Viral load tests - reducing the cost

The best indicators so far discovered for the degree of damage to the immune system from HIV and for the risk of clinical progression in AIDS are CD4 counts and HIV viral load. This was shown clearly by Mellors, in a paper first presented at the International Conference on AIDS in Vancouver the previous year, from analysis of stored blood samples taken during the Multicenter AIDS Cohort Study (MACS).

The best and clearest way to show that ARVs are working to control HIV is through monitoring viral load before and during treatment. CD4 counts should also rise, at first very rapidly (mainly through redistribution of the cells between the lymph nodes and other parts of the body) and then slowly and steadily (as numbers are regained, in much the same way that this happens after a bone marrow transplant).

The current best available standard of care includes the use of viral load tests to monitor response to treatment. A baseline test before treatment is started is needed for this, although there is some argument about whether a baseline test can or should be used in deciding whether to treat or with what.

When CD4 counts are on the borderline for treatment, a high viral load might be taken as a reason to go ahead and a very low viral load as a reason to wait.

There is some suggestion that treatment with triple nucleoside analogues, such as abacavir, AZT and 3TC, may be less effective when initial viral load is above 100,000 copies/ml. Some physicians have argued for treatment with combinations including boosted protease inhibitors or efavirenz when viral load is very high.

NAM has produced an information booklet for people with HIV on viral load which can be downloaded here.

There is currently no really low-cost viral load monitoring system. While existing systems have successfully been established in Brazil, they are generally considered too expensive and difficult to operate for large-scale regular use in most resource-limited settings.

The French National AIDS Research Agency (ANRS) and HIV-NAT in Thailand (among others) are collaborating on developing a semi-quantitative version of viral load testing, with the aim of getting the cost per test down to around US$10-$20 by the end of 2002.

Cote d'Ivoire's drug access initiative initially provided viral load tests at baseline and every three months. An evaluation concluded that CD4 counts were sufficient as a basis to decide on starting treatment: very few extra patients were treated on the grounds of having high viral load. Similarly, during treatment, it was decided that viral load tests could only be justified when CD4 was declining rather than increasing. Restricting viral load tests in this way may be essential to maximise the coverage of such programmes as they are expanded (Diomande).

HIV diversity and viral load

One issue that arises with viral load tests, because they are based on using primers chosen to match a part of the HIV viral sequence, is that some strains of the virus may be easier to detect than others. Early commercial viral load tests were effectively tests for levels of HIV-1 subtype B. While this problem has largely been overcome by careful selection of (more than one) primer, tested against HIV from various subtypes, it is still an issue that HIV-2 tests have not been commercially available.

A US-Zambian collaborative study applied four commercial assays (Amplicor 1.0 and 1.5 from Roche, Quantiplex from Chiron, and NASBA HIV-1 RNA QT from Organon Teknika) to samples from people living with HIV-1 subtype C, in a Zambian clinic population. The researchers were most interested in whether Roche's Amplicor 1.5 (which has more primers than version 1.0) was significantly better than the other tests. In this case, they concluded that any of the four tests could be used and while Amplicor 1.5 gave somewhat higher readings, this would not matter when it came to monitoring the effectiveness of therapy. The results might, however, be different if the tests had been run on another population in which different subtypes were predominant (Hoesley).

A Thai study of people identified as having HIV subtype E (strictly, recombinant AE) viruses has found that the relationship between viral load, CD4 and disease progression is closely similar to that previously reported with subtype B virus in US populations (Costello).

Quality assurance for viral monitoring

As with CD4 tests, quality assurance is vital either with viral load tests or any alternatives that may be adopted.

Brazil's AIDS care programme has adopted viral load testing and CD4 counting as part of the standard of care, setting up a network of 65 labs carrying out 220,000 viral load tests a year using the NASBA method. Ensuring consistent quality across that volume of work has been a challenge, with an external quality control programme that distributes blinded plasma samples at various known dilutions, to participating laboratories. Six rounds of testing were carried out between 1997 and 2001 by the National STD and AIDS Programme Office, which aims to distribute test samples to every laboratory, twice a year. This has picked up poor procedures, leading to 4.8% of labs being 'failed' for contaminating an HIV negative sample with HIV and reporting a viral load reading from it. Other labs were failed for getting results more than 2 standard deviations away from reference (mean) values (Vilela).

When viral load tests are not affordable

One approach, as with CD4 tests, is to make viral load tests more affordable, through developing new technologies. However, all PCR-based tests depend on a high level of skill and excellent technique on the part of their operators, especially to avoid contamination of samples.

The other approach is to look at alternative ways of measuring the amount of the virus active in blood samples.

