HIV drug resistance testing: update

by Harvey S. Bartnof, MD, for HIV&

The International AIDS Society-USA (IAS-USA) Panel of 13 expert physicians has updated their guidelines for HIV drug resistance testing. The 1998 and new guidelines were published in the Journal of the American Medical Association.

There are two clinical situations for which testing is recommended: those with drug treatment ‘failure’ and for pregnant women. The two clinical situations for which testing should be considered are acute (primary) HIV infection and for treatment-na•ve patients. The Panel also recommended expert interpretation of testing results, due to the complexity of the topic, evolving nature of the topic and limitations of the tests. The Panel’s recommendations are supported by 100% of its members. They caution that ‘definitive data to support final recommendations are not yet available.’

Since the Panel’s first set of recommendations two years ago, there have been many reports about the utility of these tests. Even though they are not yet approved by the FDA, these tests are available and are being used with increasing frequency.

Two Types of Drug Resistance Tests

HIV drug resistance tests fall into two separate types: genotype and phenotype. A genotype drug resistance test evaluates a patient’s HIV isolate (‘strain’) for the presence of gene mutations that have been previously associated with resistance to anti-HIV drugs. PCR (polymerase chain reaction) testing is used to amplify specific HIV genes from a sample of blood plasma (cells removed) from the patient. The minimum number of HIV RNA copies is 500-1,000 per millilitre of plasma. Mutations must be present in 10-50% of the HIV RNA in order to be detected. This represents a limitation of HIV drug resistance testing. If a mutation is present in less than 10% of the HIV RNA, then that RNA sequence might not be amplified. Thus, that mutation might be missed and not included in the test results.

The Panel listed different companies that currently are developing HIV genotype drug resistance tests. (Some not included by the Panel are also listed).

  • TruGene HIV-1 from Visible Genetics,
  • ABI (Advanced Biotechnologies Inc/Perkin Elmer; PE Biosystems),
  • GeneChip HIV PRT from Affymetrix,
  • INNO-LiPA (Line Probe Assay) from Innogenetics,
  • HIV-1 GentypR from Specialty Labs,
  • Chiron,
  • VircoGEN from Virco (processed in US by LabCorp).

There are two basic methods for genotype testing. The first one tests for all of the relevant gene locations (codons) in the protease and RT genes. The other uses a ‘hybridisation’ (‘link-up’ or binding) technique to detect mutations at specific codon locations. Based upon the technology used, ideally results of these tests should be available in ‘several hours to days.’ At this point in development, ‘days’ is a much more realistic endpoint.

The Panel cautions that ‘different guidelines for interpretation of results’ exist due to the ‘complexity of data generated from [gene] sequencing.’ There are ‘varying interpretations’ about the level of resistance that will occur from a particular gene mutation. The Panel adds, ‘as new data are generated, there is a risk of providing inadequate or even incorrect interpretations.’

Another concern of the Panel is of the reproducibility of the genotype tests, particularly when a sample with mutations is evaluated by different laboratories that were mostly university-based. More recent studies indicate that this may be less of a problem when experienced laboratories are used. As technicians become more experienced with using the tests, their reproducibility will continue to improve. Also, each commercial test kit will have a greater degree of standardization than university-based, ‘home-brew’ tests.

The other type of resistance test is the phenotype drug resistance test. It measures the ability of the patient’s isolate to grow in the presence of different concentrations of each anti-HIV drug. Specifically, the protease and RT (reverse transcriptase) enzymes are removed from the patient’s HIV isolate and inserted into a laboratory isolate. The result is reported as a ‘-fold change’ in the concentration of drug that is required to inhibit 50% or 90% of HIV growth, when compared to a standard, ‘wild-type’ HIV isolate without mutations. The result can also be stated as an ‘IC50’ or ‘IC90’ meaning the Inhibitory Concentration that would block 50% or 90% of HIV growth. Instead of comparing the patient’s HIV isolate to a reference ‘wild-type’ isolate, the comparison could be to a prior tested isolate from that patient.

The two companies that are performing phenotype resistance tests are:

  • PhenoSense HIV from ViroLogic; and
  • Antivirogram from Virco

Turnaround time is approximately 2-3 weeks. The PhenoSenseHIV result indicates that the patient’s HIV isolate is ‘less susceptible’ to the specified drug if the increase in the IC50 is more than 2.5-fold that of the reference, wild-type HIV. For the Antivirogram, the cut-off level for decreased susceptibility is an IC50 that is more than 4-fold that of the reference HIV isolate.

