September 2008
Feature Story

Karen Titus

In the nearly 30-year fight against AIDS, geography has held odd sway. The human immunodeficiency virus has a long, complicated history stretching back to Africa, and the present may be no different.

In Tanzania, measuring HIV-1 viral load levels has made a crucial, if late, arrival. With the help of foreign aid programs, “A lot more patients are receiving antiretroviral therapy than there were six years ago,” says John Crump, MB, ChB, assistant professor of medicine at Duke University Medical Center and director of the KCMC clinical research site in Moshi.

The ability to measure HIV-1 RNA concentrations is critical to the research work of Dr. Crump and his Tanzanian colleagues, as it is to anyone who is monitoring treatment of patients infected with HIV. KCMC (for Kilimanjaro Christian Medical Centre) acquired a real-time PCR assay system (Abbott) in January 2006, making it one of the first hospitals in Tanzania able to measure viral load, Dr. Crump says.

After lagging behind for so long, it would appear that KCMC is catching up to what has long been standard in U.S. laboratories.

Some might say it’s also time for U.S. labs to catch up to what’s happening in Tanzania and elsewhere in Africa.

Specifically, HIV-1 subtypes common in Africa may be making inroads in the United States, as they have in Europe. If nothing else, observers say, the time might be right to start asking questions about the potential impact of so-called African clades on HIV viral load monitoring.

There’s been good reason, so far, for U.S. labs and clinicians to keep “worry about HIV subtypes” toward the bottom of their to-do lists. The vast majority of HIV-1 infections in the United States are subtype B—some 98 percent according to some surveys, says Glen Hansen, PhD, director of clinical microbiology and molecular diagnostics at Hennepin County Medical Center, a public teaching hospital in Minneapolis, and an assistant professor of pathology and lab medicine, University of Minnesota. These belong to group M (major), which marches through the early part of the alphabet to contain subtypes A (divided into subtypes A1–A4), C, D, F (F1 and F2), G, H, J, and K as well.

A second major group is known as N (nonmajor, nonoutlier). The third group, O, is the outlier group. “O is outlier for a reason,” says Dr. Hansen, noting that there have been only a few confirmed cases—fewer than five—reported from African-born patients residing in the United States. Group O is also rarely found in Europe. It’s less rare, though not common, in Cameroon, where it accounts for one to six percent of HIV-1 infections. Group N is even more obscure: It’s found only in Cameroon, where 10 cases had been reported in the literature as of 2005.

Group M remains the heavyweight throughout the world, including Africa, but only about 12 percent of global infections are caused by the most-studied subtype, B (Hemelaar J, et al. AIDS. 2006; 20[16]: W13–W23). In Moshi, subtypes A, C, and D are all common, as are circulating recombinant forms. (These are what they sound like—subtypes that dually infect and recombine in an individual, then enter into the population.)

Moshi and Minneapolis may be separated by an ocean and plenty of land, but the diversity of HIV-1 subtypes in their patient populations may share some common ground.

Evidence of this at Hennepin County Medical Center was described in the Journal of Medical Virology (Cartwright CP. J Med Virol. 2006; 78: S19– S21). In the past decade, Minnesota, especially the metropolitan Minneapolis-St. Paul area, has seen sizable growth in arrival of African-born individuals, many of whom receive their medical care from the Hennepin medical center. That same period saw a large increase of newly diagnosed HIV-1-infected patients identifying as African-born, rising from four percent in 1996 to nearly 20 percent in 2004.

Something else has also expanded: viral diversity. Among the African-born population in Minnesota, subtype C was the most common, accounting for about 40 percent of cases. This was followed by A and CRF02_AG, with each account­ing for 20 to 25 percent of cases. Some 10 percent of cases were subtype D, while three to five percent were subtype G.

Suddenly Moshi and Minneapolis don’t seem quite so far apart.

