March 2012
Feature Story

Karen Titus

The latest HIV testing guideline takes into account fourth-generation screening assays but allows for use of a third-generation test. The fourth-gen versus third-gen discussion “was a swinging door that got kicked in and out a few times,“ says Dr. Mark Pandori, one of the guideline’s creators.

Fans of Hollywood movies and solid narrative structure—in other words, those who prefer their stories to have clear-cut endings and logical plots—should take satisfaction in a new set of HIV testing algorithms. The algorithms, published by the Clinical and Laboratory Standards Institute (and developed by experts who built on the work of a CDC/Association of Public Health Laboratories committee), are the first new HIV testing algorithms to emerge since 1989.

A lot has changed in 23 years.

On a practical level, the so-called window period will be narrowed even further. The CLSI guideline (M53-A) takes into account newer assays, including fourth-generation screening assays, which detect both antigen and antibody. The guideline also calls on labs to use an HIV-1/HIV-2 differentiation immunoassay.

Using more advanced testing technologies will let labs give definitive answers to physicians and patients, quickly.

Under the previous testing algorithm, laboratories could end their reports with, in essence, an answer of, “We don’t know,“ says Mark Pandori, PhD, laboratory director, San Francisco Public Health Laboratory. Now, laboratories are urged to use a tie-breaking test, so to speak, to eliminate indeterminate results.

The old algorithm was focused on specificity, says the CDC’s Bernard M. Branson, MD, who helped develop the new guideline. Given the peculiar anxieties of the early era of HIV infection and AIDS, the main diagnostic goal was to eliminate any possibility of a false-positive result. With only antibody tests available early on, results had to be right. The advent of RNA tests have made that less crucial. “We have learned that sensitivity is in fact the thing that we have sacrificed in our approach, up until now. And we need to be paying attention to it,“ says Dr. Branson, associate director for Laboratory Diagnostics, Divisions of HIV/AIDS Prevention, National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention, CDC.

(The CDC will issue its own guideline related to HIV testing within the next few months, in its Morbidity and Mortality Weekly Report, says Dr. Branson. The CDC’s report will be substantially similar to the CLSI guideline, he says. Why the repeat performance? “There are several state laws, for example, that require that they follow recommendations as published by CDC.“)

The new algorithms speak to changes in attitudes as well as to changes in diagnostics.

In 1989, HIV was a different world, one filled with shunned patients and the notion of AIDS as tantamount to a death sentence. AZT, the first major therapeutic breakthrough, had been approved only two years earlier, when the FDA’s accelerated drug approval system was still new, and when AIDS Related Complex, or ARC, was still part of the common vocabulary. “Health departments had separate AIDS bureaus,“ recalls Barbara Werner, PhD, who helped develop the new algorithms; she also serves as an infectious disease consultant with the Massachusetts Department of Public Health. “Not many other diseases have stood alone like HIV infection.“

Today, of course, HIV is part of the medical mainstream. Recommendations and laws requiring HIV testing are common, and improved treatments have moved HIV infection into the realm of chronic disease. But even as the new CLSI guideline reflects this broader shift, physicians may still need to be nudged to accept the latest testing strategies.

The CLSI guideline grew out of several needs. The most obvious, perhaps, was to respond to changes in technology.

In fact, the committee that created the guideline had to update its own work midstream, says chair Eric Rosenberg, MD. The group at first focused on algorithms that used third-generation assays for the initial test. Halfway through writing the document, it became clear that fourth-generation combination testing would become available for use in the United States. (It’s been widely used in Europe for roughly a decade.) “We hit the brakes,“ says Dr. Rosenberg, codirector, clinical microbiology laboratory, Massachusetts General Hospital, and associate professor of medicine, MGH and Harvard Medical School. “We realized that a different test would make our document old before it was out.“

“So we stopped what we were doing and went back to the drawing board,“ Dr. Rosenberg continues, “and revised all of the algorithms to accommodate fourth-generation testing. It was the right thing to do.“

While the guideline reflects the availability of fourth-generation testing (Abbott and Bio-Rad have FDA-approved fourth-gen tests), not every algorithm in the document calls for using it. That may create some confusion for laboratories.