The Panel noted a limitation of the phenotype tests that is also inherent with the genotype tests. If an HIV ‘minority species’ with mutations is less than 10-50% of the sample, resistance associated with those isolates might not be detected. Therefore, the report of the results might indicate that a certain level of resistance to a drug(s) was not present. This might be considered a ‘false negative’ test result; however, that false result is due to a biologic issue and not necessarily a problem with the test. If that particular drug was used in a combination regimen, the minority isolate with resistance could then become a dominant species, potentially leading to drug ‘failure.’

The other limitation of both types of tests that the Panel discusses is ‘unresolved technical issues.’ Specifically, the members were referring to adequate standardization and clinical validation. The standardization issue relates to the tests’ accuracy and reliability of results, between different test kits and the same test kit performed at different times by the same and different labs. The clinical validation also is an important issue. Essentially, the results of the tests must be meaningful in terms of what might or might not happen in the patient, i.e., when the results are incorporated into devising a treatment regimen and the new regimen is taken by the patient.

Other technical issues the Panel discusses included lab recommendations about collecting, storing and shipping of blood samples. In the laboratories performing the testing, ‘quality control guidelines’ are needed. This would include proper handling of the specimen after delivery, quality control testing at different steps during the testing process and including several known ‘reference’ HIV isolates, some with and others without drug resistance, in each ‘batch’ of tests.

The last technical issue the Panel addressed was that of different HIV ‘clades,’ sometimes called subtypes. When the predominant clade in North America and Europe is tested (clade B), the results of resistance tests may be more accurate than when other clades are tested. Recent studies have suggested that this has become less of an issue for clades A through H (all are within ‘group M’). However, HIV isolates that are not group M have been associated with inaccurate results with resistance testing. This would include those who are infected with HIV-2, a more common isolate in Western Africa.

Evidence for the Benefits of Resistance Tests

The Panel reviewed the studies that reveal the utility of these tests. The two types of studies include prospective and retrospective types. A prospective study outlines a hypothesis and study design before it occurs. Then the study takes place and information is analysed afterwards (sometimes parts are analysed during the study). Conversely, a retrospective study starts the analysis about a hypothesis or question after all of the events have occurred that will be used in that analysis, using information that was recorded in the patients’ charts or medical records.

Prospective Studies

The Panel reviewed three prospective studies that revealed benefits of HIV drug resistance testing. Two were genotype resistance test studies and one was a phenotype resistance study. The first genotype resistance test study was called VIRADAPT. In this European study, 108 patients were enrolled who had at least 10,000 copies per millilitre of HIV RNA (drug ‘failure’) with a protease inhibitor (PI) drug regimen for at least three months. Patients were randomised into two arms and a second-line regimen was devised by each patient’s physician. In the first arm, the patients’ physicians had access to genotype drug resistance tests before determining the regimen. In the other arm, the physicians did not have those results and decided the regimen as ‘standard-of-care,’ using the patient’s drug history and other factors. After six months, a significantly greater percentage of patients in the genotype arm had an undetectable viral load (32%, limit 200 copies per millilitre) than the other arm that did not have the benefit of genotype results (14%). This was associated with a greater decrease in the viral load among those with genotype results (-1.2 log copies per millilitre) than the other arm (-0.7 log copies per millilitre). After six months, the genotype resistance results were revealed to the physicians of patients with a detectable viral load. Six months after that, 92 of 108 patients were still available for follow-up. Those results revealed similar results among the two groups. Both arms had a 30% rate of an undetectable viral load. The original genotype arm had an identical viral load decrease at the 12-month time-point as it did at six months (-1.2 log copies per millilitre), while the other arm had a reduction of -0.98 log copies per millilitre after 12 months.

The second prospective, randomised genotype resistance study was called GART (Genotypic Antiviral Resistance Testing) and was performed in the US (CPCRA study 046). A total of 153 patients were enrolled with a median baseline viral load of approximately 4.5 log (27,000 copies) per millilitre. The main criterion for enrolment was at least a 3-fold increase in the HIV viral load on PI-drug, triple combination regimen for at least four months. The entire group was randomised to have genotype drug resistance testing or standard-of-care to determine the next drug regimen. The genotype arm also had an expert physician recommendation based upon the resistance test results. After 12 weeks, the genotype arm achieved a lower viral load (-1.2 log copies per millilitre) than the standard-of-care arm (-0.5 log copies per millilitre). While those results were not statistically significant, the differences were significant after four and eight weeks. The difference was attributed to a greater number of active drugs prescribed in the genotype test arm and expert advice. However, the level of virus undetectability in the genotype arm was 50% at eight weeks and only 33% at 12 weeks. Moreover, it was determined that only half of the treating physicians of patients in the genotype test arm followed the expert advice.