Hector Bolivar, MD, assistant professor of medicine, AIDS Clinical Research Unit, University of Miami, Fla., practices in an area with one of the highest rates of HIV infection in the country. It has been, he says, since the earliest days of the epidemic. “This ZIP code [33136] is No. 2 or No. 3 in the whole country in terms of incidence of HIV infections in the community.” The clinic sees patients from all over South and Central America as well as the Caribbean, including Haiti; a small number of patients come from Africa and Europe.

Given the broad background of his patients and the high incidence of HIV infections in Miami, it would not be unusual were Dr. Bolivar to encounter subtype diversity in his practice.

Instead he offers this story: A 25-year-old woman, a recent immigrant from Central America who was known to be HIV positive and was treatment-naive, was referred to Dr. Bolivar’s clinic because of respiratory tract infections; she also suffered from wasting syndrome.

Her CD4 count fit the clinical picture—less than 100 cells/µL. Her viral load didn’t—the level was undetectable, according to a Roche Cobas Monitor v1.5 assay done before her arrival at Dr. Bolivar’s clinic.

Puzzled by the conflicting information, Dr. Bolivar’s lab did another two viral load tests, a branched DNA (bDNA) assay (Siemens) that revealed 90,840 copies/mL and a newly available real-time PCR assay (Abbott) that found 251,189 copies/mL. Further analysis showed the virus to be subtype F; even more sequencing revealed it to be F1. Dr. Bolivar says this has been the only non-B subtype he and his colleagues have detected so far.

There could be several explanations for this. One, there simply may not be much subtype diversity in his patient population—not an impossibility, since much of the immigrant population comes from the Western hemisphere, not Africa, with its well-known subtype stew. This case may be the tip of a pencil, not of an iceberg.

On the other hand, there may be other cases lurking that have simply gone undetected. Dr. Bolivar says he and his colleagues have had similarly suspicious cases (low or undetectable viral loads paired with low CD4 counts and clinical evidence of advancing AIDS) but lacked the funds to do further testing.

Another possibility: Physicians aren’t finding subtypes because they don’t think to look for them. “I learned a lot from this case,” says Dr. Bolivar, who says he and his colleagues were “sort of disoriented for a while, until we started looking at the history of the patient.” Borrowing a page from blood bank screening questionnaires, they are now digging deeper in suspicious cases, trying to find out whether a patient may have become infected in a country where non-B subtypes are the norm.

“Probably a couple of years ago I was more or less like my colleagues, unlikely to consider that a non-B subtype might be involved,” Dr. Bolivar continues. Though primarily a clinician, he says his background in molecular biology has heightened his appreciation of HIV’s astonishing diversity. “But when I mention subtypes to my colleagues, it’s always the same: They don’t believe it’s an issue. They say it could probably happen in Africa, or Asia, but not in one of our clinics.”

“One thing that is interesting, when I mention this, is people realize they may have had cases. They say, ‘Oh, I remember I had a similar case, and the patient was from Africa.’” He’d like clinicians to connect the dots more frequently and consider the possibility of a non-B subtype when lab results don’t tally or don’t match the clinical picture. They need to see non-B subtypes as “less exotic and less strange.”

The experiences in Miami and Minneapolis raise several questions for laboratories and their clinicians. Does the patient population include individuals who are at risk for infection with a non-B strain? Is there a difference between current assay systems with respect to HIV viral load measurements?

If viral load assays were equally effective in measuring non-B strains, question No. 1 might become less important. But as Dr. Bolivar’s experience suggests, answering question No. 2 is not a slam-dunk.

“If we’re relying on systems that don’t accurately quantitate the virus,” says Dr. Hansen, “it’s tough to know whether laboratories are providing clinical information that can be used to guide treatment and monitoring of therapy.”

“If this turns out to be an issue,” he continues, “it would affect our practice here in Minneapolis, because we do see a large number of patients who are African-born.”

If, if, if.