The guideline, weighing in at 89 pages, “is a complicated document, because with all those algorithms, they’re trying to address each particular scenario that institutions may have,“ says William Koss, MD, director for clinical pathology, Scott & White Healthcare, an integrated health care system based in Temple, Tex., and professor of pathology, Texas A&M Health Science Center College of Medicine. One observer hazarded that the CLSI document contained “seven? or maybe it’s eight?“ algorithms, making it sound more like a listing of English kings named Henry.

Actually, there are six algorithms, though the first one is the most noteworthy. It calls for initial use of an antigen-antibody combination assay, that is, a fourth-generation test, followed by an HIV-1/HIV-2 differentiation assay as a second, supplemental test.

In recent years it had become apparent that even with third-generation screening EIAs that are IgM and IgG reactive, the window period was approximately 22 days, says Dr. Pandori, who helped develop the new algorithms. With other tests, most of them older, and with the Western blot, the window period is a Noah-worthy 40 days.

Under the old algorithm, screening EIA positive results that couldn’t be confirmed as positives by confirmatory tests were termed indeterminate or negative. “In fact, those could be recent infections that were in the window period of the third gen but outside the window of confirmation,“ says Dr. Pandori.

This was a fundamental problem a new algorithm would have to address. But superimposing a fourth-generation assay on the existing algorithm would have only exacerbated the problem, Dr. Pandori notes. Because fourth-gen tests detect both antigen and antibody, their window periods are only about 16 days. “So if you’re talking about running a fourth-gen and confirmation test with a 40-day window period, your gap is even larger,“ he says. “And there may now be the possibility that [people] could be positive on a screening test because they are virus positive but not antibody positive—which is quite early. And there would be no chance to confirm that with an antibody test.

“So, recognizing that, we needed to come up with an algorithm that could deal with window-period infection, so to speak—acute and/or recent HIV infections,“ says Dr. Pandori.

That’s just the first step. Both third- and fourth-generation EIAs detect antibody to HIV-1 and HIV-2, but neither one differentiates between the two.

There is, however, an FDA-approved, moderately complex rapid test on the market that detects and differentiates HIV-2 infections. Hence, the algorithm calls for that HIV-1/HIV-2 assay as a second step.

The official language, and many of the guideline’s creators, refer to it as a differentiation assay. Some call it a supplemental test or a second test.

In reality, it’s Bio-Rad’s Multispot HIV-1/HIV-2 Rapid Test. There. We said it. For all practical purposes, it’s the only test available right now. The guideline’s creators concede the difficulty of recommending a step that’s served by only one FDA-approved assay at present, and say they hope other tests will become available.

The Multispot is a little more sensitive than Western blot, Dr. Pandori says, and has a window period that may be as much as seven days shorter. It’s also less expensive than Western blot, and can be run in 15 minutes, says Dr. Branson.

Not every algorithm in the CLSI guideline calls for a differentiation assay as the second test. But anywhere between 60 and 90 percent of HIV-2 cases will have a Western blot result that can be interpreted as positive for HIV-1, Dr. Branson says, leading to misdiagnosis. Because HIV-2 does not respond to first-line medications used for HIV-1, “That’s an important miss,“ Dr. Branson says. Nor do HIV-2 infections show up on viral load tests, which target HIV-1. So patients with HIV-2 infections “end up getting diagnosed in the main because they deteriorate clinically despite having a negative viral load test.“

HIV-2 isn’t a concern only in urban areas with large populations of HIV-infected patients. Says Michael Pentella, PhD, associate director, disease control, State Hygienic Laboratory at the University of Iowa: “We’ve seen HIV-2 cases here in Iowa, which isn’t something we were expecting. But people travel, so we have to be prepared to recognize HIV-2.“ Dr. Pentella’s lab is implementing a fourth-generation assay and plans to adopt the first algorithm.

If labs use the antibody differentiation assay, the majority of people who are antibody positive will be resolved at that second testing step.