Another more recently reported prospective, randomised study using genotype testing was not included in the Panel’s recommendations. Andrea De Luca, MD, from Catholic University in Rome, Italy has reported some benefits of baseline genotype testing after three months in 168 patients, but adherence was measured to be a major co-factor. Undetectability (less than 500 copies per millilitre) was significantly associated with the number of active drugs at baseline (‘odds ratio’ of 1.69 for each drug). However, the percentage with an undetectable viral load after three months was not significantly different when comparing the genotype arm (22%) to the standard-of-care arm (19%). Yet, genotyped, adherent patients with a baseline viral load less than 4 log (10,000) copies per millilitre and who had ‘failed’ two previous regimens had a significantly greater viral load reduction (-1.5 log copies per millilitre) than those who had those characteristics, but without genotyping (-0.2 log copies per millilitre).

The Panel reported one prospective randomised study with 274 patients that used phenotypic drug resistance tests for patients who had detectable virus with a previous PI drug regimen (Cohen reference). After four months, the arm that had a second-line regimen devised based upon phenotypic resistance tests achieved a higher percentage of virus undetectability (limit 400 copies per millilitre, 38%) than the standard-of-care arm without having the benefits of testing (23%).

Retrospective Studies The Panel briefly reviewed seven retrospective studies representing a total of 618 patients that showed benefits of drug resistance testing. Three of those studies with 418 patients examined genotype and/or phenotype results of the NRTI drugs: in two of them (one used genotype testing, one used phenotype testing) with 375 patients, baseline resistance to Retrovir (zidovudine, ZDV) was predictive of progression to AIDS or death. In two of the seven studies with a total of 118 patients, baseline genotype and/or phenotype drug resistance testing was associated with a response to a regimen with saquinavir (Invirase or Fortovase) and ritonavir (Norvir). In the last two studies with 82 patients, baseline phenotype or genotype testing of the PI and NRTI drug classes, with or without the non-NRTI drug class, was predictive of virologic response. The Panel did not include an eighth retrospective study of 71 patients that revealed benefits of phenotype resistance testing (see S. Call, M. Saag reference). However, the study endpoint had a much less stringent definition of at least -0.5 log copies per millilitre reduction in viral load that was persistent for the entire follow-up duration. After a median 15 months of follow-up, the best predictor of virologic failure was the number of drugs in the new regimen to which the patient’s virus was sensitive, based upon baseline phenotype drug testing. There was an approximate 1/2 decreased risk of failure for each drug to which the virus was susceptible at baseline (‘relative risk of 0.53’).

The Panel concluded, ‘Collectively, the prospective and retrospective studies show that the presence of drug susceptibility is predictive of treatment response and adds to knowledge obtained from either drug history or viral load measurements alone.’ Yet, they caution that the presence of susceptibility (absence of resistance) of drugs does not always predict an adequate response to those drugs. As discussed above, one reason would be possible minority species not detected. Also, poor adherence, poor drug absorption or other factors leading to inadequate blood concentrations of drug could also account for a lack of response. At that point, the Panel did not comment on the often spoken converse that the presence of resistance or decreased susceptibility is more likely to predict a lack of response. However, three recent studies even counter that notion, if the drug levels achieved far exceed the IC90 of patients’ ‘resistant’ HIV isolates or if there is only low-level resistance or an intermediate level of decreased susceptibility. This relates to the concept of clinical validation of the tests such that the ‘cut-off’ values for phenotype test results correlate with clinical virologic outcome. These are still being defined.

Additional Clinical Factors

The Panel discussed several clinical and other factors related to the resistance tests that must be considered. First, the Panel cautions that the results of drug resistance test should not be used as the primary factor to start or change a drug regimen. Based upon other recommendations by the IAS-USA and the Panel convened by the US Department of Health and Human Services, the decision to start or change anti-HIV therapy should be based primarily upon the viral load and CD4 count. After drug ‘failure,’ other factors also need to be considered before selecting a new drug regimen. Those factors are: past anti-HIV drug history, tolerance to relevant medications, other (non-HIV) related conditions and medications, and the prospects for future adherence.