“Even if data come out that show there are significant differences, it may not influence practice throughout the country if you don’t suspect that you have non-North American—that is, non-B subtype—forms of the virus,” Dr. Hansen says, conceding this requires a bit of circular reasoning. “If you’re not looking for them, and the lab doesn’t provide that data on a routine basis, how do you know if you have a non-B subtype?”

Says Dr. Hansen: “It’s interesting to bring up the questions and to make people familiar with what the issues are. Unfortunately, there are no definitive answers yet to the questions people are posing.”

Ditto for comparing viral load assays. It’s a matter “around which reasonable minds disagree,” Dr. Hansen says.

They’ve been disagreeing for some time—in fact, since the earliest days of HIV testing, says David R. Hillyard, MD, professor of pathology at the University of Utah and director of molecular infectious disease testing at ARUP Laboratories, Salt Lake City. He points to a French study exploring viral load discrepancies between two more recent Roche assays (Damond F, et al. J Clin Microbiol. 2007; 45: 3436– 3438), which mentions the manufacturer modified its earlier-generation Cobas Monitor v1.0 assay to accept a broader range of viral diversity.

“The manufacturers take this very seriously,” says Dr. Hillyard. “In some cases they have developed groups that sample viruses throughout the world to prepare for evolving strains to stay ahead of the curve.”

Genetic diversity is a hallmark of HIV-1. Untreated individuals are estimated to produce up to 10 billion variants per day. Because HIV lacks proofreading ability, in theory every single base across the genome can be mutated anywhere from 10,000 to 100,000 times a day, explains John Hackett Jr., PhD, section manager in the AIDS Research and Virus Discovery Group, Abbott Diagnostics Division. “Literally, in a single day, you could have millions of variants within a single individual. So even within an individual there’s drift in the virus. It’s like a swarm of viruses, rather than one virus, that develops even within an individual.”

Again, geography plays a role. The first human infections with HIV probably occurred in Central or West Africa several decades ago. Given HIV’s high mutation rate, it doesn’t take too many years for divergence to occur. As might be expected, says Dr. Crump, the most subtype diversity is seen in areas close to where HIV was first thought to have appeared in humans; farther away from its origins, there’s been less diversity.

Thus Africa is a prime target for exploring genetic diversity. Since the mid-1990s, the Abbott HIV Global Surveillance Program, to name one group, has screened thousands of small-volume specimens to look for strains of interest. The program has characterized more than 150 group O specimens, many, though not all, from Cameroon, Dr. Hackett reports.

“We don’t expect the United States is going to be overrun with group O viruses,” he says. Rather, “It helps us to understand the total diversity, the sequence space that HIV can occupy.” Picking the extreme points represented by group O variants boosts the likelihood of developing assays that can detect the less-extreme variants in between. “We want to find the worst-case scenario,” says Dr. Hackett.

The Abbott group has been approached by its share of physicians seeking answers to puzzling clinical cases, similar to the one that Dr. Bolivar encountered—crashing CD4 counts despite undetectable viremia. These have been of extreme interest from a diagnostic and monitoring standpoint, says Dr. Hackett, providing clues about assay failure and the maximal sequence diversity of the target region.

The subtype diversity in Africa has provided research riches. But it’s not been a source of helpful comparisons between viral load assays—the first has become last, in terms of patient care.

In the United States, arguably first in terms of patient care resources, it’s the lack of subtype diversity, not resources, that has limited head-to-head assay comparisons. “I worry about labs being frightened and afraid of this, when there just hasn’t been enough data collected in North American sites yet,” says Dr. Hansen. “You’re making very large statements about assay performance based on few representative samples within a given clade of the virus.”

This will change, and fairly soon. Sites in North America are beginning to look at this, including Dr. Hansen’s lab, which, with the support of Siemens Diagnostics, Roche Molecular, and Abbott Molecular, will be comparing all three of the companies’ systems on nearly 1,000 individual viral loads of non-B strains from Hennepin County Medical Center. “All three companies have stepped up to the plate, and said, ‘If this is an issue, we want to look at it with you,’” says Dr. Hansen.

