June 2012
Feature Story

William Check, PhD

While testing for resistance to antiviral drugs has become a reality in the molecular diagnostics laboratory, with clear value in selected populations, no one is yet using words like “explosion” or “growth industry” to describe these assays. Unlike testing for antimicrobial resistance in bacteria, testing for resistance to antiviral drugs is not widespread. One major reason is that antiviral agents have been available for a much shorter time than antibacterials. Drugs against herpesviruses [herpes simplex virus 1 and 2 (HSV-1, HSV-2), cytomegalovirus (CMV), and varicella-zoster virus (VZV)], for instance, are only about 35 years old.

“Therapies against herpesviruses have not gained a lot of ground since the 1970s,” Hank Balfour, MD, professor of laboratory medicine and of pediatrics at the University of Minnesota, said in a talk on antiviral resistance testing at the 2011 meeting of the Association for Molecular Pathology. “As a result,” Dr. Balfour said, “we do not have a lot of therapeutic options for herpesviruses. That is one reason why resistance testing for these viruses has not proceeded as fast or as far as for HIV or HCV.”

Richard Whitley, MD, professor of medicine at the University of Alabama at Birmingham (UAB) School of Medicine, was one of the first clinicians to use anti-herpesvirus drugs, both the earliest—adenine arabinoside (ara-A)—and the forerunner of the modern antiherpetic class—acyclo-vir (Whitley RJ, et al. N Engl J Med. 1976;294:1193–1199; Gnann JW Jr, et al. Pharmacotherapy. 1983;3:275–283). “Herpesvirus infections are incredibly common,” Dr. Whitley told CAP TODAY. “However, herpesviruses, which are DNA viruses, are less likely to become resistant than RNA viruses like HIV and influenza. When we think about resistance in herpesviruses, the setting I worry about is the immunocompromised host.” Such patients include organ or stem cell transplant recipients; those who are immunocompromised due to underlying disease, such as HIV/AIDS; or individuals taking immunosuppressive medications. “Resistance to herpesviruses is exceedingly uncommon in someone with a normal immune system,” Dr. Whitley says.

Dr. Balfour estimated that the prevalence of resistance to herpesviruses averages two percent to five percent in immunocompromised patients. “They can’t suppress virus replication with their natural immune host defense,” he said.

In addition to discussing indications and methods for resistance testing, in his talk Dr. Balfour suggested ways in which the laboratory director can help in cases of apparent resistance. One way is to assess nonadherence, which is more common as a cause of treatment failure than true resistance. Other ways are to evaluate drug dose and possibly suggest increasing dose or switching drugs.

Pharmacotherapy for influenza A virus is at an even earlier stage than for herpesviruses, and so is resistance testing, says Karen L. Kaul, MD, PhD, who spoke on “Molecular Detection of Influenza A Variants and Antiviral Resistance” at the AMP session. Dr. Kaul’s laboratory has been developing real-time PCR assays for detection of resistance to the neuraminidase blocker oseltamivir (Tamiflu). This effort is “a work in progress,” said Dr. Kaul, who is Board of Directors chair of molecular pathology and director of the molecular diagnostics laboratory at Evanston Hospital, NorthShore University HealthSystem, and professor of pathology at the University of Chicago Medical School.

During the 2005–2006 flu season, resistance to oseltamivir rapidly emerged in the seasonal H1N1 strain and continued to grow, with 98.6 percent of seasonal H1N1 isolates becoming resistant to this antiviral treatment. Novel H1N1, which appeared in 2009, “has not shown a great deal of resistance,” Dr. Kaul said in a recent interview, “so this past flu season, with the predominant strains in circulation being H3N2 and novel H1N1, we have not worried much about antiviral resistance.”

However, Dr. Kaul noted, “As we look ahead, we expect at some point resistance will show up in swine and other seasonal flu strains, so we would like to have a specific assay ready, ideally one that detects the mutation in viral subtypes.” Current resistance to oseltamivir is due to mutation at the H275Y nucleotide site. In the future, resistance could be due to other point mutations. “As we see more resistance, the mechanism may be a changing target,” she says. “We may have to include primer pairs for other mutations, and assay design with the viral sequence variations may be challenging. This evolving virus will continue to educate us.”

In his AMP presentation, Dr. Balfour listed five situations in which to suspect antiviral resistance. Foremost is an immunocompromised host. Resistance “almost invariably happens” in these patients, he said. A second situation is a history of intermittent or low-dose therapy, which creates a favorable milieu for selecting resistant viruses. Dr. Balfour showed the case of a transplant recipient who received therapy for CMV as an outpatient. No therapy was given on the weekends. The virus developed resistance and the patient died.

Resistance should also be suspected when a cutaneous lesion due to HSV or VZV or systemic disease due to CMV does not improve or worsens after seven days of adequate dosing. Finally, Dr. Balfour said, “One of the most common reasons today for suspecting resistance is increased viral load.”