As mentioned in the previous section, adequate drug levels of anti-HIV drugs are important for anti-viral success. For example, in the VIRADAPT study above, optimal concentration of the PI drug plus genotype testing led to the highest percentage of viral undetectability. However, tests of anti-HIV drug levels have not yet been shown to be of benefit in large studies. They are available commercially, but are not FDA approved. The Panel cautions that the ‘trough [minimum concentration] level of drug needed to suppress [HIV] replication in vivo [in people] is not known for any drug, and it likely differs for each drug.’ Another factor to consider is that blood levels of drugs within cells may be low, even if blood plasma levels are adequate. This may be due to cellular ‘resistance’ mechanisms whereby the cell ‘pumps out’ the drug with ‘efflux’ factors, including ‘p-glycoprotein’ and others. The Panel also reminds patients and physicians that plasma levels of the NRTI drugs do not correlate with their efficacy: only intracellular levels of the active form (NRTI-tri-phosphate) provide meaningful information. Currently, such tests are technically challenging, even for research laboratories.

Another important factor is when the resistance test is performed. The Panel recommends that the blood sample be drawn while the patient is still on their ‘failing’ regimen. If the regimen has been stopped, the majority of patients will revert to a dominant HIV isolate that is ‘wild-type’ within 1-2 months. In this setting ‘susceptible’ HIV variants outgrow the resistant, mutant ones in the majority of patients. This would lead to a ‘false negative’ result if the blood sample is taken while off medications. Even if the mutant isolate is outgrown by ‘wild-type,’ the mutant isolate merely becomes a minority species that likely would re-emerge after the same regimen would be restarted.

The Panel cautions physicians and patients that there are likely to be interactions of several resistance mutations that lead to unknown effects, because of the many potential combinations of drugs. Naturally occurring gene variations, called ‘polymorphisms,’ also complicate the picture, since they too might lead to unknown effects, in combination with primary or secondary resistance mutations. There have been examples whereby resistance mutations led to effects on other drugs that could not have been predicted. One example that the Panel uses is the ‘M184V’ resistance mutation due to lamivudine (3TC) that partially ‘reverses’ specific zidovudine (ZDV) resistance. However, HIV subsequently can develop new NRTI drug mutations that will confer resistance to both drugs. Another example is that approximately 25% of patients with specific NRTI drug mutations have increased susceptibility to non-NRTI drugs, even when compared to wild-type (‘hypersusceptible’). Another example that the Panel did not include is that a primary resistance mutation to amprenavir, (I50V) leads to increased susceptibility to saquinavir. A last example that the Panel did not include is that the ‘N88S’ PI drug mutation that can occur after therapy with either indinavir or nelfinavir shows increased susceptibility to amprenavir (Ziermann reference).

Notwithstanding these limitations, the Panel does indicate that two or more key PI drug mutations ‘is likely to confer broad cross-resistance to most currently available PI [drug]s.’ Those key mutations are: ‘D30N, G48V, I50V, V82A/F/ or G, I84V and L90M.’ Since new information about HIV drug resistance comes available frequently, the Panel says that the results and interpretations of resistance tests will also need updating regularly.

The Panel discussed what has been referred to as a ‘virtual phenotype.’ This is the reporting of an expected phenotype result, based upon the results of genotype testing. Virco has been the pioneer in this area, since they have a large database of patient samples that have been genotyped and phenotyped. If a particular combination of genotype resistance mutations is associated with the same phenotype test result in, for example, 90% of samples, then a virtual phenotype might be reported based upon genotype test results alone. However, a virtual phenotype would not account for genetic differences in patients. The database will grow even larger with time, particularly as new drugs become available.

Considering all the factors mentioned in this section, the Panel recommends that expert interpretation of both genotype and phenotype resistance tests be included as a part of the results. This is not meant to be self-serving, since many of the Panel members would be among those experts. Given the complexity and evolving nature of drug resistance testing, the recommendation is a worthy one.

Specific Recommendations or Considerations

Given the background of information presented, the Panel considered several clinical situations for which drug resistance tests would be useful. The members used one of two terms for each clinical scenario: recommended or considered, with the latter term reflecting a more conservative, less-definitive position about potential utility.

