1. Do you think that genotypic tests are particularly useful when looking for (or ruling out) a single mutation that confers resistance? (i.e. when deciding about taking 3TC, Sustiva, Viramune, Rescriptor, Viracept)--because if you've got one of a mutation for any of these drugs, you should not take the drug? And it's much harder to interpret, say, TAMS or the six protease mutations?

2. Would it be accurate to say that pheontypic tests are particularly useful where clinical cutoffs have been established in relation to fold-resistance (and less useful where such cutoffs have not been established)? If so, what should we make of IAS-USA (Hirsch, M., et al CID 2003: 37:117):

"For example, data from the NARVAL trial show a continuous inverse relationship between fold-resistance and response rate for saquinavir and efavirenz (i.e. the higher the fold-resistance, the lower the rate of viral suppression); thresholds above which no response was observed were noted for stavudine, didanosine, abacavir, and amprenavir [cite Obry, Castigliola, Race 2001]. No correlation between fold-resistance and treatment response was observed, however, for zidovudine, lamivudine, nelfinavir, and nevirapine."

3. Thus could we say that phenotypic testing IS particularly useful for assessing whether to stop Invirase/Fortovase, Sustiva, Zerit, Videx, Ziagen, and Agenerase?

By the way did the NARVAL trial use certain genotypic tests and were their results indicative of these specific tests? Do genotypic tests manufactured by different companies vary dramatically in their results? If so, what would you recommend an informed patient do? Also if most docs have trouble interpreting resistance tests, what is an informed patient to do? How can a patient tell if their doc is informed or not? Is their some sort of second party the test can be sent to for interpretation?

And could we say that phenotypic testing is NOT useful for deciding whether to stop (or not initiate) AZT, 3TC, Viracept, or Viramune?

The same source says clinical cutoffs have been established for "relatively few drugs (e.g. abacavir, tenofovir and lopinavir-ritonavir) [cite Lanier 2001 for abacavir/phenosense; Kempf 2000 for lopinavir (unsure if ritonavir was included); Miller 2001 for tenofovir]

Thus could we add to the list where phenotypic testing IS particularly useful the following: Ziagen, Viread, Kaletra?

And are there more recent studies that establish or rule out fold-resistance cutoffs for other drugs?

Finally, does the virtual phenotype get you the "best of both worlds"? First you look for mutations--so you get clear answers where a single mutation confers resistance. (Thus, for example, you would spot a mutation that rules out 3TC, where phenotypic testing could not tell you to stop 3TC.) Then, the database does the phenotyping and it can tell you to rule out certain other drugs for which phenotypic tests can establish cutoffs though genotypic tests are harder to interpret (say, Ziagen, Agenerase or Kaletra)? 4. So would it be appropriate to say that virtual phenotyping is good BOTH for all the drugs about which genotypic tests are especially dispostivie, AND for all the drugs about which phenotypic tests are especially dispositive????

5. Also how much say does a patient actually have in determining which test their doc uses? Can a patient try to tell their doctor which resistance test they should use?

I'll take your questions in order:

1. The genotype test is useful for both of the purposed that you describe, i.e. to identify single mutations associated with drug resistance, such as the M184V or the K103 N for 3tc or NNRTI resistance; and to identify mutations that cumulatively are associated with resistance to the NRTIs, such as the TAMS, and the PIs. The latter are harder to interpret, as you suggest, but no less important than the former.

2. It is true that the most clearly defined clinical cut-offs offer the cleariest use of phenotype tests, e.g. for abacavir and lopinavir/r, and that there are limitation for other drugs, such as D4T and DDI, where the cut-ffs are quite close to the range of wild type virus.

As the quote from the IAS Resistance guidelines, there are still uncertainties with other drugs.

3. No. There is still value in the results from pheotype tests with all of the available agents. Some on your list are still difficult, eg DDI and D4T, and others are well-defined that are not on your list, e.g. lopinavir/r. Often phenotypes are best interpreted when accompanied by a genotype, as well as an accurate treatment and response history.

THere is considerable variation in genotype test results, as shown by several studies of quality review in the US and Europe, as much as 20-30% of tests showing variable results, but this has not been linked as much to specific labs as the underlying variablity of test.

Strategies for patients and their doctors in light of these uncertainties include obtaining multiple tests, to balance the results against each other and obtaining a second opinion from a physician with expertise in genotype interpretation. I discourage patients from seeking independent second opinions out of concern that all of the key information needed to interpret these complex results may not be available or communicated accurately. It is far better for a patient to work through this information with their clincian, rather than independently.

As I noted above, I agree that the clinically relevant cut-offs have been well defined for ziagen, viread, and kaletra. Other cut-offs are being defined in current studies as well.

Unfortunately, the virtual phenotype, which is in fact a genotype with an extensive data base of matched phenotypes and genotypes, does not answer all questions either. It can often be quite useful, and superior to a genotype alone, but the results offer a clinician a likelihood of susceptibility, e.g. 30% chance of resistance to drug X based on the genotype and data base of matched phenotypes. While this offers some additional information, it is not specific enough in some cases to lead to a decision, e.g. a 40-60% likelihood of resistance. It does have several advantages, i.e. lower cost, rapid turn-around, and the large database behind it...but it cannot be held up as the gold standard. There simply isn't a single best resistance test or interpretive algorithm at this point. Thus an experienced clinician to aid in the interpretation of ALL of the data - resistance test result and treatment history and response - is the best practice.