There’s a middle ground in all this. It’s called Europe.

That’s where some data have emerged to suggest—but hardly conclude—there may be a difference between current assays’ abilities to quantitate viral load.

Europe is a reasonable place to make such comparisons. “In the U.S., overall, we have a more limited perspective on the issue of non-B strains,” says Dr. Hackett. In France, on the other hand, “even in the mid-’80s they were starting to recognize they had non-B viruses present.” In a decade, he says, the percentage of non-B subtype infections rose from about four percent to 20 percent. In more recent surveillance data, from 2003, nearly 50 percent of newly identified infections were non-B strains. Dr. Hackett also points to the pan-European SPREAD program, which monitors new infections to track the spread of drug-resistant strains and HIV variability. One SPREAD study showed that some 33 percent of newly identified infections in Europe were due to non-B subtypes.

Having broader access to HIV subtypes is one thing; comparing them on viral load assays is another. For now, the data are slim.

David Zhang, MD, PhD, MPH, knows this firsthand. Dr. Zhang, associate professor, Department of Pathology, and director of molecular pathology at Mount Sinai School of Medicine, New York City, says another Journal of Clinical Microbiology paper (Holguín A, et al. Published online July 2, 2008) was a step in the right direction, comparing the Versant bDNA assay, the Cobas AmpliPrep/Cobas TaqMan test, and the Nuclisens test. Though the study identified differences in detection abilities, it’s simply too small (83 patient samples, from a single institution) to act on.

Dr. Zhang might reasonably expect to have uncovered non-B subtype virus in his patient population. He oversees labs that run two different viral load assays (Versant’s bDNA and the Roche Cobas AmpliPrep/Cobas TaqMan). In looking over his data for his interview with CAP TODAY, however, he says he’s seen no evidence of non-B strains, though there’s evidence such strains have found their way into New York City immigrant populations (Lin HH, et al. J Acquir Immune Defic Syndr. 2006; 41: 399– 404).

Dr. Zhang offers two suggestions for this apparent absence. First, he doesn’t do sequencing, which would identify subtypes. Most labs, in fact, don’t sequence. It’s just not the norm in clinical settings, here or abroad, to subtype HIV strains. Not even Dr. Crump, in Moshi, routinely subtypes his HIV isolates. His knowledge of subtype prevalence is based on studies he and others have done, rather than as a result of routine clinical care. Those who do sequence are looking for risk of resistance, so their efforts are focused on polymerase and the reverse transcriptase, not subtypes.

Secondly, Dr. Zhang says, “No physician is calling me saying the viral load I give them doesn’t match the patient’s clinical presentation.” Possibly the clinicians are not following up on discrepant results, although he says he works with a savvy, communicative bunch. “On the other hand, maybe the assays are pretty good.”

The problem isn’t unique to HIV. Dr. Zhang recalls several EBV cases where viral load didn’t match clinical representation. It took a followup PCR test and sequencing of a different region on the virus to discover significant mutations that affected detection and measurement of EBV (Clin Microbiol. 2008;46:2463–2465). “This is just the nature of viruses,” Dr. Zhang says.

And it’s the nature of medicine that physicians practice on different levels.

“Oftentimes, at least here in the U.S., I think a viral load test is looked at like a cholesterol test, that you get an answer and it’s always right,” says Dr. Hackett.

Mix the elusive nature of viruses with the exquisite complexity of molecular assays, and it’s easy to see why some clinicians might fail to grasp the nuances and limitations of viral load monitoring. Dr. Hillyard says he encounters it frequently in his discussions with colleagues, whether the topic is HIV, hepatitis C, or herpesvirus. “Molecular tests are spectacular in their sensitivity and specificity. At the same time, we need to have doctors appreciate the fact that tests are not perfect.”