But from Dr. Pandori’s perspective, the strongest part of the new algorithm is that it calls for discordant results to be resolved by a nucleic acid test, or NAT. “The previous algorithm gave no license for that whatsoever. No recommendation,“ Dr. Pandori says. “So anybody who was put in a position where they had a discordant specimen and suspected acute HIV had no recommended mechanism for resolving it with RNA, when we all knew RNA could resolve those.“

Were laboratories doing RNA even without such a recommendation? “It really depended on where you practiced,“ Dr. Pandori says.

In San Francisco, he says, physicians were highly knowledgeable about RNA’s ability to detect recent and acute HIV infection. “I think a lot of those doctors were just saying, ‘Oh, run an RNA test.’ Back in ’05, our lab actually started using RNA diagnostically.“

On the other hand, he says, he knows of other health jurisdictions where physicians lacked the knowledge or experience to know that RNA could resolve discordant results between a screening and a confirmation test.

It’s not quite as simple as Dr. Pandori makes it sound. HIV viral load tests do not have FDA approval for diagnostic use. Only one RNA test does—Gen-Probe’s Aptima. Recommendations can’t suggest off-label use of a test, so by default the guideline essentially calls for using a manual test.

Dr. Pandori, who oversaw a series of conference calls with some 60 laboratories to obtain feedback on the algorithm as it was being developed, says he heard huge concerns about confirmatory RNA testing. Labs kept asking, “How the heck are we going to confirm with an RNA? Do we have to get Aptima?“ he says.

There are other options. Labs could do their own validations of viral load tests for diagnoses. Or, makers of the tests could seek a diagnostic indication from the FDA. Dr. Branson says the CDC has asked them to pursue this. A company spokesperson confirms that Abbott Molecular has been approached by the CDC about its viral load test but gave no indication whether the company plans to seek an additional indication.

Dr. Branson says manufacturers are open to the idea, but much will depend on what the FDA requires.

Asks Dr. Pandori: “Do they want to blow another $10 million on a clinical trial to do that? I don’t know.“

What does the new emphasis on fourth-generation testing mean for laboratories that want to continue using third-gen tests? That was the other big concern labs had during the conference calls, says Dr. Pandori: “Do we have to use a fourth-gen test?“

The fourth-gen versus third-gen discussion “was a swinging door that got kicked in and out a few times,“ Dr. Pandori says. In the end, the CLSI guideline allows for use of a third-generation test.

The guideline’s creators have a clear preference for using a fourth-gen test because of its earlier sensitivity during seroconversion. But the journey from guideline to practice can be fraught. “The situation,“ Dr. Branson says, “is that a lot of labs have third-generation platforms installed. And we recognize it takes time for people to make a change.“

Dr. Branson says he frequently hears another objection: Labs claim they don’t need a fourth-gen test because they’re sure their patient population includes few if any people with acute infections. He’s taken aback by this reasoning, and recounted a conversation with an Arizona lab that had recently switched to fourth-generation screening. “They’ve been quite surprised by how many acute infections they’ve detected,“ Dr. Branson says. “They’re not seeing it because, basically, unless someone has clinical suspicion and orders a viral load test, you can’t diagnose it.“

With some 50,000 new HIV infections every year in the United States, says Dr. Rosenberg, chances are that everybody who does HIV testing will encounter a patient who’s in the window period. “There are several compelling reasons why it’s important to make the diagnosis, even if it’s only a single person.“

Those with acute HIV infections are at higher risk of transmitting virus than at any other stage of HIV infection, because the amount of circulating virus in their body is so high. “They’re almost hyper-infectious,“ says Dr. Rosenberg. The wider the window period, the greater the chance they’ll unknowingly infect someone else. There’s also intriguing—though far from conclusive—evidence that therapy initiated during acute HIV infection may be beneficial.