Primary HIV Infection

The Panel said that drug resistance testing should be considered in this setting. (Primary or Acute HIV infection is that time shortly after HIV transmission, usually associated with ‘flu’-like symptoms.) Even if a person does not have symptoms, the Panel would include those who have evidence of recent HIV infection, within several months before presenting to the physician, particularly if the ‘source patient’ (transmitter) was taking anti-HIV therapy. The Panel said that the goal in treating during primary infection would be to stop HIV replication quickly. This is because a few studies have shown that HIV-specific immune responses appear to be improved and initially enhanced in this setting. Whereas, after established infection, this is more difficult to accomplish. Studies are ongoing to determine the potential benefits of treating during or shortly after primary infection. Negative factors to consider are the difficulty of adhering to dosing for years or even decades, side effects from the drugs, including potential fat redistribution and increased blood fats, and the potential development of drug resistance. However, if the decision is made to start anti-HIV therapy in this setting, blood should be sampled for resistance testing before starting treatment. Waiting for the results before starting treatment is not recommended. If the resistance test results indicate that the patient’s HIV has significant decreased susceptibility to a drug(s) in the regimen, a change could be made at that time. The reason for treating and the type of treatment during primary infection will depend upon the geographic location of the transmission and the associated rate of HIV drug resistance. This assumes that there is constant monitoring within local communities as to whether recently transmitted isolates have drug resistance or not. The Panel reviewed the evidence for increasing rates of drug-resistant HIV transmission in several geographical locations. For example, sexual transmission of HIV with resistance to ZDV and 3TC is becoming more common. Also, ‘a surprisingly high proportion of recently infected subjects has circulation virus with some degree of reduced susceptibility (less than 10-fold) to the non-NRTI drugs, ritonavir, and nelfinavir.’ However, this would not necessarily translate into those drugs not being effective.

Even if resistance testing is not performed before therapy is started for primary infection, a slower than expected decline in the viral load might indicate resistance (or non-adherence). The Panel recommends that not achieving viral undetectability by 16-24 weeks of treatment ‘should prompt consideration of resistance testing.’ The higher the baseline viral load, the longer it would be expected to achieve undetectability. Due to the issue of new transmission of HIV with drug resistance, the Panel does state that ‘ongoing epidemiologic surveillance for PI [drug]-resistant and multidrug-resistant virus in patients presenting with acute or early HIV infection is essential.’

Established Infection

The Panel said that drug resistance testing should be considered in this setting, before starting treatment. This would especially apply in geographic locations where ‘the local prevalence [rate] of primary drug resistance is appreciable.’ A factor to consider here would be an estimate of the duration of infection. One study has shown that the rate of resistance among persons with established infection is lower than that in primary infection in the same location. However, another study did not find it to be any lower.

When Changing Therapy

The Panel said that drug resistance testing is recommended in this clinical setting, when due to a detectable viral load (never undetectable or rebound), not drug toxicity or side effects. The reasoning is because of the prospective and retrospective studies mentioned in the second section. The results would likely reveal some level of decreased susceptibility to at least one drug in the patient’s regimen. This also means that the patient’s dominant HIV isolate might not be resistant to all drugs in the regimen. Yet, changing just one drug generally would not be recommended. In general, the presence of high level resistance (high level of ‘decreased susceptibility’) would indicate that a drug would not be effective at suppressing HIV growth. There are some potential exceptions to this rule, as indicated in the previous section, particularly if the level of drug can exceed the ‘IC50’ or ‘IC90’ of the resistant isolate, as discussed above. Also as mentioned, the absence of resistance (no decrease in the susceptibility) does not automatically indicate that a drug will be effective. This is due to the potential of minority mutant HIV variants that were not detected by the resistance testing, as discussed earlier. If the resistance test results in this setting do not indicate any detectable resistance, this might indicate poor adherence, poor drug absorption into the blood stream, or some other drug-drug interaction leading to suboptimal (‘subtherapeutic’) drug levels. The utility of drug level testing (‘therapeutic drug monitoring’) at this time is on a research level only. Lastly, the Panel comments that resistance testing also may be useful after a second-line regimen fails, when fewer choices about drugs are available, and the issue of ‘drug recycling’ emerges.