Two methods that are being studied for this purpose are the PerkinElmer Life Sciences p24 antigen test and the Cavidi Tech reverse transcriptase assay.

Both of these have the advantage that they can be carried out in any laboratory that can carry out an ELISA antibody test. They do not require the specially controlled conditions (and large amount of lab space) needed for viral load tests. They are also substantially cheaper than viral load tests.

p24 antigen as an alternative to viral load

In principle, it is possible to measure the production of the viral protein p24 as an alternative to viral load.

As described above in Diagnosing HIV infection there is now a commercial test kit for p24 available from PerkinElmer Life Sciences, based on using heat treatment to separate antibodies which would otherwise mask p24 in the blood of people with HIV.

Viral load is based on detecting viral particles, each of which contains two strands of RNA. The p24 test is detecting a protein that is made inside infected cells and may be released from those cells in various ways, even if they are not producing viral particles. The results of p24 tests can be as consistent, or even more consistent than, viral load tests; but they are not quite measuring the same thing.

Which of these tests is more relevant to what is happening in a person's body when they are on treatment is a question that can only be answered through clinical studies. Those who are researching this issue stress the need to go back to first principles rather than assume that it is necessary to show equivalence for p24 with viral load.

In some circumstances, it may be that p24 is even more informative than viral load. 329 samples were taken from 55 people with HIV followed for a median of 504 days, whose plasma viral load had fallen below 50 copies/mm³. p24 decreased steadily in most patients while a significant decrease of HIV-1 RNA in plasma or peripheral blood monocytes was not seen. Lower p24 levels correlated with increasing CD4 counts (Schupbach).

In a large US study, CD4 counts (on fresh blood), RNA viral load tests and heat-denatured p24 antigen tests (both on frozen blood samples) have been compared for their ability to predict clinical disease progression. Stored samples and clinical data were available for 494 predominantly African-American adult injecting drug users with HIV, in a cohort followed for 5 years. While the best prediction was given by a combination of RNA viral load and CD4 count, a combination of p24 antigen and CD4 count gave almost equally good results. If only one test were available, then the p24 test would be as useful as either of the others. The two most substantial advantages of the p24 test are given as its lower cost (including all of the incidental costs of running the lab where it is carried out, as well as reagent costs) and the fact that samples can be transported further, in more difficult conditions, without affecting the reliability of the test (Sterling)

Work has been done at the US Centers for Disease Control to compare p24 and viral load test results in 39 newly-diagnosed people with HIV subtype B. This showed high levels of correlation, using the commercial test kit developed by PerkinElmer. Further work is still needed, and is being carried out, to validate the test in populations with non-subtype-B virus (Respess).

Researchers in Cote d'Ivoire found a strong correlation between p24 antigen decline and rebound and viral load decline and rebound (Schupbach 2001, Nkongasong 2002).

Measuring reverse transcriptase activity

ExaVir Load is a system manufactured by the Swedish company Cavidi Tech, which measures levels of reverse transcriptase, the enzyme that defines retroviruses including HIV-1^[1].

The system works by measuring the activity of the reverse transcriptase enzyme extracted from a sample of plasma using a separation gel and a system of buffer solutions. The reverse transcriptase catalyses the copying of an RNA template into DNA, which is labelled with the artificial nucleotide BrdUTP. An antibody-based system is then used to label the DNA, giving a yellow colour that is proportional to the amount of reverse transcriptase in the sample.

This test effectively measures the number of retrovirus particles in blood plasma, so it provides a closer equivalent to viral load testing than p24. It is much cheaper than RNA viral load testing, with each measurement costing around $7. In contrast, traditional viral load testing costs aroun $30 a test.

The biggest limitation is that it seems to have a lower limit of detection of 400 copies/ml. It also requires a relatively large sample: 1ml of plasma. While this may be a problem in monitoring treatment in babies, it can be solved by adding 0.1ml of the baby's plasma to 0.9ml of plasma known to be HIV negative. As viral load in HIV-infected babies is generally very high, this should not reduce the sensitivity of the test. Cavidi does not recommend that the test be used to diagnose HIV infection.

ExaVir Load involves a two-night incubation stage, which means it would require reasonably well equipped laboratories, and its interpretation requires the use of a computer. However, the complexity of this test is closer to ELISA tests than to PCR, as is the laboratory space needed for it.

Another practical requirement for the test kit is that it needs to be stored in a freezer before use, as does the plasma sample. Cavidi guarantees the stability of the kit for at least a year if it is kept frozen.

The test works equally well for HIV-1, regardless of subtype, and for HIV-2. However, it cannot distinguish between them. It is not affected by coinfection with HTLV-1 or HTLV-II, other human retroviruses which are present in some populations affected by HIV.