Some laboratories may be reluctant to let go of another test. Dr. Pandori mentions another question that arose during the conference calls—and it’s one that makes him laugh. Hard. “Do I have to use the Multispot as my confirmation, or can I keep using the Western blot? My staff love Western blot.“

Western blot doesn’t need to be banished, say the guideline’s creators. No algorithm is perfect, and Western blot can still be used to sort out diagnostic dilemmas. If a person is positive on the Multispot assay, for example, the next step would be a viral load test. If that’s negative, then what? Turn to the Western blot, Dr. Branson suggests. “We have solid evidence that in antibody-positive people—that is, people who are EIA positive and Western blot positive, between three and five percent will have negative HIV-1 RNA tests.“

But even if Western blot isn’t racing off into the sunset, it may have a diminished role as a supplemental test. Dr. Werner talks about Western blot the way one might speak of a well-loved pet whose end is drawing nigh. “It’s been a good assay,“ she says. “It had a lot of good years, and it’s still good for longstanding infections. But it’s no longer as useful as it once was in the role as a supplemental test.“

The differentiation test picks up components of the HIV-1 virus that differ from those in the screening assays, she notes. “So it’s not like you’re just doing the same test again—it helps to corroborate the finding in the screening assay, as well as differentiate -1 from -2.“

Dr. Pandori makes his own quiet case for moving away from Western blot when he says, “Recommendations are made for a reason. And the reason here is if you choose to continue running a Western blot, you probably aren’t doing your best work.“

Dr. Pentella wouldn’t mind dropping the Western blot for a less-expensive test, and is leaning toward the Multispot. But just as important to him is the reduced turnaround time. He says he’s heard of patients waiting a week or longer for a Western blot result. He finds this unconscionable. “In this day and age, that’s way too long to have to wait for a verification of a first test.“

And yet as he discussed using a new algorithm in his lab, he faced resistance from colleagues reluctant to give up routine use of Western blot. “When they thought of it going away, there was a loss there,“ he says.

The Western blot still has a mystique about it. At a recent HIV testing conference, “We joked about a Western blot support group,“ Dr. Werner says. Decades ago, when HIV screening assays were less reliable, the notion took hold that a result wasn’t “real“ until it was confirmed by Western blot. For many, that mindset still holds, regardless of technological advances—the two go together, like the French and their cheese.

While many say they welcome the algorithm and fourth-generation testing, and plan to implement both soon, Dr. Koss has already begun. His lab migrated to Abbott’s fourth-generation assay last year.

Prior to this change, the lab, running a third-gen assay as its initial test, confirmed positives with a Western blot. If results were indeterminate or negative on Western blot, the lab would send out for an HIV-2 test.

With the new algorithm, the need for that send-out HIV-2 has been eliminated. But what seems to excite Dr. Koss more is that the new algorithm has been able to reduce the number of false-positive HIV results in pregnant women. Dr. Koss suspects the false-positives were related to cross-reactivity with HLA-related antibodies. This generated a lot of subsequent workup for a result that the lab knew from the outset was likely a false-positive.

Dr. Koss says the new algorithm was relatively easy to explain to clinical colleagues.

His experience demonstrates, again, how much things have changed over the years. Up until several years ago, HIV testing was part of what was referred to as “confidential testing,“ and had been for 15-plus years. As a result, HIV testing results were not necessarily viewable in the medical record. “Anytime there was a presumptive positive or a true positive, there was a direct communication between the laboratory and the ordering physician.“ That laid the groundwork for easy communication between the lab and clinicians about HIV testing. When the CDC urged wider HIV testing, Dr. Koss’ lab seized the opportunity to make HIV test results viewable rather than confidential. That created even more openness, which in turn has paved the way for a smooth transition to the new assay and algorithm.

His advice to labs? Be prepared to explain the algorithm to clinicians of every stripe. “Some people, you have to hold their hand. No matter how good your test reports are, those always require the lab to help them with the next step, either with testing or a referral.“ At the other end are infectious disease specialists. “You have to make sure they understand that report as well. You’re managing the confirmatory testing, the supplemental testing, and also the testing that is going to be used to monitor the patient. At the end of the day, the laboratory has to handle both extremes.“

Not everything has changed in 23 years.