A leading US researcher has criticised the routine adoption of resistance testing, despite international guidelines published yesterday in the Journal of the American Medical Association.
In an accompanying editorial in the Journal, Dr Charles Flexner of Johns Hopkins University School of Medicine warns against the widespread use of resistance testing in clinical practice given the ambiguities in available data.
"Adequate evidence [of the magnitude of benefit] may never exist if resistance testing becomes standard practice now", he warns.
He cites a wide range of problems with interpretation of results, including the following issues:
- Resistance assays are not reliable in patients with viral load below 500 to 1,000 copies, yet treatment failure is commonly considered to have occurred if viral load rebounds above 50 copies.
- There is still a lack of data to allow the prediction of average drug sensitivities from studies which have conducted both genotypic and phenotypic resistance testing, and Dr Flexner points out that in a gene with 99 codons (HIV protease), the potential for variations in phenotypic sensitivity based on the variety of possible genotypic changes leads to an astronomical number of permutations.
- Results from studies such as VIRADAPT, GART and VIRA3001 may not be generalisable to patients with very low viral load
- The apparently positive results of the GART study may have been the consequence of expert analysis of resistance data and treatment history rather than resistance testing.
- Some doctors may consider it impractical or unethical to keep patients on failing therapy solely for the purpose of obtaining a blood sample for resistance testing (for example, where results of viral load tests are given by phone some time prior to the next clinic visit, and where a prescription can be changed without a clinic visit. This may be a less relevant concern in the UK).
- It is difficult to exclude with confidence factors such as poor adherence or inadequate drug concentrations (especially if these are intermittent), and this may lead to the inappropriate use of resistance testing without addressing these underlying problems.
- Just because a mother shows evidence of genotypic resistance to a drug, this doesn't mean that this drug cannot prevent mother-to-baby transmission, Flexner claims. Drugs with proven safety records should be preferred over unproven, novel combinations, he argues.
- It may be more cost-effective to invest money in improving adherence. "For some patients with secondary treatment failure [requiring a succession of genotypic or phenotypic tests], it seems that those same monies might be more effectively spent, for example, on directly observed therapy managed by a visiting home nurse", Dr Flexner remarks.
aidsmap asked Professor Clive Loveday, resistance expert at the Royal Free Hospital Medical School in London, to comment on the criticisms raised in the editorial from a UK perspective.
"I think it's an amazing achievement that we have moved in five years to be able to offer resistance testing for patients. But it's always the same for HIV - we are always being forced to use technologies to support patient care before all the data is in" he said.
"Turnaround time is slow because the systems are still run on an experimental basis, but this needn't be a problem in the future. For example, we can already offer a 96 hour turnaround for genotypic results in the event of needlestick injuries. The post-exposure prophylaxis can be altered as soon as we get the results of the genotypic test. We're starting a project with Visible Genetics to develop algorithms for high throughput, real time quality assured resistance testing.
"Regarding the question of detecting resistance shortly after a rebound of viral load above 50 copies, I've always seen that as a vulnerability of resistance testing, but I don't know how much of a problem that is in clinical practice in the UK. What currently happens is that you have to do a confirmatory at recall or follow-up, by which time viral load may have risen considerably".
When will we begin to see the improvements in reliability that Dr Flexner thinks are necessary before resistance testing can be used routinely?
"Reliability will improve as we use the assays more frequently. My best guess is that we will soon be able to get assays that can detect mutants at a lower viral load - for instance, Rob Lloyd of Visible Genetics has already shown that it is possible to detect mutants when viral load is as low as 150 copies. But I agree that it will always be very difficult to detect minority strains in different compartments. If you test the same sample ten times you will get small variations, so if you have ten compartments, for example, you may need to do 100 sequences in order to detect minority strains. It's just impractical in a clinical situation - there will always be limitations to technologies".
"We need to address all these issues within the context of high input clinical virology, not experimental settings".
What about the charge that early adoption of resistance testing could sabotage the gathering of useful data from randomised clinical trials?
"We already have three randomised clinical trials which have demonstrated significance. I know of ten randomised trials that are taking place already around the world, including ERA in the UK, NARVAL in France, Vigilance 2 in the USA, for example". Professor Loveday suggests that as testing and the data evolves, the use of resistance testing will be refined.
"If this was hypertension we would write a clinical trial protocol that would last five years and recruit 10,000 people and then we would have our answer, and that would never be tolerated in HIV. Community pressure and physician pressure are a lot more radical in HIV than in any other area of medicine. We did exactly the same with CD4 counting and viral load testing".
Nevertheless, Charles Flexner clearly thinks the results of studies conducted so far are equivocal. Are you more confident that they support resistance testing?
"These are the first studies, in patients with advanced disease; nevertheless they found a significant difference. I think he downplays the VIRADAPT study; he says the genotypic arm had a slightly better mean viral load decrease, but this difference was statistically significant."
"No tests are perfect, but added to the clinical history and other laboratory measures, resistance testing will improve the selection of next therapies".
Reference
Flexner C. HIV genotype and phenotype - arresting resistance? Journal of the American Medical Association 283 (10): 2442-44, 2000.