Two versions of this test have been produced. The second, which was made available in 2005, demonstrated tenfold increased sensitivity over the original version, through improved buffer solutions and a longer reverse transcriptase reaction time. One study found a lower limit of detection of 170 copies/ml, which could be further increased by using fluorescent detection methods^[2]. However, the use of fluorescent detection requires more complex and expensive equipment for the analysis of reverse transcriptase levels.

A number of studies have evaluated the use of the latest version of this test (version 2.0), both in labortaory settings and in the field. These have found that the test gives comparable results to the more expensive viral load tests, such as the Roche Diagnostics RT-PCR test, and the Bayer Diagnostics bDNA test. One study found 89% agreement between the ExaVir Load and standard tests, with no effect of patients having taking efavirenz on the test's accuracy^[3]. Similarly, a test using spiked plasma samples found accuracy rates between 84 and 99% with the newest version of the test^[4].

Studies in Burkina Faso and in Ethiopia have also shown that the test is feasible for the monitoring of treatment in resource-limited settings, showing that reverse transcriptase levels were detectable in over 85% of patients' samples^[5][6]. It is, however, less sensitive at lower viral load levels, although this may not be a serious concern for the monitoring of treatment effectiveness in resource-limited settings.

A further option that has become available using this method, is a low cost phenotype test for viruses that are resistant to non-nucleoside reverse transcriptase inhibitors (NNRTIs) or the thymidine analogues AZT and d4T. This kit, marketed as ExaVir Drug uses similar technology to the viral load test described above. However, by adding a range of small measured amounts of the appropriate NNRTI or thymidine analogue to the test system, it is possible to observe the degree to which reverse transcriptase activity is inhibited by the drug. This gives an indication of how resistant the patient's reverse transcriptase is to the drug in question^[7]. This system has been shown to have good agreement with standard phenotypic and genotypic resistance tests^[8].

The clinical value of this resistance test is affected by the timing and circumstances in which the blood sample was taken, in exactly the same way as other viral resistance tests are affected. It would be unlikely to detect resistant viruses where these were only a minority of the viruses in a person's blood.

Given that NNRTIs are likely to be part of first-line antiretroviral treatment in many countries, it may be especially useful to identify people who are initially infected with NNRTI-resistant viruses in order that they are put straight onto second-line treatments without giving their virus an opportunity to become resistant to nucleoside analogues too.

Beta 2 microglobulin and neopterin

Beta 2 microglobulin is a small protein which forms a constant part of the otherwise variable HLA Class I antigens which are expressed on the surface of most cells in the body including, for example, CD4 cells. Thus, raised levels of beta 2 microglobulin which are observed in people with HIV may reflect high rates of destruction of HIV-infected cells, corresponding to rates of virus production, corresponding to viral loads.

Of particular interest are reports that raised beta 2 microglobulin levels in cerebrospinal fluid correlate with AIDS dementia and that these fall with use of AZT. However, this is not a practical test for widespread use.

Blood testing for beta 2 microglobulin is antibody-based and is much cheaper than viral load testing. Unfortunately, studies which attempted to validate blood levels of beta 2 microglobulin as markers for progression risk in AIDS found no relationship, whereas low CD4 levels and high viral load did correlate with progression risk. Furthermore, beta 2 microglobulin levels do not decline with antiviral treatment as seen with p24 antigen levels, let alone viral load.

It is also possible that beta 2 microglobulin levels may be raised for reasons which have nothing to do with HIV.

For example, there are elevated levels of beta 2 microglobulin in active hepatitis C disease and also in auto-immune and inflammatory diseases, such as rheumatoid arthritis, in which cells are destroyed by the immune system.

For all these reasons, beta 2 microglobulin seems unlikely to be a satisfactory alternative to viral load for monitoring the success of ARV treatment for HIV (O'Brien).

Neopterin is a similar non-specific marker of immune activation.

A commentary by Richard Chaisson MD in 1997 concluded: Both beta2-microglobulin and neopterin are nonspecific markers of immune activation that become elevated as HIV disease progresses. Although higher serum concentrations of these markers predict more rapid progression, development of clinically applicable algorithms for their use has not been possible, and they are of little value in monitoring patients receiving therapy.

The need and potential for new technologies

Ideally, for viral load as well as for other aspects of diagnosis and monitoring of treatment, radically new approaches are needed. The development of dipstick antibody tests is an example of the level to which systems should be taken, if possible.

A series of proof-of-principle experiments have been carried out using a microchip-based system in which reagents coated on a single bead placed in a very small cell form the basis for a range of different assays (Rodriguez).

