The international treatment access NGO Médecins sans Frontières (MSF) and other advocates say that the development of a cheap, simple viral load test is an urgent priority if the current rollout of drugs to the developing world is not to falter due to first-line treatment failure and HIV drug resistance.

Viral load testing has been part of the standard of care in HIV treatment in the North since 1996, and nowadays the target for all treatment – not just first-line but second, third and fourth-line treatment too – is to suppress viral load below 50 copies/ml and keep it there.

The consequences of not doing so are the failure of first-line HAART, the economic consequences of having to provide more expensive second-line regimens, and, as we have already seen in the north, the seeding of the HIV positive population with large amounts of drug-resistant HIV.

In the developing world the problem has been the expense of current viral-load testing technology. A single viral load test done by one of the four currently available systems costs anything between $20 and $160. Doing two a year would therefore add 50% to 200% to the cost of the cheapest currently-available regimen in the least developed countries.

As a result, the World Health Organization (WHO) did not recommend viral load testing in its 2003 guidelines on the provision of antiretrovirals (ARVs) in resource-limited settings (see reference). Instead they recommended what they called a ‘public health’ approach. This involves rolling out voluntary counselling and testing, standardising and simplifying HAART regimens to a few approved ones, and monitoring CD4 counts when possible. When it comes to treatment failure, an algorithm was recommended as the way to detect virological failure and the necessity to switch to a different regimen. This combines the monitoring of clinical events (i.e. illness) with other indicators such as body weight and with simpler tests that may be available such as total lymphocyte count, haemoglobin count and platelet count.

The 2006 guidelines do say that “Viral load testing may have a role in identifying failure and indicating when to switch in some patients, and WHO advocates wider access to virological testing in tertiary centres and new, simpler virological assays.” However, they add, “Given the continued cost and complexity of the current technology, viral load monitoring is still not suitable for wide use in the public health management of HAART.”

At the Glasgow Interantional Congress on Drug Therapy in HIV Infection last November, the importance of monitoring viral load was, if not exactly dismissed,  downplayed in comparison with getting as many people on first-line therapy as possible by Charlie Gilks, head of HIV treatment and prevention scale up at the WHO.

He defended the public health approach towards treatment scale-up. “Care has to be simplified in order to get it to six million people,” he said.

In response by criticism by French treatment activist Dominique Blanc that clinical monitoring meant people accumulating drug resistance mutations, Gilks said: “This is not second-class care. The second-level standard of care is not to have it at all. We could only treat a few tens of thousands of patients if we insisted on resistance testing.”

He went further than his own guidelines in saying that he personally thought the significance of HIV drug resistance was overrated and that he was confident that the results of developing world trials such as DART in Africa proved that “the public health approach will produce results that are just as good as state-of-the-art care.” Gilks has a point. In the DART trial, only 3% of patients have had to switch therapy after three years.

Nonetheless, the consequences of not monitoring viral load may be dire on both an individual and a population level. This is shown by the results of the trial reported here in which TMC125 (etravirine), a drug that, it was hoped, would be the first non-nucleoside (NNRTI) that would still work after initial NNRTI failure, failed to work in a group of patients who were mainly from developing countries (full report here: http://www.aidsmap.com/en/news/9774EA34-B0AF-441D-B778-898104938086.asp.

The reason for frequent failure appeared to be that patients had spent a long time on a failing NNRTI-containing regimen because treatment failure was identified through clinical monitoring, not viral load monitoring.

As a consequence they had spent sufficiently long on their previous regimen with a detectable viral load to accumulate far more resistance mutations both to the NNRTI drugs and also and crucially to the nucleoside analogues (NRTIs) that would otherwise have supported TMC125. Etravirine may still be useful for patients who fail efavirenz or nevirapine – but only as long as they are switched pretty soon after they become experience viral load rebound. Without viral load testing, it will be useless in the developing world.

Children are also a group who need viral load testing. Because HIV antibody testing is not a reliable indicator of HIV status in the first 12-18 months of life, a viral load test, which can be done as early as 4-6 weeks after birth, is the only way to establish whether a baby is infected. Given that studies from Malawi, Kenya and Uganda have reported a 40-50% mortality rate in HIV-positive children within the first two years of life, many young lives may be lost due to the lack of a cheap viral load test.

