this increased sensitivity was related to the fact that in the Australian study, one physician saw all of the patients. Therefore, the inclusion criteria were based on this one physicianâs diagnoses, and it was possible to follow up with the patients to determine whether they did indeed have HSR. In the GSK study, by contrast, the cases were scattered over several dozen centers, and the only source of information was the case report forms. Thus it was impossible to go back and ask the patients whether they had taken abacavir or whether, for instance, they had ever had a fever or some other symptom. A second difference was that the GSK study included a number of ethnic groups, and the results differed among groups. The sensitivity among whites, for example, was 50Â percent, while that among Hispan- ics was only 22 percent, and there was no significant association among blacks. The problem may lie in the fact that the frequency of HLA-B*5701 varies greatly among ethnic groups, and the rate of HSR in ethnic groups varies as well. In blacks, for example, the rate is about one-half or one- third the frequency in whites. GSK is now exploring the issue of abacavir- related HSR in different ethnic groups. Applying the Biomarker The HLA-B*5701 biomarker can be used to stratify patients into groups at high and low risk of HSR. Indeed, since the biomarker was published, a number of academic groups have been screening patients for HLA-B*5701 before treating them with abacavir. At the Royal Perth Hospital in 2000â2001, before screening for the biomarker was performed, 11 of 131 patients on abacavir developed HSR. After screening became a routine practice, howeverâfrom the beginning of 2004 through July 2005âonly 1 of 49 patients exhibited HRS. That patient was known to be positive for the biomarker but chose to try abacavir anyway because there were no other options. More recently, a French study involving 137 patients found a decrease in the HSR rate from 12 percent to zero after screening for the HLA-B*5701 biomarker was implemented. Lai explained that although prospective screening with the HLA-B*5701 marker shows great promise, academic groups that have been testing its use have been conducting small stud- ies involving 50â100 cases. Therefore, it is still necessary to validate the
SCREENING TECHNOLOGIES IV: PHARMACOGENETICS 63 Abacavir-containing regimen HSR monitoring according to Standard of Care (~900) * Abacavir naive subjects Randomize (1:1) Exclude subjects (~1800) with HLA-B*5701 Abacavir-containing regimen HLA- HLA-B*5701 PGx Screening (~900) Enroll subjects without HLA-B*5701 * FIGURE 6-2â PREDICT-1 study design. The objectives of the trial are to compare HSR rates (± abacavir skin patch testing) in the two study arms marked with an asterisk (������������������������������������������������������������������������ *), and e��������������������������������������������������������������� valuate the sensitivity of HLA-B*5701 in cases in the standard- of-care arm. NOTE: PREDICT = Prospective Randomized Evaluation of DNA Screening in a Figure 6 -2 Clinical Trial; PGx = pharmacogenomic. SOURCE: Lai, 2007. screening in an adequately powered prospective clinical trial. In 2006, GSK initiated two clinical trialsâPREDICT-1 and SHAPE. PREDICT-1 (see Figure 6-2) is a highly powered prospective study that will examine the utility of HLA-B*5701 screening in a European HIV population. It will enroll 1,800 patients who have never been treated with abacavir. These patients will be randomly assigned to one of two groups, each with about 900 patients; one of the groups will be screened for HLA- B*5701, and the other will not. In the group that is screened, patients with HLA-B*5701 will be excluded from taking abacavir. By comparing the rates of HSR in the two randomized groups, GSK researchers will be able to measure the power of HLA-B*5701 screening to reduce the occurrence of HSR compared with the usual standard of care. By contrast, SHAPE is a retrospective, matched-group, case-control study intended to estimate the sensitivity of HLA-B*5701 in both black and white patients, but using skin-patch testing to supplement the clinical diagnosis of HSR. Skin-patch testing provides a much better clinical indi- Note that following the workshop, the PREDICT-1 study results were published online in Pharmaceutical Statistics on May 29, 2007, and results from the SHAPE study were presented at the 4th International AIDS Society Conference on July 22â25, 2007, in Sydney, Australia.
64 EMERGING SAFETY SCIENCE cation of HSR than does standard diagnostics, so the study will provide a clearer measure of the usefulness of the biomarker in black and white populations. Implications for the future Using this type of pharmacogenetics analysis to identify safety bio- markers prior to the approval of new drugs will demand the development of methodologies for prospectively managing drug-associated adverse events. One group at GSK, run by Clive Bowman, has begun developing a method for the real-time management of patientsâ adverse events. This method includes ⢠creating a collection of genetic markersâa thousand or moreâto examine in people who present with an adverse drug event; ⢠creating a control set of genetic markers by genotyping people who have taken the drug without adverse effects; and ⢠as patients report with HSR or some other adverse reaction, geno- typing them and comparing their genetic markers with those of the con- trol group. Calculations show that by the time 18â19 patients have reported with a particular adverse drug event, it should be possible to tell whether there is a genetic basis for the event and to identify potential markers in the genome. To test this methodology, GSK researchers designed a real-time retrospective whole genome scan study with abacavir data on 22 cases and 316 controls and worked with the data as though the cases were coming in prospectively one at a time. By the time they had 22 cases, they could identify 10 loci that correlated with HSR, and the fifth of those was the HLA-B locus. The implication is that by the time the 22nd case comes in, one will have identified that there is a problem, and one will have a number of loci that are potentially associated with a marker for HSR. Con- tinuation of the simulation for the next 100 cases that presented allowed the researchers to eliminate the nine loci other than HLA-B as false posi- tives and identify a clear marker for hypersensitivityâHLA-B*5701. Lai emphasized that the important difference between the simulation and how the marker was actually discovered is that using the simulation, it was possible to pinpoint the marker much sooner and potentially save hundreds of lives.