One known, but perhaps underappreciated, fact, even among laboratorians, is the impact of widespread adoption of real-time PCR. It raises challenges compared to earlier-generation PCR tests, says Dr. Hillyard. In early generations, hybrid probe capture was done under conditions (lower temperatures, lower stringencies) that made them much more tolerant of polymorphisms, says Dr. Hillyard. In real-time PCR, probe capture for signal generation generally is done at higher temperatures and greater stringencies. And though great effort and design expertise have been applied to compensate for this change, “Polymorphisms under the probe, under real time, at least have the potential to confound the real-time assays in ways they wouldn’t have confounded the old assays,” Dr. Hillyard says. “We are in an era when we’re using fundamentally a different biochemical approach to doing the real-time PCR. And it has its own unique vulnerabilities, which may contribute to underquantitation.”

Then there are the less technical issues. If an assay has underquantified HIV viral load, a non-B subtype may be the culprit. Or, it could just be bad sample handling. The virus is unstable, and if a specimen is drawn on a Friday but not delivered to the lab until Monday—a not-uncommon scenario—the virus may be underquantitated. Compliance to medications can also affect viral load measurements; so can the occasional virological “blip.”

So while it may not be unreasonable to suspect the presence of a non-B subtype if viral loads are inexplicably low or undetectable, it’s also reasonable to look at other explanations as well. “That’s why I’m a little hesitant to step out too far on the limb,” says Dr. Hansen. “There can be other things that can attribute to differences or funny results in the viral load that may not be 100 percent attributable to subtype diversity.”

In fact, subtype B—dull, pervasive subtype B—can be found at the heart of at least one troubling case in Dr. Hillyard’s lab.

In a recent survey, ARUP, whose testing distribution covers a variety of sites over all 50 states, found that about 2½ percent of samples submitted for resistance testing—not for viral load—are non-B samples, says Dr. Hillyard. Given the large sample volume from clients who use ARUP for viral load testing and resistance testing, Dr. Hillyard says the 2½ percent figure is probably a reasonably accurate, though not perfect, reflection of subtype diversity in the viral load samples it sees as well. Besides a preponderance of B’s, ARUP sees a half percent of A’s, one percent of C’s, and a half percent of circulating recombinant forms in the samples it tests.

This alone is not what concerns Dr. Hillyard, though he acknowledges subtype diversity cannot be ignored. “Beyond that, though, given the natural biology of HIV, labs need to be concerned about B types also. It’s just the nature of the evolving, ever-mutating HIV that there are sequences that can confound our assays.”

In fact, he says, in his lab’s initial validation of a new viral load assay, “We identified a highly polymorphic sample that was significantly underquantitated—about 500-fold—compared with the previous test.

“It turned out to be a B type.”

ARUP looked at 180 samples. “But I can’t tell you if that’s a one in 200 event or a one in 2 million event,” says Dr. Hillyard. “That’s the key question: Just how common are these events? Because we know that no assay, no matter who makes it, no matter what the chemistry, will be perfect. But what we don’t understand is the frequency of these underquantification events. It’s a hard nut to crack.”

There is one final, rather deflating question to ask: On a practical level, how might this affect clinical practice?

What might be the impact of different subtypes on treatment? “We don’t know!” says Dr. Zhang.

There are scant data, outside of small trials in Europe and sub-Sahara Africa, says Dr. Hansen, to say definitively that the different subtypes have different clinical outcomes.

Dr. Hansen is aware of one study that may suggest that subtype C could progress to disease more quickly than the other subtypes, even with low viral loads. That might say something about the nature of subtype C and its clinical course or its response to medications. Or, consistent with the circular logic that guides much HIV discussion, it may reflect an inability to pick up a non-B subtype, since it’s hard to monitor a treatment’s effectiveness if viral loads are being underquantified. “We may under-treat a patient,” Dr. Zhang acknowledged.

So where does that leave labs?

Who knows?

In the end, says Dr. Zhang, “Everything we talk about is speculation.”