Miniaturising tests can make them cheaper, since the quantities of biological reagents used in the tests is kept to a minimum. There is a limit to this, if precise measurement of low levels of a protein or small numbers of cells in a sample is needed. But if all we need to know is that a value is below a threshold - such as 200 cells per mm³ - then the sample size can be smaller than is normally used now.

A battery of different tests can be assembled in a single chip, with the idea that when one sample is prepared and exposed to the chip, several diagnoses or measurements can be made at once. These chips can be mass-manufactured at unit costs of a few US cents, and test results can be read using automated systems based on digital camera technology which implies costs in the range of hundreds rather than thousands of US dollars.

This system can be used to detect antibodies to HIV and a range of other infections and to detect HIV p24 antigens as well as CD4+ cells. It all depends on the choice of reagents attached to the beads.

The system can be used to detect HIV RNA using 'molecular beacon' methods where a single looped strand of RNA parts when it meets a matching strand, separating two molecules attached to the ends of the loop, that then become visible. However, this RNA test is unlikely to detect the amounts of HIV RNA normally present in blood samples. This means that 'amplification' which means operator training, equipment, time and lab space - is likely to remain essential.

Microchip technology of this kind is subject to a patent held by a company that has not yet been persuaded to license it for widespread use in low-resource settings. Before this could happen, there would need to be major investment in manufacturing plant and probably in better distribution systems for fragile devices that could deteriorate if badly stored (as has been reported, on occasion, with 'rapid simple' HIV antibody tests).

References

Braun J et al. A new quantitative HIV load assay based on plasma virion reverse transcriptase activity for the different types, groups and subtypes. AIDS 17:331-336, 2003.

Costello C et al. CD4 count and viral load-strong, independent predictors of time to death in clade E HIV-infected Thais. XIV International AIDS Conference, Barcelona, abstract WePeC6065, 2002.

Diomande F et al. Towards simplified laboratory monitoring of patients on antiretroviral treatment (ARV): experience from Cote d'Ivoire's drug access initiative (DAI). XIV International AIDS Conference, Barcelona, abstract WePeF6694, 2002.

Downing R et al. Evaluating the Cavidi reverse transcriptase assay to monitor response to anti-retroviral therapy. XIV International AIDS Conference, Barcelona, abstract MoPeB3106, 2002.

Hoesley CJ et al. Comparative analysis of commercial assays for the detection and quantification of human immunodeficiency virus type 1 (HIV-1) RNA in plasma from patients infected with HIV-1 subtype C. Clinical Infectious Diseases 35:323-325, 2002.

Mellors JW et al. Plasma viral load and CD4+ lymphocytes as prognostic markers of HIV-1 infection. Annals of Internal Medicine 126:946-954, 1997.

Nkengasong J et al. Laboratory monitoring of antiretroviral therapy in developing countries. Ninth Conference on Retroviruses and Opportunistic Infections, Seattle, session 30, presentation S27, 2002.

OBrien WA et al. Changes in plasma HIV-1 RNA and CD4+ lymphocyte counts and the risk of progression to AIDS. Veterans Affairs Cooperative Study Group on AIDS New England Journal of Medicine, 334: 426-31, 1996 [showed beta 2 microglobulin levels did not correlate well with progression risk].

Respess R et al. Evaluation of an ultrasensitive p24 antigen assay (UPTA) as a possible surrogate marker of HIV-1 RNA in resource-poor settings. XIV International AIDS Conference, Barcelona, abstract WeOrB1341, 2002.

Rodriguez WR et al. Development of affordable and portable HIV and CD4 diagnostic tests using microchips. XIV International AIDS Conference, Barcelona, abstract WeOrB1343, 2002.

Schø¢¡£¨ J et al. Antiretroviral treatment monitoring with an improved HIV-1 p24 antigen test: an inexpensive alternative to tests for viral RNA. J Med Virol 65(2):225-32, 2001.

Schø¢¡£¨ J et al. HIV-1 p24 antigen, an inexpensive alternative viral load marker, is an independent correlate of CD4 T-cell changes in patients on successful long-term antiretroviral therapy. XIV International AIDS Conference, Barcelona, abstract WeOrB1340, 2002.

Sterling TR et al. Heat-denatured Human Immunodeficiency Virus Type 1 protein 24 antigen: prognostic value in adults with early-stage disease. Journal of Infectious Diseases 186:1181-1185, 2002. [Previously presented, in part, at the XIV International Conference on AIDS in Barcelona, abstract MoPeB3098, 2002.]

Vilela WT et al. Establishment of a programme of external quality control in the public viral load laboratories in Brazil. XIV International AIDS Conference, Barcelona, published abstract F12276, 2002.