These points and others are made in an article urging the necessity of viral load monitoring in resource-limited areas by writers from MSF and the Clinton Foundation alongside several prominent HIV clinicians (Calmy et al.) in the latest issue of Clinical Infectious Diseases.

Need for viral load testing should be revisited

“The question of whether high-quality, effective HIV care can be provided without viral load monitoring needs to be revisited,” they say.

Regular viral load monitoring – perhaps no more than twice a year – may even be cost-effective in the long-term. This is not just because drug resistance on a population level will be minimised but because a failure to achieve undetectability can also indicate poor adherence as well as treatment failure. Detecting patients with adherence difficulties at an early stage can mean that interventions to improve adherence can be targeted at these individuals before they develop resistance and before they need to switch to a more expensive second-line regimen.

Calmy and colleagues highlight the success achieved by just such an approach at Khayelitsha in South Africa which, in 2001, became one of the first developing-world locations for an ARV treatment programme. In an update of the Khayelitsha programme presented at the Glasgow conference (Orrell), over 98% of 823 patients remained on a first-line regimen (any two of d4T, ddI, 3TC or AZT plus either nevirapine or efavirenz) after 32 months (patients who switched between first-line drugs due to side-effects were not classified as viral failures in this analysis). Although 20% of patients had at some time during the 32 months had a viral load over 1000 copies/ml, the number with viral loads over that figure at month 32 was exactly zero, though 14% did have a viral load over 50 copies/ml at this point.

This was achieved with intensive adherence support – and with six-monthly viral load testing. Patients were allocated a personal adherence peer counsellor and had to attend three treatment readiness group sessions before starting ARVs. When a patient showed clear virological failure – and 7.2% did at some point – their treatment was temporarily interrupted and they were given more counselling and attended refresher workshops. Fifty-three per cent of those with viral loads over 1000 copies/ml were re-suppressed to below 50 copies/ml, and none of remaining 47% had viral loads over 1000 copies/ml by month 32.

The frequency of viral load testing is crucial here: only 25% of patients who were left for more than six months without a viral load test and proved, upon testing, to have detectable viral load were able to re-suppress their HIV below 50 copies/ml compared with 71% of patients who had a viral load test within four months of their last one.

So, treatment can be done in a way which maximises the time on first-line therapy – but, in this case, by using viral load monitoring, which rather undermines Gilks’s point.

Clinical markers show limited ability to replace viral load for treatment monitoring

Could the WHO’s recommended algorithm achieve similar results? Not if a retrospective study of the usefulness of such indicators to predict failure in US patients is anything to go by. Shashwatee Bagchi and colleagues from the University of Birmingham, Alabama in an article in the same issue of Clinical Infectious Diseases as Calmy’s Viewpoint article describe a retrospective analysis of 466 patients who had visited the university clinic for care between January 1995 and August 2004 (its lead clinician, Michael Saag, was one of the first doctors in the world to prescribe HAART).

They defined virological failure as a viral load over 50 or over 400 (depending on which year the test was done) at 12 months after the initiation of HAART, and they defined immunological failure as a failure to achieve a rise in CD4 count of less than 50 cells by six months after the initiation of HAART. They then took four of the WHO’s recommended indicators – lymphocyte count, haemoglobin level, platelet count, and body weight – and asked: is there a significant correlation between these results and virological or immunological failure?

The answer was yes – to a degree - in the case of body weight and viral load. Weight gain of more than 10% during the first year of treatment was the only clinical or immunological factor significantly associated with a viral load below 50 copies/ml at one year (p=0.031). However the association between body weight and immunological failure was not significant.

The other tests had no significant predictive value whatsoever, with correlations between test results and virological or immunological failure in the region of zero to nine per cent. Bagchi and colleagues dismiss the relevance of body weight as a useful indicator of therapy failure in itself: “This variable is influenced by many other factors that limit its precision and utility, especially in resource-poor settings,” they say, meaning that there may be other reasons people may fail to gain weight, such as a poor harvest or a higher than usual incidence of digestive disease in the population. Equally people may gain weight despite being ARV failures, for instance if a TB treatment programme is initiated at the same time as HAART.