First Regimen ‘Failure’

The Panel said that drug resistance testing is recommended in this setting, after poor adherence has been excluded, if possible. Other factors also should be excluded, including poor drug absorption and adverse drug-drug interactions associated with inadequate blood levels of drugs. As discussed earlier, the blood sample should be obtained before discontinuing the failing regimen. The Panel said that resistance testing would not always be needed in this setting, if the anti-HIV drug history is well known and a second regimen replaces the failing one at a lower and not higher viral load (lower and higher were not defined). Four clinical situations that would apply under this category of ‘first regimen failure’ are: (1) early in the course of therapy, if there is an inadequate decline in the viral load during the first three months; (2) after the first evidence of early viral rebound, subsequent to a period of undetectability; (3) when there has been ‘prolonged viral replication in which more extensive resistance might be suspected;’ and (4) ‘prior to therapy institution to provide a baseline for longitudinal testing likely to become a more common component of patient monitoring.’

Multiple Regimen Failures

The Panel said that drug resistance testing is recommended in this setting for the reasons described in the first sections. Notwithstanding the limitations already discussed, as a general rule, the presence of resistance is predictive of not responding to that drug. Particularly for the patient with multiple treatment failures, options are limited. Therefore, resistance testing might indicate which drugs could be beneficial. This would include FDA-approved drugs and possibly, drugs under development that are available by expanded access. In the setting of multiple regimen failures, resistance testing will help to avoid the toxicity and expense of drugs that are unlikely to be effective.

Pregnant Women

The Panel said that drug resistance testing is recommended in this setting not only for the mother’s health, but to decrease the risk of HIV transmission to the unborn infant. Testing would particularly be recommended when the mother has had significant exposure to anti-HIV drugs in the past or if the level of drug resistance is known to be high in the local community. The Panel reviewed other supporting evidence. For example, heterosexual transmission of drug-resistant HIV can occur to the mother either before or during pregnancy. Multi-drug-resistant HIV has been transmitted to unborn infants when their mothers had detectable virus and an extensive history of anti-HIV therapy. While initial guidelines had recommended ZDV monotherapy to help prevent transmission from mother-to-newborn, current guidelines do not recommend withholding combination anti-HIV combination therapy when ‘otherwise indicated.’ Unfortunately, such therapy is not available to almost all pregnant HIV positive women, due to their being in a resource-poor country. Some of them do have access to ZDV monotherapy. In the future, some may gain access to the 2-dose nevirapine monotherapy regimen that has been shown to decrease transmission to newborns. [And to lead to nevirapine (cross class NNRTI) resistance in a large number of recipients – Editors note].

Future Issues

The Panel discussed some future issues in the area of drug resistance testing. First is whether resistance testing provides long-term benefits, in addition to the short-term benefits that have already been shown. Second is additional refinement or clarification of specific clinical situations for which resistance testing is beneficial. Third is the future ability to detect ‘minority’ species that are present in patients. Fourth is how the results of resistance testing for individual drugs can be applied to combination therapy. Fifth is the controversial and experimental approach of using resistance testing in the setting of Structured Treatment Interruptions (‘drug holidays’). Sixth and last is the issue of cost-effectiveness of drug resistance testing. The Panel said that the tests currently are ‘costly,’ ranging from $400-550 for genotype resistance testing and $700-900 for phenotype resistance testing. Several economic analyses have shown that drug resistance testing would be cost effective, depending upon several variables, since drugs that would not be effective would not be prescribed and toxicity from those drugs might be prevented.

Panel Members

The 14 members (including 13 physicians) of the IAS-USA Panel are: Martin S. Hirsch, MD, from Harvard Medical School; Francoise Brun-Vezinet, MD, from Hospital Bichat-Claude Bernard in Paris France; Richard T. Aquila, MD, from Harvard Medical School; Scott M. Hammer, MD, from Columbia University in New York City, New York; Victoria A. Johnson, MD, from the University of Alabama at Birmingham; Daniel R. Kuritzkes, MD, from the University of Colorado at Denver; Clive Loveday, MD, PhD, from The Royal Free Hospital in London, England; John W. Mellors, MD, from the University of Pittsburgh in Pennsylvania; Bonaventura Clotet, MD, PhD, from University Hospital Germans Trias I Pujol in Barcelona, Spain; Brian Conway, MD, from the University of British Columbia in Vancouver; Lisa M. Demeter, MD, from the University of Rochester in New York; Stefano Vella, MD, from the Istituto Superiore di Sanita in Rome, Italy; Donna M. Jacobsen, from the IAS-USA in San Francisco, California; and Douglas D. Richman, MD, from the University of California at San Diego.

Links to other websites are current at date of posting but not maintained.