As we reported in May, the results of six prospective, randomized studies that evaluated short-term virologic responses in subjects receiving a genotypic or phenotypic resistance test - compared to those who did not receive a resistance test - have been published (see related article). Initial studies compared one technology (genotype or phenotype) with no test in order to assess the clinical utility of resistance testing in general. The second round of studies compared the two testing strategies with each other. The contribution of expert advise to improved virologic outcomes has also been evaluated. Ongoing studies are comparing different interpretation approaches for genotype testing, including different rules-based algorithms and comparative database systems (e.g., Virtual Phenotype^TM); results of these studies have yet to be published.

Five studies presented at the XIV International AIDS Conference in Barcelona address similar or related questions - questions that are important for understanding the utility and limitations of resistance testing in clinical practice. In addition to determining the clinical utility of these tests with regard to short-term virologic responses, it is important to address the longer-term clinical utility of resistance testing, and how well different interpretation systems work. New findings regarding this latter issue are discussed in a separate article, entitled "How do the Experts Do in Interpreting Genotypes?"

Finally, it is very important to know how well laboratories can perform such tests. We will review findings from external proficiency testing programs/studies in our next article.

Genotypic Resistance Testing: Narval Update
We described the Narval study previously (see related article); an updated multivariable logistic regression model of predictors of virologic response or failure at 12 weeks was presented at the conference [1]. Of 541 randomized patients, 39% achieved suppression of HIV RNA levels to <200 copies/mL. Among the factors that were significantly associated with virologic success include prescription of efavirenz in NNRTI-naive patients (OR 4.37) and randomization to the genotype group (OR 2.13).

The importance of using an NNRTI in salvage regimens in NNRTI-naive patients is a recurring and very important theme. Not surprisingly, NNRTI use appears to play an even more significant role in virologic success than does access to a genotypic resistance test. Nevertheless, this analysis does suggest that there is an additional benefit - beyond reminding clinicians to remember to include an NNRTI - afforded by the resistance test itself. Factors associated with lack of virologic response included prescription of nelfinavir in this highly PI-experienced population (OR 0.30), higher baseline viral load (OR 0.37), >/= 5 PI-resistance mutations (OR 0.42), >/= 3 nucleoside analog mutations (NAMs; OR 0.61) and PI exposure > 30 months (OR 0.64).

In summary, the virologic benefit associated with the genotype test appears modest compared to the variables related to the extent of ARV experience and the new drugs selected.

Genotypic vs. Phenotypic Resistance Testing: VIHRES and CERT
A second randomized study comparing the short-term virologic effects of genotypic vs. phenotypic resistance testing was also presented in poster form (VIHRES Study) [2]. One hundred forty-four patients who had failed > 2 HAART regimens and had HIV RNA levels > 5,000 copies/mL were randomized to receive a genotype or phenotype, with results of both assessed by a committee of experts, prior to modification of their ARV regimen. At 24 weeks, the proportions of subjects in the genotype and phenotype arms with HIV RNA < 200 copies/mL were not statistically significantly different (50% and 40%, respectively; P = 0.48). This finding is different from that of the Narval study, where the genotype arm was superior to the phenotype arm. It also differs in that a no-resistance test control was not included.

Finally, the CERT trial compared the outcomes of patients randomized to receive a genotypic resistance test, a phenotypic resistance test, or no resistance test prior to modifications in ARV therapy [3]. No expert guidance was provided. The outcome variable in this study - time to persistent virologic failure despite a change in ARV therapy - was different than in all previous studies. Virologic failure was defined as less than 1.0 log10 copies/mL reduction in HIV RNA by 4 weeks, failure to achieve HIV RNA of less than 200 copies/mL 4 -6 weeks after ARV therapy change, an increase in HIV RNA to detectable levels in subjects with previously undetectable HIV RNA, or greater than 0.5 log10 copies/mL increase from nadir HIV RNA level. If subsequent ARV changes resolved the virologic failure, the subject remained on the study without reaching an endpoint.

Among the 450 study subjects, the median CD4 count was 471 cells/µL and the median HIV RNA level was 2.76 log10 copies/mL. The median follow up was 525 days. Subjects in both the genotype and phenotype arms had a delayed time to virologic failure compared to the no resistance test arm (574 days for genotype, 521 days for phenotype and 478 days for no test); there was no difference in this respect between the genotype and phenotype groups.

Of note, the genotype interpretation methodology changed over the course of the study, which complicates the interpretation of this study to some extent. Genotypes were initially interpreted with the commercially supplied rules-based algorithm employed by Virco, and then with Virco's Virtual Phenotype^TM when it became commercially available. Thus, it is not clear if this study should be considered a comparison of genotype vs. phenotype testing, or of Virtual Phenotype vs. phenotype testing.

Conclusions
The CERT study suggests that the strategy of resistance test-guided modifications in ARV therapy provides more durable virologic suppression than changes made without such guidance. In contrast to the Narval study, CERT and VIHRES suggest that phenotype testing offers a similar advantage to genotype testing. These clinical advantages exist despite a number of important limitations, not the least of which are the significant questions regarding sequence interpretation (see related article).

In our next article, we will address limitations in laboratory performance of sequencing, the ability to detect mixtures of resistant and wild-type HIV, and the adequacy of interpretations provided by laboratories participating in external quality assurance programs.

References

M Vray, JL Meynard, C Dalban, L Morand-Joubert, F Brun-Vezinet, F Clavel, D Costagliola, PM Girard. Multivariate logistic regression analysis of factors predictive of the virological response in the Narval trial. XIV International AIDS Conference. 7-12 July 2002, Barcelona, Spain. Abstract ThOrB1387.
JL Blanco, G Valdecillos, JR Arroyo, E De Lazari, A Guelar, P Domingo, H Knobel, JM Gatell. A prospective randomized study on the usefulness of Genotypic Resistance Tests versus real Phenotypic Resistance Tests in heavily pretreated patients with virological failure (VIHRES Study). XIV International AIDS Conference. 7-12 July 2002, Barcelona, Spain. Abstract TuPeB4624.
SA Wegner, M Wallace, N Aronson, D Blazes, C Tamminga, S Fraser, M Dolan, K Stephan, L Jagodzinski. Long-term clinical efficacy of resistance testing: results of the CERT trial. XIV International AIDS Conference. 7-12 July 2002, Barcelona, Spain. Abstract ThOrB1389.