Compartmentalization must be considered when sampling for a possible resistant herpesvirus. In one bone marrow transplant patient, the first resistant isolate of VZV was from CSF after the appearance of neuritis optica at day 245. At that time, virus from the blood was still sensitive. The patient died at day 250 (Brink AATP, et al. Clin Infect Dis. 2011;52:982–988).

True resistance is only one of the main reasons for treatment failure. “Nonadherence is the biggest reason,” Dr. Balfour says. “Transplant patients take home a whole box of pills.”

Treatment for herpesviruses consists of several agents administered intravenously or their orally bioavailable prodrugs: acyclovir/valacyclovir, penciclovir/famciclovir, and ganciclovir/valganciclovir, as well as foscarnet and cidofovir (Balfour HH Jr. N Engl J Med. 1999;340:1255–1268). All of these drugs, with the exception of foscarnet and cidofovir, have a similar mechanism of action: They are phosphorylated initially by kinases such as the thymidine kinase (TK) encoded by HSV-1, HSV-2, and VZV, or the CMV UL97 protein kinase, and the ultimate triphosphate metabolites are inhibitors of the virus’ DNA polymerase (pol). Resistance almost always is due to mutations in the pol gene or virus kinase genes, or in both. As a result, although there are many drugs, they don’t actually offer good alternatives for treatment of a virus that has become resistant. “If I can, I will use more than one drug at the outset to potentially avoid resistance,” Dr. Balfour says.

“There would be a huge advantage to developing drugs with different mechanisms of action,” he told CAP TODAY. Yet this has not happened for three main reasons. First, diseases caused by members of the herpesvirus family are mostly acute. “From an economic standpoint,” Dr. Balfour notes, “companies have more to gain from something like antiretroviral therapy, which is essentially a lifetime proposition.” Second, the antiherpes drugs that do exist are “fairly effective.” Also, the safety profile for most herpesvirus drugs is fairly good. “So the pressures are not there to develop new compounds, which is unfortunate,” Dr. Balfour says. This is particularly true when resistance arises.

Resistance testing can be done by phenotypic or genotypic methods. Phenotypic methods require growing the virus in cell culture, which is slow, labor-intensive, and not standardized. On the other hand, phenotyping potentially provides an integrated value for mixtures of genotypes.

Genotyping, which Dr. Balfour called the “most popular” approach, gives a precise identification of a mutation; combined with a good database, it can define resistance. For HSV, detailed genetic maps of the pol and kinase genes have been published (Piret J, Boivin G. Antimicrob Agents Chemother. 2010;55:459–472) and some point mutations can be correlated with resistance to acyclovir (Morfin F, Thouvenot D. J Clin Virol. 2003;26:29–37). In addition, Dr. Balfour said, genotyping is “reliable, rapid, and automatable.

“In the best of all possible worlds,” he continued, “we would use both approaches.” The Research Group for Antiviral Resistance, based in Leuven, Belgium, has developed algorithms integrating phenotypic and genotypic assays for HSV-1 and -2, HCV, and VZV (www.regavir.org).

As an indication of how far from ideal the current laboratory world is, Dr. Balfour himself does not do herpesvirus resistance testing. “We used to do IC50 for several herpesviruses,” he says. However, the volume was not sufficient to support testing. “We can’t afford to do [antiviral resistance testing] because the administration is all over me to be cost-effective.”

Only a limited number of reference labs do resistance testing for herpesviruses. Quest Diagnostics does phenotypic assays for acyclovir, foscarnet, and ganciclovir in HSV-1 and -2. For CMV, Quest uses genotyping assays to aid in predicting resistance to cidofovir, foscarnet, and ganciclovir. LabCorp does phenotypic assays for acyclovir and foscarnet resistance in HSV-1 and -2.

To Dr. Whitley, “The real question is, Who is going to do your resistance testing? That’s where we really get into trouble,” he says. Assays are not standardized and not available in every medical center around the country. At UAB the testing is done in-house. “When people call me, I refer them to my own people,” Dr. Whitley told CAP TODAY. “We do phenotypic tests if we have a viral isolate. Otherwise, we do genotyping.”

Mark N. Prichard, PhD, professor of pediatrics and director of the diagnostic virology laboratory, directs herpesvirus resistance testing at UAB. “We will sequence a herpesvirus genome and say whether any mutations we detect are associated with resistance,” Dr. Prichard told CAP TODAY. “That works with some mutations in some viruses, but it doesn’t work well with others.” For HSV-1 or -2 or VZV, for instance, a frameshift mutation in the TK gene will be resistant. With point mutations in TK or pol, on the other hand, “It is not clear whether they will be resistant. The data for that are weak,” Dr. Prichard says.

