The 8^th Conference on Retroviruses and Opportunistic Infections provided some interesting data regarding PI resistance. Once again, we were reminded that sweeping statements regarding cross-resistance between these agents or the proper sequencing of these drugs in salvage should be considered with great caution. We still have much to know and much to learn about resistance in this important class of drugs.

Lopinavir/ritonavir given as the combination pill Kaletra is the newest addition to this group, and resistance data regarding this drug continue to evolve. Previous work by Dale Kempf and colleagues had shed light on the correlations between the high exposure of lopinavir obtained by ritonavir boosting and the ability to provide benefit despite the presence of some degree of laboratory defined resistance (genotypic or phenotypic assay results). The question still remains open, though, just how much baseline resistance can be overcome with lopinavir/ritonavir?

Previous calculations were performed in NNRTI-naive patients who received a potent NNRTI along with lopinavir/ritonavir. Therefore, the degree to which virologic success in these patients could be attributed to lopinavir/ritonavir or to the NNRTI was unclear and precluded any real understanding of how much baseline resistance was actually overcome by the PI. The impressive clinical results updated at the 8^th CROI continue to support the good efficacy and low toxicity of combination NNRTI plus lopinavir/ritonavir in NNRTI-naive patients who are PI experienced (Abstract 525) [1].

Abstract 452 looked at resistance patterns in patients failing lopinavir/ritonavir therapy [2]. Patients' virus showed a very wide variance in terms of the degree of lopinavir/ritonavir resistance. Phenotypic resistance in these patients for whom lopinavir/ritonavir failed ranged from as high as 99-fold to as low as 9-fold in a standard phenotypic assay. Genotypic resistance ranged from more than 10 protease mutations in some patients to as few as 5 mutations in others (although many patients have been shown to fail other PI with far less than 5 mutations). Undoubtedly other factors such as resistance to other drugs in the regimen, adherence and pharmacological considerations contributed to lopinavir/ritonavir failure in these patients.

In addition, the degree of cross-resistance to the other approved PIs was studied in this group of patients. The degree of phenotypic resistance seen varied greatly for each of the other PIs. Ritonavir resistance ranged from 19- to 261-fold; indinavir from 7- to 156 fold; saquinavir from 0.4- to 52-fold; nelfinavir from 8- to 205-fold; amprenavir from 1- to 8-fold. Once again the clinical value of these data is limited by our lack of understanding of clinical relevance of these phenotypic values: will the patient with 7-fold resistance to indinavir still benefit from indinavir boosted by ritonavir? Will the patient with 8-fold resistance to amprenavir still have a virologic response if given amprenavir in the next regimen?

These questions remain open, as does our complete understanding of the genotypic resistance patterns seen in these failing patients. Although not fully understood, the patients with very large numbers of resistance mutations will probably not benefit much from a switch to another PI regimen. Continued PI therapy to maintain this large mutational pattern may still be of benefit due to impaired viral fitness as described by Robert Grant and colleagues (Abstract LB4) and recently published in the NEJM [3,4]. From the data presented, it did appear that resistance to amprenavir was less pronounced than to the other PIs, as was that to saquinavir. The clinical relevance of this remains to be determined and should be addressed in future studies as suggested by the abstract authors.

Abstract 465 also provided interesting insight regarding the relationship between drug exposure (drug levels) and resistance [5]. Although I50V is a unique mutation characteristically generated by amprenavir, other mutational patterns have been shown to develop with amprenavir therapy. The degree of cross-resistance to other PI is relatively low when I50V alone is present, suggesting that clinical benefit may still be derived from other PIs in patients failing amprenavir. But other amprenavir mutational patterns may show higher degrees of cross-resistance. I54L/M, or V32I + I47V or I84V have all been described, and positions 54 and 84 are known to participate in resistance to many if not all of the currently approved PIs. This study showed that patients receiving lower doses of amprenavir and obtaining lower drug levels often developed the 54 mutational pathway, whereas patients obtaining higher amprenavir levels often developed 50. A possible explanation is that the 50 mutation produces a greater reduction in fitness than the 54, although also conferring a higher level of resistance to amprenavir. Therefore a hypothesis might be that at lower exposure levels the virus may be able to overcome amprenavir with the weaker 54 mutation, while paying a relatively low "price" in fitness, whereas with higher, more potent exposure to amprenavir, it has no choice but to go with the 50 mutation which is now required to overcome amprenavir, although it must pay a higher "price" in fitness. Continued studies looking at drug exposure, resistance, fitness and correlating virologic outcomes are needed to confirm this interesting hypothesis.

Another abstract regarding amprenavir failure showed that some degree of cross-resistance is actually conferred by I50V to both nelfinavir and ritonavir, although to a much lesser degree than to amprenavir (Abstract 466) [6]. X-ray structure data provided insight into the mechanism. The clinical significance of this is still unknown as we do not know if this low-level resistance would preclude clinical benefit from nelfinavir or ritonavir, but might suggest that these agents may be less effective than others following amprenavir failure.

Both these abstracts and previous studies suggest that patients failing amprenavir may do so as a result of different mutational patterns, and that these patterns may confer cross-resistance to other drugs in varying degree. This would argue for considering resistance testing in patients failing amprenavir before deciding which, if any, PI would be most effective. This may be similar to the observation that patients can fail nelfinavir via either the 30 or 90 pathway (confirmed again in Abstract 453), and therefore resistance testing may be important before choosing the next regimen [7].