What sort of viral load test is needed?

However we are left with the problem that there is currently no viral load test that is practicable for use out in the field in a resource-poor setting far from laboratory facilities. Current tests require taking a vial of blood from a vein, refrigerated transport to a laboratory, mains electricity to run equipment, an assortment of scientific paraphernalia, instruments that cost in the region of $30,000-$60,000, about eight hours to produce a result and a technician with advanced training in molecular biology lab techniques.

In early 2006 MSF organised a consultation among academics and workers in ARV rollout programmes to find a specification for an ideal viral load test for resource-poor settings. They decided that it should require no more than a fingerstick’s worth of blood (about a tenth of a millilitre), a single cartridge into which a sample of blood could be inserted and which would give a result, that it should not require refrigeration, be able to be run on batteries, should not cost more than $1000 per instrument and $8 per test, should give a result within two hours and would be able to be done by a field health worker with 1-2 days’ training.

Such a test does not yet exist, though price reductions have been negotiated for the current systems, and reagents that do not require refrigeration have been developed. Costs can be greatly cut by making the test one which will give a simple either/or result to the question of whether the viral load is above a certain level, rather than measuring a continous variable between 50 copies and 1 million copies.

What this level should be is likely to be a matter for debate. Calmy and colleagues point out that evidence from the PLATO Collaboration, an international merging of 13 HIV cohort studies which contributes 2,488 patients experiencing triple-class virologic failure, suggests that, as long as the viral load remains below 10,000 copies, CD4 counts remain stable and the risk of clinical progression is low. So do other cohort studies (see Schechter). “Clearly a qualitative test with a cutoff value of 10,000 copies/mL would be of immediate practical use,” they say.

However avoiding AIDS is one thing and avoiding drug resistance and the failure of any subsequent regimen is another, and evidence both from Khayelitsha and from an older analysis of the same PLATO Collaboration (see Ledergerber) shows that if patients are allowed to maintain a viral load over 1,000 for more than four months, only a minority are able to re-suppress HIV on a new regimen.

Researchers at Cambridge University in the UK (see Dineva) are developing a prototype ‘dipstick’ that would indicate the presence of a significant viral load (over 500 is the lower limit) by the intensity of a single test line. The difference between 1,000 and 10,000 copies may not be all that crucial if the illustration of the dipstick results accompanying Calmy’s article are what it will eventually look like: whereas viral loads under 500 produce a faint line and over 10,000 a strong one, viral loads in between produce a line whose fuzziness is a matter of interpretation, so perhaps the indication should be that “any line that isn’t faint means you switch”. Dineva and colleagues say that the dipstick test can produce a viral load result “with an efficiency similar to that of a complex, expensive, and instrument-dependent method.”

What of the WHO and its public health approach? The organisation is wary of any tendency to “make the best be the enemy of the good” and would continue to oppose any demand that ARV programmes not be instituted where viral load testing is unavailable. But the organisation has shifted its position on other guidelines. At Glasgow Gilks said that single-dose nevirapine to prevent mother-to-baby transmission, another strategy supported by WHO in the past, “is now substandard”, and the WHO has started to shift its position on d4T, saying that “some governments may want to choose alternative options” due to this drug’s side effects.

However some are wishing that the WHO would move faster and would explicitly echo the same sense of urgency, before health systems in the developing world are swamped by tens of thousands of patients with NNRTI and thymidine-analogue resistance mutations. In an editorial in the same issue of Clinical Infectious Diseases Robert T Schooley of the Univeristy of California, San Diego compares the current situation to the old joke where a man jumps off a 10-floor building and is asked by someone in the fifth floor how he is as he falls past. “Doing fine so far,” is the reply. “We must avoid invoking the ‘fifth-floor syndrome’ by failing to act to prevent what we know will happen if viral load testing is not expanded in concert with the access to antiretroviral drugs,” is Schooley’s comment.