With an isolate, they do phenotyping. “We directly measure susceptibility to drugs by a traditional plaque reduction assay,” Dr. Prichard says. Resistance can be detected by this assay only if 25 percent or more of the virus population is resistant. Using deep sequencing over the past few years, Dr. Prichard and his colleagues have discovered considerable genetic variation in herpesviruses, demonstrating the need for a more sensitive measure. “With our new phenotypic resistance assay, we can detect genetic change down to 0.1 percent in a patient sample,” Dr. Prichard says. “So we can detect very early on if a patient has a genetic variant for resistance. We are no longer dependent on genotypically detecting frameshift mutations.” Increased sensitivity in the phenotypic assay is achieved by using PCR to quantitate the amount of virus produced by a plaque. If a virus is sensitive, the amount of virus produced should drop (from 105 to 102).

Since 2005, Dr. Kaul’s primary assay for detecting influenza virus has been real-time PCR, derived from a published method and based on conserved sequences of the viral matrix protein gene (Stone B, et al. J Virol Methods. 2004;117:103–112). In 2009 two things happened: Seasonal H1N1 developed a high level of resistance; and novel, swine-origin H1N1 emerged in pandemic fashion. Fortuitously, melt curves in Dr. Kaul’s assay discriminated novel H1N1 (sensitive to oseltamivir) from seasonal H1N1 (resistant) and H3N2 (sensitive), making it a de facto resistance assay (Kaul KL, et al. J Mol Diagn. 2010;12:664–669). Melt curves acted as a surrogate for resistance.

At that time Dr. Kaul and a colleague, Kathy Mangold, PhD, set out to make a real-time PCR assay that would detect resistance directly. Oseltamivir resistance is caused by a point mutation at H275Y of the neuraminidase gene. Because the region around H275Y varies among 2009 novel H1N1, the former seasonal H1N1, and H3N2, as well as within subtypes, a cocktail of primers was required. In a validation study using patient isolates, the assay correctly called all 90 sensitive strains, but only a subset of resistant isolates. “Unfortunately, we’re not picking up resistance across viral strains using a single multiplex reliably,” Dr. Kaul says. “The H275Y point mutation is a tricky area to target in a PCR assay.

“As flu drifts, we have to be cognizant of the fact that we can lose the ability to see the virus,” she continues. “We must include multiple primer pairs so we don’t get false-negative results as the virus evolves.” Dr. Kaul points to publications in which FDA-cleared assays are calling negative in viral strains that have mutations at a key spot in the primer binding site (Zheng X, et al. J Clin Microbiol. 2010;48:665–666; Klungthong C, et al. J Clin Virol. 2010;48:91–95). “Influenza continues to challenge us,” Dr. Kaul says.

At UAB, the onset of pandemic H1N1 in 2009 also provoked development of a resistance assay. “We wanted to know if treatment with oseltamivir would cause the proportion of resistant viruses to increase with time or whether patients treated with oseltamivir would develop resistance during treatment,” Dr. Prichard says. To measure this, they made a real-time PCR assay. It is an allelic discrimination assay that measures two quantities—the number of genomes and the frequency of mutations at the H275Y site. Each is measured with separate primers in a separate reaction tube. “When a virus has the mutation, it is amplified with a specific set of primers that only amplifies the mutation and not wild-type virus,” Dr. Prichard says. Analytic sensitivity is one mutation in 1,000 wild-type genomes.

At the outset of the pandemic the frequency of resistance was very low—0.001 percent of the viruses in a population, the spontaneous level that you would expect from sequence variation. In some patients the assay detected a 10 percent resistant population. “So the assay is very sensitive,” Dr. Prichard says. “The downside is that it only works with pandemic H1N1, not with seasonal H1N1 or H3N2 strains.” As with Dr. Kaul’s assay, picking up all resistant strains is difficult because of sequence variations among strains.

Initially, only one of 20 or 25 isolates from patients in UAB’s Children’s Hospital had a significant representation of H275Y mutations. “Since then the frequency has been maybe two percent to three percent,” Dr. Prichard says. So resistance appears not to be increasing.

At this point this assay’s clinical sensitivity has not been evaluated. It is being used only for surveillance.

Given the history of resistance as a clinically important factor in microbial pathogens, it seems likely that assays for antiviral resistance will assume increasing importance. Dr. Balfour notes that newer molecular technologies are making resistance assays more feasible and enabling automation. Still, volume will remain a consideration. Dr. Whitley emphasizes the incomplete state of current knowledge about resistance-causing mutations in herpesviruses. “I think sequencing will definitely play a bigger role,” he says.

For influenza virus resistance testing, Dr. Kaul raises practical considerations. Which is better—central or local testing? Does faster recognition have clinical value or decrease cost? “We need to assess the reality of clinical practice,” she says.